Highway Safety Manual - 1st Edition Volume 2 2010

HIGHWAY
SAFETY
MANUAL
1st Edition
Volume 2 • 201 0
HSM
Highway Safety Manual
~
EXECUTIVE COMMITTEE
2009-2010
Voting Members
Officers:
President: Larry L. "Butch" Brown, Mississippi
Vice President: Susan Martinovich, Nevada
Secretary-Treasurer: Carlos Braceras, Utah
Regional Representatives:
REGION I: Joseph Marie, Connecticut, One-Year Term
Gabe Klein, District of Columbia, Two-Year Term
REGION II: Dan Flowers, Arkansas, One-Year Term

Mike Hancock, Kentucky, Two-Year Term
REGION III: Nancy l Richardson, One-Year Term

Thomas K. Sorel, Minnesota, Two-Year Term
REGION IV: Paula Hammond, Washington, One-Year Term

Amadeo Saenz, Jr,, Texas, Two-Year Term
Nonvoting Members
Immediate Past President: Allen Biehler, Pennsylvania
AASHTO Executive Director: John Horsley, Washington, DC
HIGHWAY SUBCOMMITTEE ON DESIGN
CAROLANN D. WICKS, Delaware, Chair
RICHARD LAND, California, Vice Chair
DWIGHT A. HORNE, FHWA, Secretary
TIM MCDONNELL, AASHTO, Staff Liaison
ALABAMA, William Adams, MINNESOTA, Mukhtar Thakur VERMONT, Kevin Marshia

Rex Bush, Steven E. Walker
MISSISSIPPI, John M. Reese, VIRGINIA, Mohammad Mirshahi,
ALASKA, Mark Neidhold, Amy Mood, C. Keitb Purvis. Robert H. Cary, Barton A. Thrasher
Robert A. Campbell
MISSOURI, David B. Nichols, WASHINGTON, Pasco Bakotich,
ARlZONA, Mary Viparina Katbryn P. Harvey Terry L. Berends, Nancy Boyd,
Dave Olson
ARKANSAS, Phillip L. McConnell, MONTANA, Paul R. Ferry,
Charles D, Clements Lesly Tribelhom WEST VIRGINIA, Jason C. Foster,
Gregory Bailey
CALIFORNIA, Terry L. Abbott NEBRASKA, James J. Knott,
Ted Watson WlSCONSIN, Jerry H. Zogg
COLORADO, Tim Aschenbrener
NEVADA, Eric Glick, Daryl N. James, WYOMING, Paul P. Bereich,
CONNECTICUT, Michael W.
Paul K. Sinnott Tony Laird
Lonergan, James H. Norman
NEW HAMPSHIRE, Craig A. Green
DELAWARE, Michael H. Simmons,
Michael F. Balbierer, NEW JERSEY, Richard W. Dunne, U.S. DOT MEMBER
James M. Satterfield Richard Jaffe, Brian J. Strizki FAA, Rick Marinelli
DISTRICT OF COLUMBIA, NEW MEXICO,
Said Cherifi, Zahra Dorriz, Gabriela Contreras-Apodaca,
Allen Miller Joe S. Garcia ASSOCIATE MEMBER-
Bridge, Port, and Toll
FLORIDA, Lora B. Hollingsworth, NEW YORK, Daniel D'Angelo,
N.J. TURNPIKE AUTHORITY,
James Mills, David O'Hagan Richard W. Lee
J. Lawrence Williams
GEORGIA, James "Ben" Buchan, NORTH CAROLINA,
PORT AUTHORITY OF NY AND NJ,
Russell McMurry, Brent Story Deborah M. Barbour, Jay A. Bennett,
Scott D. Murrell
Art McMillan
HAWAII, Julius Fronda
NORTH DAKOTA, Roger Weigel
IDAHO, Loren D. Thomas,
Nestor Fernandez OHIO, Dirk Gross, Timotby McDonald ASSOCIATE MEMBER-
Federal
ILLINOIS, Scott E. Stitt OKLAHOMA, Tim Tegeler USDA FOREST SERVICE, Ellen G.
INDIANA, Gary Mroczka, OREGON, David Joe Polly, LaFayette
Jeff Clanton, Merril E. Dougherty Steven R. Lindland
IOWA, Michael J. Kennerly, PENNSYLVANIA~ Brian D. Hare
ASSOCIATE MEMBER-
David L. Little, Deanna Maifield
PUERTO RICO, Luis Santos, International
KANSAS, James 0. Brewer Jose E. Santana-Pimentel ALBERTA, Mob Lali
KENTUCKY, Keith Caudill, RHODE ISLAND, Vacant BRITISH COLUMBIA, Richard Voyer
Bradley S. Eldridge, Jeff D. Jasper
SOUTH CAROLINA, John V. Walsh, KOREA, Chan-Su "Chris" Reem
LOUISIANA, Nicholas Kalivoda Ill, Rob Bedenbaugh, Mark Lester
ONTARlO, Joe Bucik
Lloyd E. Porta, Jr.
SOUTH DAKOTA, Michael Behm,
SASKATCHEWAN, Sukhy Kent
MAINE, Todd Pelletier Mark A. Leiferman
MARYLAND, Kirk G. McClelland TENNESSEE, Jeff C. Jones,
Michael Agnew
MASSACHUSETTS, Helmut R. Ernst,
Stanley Wood, Jr. TEXAS, Mark A. Marek
MICHIGAN, Bradley C. Wieferich UTAH, Michael Fazio, Jesse Sweeten
HIGHWAY TRAFFIC SAFETY SUBCOMMITTEE ON SAFETY MANAGEMENT
D. W. VAUGHN, Alabama, Chair
LEANNA DEPUE, Missouri, Vice Chair
KEITH W. SINCLAIR, FHWA, Secretary
KELLY K. HARDY, AASHTO, Staff Liaison
MICHIGAN, Kathy S. Farnum, UTAH, Robert E. Hull, David Beach,

ALABAMA, Wesley Elrod,
Dale Lighthizer, Marsha L. Small Kathy T. Slagowski
Timothy E. Barnett,
Waymon Benifield MINNESOTA, Susan M. Groth VERMONT, Amy Gamble,
James V. Bush, Mario Dupigny-Giroux
ALASKA, Cindy Cashen, MISSISSIPPI, Melinda McGrath,
Jefferson C. Jeffers, Kurtis J. Smith James Willis VIRGINIA, R. Robert Rasmussen, II,
Michael B. Sawyer
ARIZONA, Reed Henry, Mike Manthey MISSOURI, Mike Curti!, John P. Miller
WASHINGTON, Mike Dornfeld,
ARKANSAS, Scott E. Bennett, MONTANA, Duane Williams LesYoung
Tony E. Evans
NEBRASKA, Daniel J. Waddle WEST VIRGINIA, Donna Hardy,
CALIFORNIA,
NEVADA, Chuck Reider, Traci Pearl Marsha Mays, Bob Tipton
Jasvinderjit "Jesse" Bhullar,
Yin-Ping Li, Christopher J. Murphy NEW HAMPSHIRE, Craig A. Green, WISCONSIN, John M. Corbin,
William Lambert, Michael P. Pillsbury Daniel W. Lonsdorf, Rebecca D. Yao
COLORADO, Mike Nugent,
Stacey Stegman, Gabriela Vidal NEW JERSEY, William Beans,
CONNECTICUT, Joseph T. Cristalli, Wilbur Dixon, Patricia Ott
U.S. DOT MEMBERS
Joseph P. Ouellette NEW MEXICO, Elias Archuleta, FHWA, Jeffrey Miller, Esther Strawder,
DELAWARE, Thomas E. Meyer Lawrence Barreras, David Harris Tony L. Young
DISTRICT OF COLUMBIA, NORTH CAROLINA, Terry Hopkins, NHTSA, Bill Watada

Carole Lewis, William McGuirk J. Kevin Lacy, David Weinstein
FLORIDA, Marianne A. Trussell NORTH DAKOTA, Christopher Holzer,

Karin Mongeon, Mark Nelson OTHER
GEORGIA, Keith Golden, AAA FTS, J. Peter Kissinger
Robert F. Dallas OHIO, Dave L. Holstein,
Jennifer Townley ATSSA, Roger Wentz
HAWAil, Sean Hiraoka
OKLAHOMA, Linda Koenig, CVSA, Stephen Keppler
IDAHO, Brent Jennings, Harold Smart ITE, Edward Stollof
Greg M. Laragan
OREGON, Troy E. Costales, NACE, David Patterson
ILLINOIS, Michael R. Stout, Douglas W. Bish, Anne P. Holder
Sandra Klein, Priscilla A. Tobias NASEMSO, Dia Gainor
PENNSYLVANIA,
INDIANA, John Nagle, Brad Steckler Girish (Gary) N. Modi, NLTAPA, Marie B. Walsh
IOWA, Steve Gent, Mary Stahlhut, Glenn C. Rowe, Scott Shenk TRB, Charles Niessner, Richard F. Pain
Tom Welch PUERTO RICO, Vacant U.S. ACCESS BOARD, Scott Windley
KANSAS, Pete Bodyk, Steve Buckley RHODE ISLAND, Janis E. Loiselle, VIRGINIA TECH
KENTUCKY, Boyd T. Sigler, Joseph A. Bucci, Robert Rocchio TRANSPORTATION INSTITUTE,
Billie Johnson, JeffWolfe SOUTH CAROLINA, Brett Harrelson, Cindy Wilkinson
LOUISIANA, Dan Magri, Darrell Munn, Phil Riley
Terri Monaghan SOUTH DAKOTA, Ben Orsbon,
MAINE, Bradford P. Foley, Darryl Belz Sonia Trautmann
MARYLAND, Ron Lipps, TENNESSEE, Kendell Poole,

Vernon Betkey Gary Ogletree
MASSACHUSETTS, TEXAS, Margaret (Meg) A. Moore,

Thomas F. Broderick III, Tony Duros Luis Gonzalez, Terry Pence
HIGHWAY SUBCOMMITTEE ON TRAFFIC ENGINEERING
DELBERT MCOMIE, Wyoming, Chair
THOMAS HICKS, Maryland, Vice Chair
MARK KEHRLI, FHWA Secretary
KEN KOBETSKY, AASHTO, Staff Liaison
ALABAMA, Stacey N. Glass MISSOURl, Eileen Rackers U.S. DOT MEMBER

ALASKA, Kurtis J. Smith MONTANA, Duane Williams FHWA, Hari Kalla
ARJZONA, Mike Manthey, NEBRASKA, Daniel J. Waddle

Richard C. Moeur
NEVADA, Fred Droes ASSOCIATE MEMBER-
ARKANSAS, Eric Phillips, Bridge, Port, and Toll
Tony Sullivan NEW HAMPSIDRE, William Lambert
GOLDEN GATE BRJDGE,
NEW JERSEY, Douglas R. Bartlett Maurice Palumbo
CALIFORNIA, Robert Copp
NEW MEXICO, Vacant N.J. TURNPIKE AUTHORJTY,
COLORADO, Gabriela Vidal
Sean M. Hill
CONNECTICUT, John F. Carey NEW YORK, David J. Clements
NORTH CAROLINA, J. Kevin Lacy
DELAWARE, Mark Luszcz,
Donald D. Weber NORTH DAKOTA, Shawn Kuntz
ASSOCIATE MEMBER-
International
DISTRJCT OF COLUMBIA, OIDO, Halle Jones Capers, MANITOBA, Glenn A. Cuthbertson
Sournya S. Dey, Yanlin Li Dave L. Holstein
NOVA SCOTIA, Bernie Clancey
FLORJDA, Mark C. Wilson OKLAHOMA, Harold Smart
SASKATCHEWAN, Sukby Kent
GEORGIA, Keith Golden OREGON, Edward L. Fischer
HAWAII, Alvin Takeshita PENNSYLVANIA, Glenn C. Rowe
AASHTO MEMBER
IDAHO, Brent Jennings PUERTO RlCO, Carlos M. Contreras MarkS. Bush
ILLINOIS, Aaron Weatherholt RHODE ISLAND, Robert Rocchio
INDIANA, James Potura!ski, SOUTH CAROLINA, Richard B. Werts
Todd Shields
SOUTH DAKOTA, Laurie Schultz
IOWA, Timothy D. Crouch
TENNESSEE, Michael L. Tugwell
KANSAS, Kenneth F. Hurst
TEXAS, Margaret (Meg) A. Moore
KENTUCKY, Jeff Wolfe
UTAH, Robert E. Hull
LOUISIANA, Peter Allain
VERMONT, Bruce Nyquist
MAINE, Bruce A. Ibarguen
VIRGINIA, Raymond J. Khoury
MASSACHUSETTS, Dave Belanger,
Neil E. Boudreau WASHINGTON, Theodore Trepanier
MICIDGAN, Mark W. Bott WEST VIRGINIA, Cindy Cramer
MINNESOTA, Susan M. Groth WISCONSIN, Thomas N. Notbohm
MISSISSIPPI, Robert "Wes" Dean WYOMING, Joel Meena

JOINT TASK FORCE ON THE HIGHWAY SAFETY MANUAL
D. W. VAUGHN, Alabama, Chair
PRISCILLA TOBIAS, Illinois, Vice Chair
KEN KOBETSKY, AASHTO, Staff Liaison
flM MCDONNELL, AASHTO, Staff Liaison
NORTH CAROLINA, J. Kevin Lacy WASHINGTON, Theodore Trepanier

FLORIDA, James Mills
NORTH DAKOTA, Mark Gaydos WEST VIRGINIA, Donna Hardy
MAINE, Bruce Ibarguen
OHIO, Timothy McDonald TRB TASK FORCE LIAISON,
MARYLAND, Kirk G. McClelland
John Milton, Washington
MICHIGAN, Mark W. Bott PENNSYLVANIA,
Girish (Gary) N. Modi U.S. DOT MEMBER,
MISSOURI, Mike Curti! FHWA, Shyuan-Ren "Clayton" Chen
UTAH, Robert E. Hull
NEW JERSEY, Wilbur Dixon
VIRGINIA, Barton A. Thrasher
NEW MEXICO, Joe S. Garcia
Acknowledgements
The publication of this Manual is the cuhnination of innumerable hours of labor by the many members and friends
of the TRB Task Force, the AASHTO Joint Task Force, and contractors and staff of the NCHRP program.
The original idea for the Highway Safety Manual (HSM) came from the deliberations and discussions offour
individuals: Ronald C. Pfefer, Douglas W Harwood, John M. Mason, Jr., and Timothy R. Neuman. They quickly
involved Michael S. Griffith and TRB staff to sponsor and develop the first workshop and formation of what is now
the Task Force for the Development of the Highway Safety Manual. From that workshop grew a long list of highway
safety professionals willing to donate many hours to the development of the Highway Safety Manual. In addition
to the volunteer Members and Friends of the TRB Task Force, numerous research projects contributed directly or
indirectly to the HSM. Several research projects sponsored by the National Cooperative Highway Research Program
resulted in the materials used to develop and implement the HSM. This research has been largely unpublished
anywhere other than the HSM, and therefore the projects and key authors are highlighted below. The TRB Task Force
Members are also highlighted below, though the list of dedicated Friends is too long to include.
Researchers
National Cooperative Highway Research Program Senior Program Officer: Charles Niessner
• NCHRP 17-18(04): Development of an HSM-DraftTable of Contents for the HSM

Bellomo-MeGee, Inc. (Warren Hughes, Principal Investigator)
• NCHRP 17-25: Crash Reduction Factors for Traffic Engineering and ITS Improvements
(Published as NCHRP Report 617)
University ofNorth Carolina-Chapel Hill (David Harkey, Principal Investigator)
• NCHRP 17-26: Methodology to Predict the Safety Performance of Urban and Suburban Arterials ·
Midwest Research Institute (Doug Harwood, Principal Investigator)
• NCHRP 17-27: Prepare Parts I and II of the HSM

iTrans Consulting Ltd. (Geni Balrar, Principal Investigator)
• NCHRP 17-29: Methodology to Predict the Safety Performance of Rural Multilane Highways
Texas A&M Research Foundation (Dominique Lord, Principal Investigator)
• NCHRP 17-34: Prepare Parts IV and V of the Highway Safety Manual

Kittelson & Associates, Inc. (John Zegeer, Principal Investigator)
• NCHRP 17-36: Production of the First Edition of the Highway Safety Manual
Kittelson & Associate, Inc. (John Zegeer, Principal Investigator)
• NCHRP 17-37: Pedestrian Predictive Crash Methodology for Urban and Suburban Arterials
Midwest Research Institute (Doug Harwood, Principal Investigator)
• NCHRP 17-38: Highway Safety Manual Implementation and Training Materials

Oregon State University (Karen Dixon, Principal Investigator)
TRANSPORTATION RESEARCH BOARD TASK FORCE ON THE DEVELOPMENT
OF THE HIGHWAY SAFETY MANUAL
Chair
John Milton, Washington State Department ofTransportation (2009-current)
Ronald Pfefer, Northwestern Traffic Institute (2000--2009)
Secretary
Elizabeth Wemple, Kittelson & Associates, Inc.
TRB Staff Representative

Richard Pain
Charles Niessner
Members
Geni Bahar, NAVIGATS Inc. John Ivan, University of Connecticut Jose Pardillo-Mayora~ Universidad
Politecnica de Madrid
Brian Barton, Department for Transport Kelly Hardy, American Association
(United Kingdom, retired) of State Highway and Transportation Bhagwant Persaud, Ryerson University
Officials
James Bonneson, Stanley Polanis, City of Winston-Salem
Texas Transportation Institute David Harkey, University of
Bruce Robinson, Transportation
North Carolina-Chapel Hill
Forrest Council, UNC Highway Safety Consultant
Research Center Douglas Harwood,
Edward Stollof, Institute of
Midwest Research Institute
Leanna Depue, Missouri Department Transportation Engineers
of Transportation Steven Kodama, City of Toronto
Larry Sutherland, Parsons
Michael Dimaiuta, GENEX Systems Francesca La Torre, Brinckerhoff~ Inc.
University of Florence
Karen Dixon, Oregon State University Daniel Turner, University of Alabama
John Mason, Auburn University
Brelend Gowan, Ca!trans (retired) Ida van Schalkwyk,
Christopher Monsere, Oregon State University
Michael Griffith, Federal Highway
Portland State University
Administration Scott Windley~ United States
Timothy Neuman, CHZM HILL Access Board
Michael Hankey, Village of
Hoffinan Estates John Nitzel, CHZM HILL John Zegeer~ Kittelson & Associates, Inc.
Highway Safety Manual
Table of Contents
VOLUME 1
Part A-Introduction, Human Factors, and Fundamentals
Chapter 1-lntroduction and Overview
Chapter 2-Human Factors
Chapter 3-Fundamentals
Part B-Roadway Safety Management Process

Chapter 4-Network Screening
Chapter 5-Diagnosis
Chapter 6-Select Countermeasures
Chapter 7-Economic Appraisal
Chapter 8-Prioritize Projects
Chapter 9-Safety Effectiveness Evaluation
VOLUME2
Part C-Predictive Method
Chapter 1a-Predictive Method for Rural Two-Lane, Two-Way Roads
Chapter 11-Predictive Method for Rural Multilane Highways
Chapter 12-Predictive Method for Urban and Suburban Arterials
VOLUME3
Part D-Crash Modification Factors
Chapter 13-Roadway Segments
Chapter 14-lntersections
Chapter 15-lnterchanges
Chapter 16-Special Facilities and Geometric Situations
Chapter 17-Road Networks

Table of Contents for Volume 2
PREFACE TO THE HIGHWAY SAFETY MANUAL ••••.......................................................•• xxvii
PART (-INTRODUCTION AND APPLICATIONS GUIDANCE .••••••••..................................... C-1

C.1. Introduction to the Highway Safety Manual Predictive Method .. ..C-1
C.2. Relationship to Parts A, B, and D of the HSM . ..C-2
C.3. Part C and the Project Development Process .. ...... C-2
C.4. Overview of the HSM Predictive Method .. ... C-3
C.5. The HSM Predictive Method .. . ....... C-5
C.6. Predictive Method Concepts .. . .... C-12

C.6.1. Roadway Limits and Faol1tyTypes .. ................. C-12
(.6.2. Definition of Roadway Segments and Intersections ............. .. ............... C-13
C.6.3. Crash Modification Factors (CMFs) .. ................. C-15
C.6.4. Calibration of Safety Performance Functions to Local Conditions ... ..C-18
C.6.5. Weighting Using the Empirical Bayes Method ..... . .. C-18
C.7. Methods for Estimating the Safety Effectiveness of a Proposed Project .. ................ C-19
C. B. Limitations of the HSM Predictive Method .. ................. C-19
C.9. Guide to Applying Part c. ................ C-20
C.10. Summary .... .... C-20
CHAPTER 10-PREDICTIVE METHOD FOR RURAL TWO-LANE, TWO-WAY ROADS ...... 10-1
10.1. Introduction ... ················· 10-1
10.2. Overview of the PrediCtrve Method .. ..... 10-1
10.3. Rural Two-Lane, Two-Way Roads-Definitions and Predictive Models In Chapter 10 .... 10-2
10.3.1. Definrtion of Chapter 10 Facilrty and Site Types... . 10-2
10.3.2. Predictive Models for Rural Two-Lane, Two-Way Roadway Segments.. . ................. 10-3
10.3.3. Predictive Models for Rural Two-Lane, Two-Way Intersections. ............ 10-4
10.4. Predictrve Method for Rural Two-Lane, Two-Way Roads ..... ... 10-4
10.5. Roadway Segments and Intersections .... ····· 10-11

10.6. Safety Performance Functions.. . 10-14
10.6.1. Safety Performance Functions for Rural Two-Lane, Two-Way Roadway Segments ...... 10-14
10.6.2. Safety Performance Functions for Intersections ... .. ············ 10-17
10.7. Crash Modification Factors ... .... 10-22

10.7.1. Crash Modification Factors for Roadway Segments ... .............. 10-23
10.7.2. Crash Modification Factors for Intersections .. ............... 10-31
10.8. Calibration of the SPFs to Local Conditions ...... . . ... 10-33
10.9. Limitations of Predrctive Method in Chapter 10 .. ............. 10-34
10.1 0. Application of Chapter 10 Predictive Method .. ······· 10-34

10.11. Summary .... .... 10-34
10.12. Sample Problems .. ............... 10-35

10.12.1. Sample Problem 1.. ····· .. 10-35
10.12.2. Sample Problem 2 .. . 10-42
10.12.3. Sample Problem 3 .. ..... ······ .. 10-49
10.12.4. Sample Problem 4 .... 10-55
10.12.5. Sample Problem 5.. .. ... 10-60
10.12.6. Sample Problem 6 .............. . .. 10-62
10.13. References .. ... 10-66

10A.1.APPENDIX lOA-WORKSHEETS FOR PREDIOIVE METHOD FOR RURAL
TWO-LANE, TWO-WAY ROADS ....................................................................................................... 10·68
CHAPTER 11-PREDICTIVE METHOD FOR RURAL MULTILANE HIGHWAYS .........•.•.••.•. 11-1

11. 1. Introduction. . 11-1
11.2. Overview of the Predictive Method .. 11-1
11.3. Rural Multilane Highways-Definitions and Predictive Models in Chapter 11 ..... .. 11-2
11.3.1. Definition of Chapter 11 Facility and Site Types ........ . ... 11-2
11.3.2. Predictive Models for Rural Multilane Roadway Segments ... 11-3
11.3.3. Predictive Models for Rural Multilane Highway Intersections .. .. .. 11-4
11.4. Predictive Method for Rural Multilane Highways .. .. ............. 11-4

11.5. Roadway Segments and Intersections .. .. .. 11-11
11.6. Safety Performance Functions.............. .. ........... . .. ... 11-13
11.6.1. Safety Performance Functions for Undivided Roadway Segments .. .. .... 11-14
11 .6.2. Safety Performance Functions for Divided Roadway Segments .... .. .. 11-17
11.6.3. Safety Performance Functions for Intersections .................... . .. 11-20
11.7. Crash Modification Factors ................................................................ ... 11-24

11.7.1. Crash Modification Factors for Undivided Roadway Segments .. . ... 11-25
11. 7.2. Crash Modification Factors for Divided Roadway Segments. 11-29
11.7.3. Crash Modification Factors for Intersections ...... . .. ... 11-32
11.8. Calibration to Local Conditions ......................... . .. 11-35

11.9. Limitations of Predictive Methods In Chapter 11 .. .. .. 11-36
11.1 0. Application of Chapter 11, Predictive Method. .. 11-36
11. 11. Summary ... 11-36
11.12. Sample Problems .. 11-37
11.12.1. Sample Problem 1 11-37
11.12.2. Sample Problem 2 .. . ... 11-43
11.12.3. Sample Problem 3 .... .. 11-49
11.12.4. Sample Problem 4 .. . 11-54
11.12.5. Sample Problem 5.... .. .. 11-56
11.12.6. Sample Problem 6.... .. .. .. 11-60
11.13. References .... .. .. 11-61

APPENDIX 11A-WORKSHEETS FOR APPLYING THE PREDICTIVE METHOD
FOR RURAL MULTILANE ROADS ...................................... ,........................................................... 11-62
CHAPTER 12-PREDICTIVE METHOD FOR URBAN AND SUBURBAN ARTERIALS ....••••• 12-1
12.1. Introduction .. .. ... 12-1
12.2. Overview of the Predictive Method .. . ...... 12-1
12.3. Urban and Suburban Arterials-Definitions and Predictive Models in Chapter 12 ... 12-2
12.3.1. Definition of Chapter 12 Facility Types ...................................... .. .. ....... 12-2
12.3.2. Predictive Models for Urban and Suburban Arterial Roadway Segments.... .... 12-4
12.3.3. Predictive Models for Urban and Suburban Arterial Intersections ...................... 12-5
12.4. Predictive Method Steps for Urban and suburban arterials .. .. 12-6
12.5. Roadway Segments and Intersections ........................... . 12-14
12.6. Safety Performance Functions .... . . ................. ................ . ... 12-16
12.6.1. Safety Performance Functions for Urban and Suburban Arterial Roadway Segments .. 12-17
12.6.2. Safety Performance Functions for Urban and Suburban Arterial intersections ............ 12-28
12.7. Crash Modification Factors ..................... . .. 12-39

12.7.1. Crash Modif1cat1on Factors for Roadway Segments ......... 12-40
12.7.2. Crash Modification Factors for Intersections .......... .. ····· 12-43
12. 7.3. Crash Modification Factors for Vehicle-Pedestrian Collisions at Signalized Intersections .... 12-46
12.8. Calibration of the SPFs to Local Conditions .. 12-47
12.9. Interim Predictive Method for Roundabouts. .... 12-47
12.1 0. Limitations of Predictive Method in Chapter 12 .... . .. 12-48
12.11. Application of Chapter 12 Predictive method .... .. ......... 12-48
12.12. Summary .......................... . 12-48
12.13. Sample Problems 12-49
12.13.1. Sample Problem 1.. .. .. .. 12-49
12.13.2. Sample Problem 2 .. .. .. ......... 12-63
12.13.3. Sample Problem 3 ..... . '12-74
12.13.4. Sample Problem 4 ... . ... 12-86
12. 13.5. Sample Problem 5.. 12-97
12.13.6. Sample Problem 6... 12-101
12 .14. References .......... . .. .. 12-106

APPENDIX 12A-WORKSHEETS FOR PREDICTIVE METHOD FOR URBAN
AND SUBURBAN ARTERIALS .•••.••••..•••.••••.••••.••••.••••.••.••••••.••••.•••••.••••.••••.•••..•••.••••.•...•••.••.••••.••••.•••• 12-108
APPENDIX A-SPECIALIZED PROCEDURES COMMON TO ALL PART C CHAPTERS ..••••••. A-1

A.1. Calibration of the Part C PrediCtive Models .......................................................... . ..A-1
A.1.1. Calibration of Predictive Models ................................................ . ..A-1
A.1.2. Development of Jurisdiction-Specific Safety Performance Functions
for Use in the Part C Predictive Method ................. .. .. ... A-9
A.1.3. Replacement of Selected Default Values in the Part C Predictive Models
to Local Conditions...... .......................................................... .. .. ........ A-10
A.2. Use of the Empirical Bayes Method to Combine Predicted Average Crash Frequency
and Observed Crash Frequency........ .. ....................... . ...A-15
A.2. 1 Determine whether the EB Method 15 Applicable ..................................... .. .... A-16
A.2.2. Determine whether Observed Crash Frequency Data are Available for the
Project or Facility and, if so, Obtain those Data ........................... .. .. ....... A-17
A.2.3. Assign Crashes to Individual Roadway Segments and Intersections
for Use in the EB Method .......... . .. ... A-17
A.2.4. Apply the Site-Specific EB Method .......................................... . .. .. A-19
A.2.5. Apply the Project-level EB Method ...................... . ...A-20
A.2.6. Adjust the Estimated Value of Expected Average Crash Frequency
to a Future Time Period, If Appropriate ........ . ... A-22
GLOSSARY •••••••••........•..••.••.•...•........•...•........•.•••••..•....•...•...•••••••••.••.....••.......••••••••..••...•..G-1
T'l"
List of Figures
PART (-INTRODUCTION AND APPLICATIONS GUIDANCE .............................................. C-1
Figure C-1. Relation betvveen Part C Predictive Method and the Project Development Process ... .. C-3
F1gure C-2. The HSM Predictive Method .... ................. C-6
Figure C-3. Definition of Roadway Segments and Intersections ... .. .......... C-14
Figure 10-1. The HSM Predictive Method ... ....... 10-5
Figure 10-2. Definition of Segments and Intersections . .. ......... 10-2
figure I 0-3. Graphical Form of SPF for Rural Two-Lane, Two-Way Roadway Segments (Equation 10-6) .. 10-16
Figure 10-4. Graphical Representation of the SPF for Three-Leg Stop-controlled (3ST)
Intersections (Equation 10-8) ... .. ...... 10-19
Figure 10-5. Graphical Representation of the SPF for Four-Leg, Stop-controlled (4SD
Intersections (Equation 10-9) .. .. ... 10-20
F1gure 10-6. Graphical Representation of the SPF for Four-Leg Signalized (4SG)
Intersections (Equation 10-1 0) ... .. ........ 10-21
F1gure 10-7. Crash Modification Factor for Lane Width on Roadway Segments ... . 10-24
Figure 10-8. Crash Modification Factor for Shoulder Width on Roadway Segments ... .. 10-26
CHAPTER 11-PREDICTIVE METHOD FOR RURAL MULTILANE HIGHWAYS .................. 11-1

Figure 11-1. The HSM Predictive Method . .. ........... 11-5
Figure 11-2. Definition of Segments and Intersections ... .. .... 11-12
Figure 11-3. Graphical Form of the SPF for Undivided Roadway Segments
(from Equation 11-7 and Table 11-3) . .. .. 11-16
Figure 11-4. Graphical Form of SPF for Rural Multilane Divided Roadway Segments
(from Equation 11·9 and Table 11-5) .. .. 11-19
Figure 11-5. Graphical Form of SPF for Three-Leg Stop-Controlled Intersections-
for Total Crashes Only (from Equation 11-11 and Table 11-7) ........ 11-22
Figure 11-6. Graphical Form of SPF for Four-Leg Stop-Controlled Intersections-
for Total Crashes Only (from Equation 11-11 and Table 11-7) . 11-23
Figure 11-7. Graphical Form of SPF for Four-leg Signalized Intersections-
for Total Crashes Only (from Equation 11-11 and Table 11-7) . .. ...... 11-23
Figure 11-8. CMFRA for Lane Width on Undivided Segments .. .. ...... 11-27
F1gure 11-9. CMF WRA for Shoulder Width on Undivided Segments ... .. 11-28
F1gure 11-10. CMFRA for Lane Width on Divided Roadway Segments ........................ . .. ....... 11-30
CHAPTER 12-PREDICTIVE METHOD FOR URBAN AND SUBURBAN ARTERIALS ......... 12-1
Figure 12-1. The HSM Predictive Method ... .. .. 12-7
Figure 12-2. Definition of Roadway Segments and Intersections .. ... 12-15

F1gure 12-3. Graphical Form of the SPF for Multiple Vehicle Non driveway collisions
(from Equation 12-10 and Table 12-3) .. .. ... 12-19
F1gure 12-4. Graphical Form of the SPF for Single-Vehicle Crashes
(from Equation 12-13 and Table 12-5) . . 12-22
Figure 12-5. Graphical Form of the SPF for Multiple Vehicle Driveway Related Collisions on
Two-Lane Undivided Arterials (2U) (from Equation 12-16 and Table 12-7) . .. 12·24
Three-Lane Undivided Artenals (3D (from Equation 12-16 and Table 12-7) . . .. 12-25
Four-Lane Undivided Arterials (4U) (from Equation 12-16 and Table 12-7) .................... 12-25
Four-Lane Divided Arterials (4D) (from Equation 12-16 and Table 12-7) ......... .. ... 12-26
Five-Lane Arterials Including a Center Two-Way lett-Turn Lane
(from Equation 12-16 and Table 12-7) ................................ ................... .. ............. 12-26
F1gure 12-1 0. Graphical Form of the Intersection SPF for Multiple Vehicle Collisions on
Three-Leg Intersections with Minor-Road Stop Control (3ST)
(from Equation 12-21 and Table 12-10) ...................... ................. ... 12-30
Figure 12-11. Graphical Form of the Intersection SPF for Multiple Vehicle Collisions on
Three-Leg Signalized Intersections (3SG) (from Equation 12-21 and Table 12-1 0) 12-31
Figure 12-12. Graphical Form of the Intersection SPF for Multiple Vehicle Collisions on
Four-Leg Intersections with Minor-Road Stop Control (4ST)
(from Equation 12-21 and Table 12-1 0) ..... .. ... 12-31
Figure 12-13. Graphical Form of the Intersection SPF for Multiple Vehicle Collisions on Four-Leg
Signalized Intersections (4SG) (from Equation 12-21 and Table 12-1 0) . ................. 12-32
Figure 12-14. Graphical Form of the Intersection SPF for Single-Vehicle Crashes on Three-leg
Intersections with Minor-Road Stop Control (3ST) (from Equat1on 12-24 and Table 12-12) .. 12-34
Figure 12-15. Graphical Form of the Intersection SPF for Single-Vehicle Crashes on Three-Leg
Signalized Intersections (3SG) (from Equation 12-24 and Table 12-12) .......................... 12-34
Figure 12-16. Graphical Form of the Intersection SPF for Single-Vehicle Crashes on Four-Leg Stop
Controlled Intersections (4SD (from Equation 12-24 and Table 12-12) . . . 12-35
F1gure 12-17. Graphical Form of the Intersection SPF for Single-Vehicle Crashes on Four-leg
Signalized Intersections (4SG) (from Equat1on 12-24 and Table 12-12) .... 12-35
APPENDIX A-SPECIALIZED PROCEDURES COMMON TO ALL PART C CHAPTERS---------A-1

Figure A-1. Definition of Roadway Segments and Intersections ... .. .. A-18
.~.. .
List of Tables
PART C-INTRODUCTION AND APPLICATIONS GUIDANCE .........................••••••............... C-1
TableC-1. Safety Performance Functions by Facility Type and Site Types in Part C .............. . .. ........ C-5
Table C-2. Constructing Confidence Intervals Us1ng CMF Standard Error ..................... . .. ...... C-17
Table 10-1. Rural Two-lane, Two-Way Road Site Type with SPFs in Chapter 10 ... .. .. 10·3
Table 10-2. Safety Performance Functions included in Chapter 10 . 10-14
Table 10-3. Default Distnbution for Crash Severity Level on Rural Two-Lane,
Two-Way Roadway Segments .. .. .. 10-17
Table 10-4. Default Distnbution by Collision Type for Specific Crash Severity Levels
on Rural Two-Lane, Two-Way Roadway Segments.. .. ................ . ... 10-17
Table 10-5. Default Distribution for Crash Severity Level at Rural Two-Lane, Two-Way Intersections .... 10-21
Table 10-6. Default Distribution for Collision Type and Manner of Collision
at Rural Two-Way Intersections.. .. ................ . .. ........... 10-22
Table I 0-7. Summary of Crash Modification Factors (CMFs) in Chapter 10
and the Corresponding Safety Periormance Functions (SPFs) .................................... 10-23
Table 10-8. CMF for Lane Width on Roadway Segments (CMF).. 10-24
Table 10-9. CMF for Shoulder Width on Roadway Segments (CMF M).. ............... 10·25
Table 10-10. Crash Modification Factors for Shoulder Types and Shoulder Widths
on Roadway Segments (CMF1, ) •. .. .... 10-26
Table 10-11. Crash Mod1f1cation Factors (CMF5) for Grade of Roadway Segments .. .. ... 10-28
Table 10-12. Nighttime Crash Proportions for Unlighted Roadway Segments .. .. 10-31
Table 10-13. Crash Modification factors (CMF,) for Installation
of Left-Turn Lanes on Intersection Approaches .. ... I 0-32
Table 10-14. Crash Modification Factors (CMF31) for Right-Turn Lanes on Approaches
to an Intersection on Rural Two-Lane, Two-Way Highways .. .. .. .. 10-33
Table 10-15. Nighttime Crash Proportions for Unlighted Intersections .. .. .. .. 10-33
Table 10-16. List of Sample Problems in Chapter 10 ... ... 10-35
CHAPTER 11-PREDICTIVE METHOD FOR RURAL MULTILANE HIGHWAYS .................. 11-1

Table 11-1. Rural Multilane Highway Site Type w1th SPFs in Chapter 11 ... . 11-3
Table 11-2. Safety Performance Functions included in Chapter 11 . . 11-14
Table 11-3. SPF Coefficients for Total and Fatal-and-Injury Crashes on Undiv1ded
Roadway Segments (for use 1n Equations 11-7 and 11-8)... .. ................................. 11-15
Table 11-4. Default Distribution of Crashes by Collision Type and Crash Severity Level for
Undivided Roadway Segments .. .. ... 11-17
Table 11-5. SPF Coefficients for Total and Fatal-and-Injury Crashes on Divided Roadway Segments
(for use in Equations 11-9 and 11-10).. .. .......... 11-18
Table 11-6. Default Distribution of Crashes by Collision Type and Crash Seventy Level for
Divided Roadway Segments... .. ........... 11-20
Table 11-7. SPF Coefficients for Three- and Four-Leg Intersections with M1nor-Road Stop Control
for Total and Fatal-and-Injury Crashes (for use in Equation 11-11) . .... 11-22
Table 11-8. SPF Coefficients for Four-Leg Signalized Intersections for Total and
fatal-and-Injury Crashes (for Use 1n Equations 11-11 and 11-12) . .. ....... 11-22
Table 11-9. Default Distribution of Intersection Crashes by Collision Type and Crash Severity.. .. .. 11-24
Table 11-1 0. Summary of CMFs in Chapter 11 and the Corresponding SPFs . .. .. 11·25
Table 11-11. CMF RA for Collision Types Related to Lane Width . 11-26
Table 11-12. CMF for Collision Types Related to Shoulder Width (CMFWR.J ... .. ...... 11-27
Table 11-13. CMFfor ColliSion Types Related to Shoulder Type and Shoulder Width (CMF,,.) ............ 11-28
Table 11-14. CMF for Sides lope on Undivided Roadway Segments (CMF3 ru). .. ........ 11-28
Table 11-15. Night-time Crash Proportions for Unlighted Roadway Segments .. ... 11-29
Table 11-16. CMF for Collision Types Related to Lane Width (CMF,,.) ........ 11-30
Table 11-17. CMF for Right Shoulder Width on Divided Roadway Segments (CMF2 ) 11-31
Table 11-18. CMFs for Median Width on Divided Roadway Segments
wrthout a Median Barrier (CMF3 ,.). .. ... 11-31
Table 11-19. Nighttrme Crash Proportions for Unlrghted Roadway Segments .................. 11-32
Table 11-20. CMFs for Three-Leg Intersections with Minor-Road Stop Control (3SD ................ 11-32
Table 11-21. CMFs for Four-Leg Intersection wrth Minor-Road Stop Control (4SD ...... 11-33
Table 11-22. Crash Modification Factors (CMF2 ) for Installation of Left-Tum Lanes
on lntersectron Approaches ........................ . .. .. 11-34
Table 11-23. Crash Modification Factors (CMF) for lnstallatron of Right-Turn Lanes
on Intersections Approaches .... ..... ........ . ................ . .. .. 11-35
Table 11-24. Default Nighttime Crash Proportions for Unlighted Intersections .. . .. .. 11-35
Table 11-25. Lrst of Sample Problems in Chapter 11 . .. ............ 11-37
Table 11-26. Summary of Results for Sample Problem 6 .. . 11-61
Table 12-1. Urban and Suburban Arterial Site Type SPFs included in Chapter 12. . 12-3
Table 12-2. Safety Performance Functions rncluded in Chapter 12. .. .......... 12-17
Table 12-3. SPF Coefficients for Multiple-Vehicle Nondriveway Collisions on Roadway Segments ........ 12-19
Table 12-4. Distribution of Multiple-Vehicle Nondnveway Collisions for Roadway Segments
by Manner of Collision Type.. .. ............................................ 12-20
Table 12-5. SPF Coefficients for Single-Vehicle Crashes on Roadway Segments ...................... . . ... 12-21
Table 12-6. Distribution of Single-Vehicle Crashes for Roadway Segments by Collision Type ................ 12-22
Table 12-7. SPF Coefficrents for Multiple-Vehicle Driveway Related Collisions ..................................... 12-24
Table 12-8. Pedestrian Crash Adjustment Factor for Roadway Segments ............................................ 12-27
Table 12-9. Bicycle Crash Adjustment Factors for Roadway Segments . 12-28
Table 12-1 0. SPF Coefficrents for Multiple-Vehicle Collisrons at lntersectrons ........................................ 12-30
Table 12-11. Distribution of Multiple-Vehicle Collisions for Intersections by Collision Type .................... 12-32
Table 12-12. SPF Coefficients for Single-Vehicle Crashes at lntersectrons .............. . .. ............... 12-33
Table 12-13. Distribution of Single-Vehicle Crashes for Intersection by Collision Type .. .. ... 12-36
Table 12-14. SPFs for Vehrcle-Pedestrian Collisions at Srgnalized Intersections .. .. ..... 12-37
Table 12-15. Estimates of Pedestrian Crossing Volumes Based on General Level of Pedestrian Actrvity .. 12-37
Table 12-16. Pedestrian Crash Adjustment Factors for Stop-Controlled Intersections.. ....... 12-38
Table 12-17. Brcycle Crash Adjustment Factors for Intersections .. .................................................... 12-38
Table 12-18. Summary of CMFs in Chapter 12 and the Corresponding SPFs .. . 12-39
Table 12-19. Values of fpk Used in Determining the Crash Modrfrcation Factor for On-Street Parking ..... 12-40
Table 12-20. Fixed-Object Offset Factor ... .. .................. 12-41
Ta'ble 12-21. Proportion of Fixed-Object Collisrons ... ....................................................................... 1~41
Table 12-22. CMFs for Median Widths on Divided Roadway Segments without a Median Barrier (CMF3) .12-42
Table 12-23. Nighttime Crash Proportions for Unlighted Roadway Segments ... .. ................ 12-42
.l...
Table 12-24. Crash Modification Factor (CMF 1) for Installation of Left-Turn Lanes
on Intersection Approaches ............................................................................................ 12.-43
Table 12-25. Crash Modification factor (CMf,) for Type of Left-Turn Signal Phasing .. ....................... 12-44
Table 12-26. Crash Modification Factor (CMF3) for lnstallatron of Right-Turn Lanes
on lntersectron Approaches...... .. .......................... . .. ......... 12-44
Table 12-27. Nighttime Crash Proportions for Unlighted Intersections .......................... .. ""' 12-45
Table 12-28. Crash Modificatron Factor (CMF 1) for the Presence of Bus Stops near the lntersectron ..... 12-46
Table 12-29. Crash Modrfrcatron Factor (CMF) for the Presence of Schools near the Intersection ......... 12-46
Table 12-30. Crash ModifiCation factor (CMf,,) for the Number of Alcohol Sales Establishments
near the Intersection ......................................................................................................... 12-47
Table 12-31. List of Sample Problems in Chapter 12 ... .. ..................... 12-49
APPENDIX A-SPECIALIZED PROCEDURES COMMON TO ALL PART C CHAPTERS ......... A-1

Table A-1. SPFs in the Part C Predictive Models that Need Calibration ..................... A-3
Table A-2. Data Needs for Calibration of Part C Predictive Models by Facility Type ............. A-5
Table A-3. Default Crash Distributions Used in Part C Predictive Models Which May Be
Calibrated by Users to Local Conditions.. ...................................... ............... A-1 1
List of Worksheets
CHAPTER 1Q-PREDICTIVE METHOD FOR RURAL TWO-LANE, TWO-WAY ROADS ...... 10·1
Worksheet SP1 A General Information and Input Data for Rural Two-Lane,
Two-Way Roadway Segments ......................... . . .................. 10·39
Worksheet SPl B. Crash Modification Factors for Rural Two-Lane, Two-Way Roadway Segments ...... 10-40
Worksheet SP1 C. Roadway Segment Crashes for Rural Two-Lane, Two-Way Roadway Segments .. 10-40
Worksheet SP1 D. Crashes by Severity Level and Collision Type for Rural Two-Lane,
Two-Way Roadway Segments.. .. 10-41
Worksheet SP1 E. Summary Results for Rural Two-Lane, Two-Way Roadway Segments .. 10-42
Worksheet SP2A. General Information and Input Data for Rural Two-Lane,
Two-Way Roadway Segments.... .. 10-46
Worksheet SP2B. Crash Mod1f1cation Factors for Rural Two·Lane, Two·Way Roadway Segments ...... 1047
Worksheet SP2C. Roadway Segment Crashes for Rural Two·Lane, Two· Way Roadway Segments .... 10·47
Worksheet SP2D. Crashes by Seventy Level and Collision Type for Rural Two~Lane,
Two~ Way Roadway Segments .. . ... 10-48
Worksheet SP2E. Summary Results for Rural Two-Lane, Two-Way Roadway Segments .... .... 10-49
Worksheet SP3A. General Information and Input Data for Rural Two~Lane,
Two·Way Road Intersections .. 10·52
Worksheet SP3B. Crash Modification Factors for Rural Two·Lane, Two· Way Road Intersections ....... 10-52
Worksheet SP3C. Intersection Crashes for Rural Two·Lane, Two·Way Road Intersections .. .. 10-53
Worksheet SP3D. Crashes by Severity Level and Collision Type for Rural Two·Lane,
Two-Way Road Intersections .. .. 10·54
Worksheet SP3E. Summary Results for Rural Two-Lane, Two~ Way Road Intersections .. .... 10-54
Worksheet SP4A. General Information and Input Data for Rural Two-Lane,
Two-Way Road Intersections .. .... 10·57
Worksheet SP4B. Crash Modification Factors for Rural Two~Lane, Two~ Way Road Intersections.... ... 10-57
Worksheet SP4C. Intersection Crashes for Rural Two~Lane, Two-Way Road Intersections.. .. ... 10-58
Worksheet SP4D. Crashes by Severity Level and Collision Type for Rural Two-Lane,
Two-Way Road Intersections.. ... ................... ... 10-59
Worksheet SP4E. Summary Results for Rural Two-Lane, Two~Way Road Intersections. ... 10-59
Worksheet SPSA. Predicted and Observed Crashes by Severity and Site Type Us1ng the Site-Specific
EB Method for Rural Two-Lane, Two· Way Roads and Multilane Highways...... .. 10~61
Worksheet SP5B. Site-Specrf1c EB Method Summary Results for Rural Two-Lane, Two-Way Roads
and Multilane Highways.. .. .. 10~62
Worksheet SP6A. Predicted and Observed Crashes by Severity and Site Type Usmg the Project-Level
EB Method for Rural Two-Lane, Two-Way Roads and Multilane Highways ............ 10-64
Worksheet SP6B. Project~Level EB Method Summary Results for Rural Two-Lane, Two~Way Roads and
Multilane Highways.. . .... 10~66
Worksheet 1A General information and Input Data for Rural Two-Lane,
Two-Way Roadway Segments .. . 10-68
Worksheet 1B. Crash Modification factors for Rural Two-Lane, Two-Way Roadway Segments.. .. 10-69
Worksheet 1C. Roadway Segment Crashes for Rural Two-Lane, Two-Way Roadway Segments . 10-69
Worksheet 1D. Crashes by Severity Level and Collision Type for Rural Two-Lane,
Two-Way Roadway Segments... .. .. 10-70
Worksheet 1E. Summary Results for Rural Two-Lane, Two-Way Roadway Segment.... . 10-70
Worksheet 2A. General information and Input Data for Rural Two-Lane, Two-Way Road lntersectrons1 0-71
Worksheet 29. Crash Modificatron factors for Rural Two-Lane, Two-Way Road Intersections ........ 10-71
Worksheet 2C. Intersection Crashes for Rural Two-Lane, Two-Way Road Intersections.... .. ....... 10-71
Worksheet 2D. Crashes by Severity Level and Collision Type for Rural Two-lane,
Two-Way Road Intersections .. .. ................... 10-72
Worksheet 2E. Summary Results for Rural Two-lane, Two-Way Road Intersections .... ... 10-72
Worksheet 3A. Predicted and Observed Crashes by Severity and Site Type Us1ng the Site-Specific
EB Method for Rural Two-Lane, Two-Way Roads and Multilane Highways ............ 10-73
Worksheet 38. Site-SpecifiC EB Method Summary Results for Rural Two-Lane, Two-Way Roads
and Multilane Highways.. . 10-73
Worksheet 4A. Predicted and Observed Crashes by Severity and S1te Type Using the Project-Level
EB Method for Rural Two-Lane, Two-Way Roads and Multilane Highways. ...... 10-74
Worksheet 48. Project-Level EB Method Summary Results for Rural Two-Lane, Two-Way Roads
and Multilane Highways.. . ..................................... 10-75
CHAPTER 11-PREDICTIVE METHOD FOR RURAL MULTILANE HIGHWAYS ........•......... 11-1

Worksheet SP1 A. General Information and Input Data for Rural Multilane Roadway Segments ......... 11-40
Worksheet SP1 B. Crash Modification Factors for Rural Multilane Divrded Roadway Segments .. . 11-40
Worksheet SP1 C. Roadway Segment Crashes for Rural Multilane Divided Roadway Segments .......... 1141
Worksheet SP1 D. Crashes by Severity Level and Collision Type for Rural Multilane Divrded
Roadway Segments... ................... . . 11-42
Worksheet SP1 E. Summary Results for Rural Multilane Roadway Segments. .. 11-42
Worksheet SP2A. General Information and Input Data for Rural Multilane Roadway Segments .. 11-46
Worksheet SP28. Crash Modification Factors for Rural Multilane Undrvided Roadway Segments ...... 11-46
Worksheet SP2C. Roadway Segment Crashes for Rural Multilane Undivided Roadway Segments 11-47
Worksheet SP2D. Crashes by Severrty Level and Collrsion Type for Rural Multilane Undivided
Roadway Segments.. . .................. 11-48
Worksheet SP2E. Summary Results for Rural Multilane Roadway Segments. .. ..... 11-48
Worksheet SP3A. General Information and Input Data for Rural Multilane Highway lntersectrons 11-51
Worksheet SP38. Crash Modification Factors for Rural Multilane Highway Intersections.. .... 11-52
Worksheet SP3C. lntersectron Crashes for Rural Multilane Highway Intersections .. . .................. 11-52
Worksheet SP3D. Crashes by Severity Level and Collision Type for Rural Multilane
Hrghway Intersections . .. ........... 11-53
Worksheet SP3E. Summary Results for Rural Multilane Highway Intersections ... .. ...... 11-53
Worksheet SP4A. Predicted and Observed Crashes by Severity and Site Type Using the Site-Specific
EB Method for Rural Two-Lane, Two-Way Roads and Multilane Highways ............. 11-55
Worksheet SP4B. Srte-Specific EB Method Summary Results for Rural Two-Lane, Two-Way Roads
and Multilane Hrghways .. ................. 11-56
Worksheet SP5A. Predrcted and Observed Crashes by Severity and Srte Type Using the Project-Level
EB Method for Rural Two-Lane, Two-Way Roads and Multilane Highways.. ....... 11-58
Worksheet SPSB. Project-Level EB Method Summaiy Results for Rural Two-Lane, Two-Way Roads
and Multilane Hrghways .. ... ........... . ................. 11-60
Worksheet 1A. General Information and Input Data for Rural Multilane Roadway Segments.. . .. 11-62
Worksheet 1B (a). Crash Modification Factors for Rural Multilane Divided Roadway Segments ... 11-62
Worksheet 1B (b). Crash Modrfication Factors for Rural Multilane Undivided Roadway Segments.. .. 11-62
Worksheet 1C (a). Roadway Segment Crashes for Rural Multilane Divided Roadway Segments ......... 11-63
Worksheet 1C (b). Roadway Segment Crashes for Rural Multilane Undivided Roadway Segments. . 11-63
Worksheet 1D (a). Crashes by Severity Level and Collisron Type for Rural Multilane Divided
Roadway Segments.. . .... 11-64
Worksheet 1D (b). Crashes by Severity Level and Collision Type for Rural Multilane Undivided
Roadway Segments.. ............... 11-64
Worksheet 1E. Summary Results for Rural Multilane Roadway Segments .. ................. 11-65
Worksheet 2A. General Information and Input Data for Rural Multilane Highway lntersectrons ..... 11-65
Worksheet 2B. Crash Modification Factors for Rural Multilane Highway Intersections....... ....... 11-65
Worksheet 2C. Intersection Crashes for Rural Multilane Highway Intersections.... . .. 11-66
Worksheet 2D. Crashes by Severity Level and Collision Type for Rural Multilane
Highway Intersections.... .......... 11-66
Worksheet 2E. Summary Results for Rural Multilane Highway Intersections... .. ... 11-67
Worksheet 3A. Predicted and Observed Crashes by Seventy and Site Type Using the
Site-Specific EB Method . .. 11-67
Worksheet 3B. Site-Specific EB Method Summary Results........ .. .. 11-68
Worksheet 4A. Predicted and Observed Crashes by Severity and Site Type Using the Project-Level
EB Method.... .. .. 11-68
Project-Level EB Method Summary Results ...... .. ..... 11-69

Worksheet 4B.
Worksheet SP1 A. General Information and Input Data for Urban and Suburban Roadway Segments 12-56
Worksheet SP1 B. Crash Modification Factors for Urban and Suburban Roadway Segments.. . ..... 12-56
Worksheet SPlC. Mult'1ple-Veh1cle Nondriveway Collisions by Severity Level for Urban and
Suburban Roadway Segments.... .. ..... 12-57
Worksheet SP1 D. Multiple-Vehicle Nondriveway Collisions by Collision Type for Urban and Suburban
Roadway Segments... .... 12-58
Worksheet SP1 E. Single-Vehicle Collisions by Seventy Level for Urban and Suburban Roadway Segments .. 12-58
Worksheet SP1 F. Single-Vehicle Collisions by Collision Type for Urban and Suburban Roadway Segments. 12-59
Worksheet SP1 G. Multiple-Vehicle Driveway-Related Collisions by Driveway Type for Urban and
Suburban Roadway Segments... ..... 12-60
Worksheet SP1 H. Multiple-Vehicle Driveway-Related Collisions by Severity Level for Urban and
Suburban Roadway Segments.. . 12-60
Worksheet SP11. Vehicle-Pedestrian Collisions for Urban and Suburban Roadway Segments. ....... 12-61
Worksheet SP1 J. Vehicle-Bicycle Collisions for Urban and Suburban Roadway Segments.... .. .. 12-61
Worksheet SP1 K. Crash Severity Distribution for Urban and Suburban Roadway Segments.. . .. 12-62
Worksheet SP1 L. Sum~ary Results for Urban and Suburban Roadway Segments.. .. ......... 12-62
Worksheet SP2A. General Information and Input Data for Urban and Suburban Roadway Segments .. 12-67
Worksheet SP2B. Crash Modification Factors for Urban and Suburban Roadway Segments ... 12-68
Worksheet SP2C. Multiple-Vehicle Nondriveway Collisions by Severity Level for Urban and
Suburban Roadway Segments .. .. ..... 12-68
Worksheet SP2D. Multiple-Vehicle Nondriveway Collisions by Collision Type for Urban and
Suburban Roadway Segments... .. 12-69
Worksheet SP2E. Single-Vehicle Collisions by Severity Level for Urban and Suburban Roadway Segments .. 12-70
Worksheet SP2F. Single-Vehicle Collisions by Collis1on Type for Urban and Suburban Roadway Segments . 12-70
Worksheet SP2G. Multiple-Vehicle Driveway-Related Collisions by Driveway Type for Urban and
Suburban Roadway Segments... . 12-71
Worksheet SP2H. Multiple-Vehicle Driveway-Related Collisions by Severity Level for Urban and
Suburban Roadway Segments.. 12-72
Worksheet SP21. Vehicle-Pedestrian Collisions... .. 12-72
Worksheet SP2J. Vehicle-Bicycle Collisions for Urban and Suburban Roadway Segments.. .......... 12-72
Worksheet SP2K. Crash Seventy D'1stribuf1on for Urban and Suburban Roadway Segments .... . 12-73
Worksheet SP2L. Summary Results for Urban and Suburban Roadway Segments ... .. ... 12-74
Worksheet SP3A. General Information and Input Data for Urban and Suburban Artenallntersedions . 12-79
Worksheet SP3B. Crash Modifkation Factors for Urban and Suburban Arterial Intersections .......... 12-80
Worksheet SP3C. Multiple-Vehicle Collisions by Severity Level for Urban and Suburban
Arterial Intersections . .. .. 12-80
.l.•
Worksheet SP3D. Multiple-Vehicle Collisions by Collision Type lor Urban and Suburban
Arterial intersections ............ . .. .................. 12-81
Worksheet SP3E. Single-Vehicle ColliSions by Severity Level lor Urban and Suburban
Arterial intersections.. ............... .. ........... . . ... 12-82
Worksheet SP3F. Smgle-Vehicle Collisions by Collision Type for Urban and Suburban
Arterial intersections.......... ................. . ...................... 12-83
Worksheet SP3G. Vehicle-Pedestrian Collisions for Urban and Suburban Arterial
Stop-Controlled Intersections ............... . 12-83
Worksheet SP3J. Vehicle-Bicycle Collisions for Urban and Suburban Arterial Intersections .. 12-84
Worksheet SP3K. Crash Severity Distribution for Urban and Suburban Arterial Intersections ... ... 12-85
Worksheet SP3L. Summary Results for Urban and Suburban Arterial intersections ... .. 12-85
Worksheet SP4A. General information and Input Data for Urban and Suburban Artenallntersections .. 12-90
Worksheet SP4B. Crash Modification Factors for Urban and Suburban Arterial Intersections.. .. 12-91
Worksheet SP4C. Multiple-Vehicle Collisions by Severity Level for Urban and
Suburban Arterial Intersections ............... .. .. ................. 12-91
Worksheet SP4D. Multiple-Vehicle Collisions by Coll'1sion Type for Urban and
Suburban Arterial Intersections.. .. ... 12-92
Worksheet SP4E. Single-Vehicle Collisions by Seventy Level for Urban and Suburban
Arterial Intersections .. .. ............... 12-93
Worksheet SP4F. Single-Vehicle Collisions by Collis'1on Type for Urban and Suburban
Arterial Intersections. .. ............ .. ... 12-94
Worksheet SP4H. Crash Modification Factors for Vehicle-Pedestrian Collisions for Urban and
Suburban Arterial Signalized Intersections . .. .. 12-94
Worksheet SP41. Vehicle-Pedestrian Collisions for Urban and Suburban Arterial Signalized Intersections ... 12-95
Worksheet SP4J. Vehicle-Bicycle Collisions for Urban and Suburban Arteriallntersect1ons ............... 12-95
Worksheet SP4K. Crash Severity Distnbution for Urban and Suburban Arterial Intersections.. . 12-96
Worksheet SP4L. Summary Results for Urban and Suburban Arterial Intersections ... .. ... 12-96
Worksheet SPSA. Predicted Crashes by Collision and Site Type and Observed Crashes Using the
Site-Specific EB Method for Urban and Suburban Arterials .. . 12-98
Worksheet SPSB. Predicted Pedestrian and Bicycle Crashes for Urban and Suburban Arterials ........ 12-1 01
Worksheet SPSC. Site-Specific EB Method Summary Results for Urban and Suburban Arterials ....... 12-1 01
Worksheet SP6A. Predicted Crashes by Collision and Site Type and Observed Crashes Us1ng the
ProJect-Level EB Method for Urban and Suburban Arterials ........... 12-103
Worksheet SP6B. Predicted Pedestnan and Bicycle Crashes for Urban and Suburban Artenals ... 12-106
Worksheet SP6C. Project-Level EB Method Summary Results for Urban and Suburban Arterials ...... 12-106
Worksheet 1A. General Information and Input Data for Urban and Suburban Roadway Segments 12-1 08
Worksheet 1B. Crash Modification Factors for Urban and Suburban Roadway Segments ............ 12-108
Worksheet 1C. Multiple-Vehicle Nondriveway Collisions by Severity Level for Urban and
Suburban Roadway Segments... .. ............. 12-1 09
Worksheet 1D. Multiple-Vehicle Non driveway Collisions by Collision Type for Urban and
Suburban Roadway Segments....... .. ... 12-109
Worksheet 1E. Single-Vehicle Collisions by Severity Level for Urban and Suburban
Roadway Segments ...... 12-11 0
Worksheet 1F. Single-Vehicle Collisions by Collision Type for Urban and Suburban
Roadway Segments... ...... .................. .. .......... 12-110
Worksheet 1G. Multiple-Vehicle Driveway-Related Collisions by Driveway Type for Urban and
Suburban Roadway Segments............. .. .. 12-111
Worksheet 1H. Multiple-Vehicle Driveway-Related Collisions by Sever'rty Level for Urban and
Suburban Roadway Segments.. .. 12-111
Worksheet 11. Vehicle-Pedestrian Collisions for Urban and Suburban Roadway Segments . 12-111
Worksheet 1J. Vehicle~Bicycle Collisions for Urban and Suburban Roadway Segments ............... 12-112
Worksheet 1K. Crash Severity Distribution for Urban and Suburban Roadway Segments ............. 12-112
Worksheet 1L. Summary Results for Urban and Suburban Roadway Segments .. .. ..... 12-113
Worksheet 2A. General information and Input Data for Urban and Suburban Arterial Intersections ... 12-113
Worksheet 28. Crash Modification Factors for Urban and Suburban Arterial Intersections.. . 12-1 14
Worksheet 2C. Multiple-Vehicle Collisions by Severity Level for Urban and
Suburban Arterial Intersections........ .. ...... 12-114
Worksheet 20. Multiple-Vehicle Collisions by Collis1on Type for Urban and
Suburban Arterial Intersections... ... 12-114
Worksheet 2E. Single-Vehicle Collisions by Severity Level for Urban and Suburban
Arterial Intersections . .. 12-115
Worksheet 2F. Single-Vehicle Collisions by Collision Type for Urban and Suburban
Arterial Intersections...... ... ............. .. 12-115
Worksheet 2G. Vehicle-Pedestrian Collisions for Urban and Suburban Arterial
Stop-Controlled Intersections. ... ............ ... 12-115
Worksheet 2H. Crash Modification Factors for Vehicle-Pedestrian Collisions for Urban and
Suburban Arterial Signalized Intersections.. ......... 12-116
Worksheet 21. Vehicle-Pedestrian Collisions for Urban and Suburban Arterial
Signalized Intersections . .. ... 12-116
Worksheet 2J. Vehicle-Bicycle Collisions for Urban and Suburban Arterial Intersections .. .. ... 12-116
Worksheet 2K. Crash Severity Distribution for Urban and Suburban Arterial Intersections ........... 12-117
Worksheet 2L. Summary Results for Urban and Suburban Arterial Intersections .... .. .. 12-117
Worksheet 3A. Predicted Crashes by Collision and S1te Type and Obse'rved Crashes
Using the Site-SpecifiC EB Method for Urban and Suburban Arterials .... ... 12-118
Worksheet 38. Predided Pedestrian and B1cycle Crashes for Urban and Suburban Arterials ........ 12-119
Worksheet 3C. Site-Specific EB Method Summary Results for Urban and Suburban Arterials ....... 12-119
Worksheet 4A. Predicted Crashes by Collision and Site Type and Observed Crashes
Using the Project-Level EB Method for Urban and Suburban Artenals .. . 12-120
Worksheet 48. Pred1cted Pedestrian and Bicycle Crashes for Urban and Suburban Arterials ........ 12-122
Worksheet 4C. ProJed-Level EB Method Summary Results for Urban and Suburban Artenals.... . 12-122
XXV
Preface to the Highway Safety Manual
PURPOSE OF THE HSM

The Highway Safety Manual (HSM) is a resource that provides safety knowledge and tools in a useful form to
facilitate improved decision making based on safety performance. The focus of the HSM is to provide quantita-
tive information for decision making. The HSM assembles currently available information and methodologies on
measuring, estimating and evaluating roadways in terms of crash frequency (number of crashes per year) and crash
severity (level of injuries due to crashes). The HSM presents tools and methodologies for consideration of"safety"
across the range of highway activities: planning, programming, project development, construction, operations, and
maintenance. The purpose is to convey present knowledge regarding highway safety information for use by a broad
array of transportation professionals.
THE NEED FOR THE HSM

Prior to this edition of the HSM, transportation professionals did not have a single national resource for quantitative
information about crash analysis and evaluation. The HSM begins to fill this gap, providing transportation profes-
sionals with current lmowledge, techniques, and methodologies to estimate future crash frequency and severity and
to identify and evaluate options to reduce crash frequency and severity.
In addition to using descriptive methods in better ways, the HSM permits use of predictive methodologies that
improve and expand the use of crash estimation methods to new and alternative design or conditions in past or
future periods. The more statistically rigorous predictive methods in the HSM reduce the vulnerability of historical
crash-based methods to random variations of crash data and provide a means to estimate crashes based on geometry,
operating characteristics, and traffic volumes. These techniques provide an opportunity to: I) improve the reliability
of common activities, such as screening a network for sites at which to reduce crashes, and 2) expand analysis to
include assessments of new or alternative geometric and operational characteristics.
THE HISTORY OF THE FIRST EDITION OF THE HSM

A special conference session was held at the annual meeting of the Transportation Research Board (TRB) in January
1999 on the subject of predicting highway safety impacts of highway design and operation. The session participants
concluded that one reason for a lack of quantitative safety emphasis in decision making is the absence of a single au-
thoritative document to use for quantitatively estimating "safety." In December of 1999, a workshop was held under
sponsorship of eight TRB committees and funded by FHWA for the purpose of determining the need for, nature of,
and feasibility of producing a highway safety manual. An initial outline and plan for an HSM was produced. This led
to the formation of a TRB Joint Subcommittee in May of 2000. Subsequently, the Subcommittee became the Task
Force for the Development of a Highway Safety Manual (ANB25T). It was under the direction of this Task Force of
volunteers that the materials for this edition were produced. The Task Force formed several subcommittees to over-
see various research and development aspects of the task. They also employed independent review groups to assess
xxvii
research results before proceeding with final preparation of materials. The majority of the research and development
was funded by the NCHRP, with significant supplementary funding and research support provided by the FHWA.
In 2006, the decision was made to publish the HSM as an AASHTO document. A Joint Task Force (JTF) was formed
with representatives from the Subcommittees on Design, Traffic Engineering, and Safety Management. The JTF
members were tasked with ensuring the HSM meets the needs of the state Departments of Transportation, and with
promoting the HSM to their respective subcommittees. In 2009, the subcommittees and parent committees, the
Standing Committee on Higbways and the Standing Committee on Higbway Traffic Safety, balloted and approved
the HSM. The AASHTO Board of Directors then approved the HSM.
CONSIDERATIONS AND CAUTIONS WHEN USING THE HSM

The HSM translates analytical tools based upon scientifically based knowledge, methods, and processes into a form
that is usable by transportation professionals.
The HSM will be used by individuals with a variety of professional and technical backgrounds, including
engineering, planning, field operations, enforcement, and education. They will come to the HSM with different
levels of understanding of the fundamentals of roadway safety. Chapter I, "Introduction and Overview," provides
key information and the context for understanding how to apply and integrate safety analysis related to the common
activities within higbway planning, design, and operations. The HSM includes traditional "safety" analysis
techniques and also applies recent developments in crash estimation and evaluation methodologies. A majority of
the analytical techniques are new; it is important to fully understand the material presented in Chapter 2, "Human
Factors," and Chapter 3, "Fundamentals," to understand reasons for development and use of these techniques.
Because the HSM does not account for jurisdiction-specific differences, it contains calibration techniques to modify
tools for local use. This is necessary because of differences in factors, such as driver populations, local roadway and
roadside conditions, traffic composition, typical geometries, and traffic control measures. There are also variations
in how each state or jurisdiction reports crashes and manages crash data. Chapter 3, "Fundamentals," discusses this
topic and others related to the reliability of crash data. Calibration does not make the crash data uniform across
states. Similarly, applying the HSM outside the United States and Canada should be done with caution. The models
and research findings presented in this document may not be applicable in other countries as the roadway systems,
driver training and behavior, and crash frequencies and severity patterns may be widely different. At a minimum,
techniques presented in the HSM should be properly calibrated.
The HSM is not a legal standard of care as to the information contained herein. Instead, the HSM provides analytical
tools and techniques for quantifying the potential effects of decisions made in planning, design, operations, and
maintenance. There is no such thing as "absolute safety," noTwithstanding efforts by government to maintain,
improve and operate highway facilities to the higbest level that government funding allows. There is risk in all
highway transportation. That risk is inherent due to the variability of user behaviors, environmental conditions, and
other factors over which the government has no control. A universal objective is to reduce the number and severity of
crashes within the limits of available resources, science, technology, and legislatively mandated priorities. Because
these considerations are constantly changing, it is unlikely, if not impossible, that any highway facility can be
"state of the art". The information in the HSM is provided to assist agencies in their effort to integrate safety into
their decision-making processes. The HSM is not intended to be a substitute for the exercise of sound engineering
judgment. No standard of conduct or any duty toward the public or any person shall be created or imposed by the
publication and use or nonuse of the HSM.
As a resource, the HSM does not supersede publications such as the Manual on Uniform Traffic Control Devices
(MUTCD), American Association of State Highway Transportation Official's (AASHTO) "Green Book" titled
A Policy on Geometric Design ofHighways and Streets, or other AASHTO and agency guidelines, manuals, and
policies. If conflicts arise between these publications and the HSM, the previously established publications should be
given the weigbt they would otherwise be entitled, in accordance with sound engineering judgment. The HSM may
provide needed justification for an exception from previously established publications.
xxviii
FUTURE EDITIONS OF THE HSM
This first edition of the HSM provides the most current and accepted knowledge and practices relating to roadway
safety management. The TRB and AASHTO HSM Task Forces recognize that knowledge and methods of analysis
are evolving and improving with new research and lessons learned in practice.
The· evolution in professional practice and knowledge will be influenced by this first edition of the HSM because it
introduces new methods, techniques, and information to transportation professionals. The knowledge base will also
continue to grow and to enhance transportation professionals' understanding of how decisions related to planning,
design, operations, and maintenance affect crash frequency and severity. The transportation profession will continue
to take the opportunity to learn more about the relationships between crash occurrences on various types of facilities
and the corresponding geometry and operational characteristics of those facilities that may affect crash frequency
and severity. This will be facilitated as agencies improve the processes used to collect and maintain data for crashes,
roadway geometry, traffic volumes, land uses, and many other useful data to assess the roadway environment and
context in which crashes are occurring. These or other potential enhancements in analysis techniques and knowledge
will be reflected in future editions of the HSM.
x.xix
Part (-Introduction and
Applications Guidance
C.1. INTRODUCTION TO THE HIGHWAY SAFETY MANUAL PREDICTIVE METHOD

Part C of the Highway Safety Manual (HSM) provides a predictive method for estimating expected average crash
frequency (including by crash severity and collision types) of a network, facility, or individual site. The estimate can be
made for existing conditions, alternatives to existing conditions (e.g., proposed upgrades or treatments), or proposed
new roadways. The predictive method is applied to a given time period, traffic volume, and constant geometric design
characteristics of the roadway.
The predictive method provides a quantitative measure of expected average crash frequency under both existing
conditions and conditions which have not yet occurred. This allows proposed roadway conditions to be quantitatively
assessed along with other considerations such as community needs, capacity, delay, cost, right-of-way, and
environmental considerations.
The predictive method can be used for evaluating and comparing the expected average crash frequency of situations such as:
• Existing facilities under past or future traffic volumes;
• Alternative designs for an existing facility under past or future traffic volumes;
• Designs for a new facility under future (forecast) traffic volumes;
• The estimated effectiveness of countermeasures after a period of implementation; and
• The estimated effectiveness of proposed countermeasures on an existing facility (prior to implementation).
Part C-Introduction and Applications Guidance presents the predictive method in general terms for the first -time user
to understand the concepts applied in each of the Part C chapters. Each chapter in Part C provides the detailed steps of
the predictive method and the predictive models required to estimate the expected average crash frequency for a specific
facility type. The following roadway facility types are included in Part C:
• Chapter 10-Rural Two-Lane, Two-Way Roads
• Chapter 11-Rural Multilane Highways
• Chapter 12-Urban and Suburban Arterials
The Part C-lntroduction and Applications Guidance also provides:

• Relationships between Part C and Parts A, B, and D of the HSM;
• Relationship between Part C and the Project Development Process;

• An overview of the predictive method;
C-1
C-2 HIGHWAY SAFETY MANUAL
• A summary of the predictive method;
• Detailed information needed to understand the concepts and elements in each of the steps of the predictive method;
• Methods for estimating the change in crash frequency due to a treatment;
• Limitations of the predictive method; and

• Guidance for applying the predictive method.
C.2. RELATIONSHIP TO PARTS A, B, AND D OF THE HSM

All information needed to apply the predictive method is presented in Part C. The relationships of the predictive
method in Part C to the contents of Parts A, B, and Dare summarized below.
• Part A introduces concepts that are fundamental to understanding the methods provided in the HSM to analyze
and evaluate crash frequencies. Part A introduces the key components of the predictive method, including safety
performance functions (SPFs) and crash modification factors (CMFs). Prior to using the information in Part C, an
understanding of the material in Part A: Chapter 3, Fundamentals is recommended.
• Part B presents the six basic components of a roadway safety management process. The material is useful for
monitoring, improving, and maintaining an existing roadway network. Applying the methods and information
presented in Part B can help to identify sites most likely to benefit from an improvement, diagnose crash patterns
at specific sites, select appropriate countermeasures likely to reduce crashes, and anticipate the benefits and costs
of potential improvements. In addition, it helps agencies determine whether potential improvements are economi-
cally justified, establish priorities for potential improvements, and assess the effectiveness of improvements that
have been implemented. The predictive method in Part C provides tools to estimate the expected average crash
frequency for application in Part B: Chapter 4, Network Screening and Chapter 7, Economic Appraisal.
• Part D contains all CMFs in the HSM. The CMFs in Part Dare used to estimate the change in expected average
crash frequency as a result of implementing a counterrneasure(s). Some Part D CMFs are included in Part C for
use with specific SPFs. Other Part D CMFs are not presented in Part C but can be used in the methods to estimate
change in crash frequency described in Section C. 7.
C.3. PART C AND THE PROJECT DEVELOPMENT PROCESS

• Figure C-1 illustrates the relationship of the Part C predictive method to the project development process. As dis-
cussed in Chapter I, the project development process is the framework used in the HSM to relate crash analysis to
activities within planning, design, construction, operations, and maintenance.
PART C-INTRODUCfiON AND APPLICATIONS GUIDANCE C-3
System Planning
Project Planning
v em'"""'""'"""';"'"'""
performance of an existing facility.
During this process, Part D and
Chapters 5 through 7 (Part B) can
~ '''" "' ""''"

frequency "''"""select
and severity, """
countermeasures, and conduct an
economrc evaluation.
Preliminary Design Part C can be used to predict future

performance. During this process,
Part D and Chapters 6 through 7
Final Design ~ (Part B) can be used for selection
and economic evaluation of
countermeasures.
Construction
Operations
Maintenance
Figure C-1. Relation between Part C Predictive Method and the Project Development Process
C.4. OVERVIEW OF THE HSM PREDICTIVE METHOD

The predictive method provides an 18-step procednre to estimate the "expected average crash frequency'" (by
total crashes, crash severity, or collision type) of a roadway network, facility, or site. In the predictive method the
roadway is divided into individual sites that are either homogenous roadway segments or intersections. A facility
consists of a contiguous set of individual intersections and roadway segments, each referred to as "sites." Dif-
ferent facility types are determined by surrounding land use, roadway cross-section, and degree of access. For
each facility type a number of different site types may exist, such as divided and undivided roadway segments or
signalized and unsignalized intersections. A roadway network consists of a number of contiguous facilities.
The predictive method is used to estimate the expected average crash frequency of an individual site. The cumula-
tive sum of all sites is used as the estimate for an entire facility or network. The estimate is for a given time period of
interest (in years) during which the geometric design and traffic control features are unchanged and traffic volumes
are known or forecast. The estimate relies upon regression models developed from observed crash data for a number
of similar sites .
•L...
TT
The predicted average crash frequency of an individual site, N ,. __., is estimated based on the geometric design
pre 1""'" '
traffic control features, and traffic volumes of that site. For an existing site or facility, the observed crash frequency,
No•serve,, for that specific site or facility is then combined with Npre,.tcle,, to improve the statistical reliability of the
estimate. The result from the predictive method is the expected average crash frequency, N expecte,. This is an estimate
of the long-term average crash frequency that would be expected, given sufficient time to make a controlled observa-
tion, which is rarely possible. Once the expected average crash frequencies have been determined for all the individ-
ual sites that make up a facility or network, the sum of the crash frequencies for all of the sites is used as the estimate
of the expected average crash frequency for an entire facility or network.
As discussed in Section 3.3.3 in Chapter 3, the observed crash frequency (number of crashes per year) will
fluctuate randomly over any period and, therefore, using averages based on short-term periods (e.g., 1 to 3 years)
may give misleading estimates and create problems associated with regression-to-the-mean bias. The predictive
method addresses these concerns by providing an estimate oflong-term average crash frequency, which allows for
sound decisions about improvement programs.
In the HSM, predictive models are used to estimate the predicted average crash frequency, Npredictcd' for a particular
site type using a regression model developed from data for a number of similar sites. These regression models,
called safety performance functions (SPFs), have been developed for specific site types and "base conditions" that
are the specific geometric design and traffic control features of a "base" site. SPFs are typically a function of only
a few variables, primarily average annual daily traffic (AADT) volumes.
Adjustment to the prediction made by an SPF is required to account for the difference between base conditions,
specific site conditions, and local/state conditions. Crash modification factors (CMFs) are used to account for
the specific site conditions which vary from the base conditions. For example, the SPF for roadway segments in
Chapter 10 has a base condition of 12-ft lane width, but the specific site may be a roadway segment with a 10-ft
lane width. A general discussion of CMFs is provided in Section C.6.4.
CMFs included in Part C chapters have the same base conditions as the SPFs in Part C and, therefore, the
CMF ~ 1.00 when the specific site conditions are the same as the SPF base conditions.
A calibration factor (C ) is used to account for differences between the jurisdiction( s) for which the models were
developed and the juri;diction for which the predictive method is applied. The use of calibration factors is described
in Section C.6.5 and the procedure to determine calibration factors for a specific jurisdiction is described in the Part
C, Appendix A. I.
The predictive models used in Part C to determine the predicted average crash frequency, N'"""''' are of the general
form shown in Equation C-1.
(C-1)
Where:
predicted average crash frequency for a specific year for site type x;
predicted average crash frequency determined for base conditions of the SPF developed for site type x;
crash modification factors specific to SPF for site type x; and
calibration factor to adjust SPF for local conditions for site type x.
For existing sites, facilities, or roadway networks, the Empirical Bayes (EB) Method is applied within the predic-
tive method to combine predicted average crash frequency determined using a predictive model, Npn:,·.tete,, with the
observed crash frequency, N,•.,N,, (where applicable). A weighting is applied to the two estimates which reflects
PART (-INTRODUCTION AND APPLICATIONS GUIDANCE C-5
the statistical reliability of the SPF. The EB Method applies only when observed crash data are available. A discus-
sion of the EB Method is presented in the Part C, Appendix A.2. The EB Method may be applied at the site-specific
level when crashes can be assigned to individual sites (i.e., detailed geographic location of the observed crashes is
!mown). Alternatively, the EB Method can be applied at the project-specific level (i.e., to an entire facility or net-
work) when crashes cannot be assigned to individual sites but are !mown to occur within general geographic limits
(i.e., detailed geographic locations of crashes are not available). As part of the EB Method, the expected average
crash frequency can also be estimated for a future time period, when AADT may have changed or specific treatments
or countermeasures may have been implemented.
Advantages of the predictive method are that:

• Regression-to-the-mean bias is addressed as the method concentrates on long-term expected average crash fre-
quency rather than short-term observed crash frequency.
• Reliance on availability of crash data for any one site is reduced by incorporating predictive relationships based on
data from many similar sites.
• The SPF models in the HSM are based on the negative binomial distribution, which are better suited to modeling the
high natural variability of crash data than traditional modeling techniques, which are based on the normal distnbution.
• The predictive method provides a method of crash estimation for sites or facilities that have not been constructed
or have not been in operation long enough to make an estimate based on observed crash data.
The following sections provide the generaliS steps of the predictive method and detailed information about each of
the concepts or elements presented in the predictive method. The information in the Part C-Introduction and Appli-
cations Guidance chapter provides a brief summary of each step. Detailed information on each step and the associ-
ated predictive models are provided in the Part C chapters for each of the following facility types:
• Chapter 10-Rural Two-Lane, Two-Way Roads
C.5. THE HSM PREDIGIVE METHOD

While the general form of the predictive method is consistent across the chapters, the predictive models vary by
chapter and therefore the detailed methodology for each step may vary. The generic overview of the predictive
method presented here is intended to provide the first time or infrequent user with a high level review of the steps in
the method and the concepts associated with the predictive method. The detailed information for each step and the
associated predictive models for each facility type are provided in Chapters 10, 11, and 12. Table C-1 identifies the
specific facility and site types for which safety performance fimctions have been developed for the HSM.
Table C-1. Safety Performance Functions by Facility Type and Site Types in Part C
Intersections
Undivided Divided
HSM Chapter/
Roadway Roadway Stop Control on Minor Leg(s) Signalized
Facility Type
Segments Segments
3-Leg 4-Leg 3-Leg 4-Leg
10-Rural Two-Lane,
.I ,/ ,/ ,/
Two-Way Roads
11-Rural Multilane
,/ ,/ ,/ ,/ ,/
Highways
12-Urban and
,/ ,/ ,/ ,/ ,/ ,/
Suburban Arterials
The predictive method in Chapters 10, II, and 12 consists of 18 steps. The elements of the predictive models that
were discussed in Section C.4 are determined and applied in Steps 9, 10, and II of the predictive method. The 18
steps of the HSM predictive method are detailed below and shown graphically in Figure C-2. Brief detail is provided
for each step, and material outlining the concepts and elements of the predictive method is provided in the following
sections of the Part C-Introduction and Applications Guidance or in the Part C, Appendix A. In some situations,
certain steps will not require any action. For example, a new site or facility will not have observed crash data and,
therefore, steps relating to the EB Method are not performed.
Where a facility consists of a number of contiguous sites or crash estimation is desired for a period of several years,
some steps are repeated. The predictive method can be repeated as necessary to estimate crashes for each alternative
design, traffic volume scenario, or proposed treatment option within the same period to allow for comparison.
Step 1 I Define roadway limits and facility type . I

.j.
Step 2
I Define the period of study. I
.j.
Determine AADT ;md avaii<Jbility of crash
Step 3
I data for every year in the period of interest.
.j.
Step 4 Determine geometric conditions.
t
Divide roadway into individual
Step 5
I roadway segments and inte~ctions
I
.j.
Step 6 I Assign observed crashes to individual sites (it applicable).
.j.
Step 7 -1 Select a roadway segment or intersection.
.j.
Step 8 r+l Select first or neX'! year of the evaluation period. I
.j.
Step 9 Select and apply SPF.
t
Step 10 Apply CMFs.
I
Step 11
I Apply a calibration factor. I
~
Step 12
YES
1
Step 13 Apply site-specific EB method {if applicable).
Step 14
Step 15 I
YES
<$> 1
Apply project-level EB method (if applicable}.
Step 16
I Sum all sites and years.
Is there
an alternati~
design, tre.atmem, or YES
Step 17
foreGJst AAOT to
be evaluated?
Step 18
I Compare and evaluate results, I
Figure C-2. The HSM Predictive Method

PART C-INTRODUCfiON AND APPLICATIONS GUIDANCE C-7
Step 1-Define the limits of the roadway and facility types in the study network, facility, or site for
which the expected average crash frequency, severity, and collision types are to be estimated.
The predictive method can be undertaken for a roadway network, a facility, or an individual site. The facility types
included in the HSM are outlined in Section C.6.1. A site is either an intersection or homogeneous roadway segment.
There are a number of different types of sites, such as signalized and unsignalized intersections or divided and undi-
vided roadway segments. The site types included in the HSM are indicated in Table C-1.
The predictive method can be applied to an existing roadway, a design alternative for an existing roadway, or a design alter-
native for new roadway (that may be either unconstructed or yet to experieoce enough traffic to have observed crash data).
The limits of the roadway of interest will depend on the nature of the study. The study may be limited to only one specific
site or a group of contiguous sites. Alternatively, the predictive method can be applied to a long corridor for the purposes
of network screening (determining which sites require upgrading to reduce crashes) which is discussed in Chapter 4.
Step 2-Define the period of interest.

The predictive method can be undertaken for a past period or a future period. All periods are measured in years.
Years of interest will be determined by the availability of observed or forecast AADTs, observed crash data, and
geometric design data. Whether the predictive method is used for a past or future period depends upon the purpose
of the study. The period of study may be:
• A past period (based on observed AADTs) for:
• An existing roadway network, facility, or site. If observed crash data are available, the period of study is the
period of time for which the observed crash data are available and for which (during that period) the site
geometric design features, traffic control features, and traffic volumes are known.
Cll An existing roadway network, facility, or site for which alternative geometric design features or traffic con-
trol features are proposed (for near term conditions).
• A future period (based on forecast AADTs) for:
• An existing roadway network, facility, or site for a future period where forecast traffic volumes are available.
• An existing roadway network, facility, or site for which alternative geometric design or traffic control fea-
tures are proposed for implementation in the future.
• A new roadway network, facility, or site that does not currently exist, but is proposed for construction during
some future period.
Step 3-For the study period, determine the availability of annual average daily traffic volumes
and, for an existing roadway network, the availability of observed crash data to determine wheth-
er the EB Method is applicable.
Determining Traffic Volumes
The SPFs used in Step 9 (and some CMFs in Step 10), require AADT volumes (vehicles per day). For a past period,
the AADT may be determined by automated recording or estimated by a sample survey. For a future period, the
AADT may be a forecast estimate based on appropriate land use planning and traffic volume forecasting models,
or based on the assumption that current traffic volumes will remain relatively constant.
For each roadway segment, the AADT is the average daily two-way, 24-hour traffic volume on that roadway seg-
ment in each year of the period to be evaluated (selected in Step 8).
For each intersection, two values are required in each predictive model. These are the AADT of the major street,
AADT rna}., and the AADT of the minor street, AADT mlrl.. The method for determining AADT maj and AADT mm. varies
between chapters because the predictive models in Chapters 10, 11, and 12 were developed independently.
In many cases, it is expected that AADT data will not be available for all years of the evaluation period. In that
case, an estimate of AADT for each year of the evaluation period is determined by interpolation or extrapolation
as appropriate. If there is not an established procedure for doing this, the following default rules can be applied:
• If AADT data are available for only a single year, that same value is assumed to apply to all years of the before period.
• If two or more years of AADT data are available, the AADTs for intervening years are computed by
interpolation.
• The AADTs for years before the first year for which data are available are assumed to be equal to the AADT for
that first year.
• The AADTs for years after the last year for which data are available are assumed to be equal to the last year.
If the EB Method is to be used (discussed below), AADT data are needed for each year of the period for which
observed crash frequency data are available. If the EB Method will not be used, AADT data for the appropriate time
period-past, present, or future-determined in Step 2 are used.
Determining Availability of Observed Crash Data

Where an existing site or alternative conditions to an existing site are being considered, the EB Method is used.
The EB Method is only applicable when reliable, observed crash data are available for the specific study roadway
network, facility, or site. Observed data may be obtained directly from the jurisdiction's crash report system. At
least two years of observed crash frequency data are desirable to apply the EB Method. The EB Method and crite-
ria to determine whether the EB Method is applicable are presented in Section A.2.1 in the Appendix A to Part C.
The EB Method can be applied at the site-specific level (i.e., observed crashes are assigned to specific intersections
or roadway segments in Step 6) or at the project level (i.e., observed crashes are assigned to a facility as a whole).
The site-specific EB Method is applied in Step 13. Alternatively, if observed crash data are available but can not be
assigned to individual roadway segments and intersections, the project-level EB Method is applied (in Step 15).
If observed crash frequency data are not available, then Steps 6, 13, and 15 of the predictive method would not be
performed. In this case, the estimate of expected average crash frequency is limited to using a predictive model
(i.e., the predicted average crash frequency).
Step 4-Determine geometric design features, traffic control features, and site characteristics for
all sites in the study network.
In order to determine the relevant data required and avoid unnecessary collection of data, it is necessary to under-
stand the base conditions of the SPFs in Step 9, and the CMFs in Step 10. The base conditions for the SPFs for each
of the facility types in the HSM are detailed in Chapters 10, II, and 12.
Step 5-Divide the roadway network or facility under consideration into individual roadway seg-
ments and intersections, which are referred to as sites.
Using the information from Step I and Step 4, the roadway is divided into individual sites, consisting of individual
homogenous roadway segments and intersections. Section C.6.2 provides the general definitions of roadway seg-
ments and intersections used in the predictive method. When dividing roadway facilities into small homogenous
roadway segments, limiting the segment length to no less than 0.10 miles will minimize calculation efforts and not
affect results.
Step 6-Assign observed crashes to the individual sites (if applicable).

Step 6 only applies if it was determined in Step 3 that the site-specific EB Method was applicable. If the site-specific
EB Method is not applicable, proceed to Step 7. In Step 3, the availability of observed data and whether the data
could be assigned to specific locations was determined. The specific criteria for assigning crashes to individual road-
way segments or intersections are presented in Section A.2.3 of the Appendix A to Part C.
Crashes that occur at an intersection or on an intersection leg, and are related to the presence of an intersection,
are assigned to the intersection and used in the EB Method together with the predicted average crash frequency
for the intersection_ Crashes that occur between intersections and are not related to the presence of an intersection
are assigned to the roadway segment on which they occur, this includes crashes that occur within the intersection
limits but are unrelated to the presence of the intersection. Such crashes are used in the EB Method together with the
predicted average crash frequency for the roadway segment.
Step 7-Select the first or next individual site in the study network. If there are no more sites to be
evaluated, go to Step 15.
In Step 5 the roadway network within the study limits is divided into a number of individual homogenous sites
(intersections and roadway segments). At each site, all geometric design features, traffic control features, AADTs,
and observed crash data are determined in Steps I through 4. For studies with a large number of sites, it may be
practical to assign a number to each site.
The outcome of the HSM predictive method is the expected average crash frequency of the entire study network,
i.e., the sum of the all of the individual sites for each year in the study. Note that this value will be the total number
of crashes expected to occur over all sites during the period of interest. If a crash frequency is desired, the total can
be divided by the number of years in the period of interest.
The estimate for each site (roadway segments or intersection) is undertaken one at a time. Steps 8 through 14,
described below, are repeated for each site.
Step 8-For the selected site, select the first or next year in the period of interest. If there are no
more years to be evaluated for that site, proceed to Step 15.
Steps 8 through 14 are repeated for each site in the study and for each year in the study period.
The individual years of the evaluation period may have to be analyzed one year at a time for any particular roadway
segment or intersection because SPFs and some CMFs (e.g., lane and shoulder widths) are dependent on AADT,
which may change from year to year.
Step 9-For the selected site, determine and apply the appropriate Safety Performance
Function (SPF) for the site's facility type and traffic control features.
Steps 9 through 13, described below, are repeated for each year of the evaluation period as part of the evalua·
tion of any particular roadway segment or intersection.
Each predictive model in the HSM consists of a safety performance function (SPF), that is adjusted to site-
specific conditions (in Step 10) using crash modification factors (CMFs) and adjusted to local jurisdiction
conditions (in Step II) using a calibration factor (C). The SPFs, CMFs, and calibration factor obtained in Steps
9, 10, and II are applied to calculate the predicted average crash frequency for the selected year of the selected
site. The resultant value is the predicted average crash frequency for the selected year.
The SPF (which is a statistical regression model based on observed crash data for a set of similar sites) esti-
mates the predicted average crash frequency for a site with the base conditions (i.e., a specific set of geometric
design and traffic control features). The base conditions for each SPF are specified in each of the Part C chap-
ters. A detailed explanation and overview of the SPFs in Part Cis provided in Section C.6.3.
The facility types for which SPFs were developed for the HSM are shown in Table C-1. The predicted aver-
age crash frequency for base conditions is calculated using the traffic volume determined in Step 3 (AADT for
roadway segments or AADT maJ. and -AADT mm. for intersections) for the selected year.
The predicted average crash frequency may be separated into components by crash severity level and colli-
sion type. Default distributions of crash severity and collision types are provided in the Part C chapters. These
default distributions can benefit from being updated based on local data as part of the calibration process pre-
sented in Appendix A.l.l to Part C.
Step 10-Multiply the result obtained in Step 9 by the appropriate CMFs to adjust the predicted
average crash frequency to site-specific geometric design and traffic control features.
Each SPF is applicable to a set of base geometric design and traffic control features, which are identified for
each site type in the Part C chapters. In order to account for differences between the base geometric design and
the specific geometric design of the site, CMFs are used to adjust the SPF estimate. An overview ofCMFs and
guidance for their use is provided in Section C.6.4 including the limitations of current knowledge regarding
the effects of simultaneous application of multiple CMFs. In using multiple CMFs, engineering judgment is
required to assess the interrelationships. or independence. or both, of individual elements or treatments being
considered for implementation within the same project.
All CMFs used in Part C have the same base conditions as the SPFs used in the Part C chapter in which the CMF
is presented (i.e .• when the specific site has the same condition as the SPF base condition, the CMF value for that
condition is 1.00). Only the CMFs presented in Part C may be used as part of the Part C predictive method.
Part D contains all CMFs in the HSM. Some Part D CMFs are included in Part C for use with specific SPFs. Other
Part D CMFs are not presented in Part C, but can be used in the methods to estimate change in crash frequency
described in Section C.7.
For urban and suburban arterials (Chapter 12), the average crash frequency for pedestrian- and bicycle-base crashes
is calculated at the end of this step.
Step 11-Multiply the result obtained in Step 10 by the appropriate calibration factor.
The SPFs used in the predictive method have each been developed with data from specific jurisdictions and time
periods. Calibration ofSPFs to local conditions will account for differences. A calibration factor (C, for roadway
segments or C1 for intersections) is applied to each SPF in the predictive method. An overview of the use of calibra-
tion factors is provided in Section C.6.5. Detailed guidance for the development of calibration factors is included in
Part C, Appendix A.l.l.
Step 12-lf there is another year to be evaluated in the study period for the selected site, return to
Step 8. Otherwise, proceed to Step 13.
This step creates a loop through Steps 8 to 12 that is repeated for each year of the evaluation period for the selected site.
Step 13-Apply site-specific EB Method (if applicable).

Whether the site-specific EB Method is applicable is determined in Step 3 using criteria in Part C, Appendix A.2.1.
If it is not applicable, then proceed to Step 14.
If the site-specific EB Method is applicable, Step 6 EB Method criteria (detailed in Part C, Appendix A.2.4.) is used
to assign observed crashes to each individual site.
The site-specific EB Method combines the predictive model estimate of predicted average crash frequency, N'"""'d'
with the observed crash frequency of the specific site, N,b~N.,. This provides a more statistically reliable estimate of
the expected average crash frequency of the selected site.
In order to apply the site-specific EB Method, in addition to the material in Part C, Appendix A.2.4, the overdisper-
sion parameter, k, for the SPF is also used. The overdispersion parameter provides an indication of the statistical
reliability of the SPF. The closer the overdispersion parameter is to zero, the more statistically reliable the SPF. This
parameter is used in the site-specific EB Method to provide a weighting ..... toN pre" cte d and N o.,erve
"· d" Overdispersion pa-
rameters are provided for each SPF in the Part C chapters.
Apply the site-specific EB Method to a future time period, if appropriate.
The estimated expected average crash frequency obtained in this section applies to the time period in the past for
which the observed crash data were collected. Section A.2.6 in Appendix A to Part C provides a method to convert
the estimate of expected average crash frequency for a past time period to a future time period.
Step 14-lf there is another site to be evaluated, return to Step 7, otherwise, proceed to Step 15.
This step creates a loop for Steps 7 to 13 that is repeated for each roadwey segment or intersection within the study area.
Step 15-Apply the project level EB Method (if the site-specific EB Method is not applicable).
This step is applicable to existing conditions when observed crash data are available, but cannot be accurately
assigned to specific sites (e.g., the crash report may identify crashes as occurring between two intersections,
but is not accurate to determine a precise location on the segment). The EB Method is discussed in Section C.6.6.
Detailed description of the project-level EB Method is provided in Part C, Appendix A.2.5.
Step 16-Sum all sites and years in the study to estimate total crashes or average crash frequency
for the network
The total estimated number of crashes within the network or facility limits during the study period years is calcu-
lated using Equation C-2:
Ntotal = L all
Nrs + L all
Nint (C-2)
roadway intersections
segments
Where:
N,01 total expected number of crashes within the roadway limits of the study for all years in the period of
interest. Or, the sum of the expected average crash frequency for each year for each site within the
defined roadway limits within the study period;
N, expected average crash frequency for a roadway segment using the predictive method for one year; and
N1n1 expected average crash frequency for an intersection using the predictive method for one year.
Equation C-2 represents the total expected number of crashes estimated to occur during the study period. Equation
C-3 is used to estimate the total expected average crash frequency within the network or facility limits during the
study period.
N NtoW
(C-3)
total aveage = - -
n
Where:
Nlotal avcr:lge total expected average crash frequency estimated to occur within the defined roadway limits
during the study period; and
n number of years in the study period.
Regardless of whether the total or the total average is used, a consistent approach in the methods will produce
reliable comparisons.
Step 17-Determine if there is an alternative design, treatment, or forecast AADT to be evaluated.

Steps 3 through 16 of the predictive method are repeated, as appropriate, not only for the same roadway limits, but
also for alternative geometric design, treatments, or periods of interest or forecast AADTs.
Step 18-Evaluate and compare results.

The predictive method is used to provide a statistically reliable estimate of the expected average crash frequency
within defined network or facility limits over a given period of time for given geometric design and traffic con-
trol features and known or estimated AADT. The predictive method results may be used for a number of different
purposes. Methods for estimating the effectiveness of a project are presented in Section C.7. Part B of the HSM
includes a number of methods for effectiveness evaluation and network screening, many of which use of the predic-
tive method. Example uses include:
• Screening a network to rank sites and identify those sites likely to respond to a safety improvement;
• Evaluating the effectiveness of countermeasures after a period of implementation; and
• Estimating the effectiveness of proposed countermeasures on an existing facility.
C-6. PREDICTIVE METHOD CONCEPTS

The 18 steps of the predictive method are sununarized in Section C.5. Section C.6 provides additional explanation of
the some of the steps of the predictive method. Detail regarding the procedure for determining a calibration factor to
apply in Step II is provided in the Part C, Appendix A. I. Detail regarding the EB Method, which is required in Steps
6, 13, and 15, is provided in the Part C, AppendixA.2.
C-6.1. Roadway Limits and Facility Types

In Step I of the predictive method, the extent or limits of the roadway network under consideration are defined and
the facility type or types within those limits is determined. Part C provides three facility types: Rural Two-Lane,
Two-Way Roads, Rural Multilane Highways, and Urban and Suburban Arterials. In Step 5 of the predictive method,
the roadway within the defined roadway limits is divided into individual sites, which are either homogenous roadway
segments or intersections. A facility consists of a contiguous set of individual intersections and roadway segments,
referred to as "sites." A roadway network consists of a number of contiguous facilities.
Classifying an area as urban, suburban, or rural is subject to the roadway characteristics, surrounding population,
and land uses, and is at the user's discretion. In the HSM, the definition of ''urban" and "rural" areas is based on
Federal Highway Administration (FHWA) guidelines which classify "urban" areas as places inside urban boundaries
where the population is greater than 5,000 persons. "Rural" areas are defined as places outside mban areas where the
population is less than 5,000. The HSM uses the term "suburban" to refer to outlying portions of an urban area; the
predictive method does not distinguish between urban and suburban portions of a developed area.
PART C-INTRODUCTION AND APPLICATIONS GUIDANCE C-13
For each facility type, SPFs and CMFs for specific individual site types (i.e., intersections and roadway segments)
are provided. The predictive method is used to determine the expected average crash frequency for each individual
site in the study for all years in the period of interest, and the overall crash estimation is the cumulative sum of all
sites for all years.
The facility types and facility site types in the HSM Part C are defined below. Table C-1 summarizes the site
types for each of the facility types that are included in each of the Part C chapters:
• Chapter 10-Rural Two-Lane, Two-Way Roads-includes all rural highways with two-lanes and two-way
traffic operation. Chapter 10 also addresses two-lane, two-way highways with center two-way left-turn lanes
and two-lane highways with added passing or climbing lanes or with short segments of four-lane cross-sec-
tions (up to two miles in length) where the added lanes in each direction are provided specifically to enhance
passing opportunities. Short lengths of highway with four-lane cross-sections essentially function as two-
lane highways with side-by-side passing lanes and, therefore, are within the scope of the two-lane, two-way
highway methodology. Rural highways with longer sections of four-lane cross-sections can be addressed with
the rural multilane highway procedures in Chapter II. Chapter 10 includes three- and four-leg intersections
with minor-road stop control and four-leg signalized intersections on all the roadway cross-sections to which
the chapter applies.
• Chapter 11-Rural Multilane Highways-includes rural multilane highways without full access control. This
includes all rural nonfreeways with four through travel lanes, except for two-lane highways with side-by-side
passing lanes, as described above. Chapter II includes three- and four-leg intersections with minor-road stop
control and four-leg signalized intersections on all the roadway cross-sections to which the chapter applies.
• Chapter 12-Urban and Suburban Arterial Highways-includes arterials without full access control, other
than freeways, with two or four through lanes in urban and suburban areas. Chapter 12 includes three- and
four-leg intersections with minor-road stop control or traffic signal control and roundabouts on all of the
roadway cross-sections to which the chapter applies.
C.6.2. Definition of Roadway Segments and Intersections

The predictive models for roadway segments estimate the frequency of crashes that would occur on the road-
way if no intersection were present. The predictive models for an intersection estimate the frequency of addi-
tional crashes that occur because of the presence of the intersection.
A roadway segment is a section of continuous traveled way that provides two-way operation of traffic, that
is not interrupted by an intersection, and consists of homogenous geometric and traffic control features. A
roadway segment begins at the center of an intersection and ends at either the center of the next intersection,
or where there is a change from one homogeneous roadway segment to another homogenous segment. The
roadway segment model estimates the frequency of roadway segment related crashes which occur in Region B
in Figure C-3. When a roadway segments begins or ends at an intersection, the length of the roadway segment
is measured from the center of the intersection.
Intersections are defined as the junction of two or more roadway segments. The intersection models estimate
the predicted average frequency of crashes that occur within the limits of an intersection (Region A of Figure
C-3) and intersection-related crashes that occur on the intersection legs (Region B in Figure C-3).
When the EB Method is applicable at the site-specific level (see Section C.6.6), observed crashes are assigned
to individual sites. Some observed crashes that occur at intersections may have characteristics of roadway seg-
ment crashes and some roadway segment crashes may be attributed to intersections. These crashes are individu-
ally assigned to the appropriate site. The method for assigning and classifying crashes as individual roadway
segment crashes and intersection crashes for use with the EB Method is described in Part C, Appendix A.2.3.
In Figure C-3, all observed crashes that occur in Region A are assigned as intersection crashes, but crashes that
occur in Region B may be assigned as either roadway segment crashes or intersection crashes depending on the
characteristics of the crash.
Using these definitions, the roadway segment predictive models estimate the frequency of crashes that would
occur on the roadway if no intersection were present. The intersection predictive models estimate the frequency
of additional crashes that occur because of the presence of the intersection.
Segment Length
(center of intersection to center of intersection)
A All crashes that occur within this region are classified as intersection crashes.
B Crashes in this reg1on may be segment or intersection related depending on

the characteristics of the crash.
Figure C-3. Definition of Roadway Segments and Intersections
SPFs are regression models for estimating the predicted average crash frequency of individual roadway segments or
intersections. In Step 9 of the predictive method, the appropriate SPFs are used to determine the predicted average
crash frequency for the selected year for specific base conditions. Each SPF in the predictive method was devel-
oped with observed crash data for a set of similar sites. In the SPFs developed for the HSM, the dependent variable
estimated is the predicted average crash frequency for a roadway segment or intersection under base conditions and
the independent variables are the AADTs of the roadway segment or intersection legs (and, in some cases a few ad-
ditional variables such as the length of the roadway segment).
An example of an SPF (for rural two-way two-lane roadway segments from Chapter 10) is shown in Equation C-4.
N ~ (AADT) X (L) X (365) X I o<-6) X e<-O 4865) (C-4)

sp[r.<
Where:
predicted average crash frequency estimated for base conditions using a statistical regression model;
AADT annual average daily traffic volume (vehicles/day) on roadway segment; and
L length of roadway segment (miles).
SPFs are developed through statistical multiple regression techniques using historic crash data co!lected over a
number of years at sites with similar characteristics and covering a wide range of AADTs. The regression parameters
of the SPFs are determined by assuming that crash frequencies follow a negative binomial distribution. The negative
binomial distribution is an extension of the Poisson distribution which is typically used for crash frequencies. How·
ever, the mean and the variance of the Poisson distribution are equal. This is often not the case for crash frequencies
where the variance typically exceeds the mean.
The negative binomial distribution incorporates an additional statistical parameter, the overdispersion parameter that
is estimated along with the parameters of the regression equation. The overdispersion parameter has p~;5itive values.
The greater the overdispersion parameter, the more that crash data vary as compared to a Poisson distribution with
the same mean. The overdispersion parameter is used to determine a weighted adjustment factor for use in the EB
Method described in Section C.6.6.
Crash modification factors (CMFs) are applied to the SPF estimate to account for geometric or geographic differ-
ences between the base conditions of the model and local conditions of the site under consideration. CMFs and their
application to SPFs are described in Section C.6.4.
In order to apply an SPF, the following information relating to the site under consideration is necessary:
• Basic geometric design and geographic information of the site to determine the facility type and whether an SPF
is available for that site type;
• AADT information for estimation of past periods, or forecast estimates of AADT for estimation of future periods; and
• Detailed geometric design of the site and base conditions (detailed in each of the Part C chapters) to determine
whether the site conditions vary from the base conditions and therefore a CMF is applicable.
Updating Default Values of Crash Severity and Collision Type Distribution for Local Conditions
In addition to estimating the predicted average crash frequency for all crashes, SPFs can be used to estimate the
distribution of crash frequency by crash severity types and by collision types (such as single-vehicle or driveway
crashes). The distribution models in the HSM are default distributions.
Where sufficient and appropriate local data are available, the default values (for crash severity types and collision
types and the proportion of night-time crashes) can be replaced with locally derived values when it is explicitly
stated in Chapters 10, 11, and 12. Calibration of default distributions to local conditions is described in detail in the
Part C, Appendix A.l.l.
Development of Local SPFs

Some HSM users may prefer to develop SPFs with data from their own jurisdiction for use with the predictive meth-
od rather than calibrating the SPFs presented in the HSM. Appendix A to Part C provides guidance on developing
jurisdiction-specific SPFs that are suitable for use with the predictive method. Development of jurisdiction-specific
SPFs is not required.
C.6.3. Crash Modification Factors (CMFs)

In Step 10 of the predictive method, CMFs are determined and applied to the results of Step 9. The CMFs are used
in Part C to adjust the predicted average crash frequency estimated by the SPF for a site with base conditions to the
predicted average crash frequency for the specific conditions of the selected site.
CMFs are the ratio of the estimated average crash frequency of a site under two different conditions. Therefore, a
CMF represents the relative change in estimated average crash frequency due to a change in one specific condition
(when all other conditions and site characteristics remain constant).
Equation C-5 shows the calculation of a CMF for the change in estimated average crash frequency from site condi-
tion 'a' to site condition 'b'.
CMF = estimated average crash frequency with condition "b"

(C-5)
estimated average crash frequency with condition "a"
CMFs defined in this way for expected crashes can also be applied to the comparison of predicted crashes between
site condition 'a' and site condition 'b'.
CMFs are an estimate of the effectiveness of the implementation of a particular treatment, also known as a coun-
termeasure, intervention, action, or alternative design. Examples include: illwninating an unlighted road segment,
paving gravel shoulders, signalizing a stop-controlled intersection, increasing the radius of a h01jzontal curve, or
choosing a signal cycle time of 70 seconds instead of 80 seconds. CMFs have also been developed for conditions
!'1"
that are not associated with the roadway, but represent geographic conditions surrounding the site or demographic
conditions with users of the site. For example, the number ofliquor outlets in proximity to a site.
The values ofCMFs in the HSM are determined for a specified set of base conditions. These base conditions serve
the role of site condition 'a' in Equation C-5. This allows comparison of treatment options against a specified refer-
ence condition. For example, CMF values for the effect oflane width changes are determined in comparison to a
base condition of 12-ft lane width. Under the base conditions (i.e., with no change in the conditions), the value of
a CMF is 1.00. CMF values less than 1.00 indicate the alternative treatment reduces the estimated average crash
frequency in comparison to the base condition. CMF values greater than 1.00 indicate the alternative treatment in-
creases the estimated crash frequency in comparison to the base condition. The relationship between a CMF and the
expected percent change in crash frequency is shown in Equation C-6.
Percent Reduction in Accidents= 100% x (1.00- CMF) (C-6)
For example,
• If a CMF = 0.90 then the expected percent change is 100% x (I - 0.90) = 10%, indicating a 10% change in esti-
mated average crash frequency.
• If a CMF = 1.20 then the expected percent change is l 00% x (I - 1.20) = -20%, indicating a -20% change in
estimated average crash frequency.
Application of CMFs to Adjust Crash Frequencies for Specific Site Conditions

In the Part C predictive models, an SPF estimate is multiplied by a series of CMFs to adjust the estimate of
average crash frequency from the base conditions to the specific conditions present at that site (see, for example,
Equation C-1). The CMFs are multiplicative because the most reasonable assumption based on current knowledge
is to assume independence of the effects of the features they represent. Little research exists regarding the inde-
pendence of these effects. The use of observed crash data in the EB Method (see Section C.6.6 and Appendix A to
Part C) can help to compensate for any bias which may be caused by lack of independence of the CMFs. As new
research is completed, future HSM editions may be able to address the independence (or lack thereof) of CMF
effects more fully.
Application of CMFs in Estimating the Effect on Crash Frequencies of Proposed Treatments or Countermeasures
CMFs are also used in estimating the anticipated effects of proposed future treatments or countermeasures (e.g., in
some of the methods discussed in Section C.7). Where multiple treatments or countermeasures will be applied con-
currently and are presumed to have independent effects, the CMFs for the combined treatments are multiplicative.
As discussed above, limited research exists regarding the independence of the effects of individual treatments from
one another. However, in the case of proposed treatments that have not yet been implemented, there are no observed
crash data for the future condition to provide any compensation for overestimating forecast effectiveness of multiple
treatments. Thus, engineering judgment is required to assess the interrelationships and independence for multiple
treatments at a site.
The limited understanding of interrelationships among various treatments requires consideration, especially when
several CMFs are being multiplied. It is possible to overestimate the combined effect of multiple treatments when
it is expected that more than one of the treatments may affect the same type of crash. The implementation of wider
lanes and shoulders along a corridor is an example of a combined treatment where the independence of the individ-
ual treatments is unclear because both treatments are expected to reduce the same crash types. \Vhen implementing
potentially interdependent treatments, users should exercise engineering judgment to assess the interrelationship and/
or independence of individual elements or treatments being considered for implementation within the same project.
These assumptions may or may not be met by multiplying the CMFs under consideration together with either an SPF
or with observed crash frequency of an existing site.
Engineering judgment is also necessary in the use of combined CMFs where multiple treatments change the over-
all nature or character of the site. In this case, certain CMFs used in the analysis of the existing site conditions and
the proposed trea1ment may not be compatible. An example of this concern is the installation of a roundabout at an
urban two-way, stop-controlled or signalized intersection. Since an SPF for roundabouts is currently unavailable,
the procedure for estimating the crash frequency after installing a roundabout (see Chapter 12) is to first estimate
the average crash frequency for the existing site conditions and then apply a CMF for conversion of a conventional
intersection to a roundabout. Clearly, installing a roundabout changes the nature of the site so that other CMFs which
may be applied to address other conditions at the two-way, stop-controlled location may no longer be relevant.
CMFs and Standard Error

Standard error is defined as the estimated standard deviation of the difference between estimated values and values
from sample data. It is a method of evaluating the error of an estimated value or model. The smaller the standard er-
ror, the more reliable (less error) the estimate. All CMF values are estimates of the change in expected average crash
frequency due to a change in one specific condition plus or minus a standard error. Some CMFs in the HSM include
a standard error value, indicating the variability of the CMF estimation in relation to sample data values.
Standard error can also be used to calculate a confidence interval for the estimated change in expected average crash
frequency. Confidence intervals can be calculated using multiples of standard error using Equation C-7 and values
from Table C-2.
CI(XO/o) ~ CMF ±(SEx MSE) (C-7)
Where:
CI(X%) confidence interval, or range of estimate values within which it is XO/o probable the true value will occur;
CMF crash modification factor;
SE standard error of the CMF; and
MSE multiple of standard error.
Table C-2. Constructing Confidence Intervals Using CMF Standard Error

Confidence Interval
(probability that the true value is within Multiple of Standard Error (1\tiSE)
Desired Level of Confidence the estimated intervals) to Use in Equation C~7
Low 65-70%
Medium 95% 2
High 99.9% 3
CMFs in the HSM Part C

CMF values in the HSM are either explained in the text (typically where there are a limited range of options for a
particular trea1ment), in a formula (where trea1ment options are continuous variables) or in tables (where the CMF
values vary by facility type or are in discrete categories). The differences between CMFs in Part C and D CMFs
are explained below.
Part D contains all CMFs in the HSM. Some Part D CMFs are included in Part C for use with specific SPFs. Oth-
er Part D CMF s are not presented in Part C but can be used in the methods to estimate change in crash frequency
described in Section C. 7.
r C-18 HIGHWAY SAFETY MANUAL
C-6.4. Calibration of Safety Performance Functions to Local Conditions

The predictive models in Chapters 10, 11, and 12 have three basic elements: safety performance functions, crash
modification factors, and a calibration factor. The SPFs were developed as part ofHSM-related research from the
most complete and consistent available data sets. However, the general level of crash frequencies may vary sub-
stantially from one jurisdiction to another for a variety of reasons including crash reporting thresholds and crash
reporting system procedures. These variations may result in some jurisdictions experiencing substantially more
reported traffic crashes on a particular facility type than in other jurisdictions. In addition, some jurisdictions may
have substantial variations in conditions between areas within the jurisdiction (e.g., snowy winter driving condi-
tions in one part of the state and only wet winter driving conditions in another part of the state). Therefore, for the
predictive method to provide results that are reliable for each jurisdiction that uses them, it is important that the
SPFs in Part C be calibrated for application in each jurisdiction. Methods for calculating calibration factors for
roadway segments, C,, and intersections, C;o are included in the Part C, Appendix A to allow highway agencies to
adjust the SPF to match local conditions.
The calibration factors will have values greater than 1.0 for roadways that, on average, experience more crashes
than the roadways used in developing the SPFs. Roadways that, on average, experience fewer crashes than the
roadways used in the development of the SPF, will have calibration factors less than 1.0.
C-6.5. Weighting Using the Empirical Bayes Method

Step 13 or Step 15 of the predictive method are optional steps that are applicable only when observed crash
data are available for either the specific site or the entire facility of interest. Where observed crash data and a
predictive model are available, the reliability of the estimation is improved by combining both estimates. The
predictive method in Part C uses the Empirical Bayes method, herein referred to as the EB Method.
The EB Method can be used to estimate expected average crash frequency for past and future periods and used
at either the site-specific level or the project-specific level (where observed data may be known for a particular
facility, but not at the site-specific level).
For an individual site (i.e., the site-specific EB Method) the EB Method combines the observed crash frequency
with the predictive model estimate using Equation C-8. The EB Method uses a weighted factor, w, which is a
function of the SPFs overdispersion parameter, k, to combine the two estimates. The weighted adjustment is
therefore dependant only on the variance of the SPF model. The weighted adjustment factor, w, is calculated
using Equation C-9.
N expected ~ . + (1.00- w) x N observed

w x N predicted (C-8)
w= - - - ; - - - - - - ,
(C-9)
l+kx[ L
all study
Npredicted]
y=s
Where:
estimate of expected average crash frequency for the study period;
Npredicred
predictive model estimate of predicted average crash frequency for the study period;
Nobscrvcd
observed crash frequency at the site over the study period;
w weighted adjustment to be placed on the SPF prediction; and
k overdispersion parameter from the associated SPF.

As the value of the overdispersion parameter increases, the value of the weighted adjustment factor decreases,
and thus more emphasis is placed on the observed rather than the SPF predicted crash frequency. When the
data used to develop a model are greatly dispersed, the precision of the resulting SPF is likely to be lower; in
this case, it is reasonable to place less weight on the SPF estimation and more weight on the observed crash
frequency. On the other hand, when the data used to develop a model have little overdispersion, the reliability
of the resulting SPF is likely to be higher; in this case, it is reasonable to place more weight on the SPF estima-
tion and less weight on the observed crash frequency. A more detailed discussion of the EB Method is included
in Appendix A to Part C.
The EB Method cannot be applied without an applicable SPF and observed crash data. There may be circum-
stances where an SPF may not be available or cannot be calibrated to local conditions or circumstances where
crash data are not available or applicable to current conditions. If the EB Method is not applicable, Steps 6, 13,
and 15 are not conducted.
C.7. METHODS FOR ESTIMATING THE SAFETY EFFECTIVENESS OF A PROPOSED PROJECT

The Part C predictive method provides a structured methodology to estimate the expected average crash frequency
where geometric design and traffic control features are specified. There are four methods for estimating the change
in expected average crash frequency of a proposed project or project design alternative (i.e., the effectiveness of the
proposed changes in terms of crash reduction). In order of predictive reliability (high to low) these are:
• Method 1-Apply the Part C predictive method to estimate the expected average crash frequency ofboth the exist-
ing and proposed conditions.
• Method 2-Apply the Part C predictive method to estimate the expected average crash frequency of the existing
condition and apply an appropriate project CMF from Part D (i.e., a CMF that represents a project which changes
the character of a site) to estimate the safety performance of the proposed condition.
• Method 3-!f the Part C predictive method is not available, but a Safety Performance Function (SPF) applicable
to the existing roadway condition is available (i.e., an SPF developed for a facility type that is not included in Part
C of the HSM), use that SPF to estimate the expected average crash frequency of the existing condition. Apply an
appropriate project CMF from Part D to estimate the expected average crash frequency of the proposed condition.
A locally-derived project CMF can also be used in Method 3.
• Method 4-Use observed crash frequency to estimate the expected average crash frequency of the existing condi-
tion and apply an appropriate project CMF from Part D to the estimated expected average crash frequency of the
existing condition to obtain the estimated expected average crash frequency for the proposed condition.
In all four of the above methods, the difference in estimated expected average crash frequency between the existing
and proposed conditions/projects is used as the project effectiveness estimate.
C.S. LlMlTATlONS OF THE HSM PREDICTIVE METHOD

The predictive method is based on research using available data describing geometric and traffic characteristics
of road systems in the United States. The predictive models incorporate the effects of many, but not all, geometric
designs and traffic control features of potential interest. The absence of a factor from the predictive models does not
necessarily mean that the factor has no effect on crash frequency; it may merely indicate that the effect is not fully
known or has not been quantified at this time.
While the predictive method addresses the effects of physical characteristics of a facility, it considers effect of non-
geometric factors only in a general sense. Primary examples of this limitation are:
• Driver populations vary substantially from site to site in age distribution, years of driving experience, seat belt us-
age, alcohol usage, and other behavioral factors. The predictive method accounts for the statewide or community-
wide influence of these factors on crash frequencies through calibration, but not site-specific variations in these
factors, which may be substantial.
• 1'"
r C-20 HIGHWAY SAFETY MANUAL
• The effects of climate conditions may be addressed indirectly through the calibration process, but the effects of
weather are not explicitly addressed.
• The predictive method considers annual average daily traffic volumes, but does not consider the effects of traffic
volume variations during the day or the proportions of trucks or motorcycles; the effects of these traffic factors are
not fully understood.
Furthermore, the predictive method treats the effects of individual geometric design and traffic control features as
independent of one another and ignores potential interactions between them. It is likely that such interactions exist,
and ideally, they should be accounted for in the predictive models. At present, such interactions are not fully under-
stood and are difficult to quantify.
C.9. GUIDE TO APPLYING PART C

The HSM provides a predictive method for crash estimation which can be used for the purposes of making decisions
relating to designing, planning, operating, and maintaining roadway networks.
These methods focus on the use of statistical methods in order to address the inherent randomness in crashes. The
use of the HSM requires an understanding of the following general principles:
• Observed crash frequency is an inherently random variable. It is not possible to precisely predict the value for a
specific one year period-the estimates in the HSM refer to the expected average crash frequency that would be
observed if the site could be maintained under consistent conditions for a long-term period, which is rarely possible.
• Calibration of an SPF to local state conditions is an important step in the predictive method.
• Engineering judgment is required in the use of all HSM procedures and methods, particularly selection and
application of SPFs and CMFs to a given site condition.
• Errors and limitations exist in all crash data which affects both the observed crash data for a specific site, and also
the models developed. Chapter 3 provides additional explanation on this subject.
• Development of SPFs and CMFs requires understanding of statistical regression modeling and crash analysis
techniques. Appendix A to Part C provides guidance on developing jurisdiction-specific SPFs that are suitable
for use with the predictive method. Development of jurisdiction-specific SPFs is not required.
• In general, a new roadway segment is applicable when there is a change in the condition of a roadway segment
that requires application of a new or different CMF value, but where a value changes frequently within a minimum
segment length, engineering judgment is required to determine an appropriate average value across the minimum
segment length. When dividing roadway facilities into small homogenous roadway segments, limiting the segment
length to greater than or equal to 0.10 miles will decrease data collection and management efforts.
• Where the EB Method is applied, a minimum of two years of observed data is recommended. The use of observed
data is only applicable if geometric design and AADTs are known during the period for which observed data are
available.
C.10. SUMMARY
The predictive method consists of 18 steps which provide detailed guidance for dividing a facility into individual
sites, selecting an appropriate period of interest, obtaining appropriate geometric data, traffic volume data, and
observed crash data, and applying the predictive models and the EB Method. By following the predictive method
steps, the expected average crash frequency of a facility can be estimated for a given geometric design, traffic
volumes, and period of time. This allows comparison to be made between alternatives in design and traffic volume
forecast scenarios. The HSM predictive method allows the estimate to be made between crash frequency and
treatment effectiveness to be considered along with community needs, capacity, delay, cost, right-of-way and
environmental considerations in decision making for highway improvement projects.
The predictive method can be applied to either a past or a future period of time and used to estimate total expected
average crash frequency or crash frequencies by crash severity and collision type. The estimate may be for an exist-
ing facility, for proposed design alternatives for an existing facility, or for a new (unconstructed) facility. Predictive
models are used to determine the predicted average crash frequencies based on site conditions and traffic volumes.
The predictive models in the HSM consist of three basic elements: safety performance functions, crash modifica-
tion factors, and a calibration factor. These are applied in Steps 9, 10, and 11 of the predictive method to determine
the predicted average crash frequency of a specific individual intersection or homogenous roadway segment for a
specific year.
Where observed crash data are available, observed crash frequencies are combined with the predictive model es-
timates using the EB Method to obtain a statistically reliable estimate. The EB Method may be applied in Step 13
or 15 of the predictive method. The EB Method can be applied at the site-specific level (Step 13) or at the project-
specific level (Step 15). It may also be applied to a future time period if site conditions will not change in the future
period. The EB Method is described in the Part C, Appendix A.2.
The following chapters in Part C provide the detailed predictive method steps for estimating expected average crash
frequency for the following facility types:
• Chapter 10--Rural Two-Lane, Two-Way Roads
Chapter 10-Predictive Method for
Rural Two-Lane, Two-Way Roads
10.11NTRODUCTION
This chapter presents the predictive method for rural two-lane, two-way roads. A general introduction to the Highway
Safety Manual (HSM) predictive method is provided in the Part C-Introduction and Applications Guidance.
The predictive method for rural two-lane, two-way roads provides a structured methodology to estimate the
expected average crash frequency, crash severity, and collision types for a rural two-lane, two-way facility with
known characteristics. All types of crashes involving vehicles of all types, bicycles, and pedestrians are included,
with the exception of crashes between bicycles and pedestrians. The predictive method can be applied to existing
sites, design alternatives to existing sites, new sites, or for alternative traffic volume projections. An estimate can be
made for crash frequency of a prior time period (i.e., what did or would have occurred) or in the future (i.e., what is
expected to occur). The development of the predictive method in Chapter 10 is documented by Harwood et al. (5).
This chapter presents the following information about the predictive method for rural two-lane, two-way roads:
• A concise overview of the predictive method.
• The definitions of the facility types included in Chapter 10 and site types for which predictive models have been
developed for Chapter 10.
• The steps of the predictive method in graphical and descriptive forms.
• Details for dividing a rural two-lane, two-way facility into individual sites consisting of intersections and
roadway segments.
• Safety performance functions (SPFs) for rural two-lane, two-way roads.
• Crash modification factors (CMFs) applicable to the SPFs in Chapter 10.
• Guidance for applying the Chapter 10 predictive method and limitations of the predictive method specific to
Chapter 10.
• Sample problems illustrating the Chapter 10 predictive method for rural two-lane, two-way roads.
10.2. OVERVIEW OF THE PREDICTIVE METHOD

The predictive method provides an 18-step procedure to estimate the "expected average crash frequency," N expecte,
(by total crashes, crash severity, or collision type), of a roadway network, facility, or site. In the predictive method,
the roadway is divided into individual sites which are homogenous roadway segments and intersections. A facility
consists of a contiguous set of individual intersections and roadway segments referred to as "sites." Different facility
types are determined by surrounding land use, roadway cross-section, and degree of access. For each facility type,
10-1
.l:..•
10-2 HIGHWAY SAFETY MANUAL
a number of different site types may exist, such as divided and undivided roadway segments and signalized and
unsignalized intersections. A roadway network consists of a number of contiguous facilities.
The method is used to estimate the expected average crash frequency of an individual site, with the cumulative sum
of all sites used as the estimate for an entire facility or network. The estimate is for a given time period of interest
(in years) during which the geometric design and traffic control features are unchanged and traffic volumes are
known or forecasted. The estimate relies on estimates made using predictive models which are combined with
observed crash data using the Empirical Bayes (EB) Method.
The predictive models used within the Chapter 10 predictive method are described in detail in Section 10.3.
The predictive models used in Chapter 10 to determine the predicted average crash frequency, N '"""''' are of the
general form shown in Equation I 0-1.
Npredicted = N..,pfx X (CMFh X CMF2x X ••• X CMF) X ex (10-1)
Where:
Npredicted = predicted average crash frequency for a specific year for site type x;
N,Pf' = predicted average crash frequency determined for base conditions of the SPF developed for site type x;
CMF~> = crash modification factors specific to site type x and specific geometric design and traffic control features
y; and
C, = calibration factor to adjust SPF for local conditions for site type x.
10.3. RURAL TWO-LANE, TWO-WAY ROAD5-DEFINITIONS AND PREDICTIVE MODELS IN CHAPTER 10

This section provides the definitions of the facility and site types, and the predictive models for each of the site types
included in Chapter 10. These predictive models are applied following the steps of the predictive method presented
in Section I 0.4.
10.3.1. Definition of Chapter 10 Facility and Site Types

The predictive method in Chapter 10 addresses all types of rural two-lane, two-way highway facilities, including
rural two-lane, two-way highways with center two-way left-tum lanes or added passing lanes, and rural two-lane,
two-way highways containing short sections of rural four-lane highway that serve exclusively to increase passing
opportunities (i.e., side-by-side passing lanes). Facilities with four or more lanes are not covered in Chapter 10.
The terms "highway" and "road" are used interchangeably in this chapter and apply to all rural two-lane, two-way
facilities independent of official state or local highway designation.
Classifying an area as urban, suburban, or rural is subject to the roadway characteristics, surrounding population and
land uses and is at the user's discretion. In the HSM, the definition of "urban" and "rural" areas is based on Federal
Highway Administration (FHWA) guidelines which classify "urban" areas as places inside urban boundaries where
the population is greater than 5,000 persons. "Rural" areas are defined as places outside urban areas which have a
population less than 5,000 persons. The HSM uses the term "suburban" to refer to outlying portions of an urban
area; the predictive method does not distinguish between urban and suburban portions of a developed area.
Table 10-1 identifies the site types on rural two-lane, two-way roads for which SPFs have been developed for
predicting average crash frequency, severity, and collision type.
CHAPTER 1Q-PREDICTIVE METHOD FOR RURAL TWO-LANE, TWO-WAY ROADS 10-3
Table 10-1. Rural Two-Lane, Two-Way Road Site Type with SPFs in Chapter 10
Site Type Site Types with SPFs in Chapter 10
Roadway Segments Undivided rural two-lane. two-way roadway segments (2U)

Unsignalized three-leg (stop control on minor-road approaches) (3ST)
Intersections Unsignalized four-leg (stop control on minor-road approaches) (4ST)
Signalized four-leg ( 4SG)
These specific site types are defined as follows:

• Undivided roadway segment (2U)-a roadway consisting of two lanes with a continuous cross-section providing
two directions of travel in which the lanes are not physically separated by either distance or a barrier. In addition,
the definition includes a section with three lanes where the center lane is a two-way left-tum lane (TWLTL) or a
section with added lanes in one or both directions of travel to provide increased passing opportunities (e.g., pass-
ing lanes, climbing lanes, and short four-lane sections).
• Three-leg intersection with stop control (3ST)-an intersection of a rural two-lane, two-way road and a minor road.
A stop sign is provided on the minor road approach to the intersection only.
• Four-leg intersection with stop control (4ST)-an intersection of a rural two-lane, two-way road and two minor
roads. A stop sign is provided on both minor road approaches to the intersection.
• Four-leg signalized intersection (4SG)-an intersection of a rural two-lane, two-way road and two other rural two·
lane, two-way roads. Signalized control is provided at the intersection by traffic lights.
10.3.2. Predictive Models for Rural Two-Lane, Two-Way Roadway Segments

The predictive models can be used to estimate total predicted average crash frequency (i.e., all crash severities and
collision types) or can be used to predict average crash frequency of specific crash severity types or specific collision
types. The predictive model for an individual roadway segment or intersection combines a SPF with CMFs and a
calibration factor.
For rural two-lane, two-way undivided roadway segments the predictive model is shown in Equation I 0-2:
(10-2)
Where:
Npredicted rs ~ predicted average crash frequency for an individual roadway segment for a specific year;
~ predicted average crash frequency for base conditions for an individual roadway segment;
c, calibration factor for roadway segments of a specific type developed for a particular jurisdiction
or geographical area; and
CMF1, ••• CMF12, ~ crash modification factors for rural two-lane, two-way roadway segments.
This model estimates the predicted average crash frequency of non-intersection related crashes (i.e., crashes that
would occur regardless of the presence of an intersection).
10.3.3. Predictive Models for Rural Two-Lane, Two-Way Intersections

The predictive models for intersections estimate the predicted average crash frequency of crashes occurring within
the limits of an intersection (i.e., at-intersection crashes) and crashes that occur on the intersection legs and are
attributed to the presence of an intersection (i.e., intersection-related crashes).
For all intersection types in Chapter I 0 the predictive model is shown in Equation I 0-3:
(10-3)
Where:
= predicted average crash frequency for an individual intersection for the selected year;
= predicted average crash frequency for an intersection with base conditions;
CMF11 ••• CMF41 = crash modification factors for intersections; and
c, = calibration factor for intersections of a specific type developed for use for a particular
jurisdiction or geographical area.
The SPFs for rural two-lane, two-way roads are presented in Section 10.6. The associated CMFs for each of the SPFs are
presented in Section 10.7 and summarized in Table I 0-7. Only the specific CMFs associated with each SPF are applicable
to that SPF (as these CMFs have base conditions which are identical to the base conditions of the SPF). The calibration
factors, C, and C,, are detennined in the Part C, AppendixA.l.l. Due to continual change in the crash frequency and sever-
ity distributions with time, the value of the calibration factors may change for the selected year of the study period.
10.4. PREDIGIVE METHOD FOR RURAL TWO-LANE, TWO-WAY ROADS

The predictive method for rural two-lane, two-way road is shown in Figure I 0-1. Applying the predictive method
yields an estimate of the expected average crash frequency (and/or crash severity and collision types) for a rural
two-lane, two-way facility. The components of the predictive models in Chapter 10 are determined and applied in
Steps 9, 10, and II of the predictive method. The information that is needed to apply each step is provided in the
following sections and in the Part C, Appendix A.
There are 18 steps in the predictive method. In some situations, certain steps will not be needed because the data is
not available or the step is not applicable to the situation at hand. In other situations, steps may be repeated, such as
if an estimate is desired for several sites or for a period of several years. In addition, the predictive method can be
repeated as necessary to undertake crash estimation for each alternative design, traffic volume scenario, or proposed
treatment option within the same period to allow for comparison.
The following explains the details of each step of the method as applied to two-lane, two-way rural roads.
Step 1-Define the limits of the roadway and facility types in the study network, facility, or site for which the
expected average crash frequency, severity, and collision types are to be estimated.
The predictive method can be undertaken for a roadway network, a facility, or an individual site. A site is either an
intersection or a homogeneous roadway segment. There are a number of different types of sites, such as signalized
and unsignalized intersections. The definitions of a rural two-lane, two-way road, an intersection, and a roadway
segment, along with the site types for which SPFs are included in Chapter 10, are provided in Section 10.3.
The predictive method can be applied to an existing roadway, a design alternative for an existing roadway, or a design
alternative for new roadway (which may be either unconstructed or yet to experience enough traffic to have observed
crash data).
The limits of the roadway of interest will depend on the nature of the study. The study may be limited to only one specific
site or a group of contiguous sites. Alternatively, the predictive method can be applied to a long corridor for the purposes
of network screening (detennining which sites require upgrading to reduce crashes) which is discussed in Chapter 4.
CHAPTER 10-PREDICTIVE METHOD FOR RURAL TWO-LANE, TWO-WAY ROADS 10-5
Step 1 Define roadway limits and facility type.
Step 2 Define the period of study.
Determine AADT and availability of crash data

Step 3
for every year 1n the period of interest.
Step 4 Determine geometnc conditions.
D1vide roadway into individual

Step 5
roadway segments and intersections.
Step 6 Ass1gn observed crashes to indiv1dual sites (if applicable).
Step 7 Select a roadway segment or intersection.
Step 8 Select first or next year of the evaluat1on penod.
Step 10 Apply CMFs.
Step 11 Apply a calibration factor.
YES
Step 12
Step 13 Apply site-specific EB method (1f applicable).
YES
Step 14
Step 15 Apply project-level EB method (if applicable).
Step 16 Sum all sites and years.
Is there
an alternative
YES
Step 17 design, treatment, or
forecast AADT to be
evaluated?
Step 18 Compare and evaluate results.
Figure 10-1. The HSM Predictive Method
-~- .

The predictive method can be undertaken for either a past or future period measured in years. Years of interest will be
determined by the availability of observed or forecast average aruma! daily traffic (AADT) volwnes, observed crash data,
and geometric design data. Whether the predictive method is used for a past or future period depends upon the purpose of
the study. The period of study may be:
• A past period (based on observedAADTs) for:
• An existing roadway network, facility, or site. If observed crash data are available, the period of study is the pe-
riod of time for which the observed crash data are available and for which (during that period) the site geometric
design features, traffic control features, and traffic volumes are known.
• An existing roadway network, facility, or site for which alternative geometric design features or traffic control
features are proposed (for near term conditions).
z An existing roadway network, facility, or site for which alternative geometric design or traffic control features
are proposed for implementation in the future.
o A new roadway network, facility, or site that does not currently exist, but is proposed for construction during
some future period.
Step 3-For the stndy period, determine the availability of annual average daily traffic volumes and, for an existing
roadway network, the availability of observed crash data to determine whether the EB Method is applicable.
The SPFs used in Step 9 (and some CMFs in Step 10), include AADT volumes (vehicles per day) as a variable. For
a past period, the AADT may be determined by automated recording or estimated from a sample survey. For a future
period the AADT may be a forecast estimate based on appropriate land use planning and traffic volume forecasting
models, or based on the assumption that current traffic volumes will remain relatively constant
For each roadway segment, the AADT is the average daily two-way, 24-hour traffic volume on that roadway segment
in each year of the evaluation period selected in Step 8.
For each intersection, two values are required in each predictive modeL These are the AADT of the major street,
AADTmoJ' and the two-way AADT of the minor street, AADT.,".
In Chapter I 0, AADT moJ and AADT•'" are determined as follows. If the AADTs on the two major road legs of an inter-
section differ, the larger of the two AADT values is used for the intersection. For a three-leg intersection, the minor road
AADT is the AADT of the single minor road leg. For a four-leg intersection, if the AADTs of the two minor road legs
differ, the larger of the two AADTs values is used for the intersection. If AADTs are available for every roadway seg-
ment along a facility, the major roadAADTs for intersection legs can be determined without additional data.
In many cases, it is expected that AADT data will not be available for all years of the evaluation period. In that case,
an estimate of AADT for each year of the evaluation period is interpolated or extrapolated as appropriate. If there is
no established procedure for doing this, the following default rules may be applied within the predictive method to
estimate the AADTs for years for which data are not available.
• IfAADT data are available for only a single year, that same value is assumed to apply to all years of the before period.
• If two or more years of AADT data are available, the AADTs for intervening years are computed by interpolation.
that first year.
If the EB Method is used (discussed below), AADT data are needed for each year of the period for which observed
crash frequency data are available, If the EB Method will not be used, AADT data for the appropriate time period-
past, present, or future-determined in Step 2 are used.

Where an existing site or alternative conditions to an existing site are being considered, the EB Method is used. The
EB Method is only applicable when reliable observed crash data are available for the specific study roadway net-
work, facility, or site. Observed data may be obtained directly from the jurisdiction's crash report system. At least
two years of observed crash frequency data are desirable to apply the EB Method. The EB Method and criteria to
determine whether the EB Method is applicable are presented in Section A.2.1 in Appendix A to Part C.
The site-specific EB Method is applied in Step 13. Alternatively, if observed crash data are available but cannot be
assigned to individual roadway segments and intersections, the project level EB Method is applied (in Step 15).
If observed crash data are not available, then Steps 6, 13, and 15 of the predictive method are not conducted. In this
case, the estimate of expected average crash frequency is limited to using a predictive model (i'.e., the predicted aver-
age crash frequency).
Step 4-Determine geometric design features, traffic control features, and site characteristics for all sites in
the study network.
In order to determine the relevant data needs and avoid unnecessary data collection, it is necessary to understand the
base conditions of the SPFs in Step 9 and the CMFs in Step 10. The base conditions are defined in Section 10.6.1 for
roadway segments and in Section 10.6.2 for intersections.
The following geometric design and traffic control features are used to select a SPF and to determine whether the
site specific conditions vary from the base conditions and, therefore, whether a CMF is applicable:
• Length of segment (miles)
• AADT (vehicles per day)
• Lane width (feet)
• Shoulder width (feet)

• Shoulder type (paved/gravel/composite/turf)
• Presence or absence of horizontal curve (curve/tangent). If the segment has one or more curve:
• Length of horizontal curve (miles), (this represents the total length of the horizontal curve and includes spiral
transition curves, even if the curve extends beyond the limits of the roadway segment being analyzed);
• Radius of horizontal curve (feet);

• Presence or absence of spiral transition curve, (this represents the presence or absence of a spiral transition
curve at the beginning and end of the horizontal curve, even if the beginning and/or end of the horizontal curve
are beyond the limits of the segment being analyzed); and
• Superelevation of horizontal curve and the maximum superelevation (e ) used according to policy for the
jurisdiction, if available. ·~
• Grade (percent), considering each grade as a straight grade from Point ofVertical Intersection (PVI) to PVI
(i.e., ignoring the presence of vertical curves)
• Driveway density (driveways per mile)

• Presence or absence of centerline rumble strips
I
• Presence or absence of a passing lane
• Presence or absence of a short four-lane section
• Presence or absence of a two-way left-tum lane
• Roadside hazard rating

• Presence or absence of roadway segment lighting
• Presence or absence of automated speed enforcement
For all intersections within the study area, the following geometric design and traffic control features are identified:
• Number of intersection legs (3 or 4)
• Type of traffic control (minor road stop or signal control)
• Intersection skew angle (degrees departure from 90 degrees)
• Number of approaches with intersection left-tum lanes (0, 1, 2, 3, or 4), not including stop-controlled approaches
• Number of approaches with intersection right-tum lanes (0, 1, 2, 3, or 4), not including stop-controlled approaches
• Presence or absence of intersection lighting
Step 5---Divide the roadway network or facility under consideration into individual homogenous roadway
segments and intersections which are referred to as sites.
Using the information from Step 1 and Step 4, the roadway is divided into individual sites, consisting of individual
homogenous roadway segments and intersections. The definitions and methodology for dividing the roadway into
individual intersections and homogenous roadway segments for use with the Chapter 10 predictive models are
provided in Section 10.5. When dividing roadway facilities into small homogenous roadway segments, limiting
the segment length to a minimum of 0.10 miles will decrease data collection and management efforts.
Step 6--Assign observed crashes to the individual sites (if applicable).

way segments or intersections are presented in Section A.2.3 of Appendix A to Part C.
are assigned to the intersection and used in the EB Method together with the predicted average crash frequency for
the intersection. Crashes that occur between intersections and are not related to the presence of an intersection are
assigned to the roadway segment on which they occur; such crashes are used in the EB Method together with the
Step 7-Select the first or next individual site in the study network. If there are no more sites to be evaluated,
proceed to Step 15.
In Step 5, the roadway network within the study limits is divided into a number of individual homogenous sites
(intersections and roadway segments).
which is the sum of the all of the individual sites, for each year in the study. Note that this value will be the total
number of crashes expected to occur over all sites during the period of interest. If a crash frequency (crashes per
year) is desired, the total can be divided by the number of years in the period of interest.
The estimation for each site (roadway segments or intersection) is conducted one at a time. Steps 8 through 14,
CHAPTER 10-PREDiaiVE METHOD FOR RURAL TWO-LANE, TWO-WAY ROADS 10-9
Step 8--For the selected site, select the first or next year in the period of interest. If there are no more years to
be evaluated for that site, proceed to Step 15,
Steps 8 through 14 are repeated for each site in the study aod for each year in the study period.
The individual years of the evaluation period may have to be aoalyzed one year at a time for aoy particular roadway
segment or intersection because SPFs aod some CMFs (e.g., laoe aod shoulder widths) are dependent onAADT
which may chaoge from year to year.
Step 9-For the selected site, determine and apply the appropriate safety performance function (SPF) for the
site's facility type and traffic control features.
Steps 9 through 13 are repeated for each year of the evaluation period as part of the evaluation of aoy particular
roadway segment or intersection. The predictive models in Chapter I 0 follow the general form shown in Equation
I 0-1. Each predictive model consists of ao SPF, which is adjusted to site specific conditions using CMFs (in Step
I 0) and adjusted to local jurisdiction conditions (in Step II) using a calibration factor (C). The SPFs, CMFs, aod
calibration factor obtained in Steps 9, 10, aod II are applied to calculate the predicted average crash frequency
for the selected year of the selected site. The resultaot value is the predicted average crash frequency for the
selected year. The SPFs available for rural two-laoe, two-way highways are presented in Section 10.6.
The SPF (which is a statistical regression model based on observed crash data for a set of similar sites) determines
the predicted average crash frequency for a site with the base conditions (i.e., a specific set of geometric design aod
traffic control features). The base conditions for each SPF are specified in Section 10.6. A detailed explaoation aod
overview of the SPFs in Part Cis provided in Section C.6.3 of the Part C-lntroduction aodApplications Guidance.
The SPFs for specific site types (and base conditions) developed for Chapter 10 are summarized in Table 10-2
in Section I 0.6. For the selected site, determine the appropriate SPF for the site type (roadway segment or one of
three intersection types). The SPF is calculated using the AADT volume determined in Step 3 (AADT for roadway
segments or AADT maJ. aodAADT mm.. for intersections) for the selected year.
Each SPF determined in Step 9 is provided with default distributions of crash severity aod collision type. The default
distributions are presented in Tables 10-3 aod 10-4 for roadway segments aod in Tables 10-5 aod 10-6 for intersections.
These default distributions cao benefit from being updated based on local data as part of the calibration process
presented in Appendix A.l.l.
Step 10-Multiply the result obtained in Step 9 by the appropriate CMFs to adjust the estimated crash
frequency for base conditions to the site specific geometric design and traffic control features.
In order to account for differences between the base conditions (Section 10.6) aod site specific conditions,
CMFs are used to adjust the SPF estimate. An overview of CMFs and guidance for their use is provided in
Section C.6.4 of the Part C-Introduction and Applications Guidaoce. This overview includes the limitations
of current knowledge related to the effects of simultaoeous application of multiple CMFs. In using multiple
CMFs, engineering judgment is required to assess the interrelationships and/or independence of individual
elements or treatments being considered for implementation within the same project.
All CMFs used in Chapter 10 have the same base conditions as the SPFs used in Chapter 10 (i.e., when the specific
site has the same condition as the SPF base condition, the CMF value for that condition is 1.00). Only the CMFs
presented in Section I 0. 7 may be used as part of the Chapter I 0 predictive method. Table I 0-7 indicates which
CMFs are applicable to the SPFs in Section 10.6.
Step 11-Multiply the result obtained in Step 10 by the appropriate calibration factor,
periods. Calibration of the SPFs to local conditions will account for differences. A calibration factor (C, for
roadway segments or C, for intersections) is applied to each SPF in the predictive method. An overview of the use
of calibration factors is provided in the Part C-Introduction and Applications Guidaoce, Section C.6.5. Detailed
guidaoce for the development of calibration factors is included in Part C, Appendix A.!.!.
Steps 9, 10, and II together implement the predictive models in Equations 10-2 and 10-3 to determine predicted
average crash frequency.
Step 12-If there is another year to be evaluated in tbe study period for the selected site, return to Step 8.
Otherwise, proceed to Step 13.
Step 13---Apply site-specific EB Method (if applicable).

Whether the site-specific EB Method is applicable is determined in Step 3. The site-specific EB Method combines
the Chapter I 0 predictive model estimate of predicted average crash frequency, Npre,.JC!e,, with the observed crash
frequency of the specific site, Nobserved. This provides a more statistically reliable estimate of the expected average
crash frequency of the selected site.
In order to apply the site-specific EB Method, overdispersion parameter, k, for the SPF is used. This is in addition to
the material in Part C, Appendix A.2.4. The overdispersion parameter provides an indication of the statistical reliabil-
ity of the SPF. The closer the overdispersion parameter is to zero, the more statistically reliable the SPF. This param-
eter is used in the site-specific EB Method to provide a weightin2.....,
to N pre,.tete, and N ob serve,. Overdispersion parameters
are provided for each SPF in Section 10.6.
The estimated expected average crash frequency obtained above applies to the time period in the past for which
the observed crash data were obtained. Section A.2.6 in Appendix A to Part C provides method to convert the past
period estimate of expected average crash frequency into to a future time period.
Step 14-Ifthere is another site to be evaluated, return to Step 7, otherwise, proceed to Step 15.
This step creates a loop through Steps 7 to 13 that is repeated for each roadway segment or intersection within the facility.
Step 15-Apply the project level EB Method (if the site-specific EB Method is not applicable).
This step is only applicable to existing conditions when observed crash data are available, but cannot be accurately
assigned to specific sites (e.g., the crash report may identify crashes as occurring between two intersections, but is
not accurate to determine a precise location on the segment). Detailed description of the project level EB Method is
provided in Part C, Appendix A.2.5.
Step 16-Sum all sites and years in the study to estimate total crash frequency.
The total estimated number of crashes within the network or facility limits during a study period of n years is calcu-
lated using Equation I 0-4:
(10-4)
segments
Where:
N.,., = total expected number of crashes within the limits of a rural two-lane, two-way facility for the period of
interest. Or, the sum of the expected average crash frequency for each year for each site within the defined
roadway limits within the study period;
N, expected average crash frequency for a roadway segment using the predictive method for one specific year; and
N,, expected average crash frequency for an intersection using the predictive method for one specific year.
CHAPTER 10-PREDICTIVE METHOD FOR RURAL lWO-LANE, lWO-WAY ROADS 10-11
Equation I 0-4 represents the total expected nwnber of crashes estimated to occur during the study period. Equation
I 0-5 is used to estimate the total expected average crash frequency within the network or facility limits during the
study period.
Ntotal
Ntotal average = -n- (10-5)
Where:
~ total expected average crash frequency estimated to occur within the defined network or facility limits
n ~ nwnber of years in the study period.

Steps 3 through 16 of the predictive method are repeated, as appropriate, not only for the same roadway limits, but
also for alternative conditions, treatments, periods of interest, or forecast AADTs.

within defined network or facility limits over a given period of time, for given geometric design and traffic control
features, and known or estimated AADT. In addition to estimating total crashes, the estimate can be made for dif-
ferent crash severity types and different collision types. Default distributions of crash severity and collision type are
provided with each SPF in Section I 0.6. These default distributions can benefit from being updated based on local
data as part of the calibration process presented in Part C, Appendix A.l.l.
10.5. ROADWAY SEGMENTS AND INTERSECTIONS

Section 10.4 provides an explanation of the predictive method. Sections 10.5 through 10.8 provide the specific detail
necessary to apply the predictive method steps in a rural two-lane, two-way road environment. Detail regarding the
procedure for determining a calibration factor to apply in Step II is provided in the Part C, Appendix A. I. Detail
regarding the EB Method, which is applied in Steps 6, 13, and 15, is provided in the Part C, Appendix A.2.
In Step 5 of the predictive method, the roadway within the defined roadway limits is divided into individual sites,
which are homogenous roadway segments and intersections. A facility consists of a contignous set of individual
intersections and roadway segments, referred to as "sites." A roadway network consists of a number of Contiguous
facilities. Predictive models have been developed to estimate crash frequencies separately for roadway segments and
intersections. The definitions of roadway segments and intersections presented below are the same as those used in
the FHWA Interactive Highway Safety Design Model (IHSDM) (3).
Roadway segments begin at the center of an intersection and end at either the center of the next intersection, or
where there is a change from one homogeneous roadway segment to another homogenous segment. The roadway
segment model estimates the frequency of roadway-segment-related crashes which occur in Region B in Figure 10-2.
When a roadway segment begins or ends at an intersection, the length of the roadway segment is measured from the
center of the intersection.
The Chapter 10 predictive method addresses stop controlled (three- and four-leg) and signalized (four-leg) intersec-
tions. The intersection models estimate the predicted average frequency of crashes that occur within the limits of an
intersection (Region A of Figure 10-2) and intersection-related crashes that occur on the intersection legs (Region B
in Figure 10-2).
Segment Length
A All crashes that occur withm this region are classified as rntersection crashes.
B Crashes in this region may be segment or intersection related, depending on

Figure 10-2. Definition of Segments and Intersections
The segmentation process produces a set of roadway segments of varying length, each of which is homogeneous
with respect to characteristics such as traffic volumes, roadway design characteristics, and traffic control features.
Figure I 0-2 shows the segment length, L, for a single homogenous roadway segment occurring between two inter-
sections. However, it is likely that several homogenous roadway segments will occur between two intersections. A
new (unique) homogeneous segment begins at the center of each intersection or at any of the following:
• Beginning or end of a horizontal curve (spiral transitions are considered part of the curve).
• Point of vertical intersection (PVI) for a crest vertical curve, a sag vertical curve, or an angle point at which two
different roadway grades meet. Spiral transitions are considered part of the horizontal curve they adjoin and
vertical curves are considered part of the grades they adjoin (i.e., grades run from PVI to PVI with no explicit
consideration of any vertical curve that may be present).
• Beginning or end of a passing lane or short four-lane section provided for the purpose of increasing passing
opportunities.
• Beginning or end of a center two-way left-tum lane.
Also, a new roadway segment starts where there is a change in at least one of the following characteristics of the roadway:
• Average annual daily traffic volume (vehicles per day)
• Lane width
For lane widths measured to a 0.1-ft level of precision or similar, the following rounded lane widths are
recommended before determining "homogeneous" segments:
Measured Lane Width Rounded Lane Width
9.2 ft orless 9ft or less

9.3 ft to 9.7 ft 9.5 ft
9.8 ft to 10.2 ft 10ft
10.3 ft to 10.7 ft 10.5ft
10.8 ft to 11.2 ft lift
11.3ftto11.7ft 11.5 ft
II.8ftormore 12ft or more
• Shoulder width
For shoulder widths measures to a 0.1-ft level of precision or similar, the following rounded paved shoulderwidths
are recommended before determining "homogeneous" segments:
Measured Shoulder Width Rounded Shoulder Width
0.5 ft or less Oft
0.6 ft to 1.5 ft I ft
1.6 ft to 2.5 ft 2ft
7.6 ft or more 8ft or more
• Shoulder type
• Driveway density (driveways per mile)
For very short segment lengths (less than 0.5-miles), the use of driveway density for the single segment length may
result in an inflated value since driveway density is determined based on length. As a result, the driveway density
used for determining homogeneous segments should be for the facility (as defined in Section 10.2) length rather
than the segment length.
• Roadside hazard rating

As described later in Section 10.7.1, the roadside hazard rating (a scale from I to 7) will be used to determine a
roadside design CMF. Since this rating is a subjective value and can differ marginally based on the opinion of the
assessor, it is reasonable to asswne that a "homogeneous" segment can have a roadside hazard rating that varies
by as much as 2 rating levels. An average of the roadside hazard ratings can be used to compile a "homogeneous"
segment as long as the minimum and maximum values are not separated by a value greater than 2.
For example, if the roadside hazard rating ranges from 5 to 7 for a specific road, an average value of 6 can be
assumed and this would be considered one homogeneous roadside design condition. If, on the other hand, the
roadside hazard ratings ranged from 2 to 5 (a range greater than 2) these would not be considered "homogeneous"
roadside conditions and smaller segments may be appropriate.
• Presence/absence of centerline rumble strip
• Presence/absence oflighting
• Presence/absence of automated speed enforcement
There is no minimum roadway segment length for application of the predictive models for roadway segments. When
dividing roadway facilities into small homogenous roadway segments, limiting the segment length to a minimum of
0.10 miles will minimize calculation efforts and not affect results.
In order to apply the site-specific EB Method, observed crashes are assigned to the individual roadway segments
and intersections. Observed crashes that occur between intersections are classified as either intersection-related or
roadway-segment-related. The methodology for assignment of crashes to roadway segments and intersections for use
in the site-specific EB Method is presented in Section A.2.3 in Appendix A to Part C.
10.6. SAFETY PERFORMANCE FUNCTIONS

In Step 9 of the predictive method, the appropriate safety performance functions (SPFs) are used to predict average
crash frequency for the selected year for specific base conditions. SPFs are regression models for estimating the pre-
dicted average crash frequency of individual roadway segments or intersections. Each SPF in the predictive method
was developed with observed crash data for a set of similar sites. The SPFs, like all regression models, estimate the
value of a dependent variable as a function of a set of independent variables. In the SPFs developed for the HSM, the
dependent variable estimated is the predicted average crash frequency for a roadway segment or intersection under
base conditions and the independent variables are the AADTs of the roadway segment or intersection legs (and, for
roadway segments, the length of the roadway segment).
The SPFs used in Chapter 10 were originally formulated by Vogt and Bared (13, 14, 15). A few aspects of the
Harwood eta!. (5) and Vogt and Bared (13, 14, 15) work have been updated to match recent changes to the crash
prediction module of the FHWA Interactive Highway Safety Design Model (3) software. The SPF coefficients, de-
fault crash severity and collision type distributions, and default nighttime crash proportions have been adjusted to a
consistent basis by Srinivasan et al. (12).
The predicted crash frequencies for base conditions are calculated from the predictive models in Equations I 0-2
and 10-3. A detailed discussion of SPFs and their use in the HSM is presented in Chapter 3, Section 3.5.2, and the
Part C-Introduction and Applications Guidance, Section C.6.3.
Each SPF also has an associated overdispersion parameter, k. The overdispersion parameter provides an indication of
the statistical reliability of the SPF. The closer the overdispersion parameter is to zero, the more statistically reliable
the SPF. This parameter is used in the EB Method discussed in Part C, Appendix A. The SPFs in Chapter 10 are sum-
marized in Table l 0-2.
Table 10-2. Safety Performance Functions included in Chapter 10

Chapter 10 SPFs for Rural Two-Lane, Two-Way Roads SPF Equations and Figures
Rural two-lane, two-way roadway segments Equation 10-6 , Figure 10-3
Tbree-leg stop controlled intersections Equation 10-8 , Figure 10-4
Four-leg stop controlled intersections Equation 10-9, Figure 10-5
Four-leg signalized intersections Equation 10-10, Figure 10-6
Some highway agencies may have performed statistically-sound studies to develop their own jurisdiction-specific
SPFs derived from local conditions and crash experience. These models may be substituted for models presented in
this chapter. Criteria for the development of SPFs for use in the predictive method are addressed in the calibration
procedure presented in Appendix A to Part C.
10.6.1. Safety Performance Functions for Rural Two-Lane, Two-Way Roadway Segments
The predictive model for predicting average crash frequency for base conditions on a particular rural two-lane,
two-way roadway segment was presented in Equation I 0-2. The effect of traffic volume (AADT) on crash frequency
is incorporated through an SPF, while the effects of geometric design and traffic control features are incorporated
through the CMFs.
The base conditions for roadway segments on rural two-lane, two-way roads are:
• Lane width (LW) 12 feet
• Shoulder width (SW) 6 feet

• Shoulder type Paved
CHAPTER 1()-PREDICTIVE METHOD FOR RURAL TWO-LANE, TWO-WAY ROADS 10-15
• Roadside hazard rating (RHR) 3
• Driveway density (DD) 5 driveways per mile
• Horizontal curvature None
• Vertical curvature None
• Centerline rumble strips None
• Passing lanes None
• Two-way left-tum lanes None
• Lighting None
• Automated speed enforcement None
• Grade Level 0% (see note below)
A zero percent grade is not allowed by most states and presents issues such as drainage. The SPF uses zero percent
as a numerical base condition that must always be modified based on the actual grade.
The SPF for predicted average crash frequency for rural two-lane, two-way roadway segments is shown in Equation
10-6 and presented graphically in Figure 10-3:
Nspjrs = AADT x L x 365 x !Q-6 x ei·0.312J (10-6)
Where:
N,,1 ,. = predicted total crash frequency for roadway segment base conditions;
AADT = average annual daily traffic volume (vehicles per day); and
L = length of roadway segment (miles).
Guidance on the estimation of traffic volumes for roadway segments for use in the SPFs is presented in Step 3 of
the predictive method described in Section 10.4. The SPFs for roadway segments on rural two-lane highways are
applicable to the AADT range from zero to 17,800 vehicles per day. Application to sites with AADTs substantially
outside this range may not provide reliable results.
6~-----------------------------------------------------------,
~
~ 4
~
c.
>.
u
"
""
C"
3
"
..._
~
-"
~
~
u"' 2
-c
~
u"
:;;
~
"
"-
0 5,000 10,000 15,000

AADT (veh/day)
Figure 10-3, Graphical Form of SPF for Rural Two-Lane, Two-Way Roadway Segments (Equation 10-6)
The value of the overdispersion parameter associated with the SPF for rural two-lane, two-way roadway segments is
determined as a function of the roadway segment length using Equation 10-7. The closer the overdispersion param-
eter is to zero, the more statistically reliable the SPF. The value is determined as:
k= 0.236
L
(10-7)
Where:
k = overdispersion parameter; and
L = length of roadway segment (miles).
Tables l 0-3 and 10-4 provide the default proportions for crash severity and for collision type by crash severity level,
respectively. These tables may be used to separate the crash frequencies from Equation 10-6 into components by
crash severity level and collision type. Tables 10-3 and 10-4 are applied sequentially. First, Table 10-3 is used to
estimate crash frequencies by crash severity level, and then Table 10-4 is used to estimate crash frequencies by col-
lision type for a particular crash severity level. The default proportions for severity levels and collision types shown
in Tables 10-3 and 10-4 may be updated based on local data for a particular jurisdiction as part of the calibration
process described in Appendix A to Part C.
CHAPTER 1G-PREDICTIVE METHOD FOR RURAL TWO-LANE, TWO-WAY ROADS 10-17
Table 10-3, Default Distribution for Crash Severity Level on Rural Two-Lane, Two-Way Roadway Segments
Crash Severity Level Percentage of Total Roadway Segment Crashes"
Fatal 1.3
Incapacitating Injury 5.4

Nonincapacitating injury 10.9
Possible injury 14.5
Total fatal plus injury 32.1
Property damage only 67.9
Total 100.0
Note: Accident severity distributions are estimated for rural two-lane roadway segments in Exhibit 10-4.
• Based on HSIS data for Washington (2002-2006)
Table 10-4. Default Distribution by Collision Type for Specific Crash Severity Levels on Rural Two-Lane, Two-Way
Roadway Segments
Percentage of Total Roadway Segment Crashes by Crash Severity Level•
Collision Type Total Fatal and Injury Property Damage Only Total (All Severity Levels Combined)
SINGLE-VEHICLE CRASHES
Collision with animal 3.8 18.4 12.1
Collision with bicycle 0.4 0.1 0.2
Collision with pedestrian 0.7 0.1 0.3
Overturned 3.7 1.5 2.5
Ran off road 54.5 50.5 52.1
Other single-vehicle crash 0.7 2.9 2.1
Total single-vehicle crashes 63.8 73.5 69.3
MULTIPLE-VEHICLE CRASHES
Angle collision 10.0 7.2 8.5
Head-on collision 3.4 0.3 1.6
Rear-end collision 16.4 12.2 14.2
Sideswipe collision" 3.8 3.8 3.7
Other multiple-vehicle collision 2.6 3.0 2.7
Total multiple-vehicle crashes 36.2 26.5 30.7
Total Crashes 100.0 100.0 100.0
• Based on HSlS data for Washington (2002-2006)

b Includes approximately 70 percent opposite-direction sideswipe collisions and 30 percent same-direction sideswipe collisions
1 0.6.2. Safety Performance Functions for Intersections

The predictive model for predicting average crash frequency at particular rural two-lane, two-way road intersections
was presented in Equation 10-3. The effect of the major and minor road traffic volumes (AADTs) on crash frequency
is incorporated through SPFs, while the effects of geometric design and traffic control features are incorporated
through the CMFs. The SPFs for rural two-lane, two-way highway intersections are presented in this section.
SPFs have been developed for three types of intersections on rural two-lane, two-way roads. The three types of inter-
sections are:
• Three-leg intersections with minor-road stop control (3ST)

• Four-leg intersections with minor-road stop control (4ST)
• Four-leg signalized intersections (4SG)
SPFs for three-leg signalized intersections on rural two-lane, two-way roads are not available. Other types of inter-
sections may be found on rural two-lane, two-way highways but are not addressed by these procedures.
The SPFs for each of the intersection types listed above estimates total predicted average crash frequency for
intersection-related crashes within the limits of a particular intersection and on the intersection legs. The distinction
between roadway segment and intersection crashes is discussed in Section I 0.5 and a detailed procedure for distin-
guishing between roadway-segment-related and intersection-related crashes is presented in Section A.2.3 in Appen-
dix A to Part C. These SPFs address intersections that have only two lanes on both the major and minor road legs,
not including turn lanes. The SPFs for each of the three intersection types are presented below in Equations 10-8,
10-9, and 10-10. Guidance on the estimation of traffic volumes for the major and minor road legs for use in the SPFs
is presented in Section 10.4, Step 3.
The base conditions which apply to the SPFs in Equations 10-8, 10-9, and 10-10 are:
• Intersection skew angle 0'
• Intersection left-tum lanes None on approaches without stop control
• Intersection right-tum lanes None on approaches without stop control
• Lighting None
Three-Leg Stop-Controlled Intersections

The SPF for three-leg stop-controlled intersections is shown in Equation 10-8 and presented graphically in Figure I 0-4.
N,,fJST = exp[ -9.86 + 0.79 x ln(AADT=) + 0.49 x In(AADT.,,)] (10-8)
Where:
N,,r 3sr = estimate of intersection-related predicted average crash frequency for base conditions for three-leg stop-
controlled intersections;
AADT.,1 = AADT (vehicles per day) on the major road; and
AADT.,, = AADT (vehicles per day) on the minor road.
The overdispersion parameter (k) for this SPF is 0.54. This SPF is applicable to an AADT.,1 range from zero to
19,500 vehicles per day andAADT.,, range from zero to 4,300 vehicles per day. Application to sites withAADTs
substantially outside these ranges may not provide reliable results.
10
8 AADTmin=4,300
AADTmin=4,000
7
AADTmln=3,000
~
1: 6
"""" AADTmln=2,000
.._"
~
5
..::
"'
!!! 4
u AADTmln=1 ,000
"'C
~ 3
:c
"
~
0.
2
0 5,000 10,000 15,000 20,000

AADT moJo• (veh/day)
Figure 10-4. Graphical Representation of the SPF for Three-leg Stop-controlled (3ST) Intersections (Equation 10-8)
Four-Leg Stop-Controlled Intersections

The SPF for four-leg stop controlled intersections is shown in Equation 10-9 and presented graphically in Figure 10-5.
N,pf<ST = e.xp[-8.56 + 0.60 x In(AADT••) + 0.61 x In(AADT.,,)] (10-9)
Where:
= estimate of intersection-related predicted average crash frequency for base conditions for four-leg stop
controlled intersections;
AADT••1 = AADT (vehicles per day) on the major road; and
The overdispersion parameter (k) for this SPF is 0.24. This SPF is applicable to an AADT••J range from zero to
14,700 vehicles per day andAADT.,, range from zero to 3,500 vehicles per day. Application to sites with AADTs
substantially outside these ranges may not provide accurate results.
10
9 AADTmln=3,500
AADTmln=3,000
8
7
>.
u ~ AADT.1,=2,000
"" 6
"C"
..."
~
5
.:
~
"
~
u 4 =-j AADT.1,=1 ,000
""C
"
~
u 3
:0
"
~
1>.
2
0 J---+---~--+---r---~--~--~------~~--~--~--~--~--~
0 5,000 10,000 15,000

AADT m•i•• (veh/day)
Figure 10-5. Graphical Representation of the SPF for Four-leg, Stop-controlled (4ST) Intersections (Equation 10·9)
Four-Leg Signalized Intersections

The SPF for four-leg signalized intersections is shown in Equation I 0-10 and presented graphically in Figure I 0-6.
N,Pf,SG = exp[-5.13 + 0.60 x ln(AADT.") + 0.20 x In(AADT.)] (10-10)
Where:
= SPF estimate of intersection-related predicted average crash frequency for base conditions;
AADT."1 = AADT (vehicles per day) on the major road; and

The overdispersion parameter (k) for this SPF is 0.11. This SPF is applicable to an AADT moJ range from zero to
25,200 vehicles per day and AADT . range from zero to 12,500 vehicles per day. For instances when application is
made to sites with AADT substantiilly outside these ranges, the reliability is unknown.
18
17
16
15
14
13
12
>.
u
11
.,.""
:l
10
"~
LL 9
.<::
~ 8
u"'
~
7
"'0
6
"u
~
'5 5
"~
"- 4
3
2
1
0
0 5,000 10,000 15,000 20,000 25,000
AADT m•io• (veh/day)
Figure 10-6. Graphical Representation ofthe SPF for Four-leg Signalized (4SG) Intersections (Equation 10-10)
Tables I 0-5 and I 0-6 provide the default proportions for crash severity levels and collision types, respectively. These
tables may be used to separate the crash frequencies from Equations I 0-8 through I 0-10 into components by sever-
ity level and collision type. The default proportions for severity levels and collision types shown in Tables 10-5 and
I 0-6 may be updated based on local data for a particular jurisdiction as part of the calibration process described in
Appendix A to Part C.
Table 10-5. Default Distribution for Crash Severity Level at Rural Two-Lane, Two-Way Intersections
Percentage of Total Crashes
Three-Leg Four-Leg Four-Leg

Crash Severity Level Stop-Controlled Intersections Stop-Controlled Intersections Signalized Intersections
Fatal 1.7 1.8 0.9

Incapacitating Injury 4.0 4.3 2.1
Nonincapacitating injury 16.6 16.2 10.5
Possible injury 19.2 20.8 20.5
Total fatal plus injury 41.5 43.1 34.0
Property damage only 58.5 56.9 66.0
Tmal 100.0 100.0 100.0
Note: Based on HSIS data for California (2002-2006).

Table 10-6. Default Distribution for Collision Type and Manner of Collision at Rural Two-Way Intersections
Percentage of Total Crashes by CoJlision Type
Three-Leg Stop-Controlled Four-Leg Stop-Controlled Four.-Leg Signalized
Intersections Intersections Intersections
Fatal Property Fatal Property Fatal Property
and Damage and Damage and Damage
Collision Type Injury Only Total Injury Only Total Injury Only Total
SINGLE-VEHICLE CRASHES
Collision with animal 0.8 2.6 1.9 0.6 1.4 1.0 0.0 0.3 0.2
Collision with bicycle 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1
Collision with pedestrian 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1
Overturned 2.2 0.7 1.3 0.6 0.4 0.5 0.3 0.3 0.3
Ran off road 24.0 24.7 24.4 9.4 14.4 12.2 3.2 8.1 6.4
Other single-vehicle crash 1.1 2.0 1.6 0.4 1.0 0.8 0.3 1.8 0.5
Total single-vehicle crashes 28.3 30.2 29.4 11.2 17.4 14.7 4.0 10.7 7.6
MULTIPLE-VEHICLE CRASHES
Angle collision 27.5 21.0 23.7 53.2 35.4 43.1 33.6 24.2 27.4
Head-on collision 8.1 3.2 5.2 6.0 2.5 4.0 8.0 4.0 5.4
Rear-end collision 26.0 29.2 27.8 21.0 26.6 24.2 40.3 43.8 42.6
Sideswipe collision 5.1 13.1 9.7 4.4 14.4 10.1 5.1 15.3 ll.8
Other multiple-vehicle collision 5.0 3.3 4.2 4.2 3.7 3.9 9.0 2.0 5.2
Total multiple-vehicle crashes 71.7 69.8 70.6 88.8 82.6 85.3 96.0 89.3 92.4
Total Crashes 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
Note: Based on HSIS data for California (2002-2006).
10.7. CRASH MODIFICATION FACTORS

In Step 10 of the predictive method shown in Section 10.4, crash modification factors (CMFs) are applied to account
for the effects of site-specific geometric design and traffic control features. CMFs are used in the predictive method
in Equations 10-2 and 10-3. A general overview of crash modification factors (CMFs) is presented in Chapter 3, Sec-
tion 3.5.3. The Part C-Introduction and Applications Guidance provides further discussion on the relationship of
CMFs to the predictive method. This section provides details of the specific CMFs applicable to the safety perfor-
mance functions presented in Section I 0.6.
Crash modification factors (CMFs) are used to adjust the SPF estimate of predicted average crash frequency for
the effect of individual geometric design and traffic control features, as shown in the general predictive model for
Chapter I 0 shown in Equation I 0-1. The CMF for the SPF base condition of each geometric design or traffic control
feature has a value of 1.00. Any feature associated with higher crash frequency than the base condition has a CMF
with a value greater than 1.00. Any feature associated with lower crash frequency than the base condition has a CMF
with a value less than 1.00.
The CMFs used in Chapter I 0 are consistent with the CMFs in Pan D, although they have, in some cases, been
expressed in a different form to be applicable to the base conditions. The CMFs presented in Chapter 10 and the
specific site types to which they apply are summarized in Table 10-7.
CHAPTER 10--PREDICTIVE METHOD FOR RURAL TWO-LANE, TWO-WAY ROADS 10-23
Table 10-7. Summary of Crash Modification Factors (CMFs) in Chapter IQ and the Corresponding Safety
Performance Functions (SPFs)
Facility Type CMF CMF Description CMF Equations and Tables
cwlr Lane Width Table 10-8, Figure 10-7, Equation 10-11
CMF 2, Shoulder Width and Type Table 10-9, Figure 10-8, Table 10-10,
Equation 10-12
CMF3, Horizontal Curves: Length, Radius, and Table 10-7
Presence or Absence of Spiral Transitions
CJ\1F4r Horizontal Curves: Superelevation Equation 10-14, 10-15, 10-16,
CMF5, Grades Table 10-11
Rural Two-Lane Two-Way eMF. Driveway Density Table 10-11

Roadway Segments
CMF7, Centerline Rwnble Strips See text
CMFs, Passing Lanes See text
CMF9, Two-Way Left-Tum Lanes Equation 10-18, 10-19
CMFIOr Roadside Design Equation 10-20
CMF 11 , Lighting Equation 10-21, Table 10-12
CMF 12, Automated Speed Enforcement See text
CMFJI Intersection Skew Angle Equation 10-22, 10-23

Three- and four-leg stop control Cill2, Intersection Left-Turn Lanes Table 10-13
intersections and four-leg
signalized intersections CMFJI Intersection Right-Tum Lanes Table 10-14
CNIF 4 , Lighting Equation 10-24, Table 10-15
10.7.1. Crash Modification Factors for Roadway Segments

The CMFs for geometric design and traffic control features of rural two-lane, two-way roadway segments are pre-
sented below. These CMFs are applied in Step IQ of the predictive method and used in Equation IQ-2 to adjust the
SPF for rural two-lane, two-way roadway segments presented in Equation IQ-6, to account for differences between
the base conditions and the local site conditions.
CMF,,-Lane Width
The CMF for lane width on two-lane highway segments is presented in Table I Q-8 and illustrated by the graph in
Figure I Q. 7. This CMF was developed from the work of Zegeer et aL (16) and Griffin and Mak (4). The base value
for the lane width CMF is 12ft. In other words, the roadway segment SPF will predict safety performance of a road-
way segment with 12-ft lanes. To predict the safety performance of the aetna! segment in question (e.g., one with
lane widths different than 12 ft), CMFs are used to account for differences between base and aetna! conditions, Thus,
12-ft lanes are assigned a CMF of ],QQ. CMF 1, is determined from Table IQ-8 based on the applicable lane width and
traffic volume range, The relationships shown in Table IQ-8 are illustrated in Figure IQ-7. Lanes with widths greater
than 12 ft are assigned a CMF equal to that for 12-ft lanes,
For lane widths with Q.5-ft increments that are not depicted specifically in Table IQ-8 or Figure IQ-7, a CMF value
can be interpolated using either of these exhibits since there is a linear transition between the various AADT-effects,
I
I
Table 10-8. CMF for Lane Width on Roadway Segments (CMFro)

AADT (vehicles per day)
Lane Width <400 400 to 2000 >2000
9ft or less 1.05 1.05 + 2.81 X 1o-'(AADT- 400) 1.50

10ft 1.02 1.02 + 1.75 X 10~(AADT- 400) 1.30
11 ft 1.01 1.01 +2.5 X lO''(AADT-400) 1.05
12 ftormore 1.00 1.00 1.00
Note: The collision types related to lane width to which this CMF applies include single-vehicle run-off-the-road and multiple-vehicle head-on,
opposite-direction sideswipe, and same-direction sideswipe crashes.
1.70
The factor applies to

single-vehicle run-off-the-road
1.60 and multiple-vehicle head-on,
opposite-direction sideswipe,
and same-direction sideswipe
crashes.
1.50 9-ft Lanes
1.50
~
-
0
u
~
"'
r::
1.40
..,
0
"'
u 1.30 10-ft Lanes
"''ti
0
1.30
:;;
.r::
~
~ 1.20
u
1.10 1.05
1.05 11-ft Lanes
1.00 12-ft Lanes

1.00
0 400 800 1,200 1,600 2,000 2,400
AADT (veh/day)
Figure 10-7, Crash Modification Factor for Lane Width on Roadway Segments
If the lane widths for the two directions of travel on a roadway segment differ, the CMF are determined separately
for the lane width in each direction of travel and the resulting CMFs are then be averaged.
The CMFs shown in Table 10-8 and Figure 10-7 apply only to the crash types that are most likely to be affected by
lane width: single-vehicle run-off-the-road and multiple-vehicle head-on, opposite-direction sideswipe, and same-di-
rection sideswipe crashes. These are the only crash types assumed to be affected by variation in lane width, and other
crash types are assumed to remain unchanged due to the lane width variation. The CMFs expressed on this basis are,
therefore, adjusted to total crashes within the predictive method. This is accomplished using Equation 10-11:
CMF1, = (CMFro- 1.0) x Pro+ 1.0 (10-11)

Where:
CMF1, crash modification factor for the effect of lane width on total crashes;
CMF~ crash modification factor for the effect oflane width on related crashes (i.e., single-vehicle run-off-the-
road and multiple-vehicle head-on, opposite-direction sideswipe, and same-direction sideswipe crashes),
such as the crash modification factor for lane width shown in Table 10-8; and
p~ ~ proportion of total crashes constituted by related crashes.
The proportion of related crashes, p ~' (i.e., single-vehicle run-off-the-road, and multiple-vehicle head-on, opposite-
direction sideswipe, and same-direction sideswipes crashes) is estimated as 0.574 (i.e., 57.4 percent) based on the
default distribution of crash types presented in Table I 0-4. This default crash type distribution, and therefore the
value of pro' may be updated from local data as part of the calibration process.
CMF,,-Shoulder Width and Type

The CMF for shoulders has a CMF for shoulder width (CMFwro) and a CMF for shoulder type (CMF,). The CMFs
for both shoulder width and shoulder type are based on the results of Zegeer et al. (16, 17). The base value of shoul-
der width and type is a 6-foot paved shoulder, which is assigned a CMF value of 1.00.
CMF wro for shoulder width on two-lane highway segments is determined from Table I 0-9 based on the applicable
shoulder width and traffic volume range. The relationships shown in Table 10-9 are illustrated in Figure I 0-8.
Shoulders over 8-ft wide are assigned a CMFwro equal to that for 8-ft shoulders. The CMFs shown in Table 10-9 and
Figure 10-8 apply only to single-vehicle run-off the-road and multiple-vehicle head-on, opposite-direction sideswipe,
and same-direction sideswipe crashes.
Table 10-9. CMF for Shoulder Width on Roadway Segments (CMF wro)
AADT (vehicles per day)
Shoulder Width <400 400 to 2000 > 2000
Oft 1.10 1.10 + 2.5 X 10~ (AADT- 400) 1.50

2ft 1.07 1.07 + 1.43 X 1~ (AADT- 400) 1.30
4ft 1.02 1.02 + 8.125 X 1~ (AADT- 400) 1.15
6ft 1.00 1.00 1.00
8ft or more 0.98 0.98 + 6.875 X 1Q-' (AADT- 400) 0.87
Note: The collision types related to shoulder width to which this CMF applies include single-vehicle run-off the-road and multiple-vehicle
head-on, opposite-direction sideswipe, and same-direction sideswipe crashes.
1.60
1.50 0-ftShoulders
1.50 This factor applies to single-vehicle
run-off·the-road and multiple-vehicle
head-on. opposite-direction sideswipe,
1.40
-.
~
0
u
"-
1.30
130 2-ftShoulders
c
..
0
·;::;
u
1.20
1.15 4-ftShoulders
"''5
0
1.10
1.10
:;:
.c
~
l! 6-ft Shoulders
u 1.00
1.00
0.98
0.90 0,87 8-ft Shoulders
0.80
0 200 400 600 800 1,ODD 1,200 1,400 1,600 1,800 2,000 2,200 2,400
AADT (veh/day)
Figure 10-8. Crash Modification Factor for Shoulder Width on Roadway Segments
The base condition for shoulder type is paved. Table 10-10 presents values for CMF,, which adjusts for the safety
effects of gravel, turf, and composite shoulders as a function of shoulder width.
Table 10-10. Crash Modification Factors for Shoulder Types and Shoulder Widths on Roadway Segments (CMF ,)
Shoulder Width (ft)
Shoulder Type 0 1 2 3 4 6 8
Paved 1.00 1.00 1.00 1.00 1.00 1.00 1.00
Gravel 1.00 1.00 1.01 ].01 1.01 1.02 1.02
Composite 1.00 1.01 1.02 1.02 1.03 1.04 1.06
Turf 1.00 1.01 1.03 1.04 1.05 1.08 1.11
Note: The values for composite shoulders in this table represent a shoulder for which 50 percent of the shoulder width is paved and 50 percent
of the shoulder width is turf.
If the shoulder types and/or widths for the two directions of a roadway segment differ, the CMF are determined
separately for the shoulder type and width in each direction of travel and the resulting CMFs are then be averaged.
The CMFs for shoulder width and type shown in Table 9, Figure 8, and Table 10 apply only to the collision types
that are most likely to be affected by shoulder width and type: single-vehicle run-offthe-road and multiple-vehicle
head-on, opposite-direction sideswipe, and same-direction sideswipe crashes. The CMFs expressed on this basis are,
therefore, adjusted to total crashes using Equation I 0-12.
CHAPTER 10-PREDICTIVE METHOD FOR RURAL 1WO-LANE, 1WO-WAY ROADS 10-27
CMF2, = (CMFwro X CMF,ro- 1.0) X Pro+ 1.0 (10-12)
Where:
CMF,, = crash modification factor for the effect of shoulder width and type on total crashes;
CMFwro = crash modification factor for related crashes (i.e., single-vehicle run-off-the-road and multiple-vehicle
head-on, opposite-direction sideswipe, and same-direction sideswipe crashes), based on shoulder width
(from Table 10-9);
CMF,ro = crash modification factor for related crashes based on shoulder type (from Table 10-10); and
pro = proportion of total crashes constituted by related crashes.
The proportion of related crashes, pro, (i.e., single-vehicle run-off-the-road, and multiple-vehicle head-on, opposite-
direction sideswipe, and same-direction sideswipes crashes) is estimated as 0.574 (i.e., 57.4 percent) based on the
default distribution of crash types presented in Table I 0-4. This default crash type distribution, and therefore the
value of pra, may be updated from local data by a highway agency as part of the calibration process.
CMF3,-Horizontal Curves: Length, Radius, and Presence or Absence of Spiral Transitions

The base condition for horizontal aligmnent is a tangent roadway segment. A CMF has been developed to represent
the manner in which crash experience on curved aligmnents differs from that of tangents. This CMF applies to total
roadway segment crashes.
The CMF for horizontal curves has been determined from the regression model developed by Zegeer eta!. (18).
The CMF for horizontal curvature is in the form of an equation and yields a factor similar to the other CMFs in this
chapter. The CMF for length, radius, and presence or absence of spiral transitions on horizontal curves is determined
using Equation I 0-13.
(1.55 X Lc) + (~)- (0.012 X S)

(10-13)
(1.55 x Lo)
Where:
CMF, = crash modification factor for the effect of horizontal aligmnent on total crashes;
L, = length of horizontal curve (miles) which includes spiral transitions, if present;
R = radius of curvature (feet); and
S = 1 if spiral transition curve is present; 0 if spiral transition curve is not present; 0.5 if a spiral transition
curve is present at one but not both ends of the horizontal curve.
Some roadway segments being analyzed may include only a portion of a horizontal curve. In this case, L, represents
the length of the entire horizontal curve, including portions of the horizontal curve that may lie outside the roadway
segment of interest.
In applying Equation 10-13, if the radius of curvature (R) is less than 100-ft, R is set to equal to 100ft. If the length
of the horizontal curve (L,) is less than 100 feet, L, is set to equallOO ft.
CMF values are computed separately for each horizontal curve in a horizontal curve set (a curve set consists of a
series of consecutive curve elements). For each individual curve, the value of L, used in Equation 10-13 is the total
length of the compound curve set and the value ofR is the radius of the individual curve.
If the value of CMF,,is less than 1.00, the value of CMF, is set equal to 1.00.
CMF,,-Horizontal Curves: Superelevation

The base condition for the CMF for the superelevation of a horizontal curve is the amount of superelevation identi-
fied in A Policy on Geometric Design of Highways and Streets-also called the AASHTO Green Book (I). The su-
perelevation in the AASHTO Green Book is determined by taking into account the value of maximum superelevation
rate, emax' established by highway agency policies. Policies concerning maximum superelevation rates for horizontal
curves vary betvveen highway agencies based on climate and other considerations.
The CMF for superelevation is based on the superelevation variance of a horizontal curve (i.e., the difference
between the actual superelevation and the superelevation identified by AASHTO policy). When the actual superel-
evation meets or exceeds that in the AASHTO policy, the value of the superelevation CMF is 1.00. There is no effect
of superelevation variance on crash frequency until the superelevation variance exceeds 0.0 I. The general functional
form of a CMF for superelevation variance is based on the work of Zegeer et a!. (18, 19).
The following relationships present the CMF for superelevation variance:
CMF4, ~ 1.00 for SV < 0.01 (10-14)
CMF4, ~ 1.00 + 6 x (SV- 0.01) for 0.01 ~ SV < 0.02 (10-15)
CMF,, ~ 1.06 + 3 x (SV- 0.02) for SV<: 0.02 (10-!6)
Where:
CMF4, ~ crash modification factor for the effect of superelevation variance on total crashes; and
SV superelevation variance (ft/ft), which represents the superelevation rate contained in the AASHTO
Green Book minus the actual superelevation of the curve.
CMF4, applies to total roadway segment crashes for roadway segments located on horizontal curves.
CMF,,-Grades
The base condition for grade is a generally level roadway. Table I 0-11 presents the CMF for grades based on an
analysis of rural two-lane, two-way highway grades in Utah conducted by Miaou (8). The CMFs in Table 10-!1 are
applied to each individual grade segment on the roadway being evaluated without respect to the sign of the grade.
The sign of the grade is irrelevant because each grade on a rural two-lane, two-way highway is an upgrade for one
direction of travel and a downgrade for the other. The grade factors are applied to the entire grade from one point of
vertical intersection (PVI) to the next (i.e., there is no special account taken of vertical curves). The CMFs in Table
10-11 apply to total roadway segment crashes.
Table 10-11. Crash Modification Factors (CMF ,) for Grade of Roadway Segments
Approximate Grade (%)
Level Grade Moderate Terrain Steep Terrain
(:53%) (3%< grade :56%) (> 6%)
1.00 1.10 1.16
CMF6,-Driveway Density
The base condition for driveway density is five driveways per mile. As with the other CMFs, the model for the base
condition was established for roadways with this driveway density. The CMF for driveway density is determined
using Equation 10-17, derived from the work of Muskaug (9).
CMF ~ 0.322 + DD X [0.05 - 0.005 x In(AADT)]

6 (10-17)
' 0.322 + 5 X [0.05- 0.005 X Jn(AADT)]
CHAPTER lG--PREDICTIVE METHOD FOR RURAL TWO-LANE, TWO-WAY ROADS 10-29
Where:
CMF6, crash modification factor for the effect of driveway density on total crashes;
AADT average annual daily traffic volume of the roadway being evaluated (vehicles per day); and
DD = driveway density considering driveways on both sides of the highway (driveways/mile).
If driveway density is less than 5 driveways per mile, CMF 6, is 1.00. Equation I 0-17 can be applied to total
roadway crashes of all severity levels.
Driveways serving all types of land use are considered in determining the driveway density. All driveways that
are used by traffic on at least a daily basis for entering or leaving the highway are considered. Driveways that
receive only occasional use (less than daily), such as field entrances are not considered.
CMF ,,-Centerline Rnmble Strips

Centerline rumble strips are installed on undivided highways along the centerline of the roadway which divides
opposing directions of traffic flow. Centerline rumble strips are incorporated in the roadway surface to alert
drivers who unintentionally cross, or begin to cross, the roadway centerline. The base condition for centerline
rumble strips is the absence of rumble strips.
The value of CMF ,,. for the effect of centerline rumble strips for total crashes on rural two-lane, two-way
highways is derived as 0.94 from the CMF value presented in Chapter 13 and crash type percentages found in
Chapter 10. Details of this derivation are not provided.
The CMF for centerline rumble strips applies only to two-lane undivided highways with no separation other than
a centerline marking between the lanes in opposite directions of travel. Otherwise the value of this CMF is 1.00.
CMF,,-Passing Lanes
The base condition for passing lanes is the absence of a lane (i.e., the normal two-lane cross section). The CMF
for a conventional passing or climbing lane added in one direction of travel on a rural two-lane, two-way highway
is 0.75 for total crashes in both directions of travel over the length of the passing lane from the upstream end of
the lane addition taper to the downstream end of the lane drop taper. This value assumes that the passing lane is
operationally warranted and that the length of the passing lane is appropriate for the operational conditions on the
roadway. There may also be some safety benefit on the roadway downstream of a passing lane, but this effect has
not been quantified.
The CMF for short four-lane sections (i.e., side-by-side passing lanes provided in opposite directions on the
same section of roadway) is 0.65 for total crashes over the length of the short four-lane section. This CMF
applies to any portion of roadway where the cross section has four lanes and where both added lanes have been
provided over a limited distance to increase passing opportunities. This CMF does not apply to extended four-
lane highway sections.
The CMF for passing lanes is based primarily on the work of Harwood and St.John (6), with consideration also
given to the results of Rinde (11) and N ettelblad (I 0). The CMF for short four-lane sections is based on the
work of Harwood and St. John (6).
CMF9,-Two-Way Left-Tnrn Lanes

The installation of a center two-way left-turn lane (TWLTL) on a rural two-lane, two-way highway to create a
three-lane cross-section can reduce crashes related to turning maneuvers at driveways. The base condition for
two-way left-turn lanes is the absence of a TWLTL. The CMF for installation of a TWLTL is:
: "T
CMF9, ~ 1.0- (0.7 x Pdwy x Pwd (10-18)
Where:
CMF9, ~ crash modification factor for the effect of two-way left-tum lanes on total crashes;
Pdwy ~ driveway-related crashes as a proportion of total crashes; and
PwD ~ left-tum crashes susceptible to correction by a TWLTL as a proportion of driveway-related crashes.
The value ofpdwy can be estimated using Equation 10-19 (6).
2
(0.0047 X DD) + (0.0024 x DD( ))
Pdwy = ---------'----,-----'-c=
(10-19)
1.199 + (0.0047 x DD) + (0.0024 x DD( 2 ))
Where:
Pdwy ~ driveway-related crashes as a proportion of total crashes; and
DD ~ driveway density considering driveways on both sides of the highway (driveways/mile).
The value ofpLTID is estimated as 0.5 (6).
Equation 10-18 provides the best estimate of the CMF forTWLTL installation that can be made without data on
the left-tum volumes within the TWLTL. Realistically, such volumes are seldom available for use in such analyses
though Section A. I. of Appendix A to Part C describes how to appropriately calibrate this value. This CMF applies
to total roadway segment crashes.
The CMF for TWLTL installation is not applied unless the driveway density is greater than or equal to five driveways
per mile. If the driveway density is less than five driveways per mile, the CMF for TWLTL installation is 1.00.
CMF10,-Roadside Design
For purposes of the HSM predictive method, the level of roadside design is represented by the roadside hazard rating
(1-7 scale) developed by Zegeer et al. (16). The CMF for roadside design was developed in research by Harwood et
a!. (5). The base value of roadside hazard rating for roadway segments is 3. The CMF is:
(-0.6869 + 0.0668 XRHR)

CMF - e
Io,- ---e'(-{)~.4"'86:;-:s")-- (10-20)
Where:
CMF,., ~ crash modification factor for the effect of roadside design; and
RHR ~ roadside hazard rating.
This CMF applies to total roadway segment crashes. Photographic examples and quantitative definitions for each
roadside hazard rating (1-7) as a function of roadside design features such as sideslope and clear zone width are
presented in Chapter 13, Appendix 13A.
CMF 11 ,-Lighting
The base condition for lighting is the absence of roadway segment lighting. The CMF for lighted roadway segments
is determined, based on the work ofElvik and Vaa (2), as:
(10-21)
Where:
CMF11 , ~ crash modification factor for the effect oflighting on total crashes;
p 1, ~ proportion of total nighttime crashes for unlighted roadway segments that involve a fatality or injury;
P,, ~ proportion of total nighttime crashes for unlighted roadway segments that involve property damage only; and
P, ~ proportion of total crashes for unlighted roadway segments that occur at night.
This CM:F applies to total roadway segment crashes. Table I 0-12 presents default values for tbe nighttime crash propor-
tionsp.mr,ppnr, andp nr. HSM users are encouraged to replace tbe estimates in Table 10-12 witb locally derived values. If
lighting installation increases tbe density of roadside fixed objects, tbe value ofCMF10, is adjusted accordingly.
Table 10-12. Nighttime Crash Proportions for Unlighted Roadway Segments

Proportion of Total Nighttime Crashes by Severity Level Proportion of Crashes that Occur at Night
Roadway Type Fatal and Injury p 1,, PDOp ~' P_.
2U 0.382 0.618 0.370
Note: Based on HSIS data for Washington (2002-2006)
CMF12,-Automated Speed Enforcement

Automated speed enforcement systems use video or photographic identification in conjunction with radar or
lasers to detect speeding drivers. These systems automatically record vehicle identification information without
the need for police officers at the scene. The base condition for automated speed enforcement is that it is absent.
The value of CM:F 1,, for tbe effect of automated speed enforcement for total crashes on rural two-lane, two-way
highways is derived as 0.93 from tbe CMF value presented in Chapter 17 and crash type percentages found in
Chapter 10. Details oftbis derivation are not provided.
10.7.2. Crash Modification Factors for Intersections

The effects of individual geometric design and traffic control features of intersections are represented in tbe
predictive models by CMFs. The CMFs for intersection skew angle, left-tum lanes, right-turn lanes, and lighting
are presented below. Each of tbe CMFs applies to total crashes.
CMF11-Intersection Skew Angle

The base condition for intersection skew angle is zero degrees of skew (i.e., an intersection angle of 90 degrees).
The skew angle for an intersection was defined as the absolute value of !be deviation from an intersection angle of
90 degrees. The absolute value is used in tbe definition of skew angle because positive and negative skew angles are
considered to have similar detrimental effect (4). This is illustrated in Chapter 14, Section 14.6.2.
Three-Leg Intersections with Stop-Control on the Minor Approach

The CMF for intersection angle at tbree-leg intersections witb stop-control on tbe minor approach is:
CMF II = e (0.004 X skew) (10-22)
Where:
CMFn ~ crash modification factor for tbe effect of intersection skew on total crashes; and
• '11"
skew intersection skew angle (in degrees); the absolute value of the difference between 90 degrees and the
actual intersection angle.
This CMF applies to total intersection crashes.
Four-Leg Intersections with Stop-Control on the Minor Approaches

The CMF for intersection angle at four-leg intersection with stop-control on the minor approaches is:
CMF
li
= e (o.oo54 "skew) (10-23)
Where:
crash modification factor for the effect of intersection skew on total crashes; and
intersection skew angle (in degrees); the absolute value of the difference between 90 degrees and the
This CMF applies to total intersection crashes.
If the skew angle differs for the two minor road legs at a four-leg stop-controlled intersection, values ofCMF 11 is
computed separately for each minor road leg and then averaged.
Four-Leg Signalized Intersections

Since the traffic signal separates most movements from conflicting approaches, the risk of collisions related to the
skew angle between the intersecting approaches is limited at a signalized intersection. Therefore, the CMF for skew
angle at four-leg signalized intersections is 1.00 for all cases.
CMF11-Intersection Left-Turn Lanes

The base condition for intersection left-turn lanes is the absence of left-tum lanes on the intersection approaches.
The CMFs for the presence of left-tum lanes are presented in Table 10-13. These CMFs apply to installation of
left-turn lanes on any approach to a signalized intersection, but only on uncontrolled major road approaches to a
stop-controlled intersection. The CMFs for installation of left-turn lanes on multiple approaches to an intersec-
tion are equal to the corresponding CMF for the installation of a left-turn lane on one approach raised to a power
equal to the number of approaches with left-tum lanes. There is no indication of any safety effect of providing
a left-turn lane on an approach controlled by a stop sign, so the presence of a left-turn lane on a stop-controlled
approach is not considered in applying Table I 0-13. The CMFs for installation of left-tum lanes are based on re-
search by Harwood eta!. (5) and are consistent with the CMFs presented in Chapter 14. A CMF of 1.00 is always
be used when no left-tum lanes are present.
Table 10-13. Crash Modification Factors (CMF21) for Installation of Left-Tum Lanes on Intersection Approaches
Number of Approaches with Left-Turn Lanes•
Intersection Type Intersection Traffic Control One Approach Two Approaches Three Approaches Four Approaches
Three-leg Intersection Minor road stop controlb 0.56 0.31
Minor road stop controJb 0.72 0.52
Four-leg Intersection
Traffic signal 0.82 0.67 0.55 0.45
• Stop-controlled approaches are not considered in determining the number of approaches with left-turn lanes
b Stop signs present on minor road approaches only.
CMF,,-Intersection Right-Turn Lanes

The base condition for intersection right-turn lanes is the absence of right-turn lanes on the intersection approaches.
The CMF for the presence of right-turn lanes is based on research by Harwood eta!. (5) and is consistent with the
CMFs in Chapter 14. These CMFs apply to installation of right-turn lanes on any approach to a signalized intersec-
lion, but only on uncontrolled major road approaches to stop-controlled intersections. The CMFs for installation of
right-turn Janes on multiple approaches to an intersection are equal to the corresponding CMF for installation of a
right-turn lane on one approach raised to a power equal to the number of approaches with right-turn lanes. There
is no indication of any safety effect for providing a right-turn lane on an approach controlled by a stop sign, so the
presence of a right-turn lane on a stop-controlled approach is not considered in applying Table 10-14. The CMFs
in the table apply to total intersection crashes. A CMF value of 1.00 is always be used when no right-turn lanes are
present. This CMF applies only to right-turn lanes that are identified by marking or signing. The CMF is not appli-
cable to long tapers, flares, or paved shoulders that may be used informally by right-turn traffic.
Table 10-14, Crash Modification Factors (CMF 3,) for Right-Turn Lanes on Approaches to an Intersection on Rural
Two-Lane, Two-Way Highways
Number of Approaches with Right-Turn Lanesa
Three-Leg Intersection Minor road stop controJb 0.86 0.74

Minor road stop controlb 0.86 0.74
Four-Leg Intersection
Traffic signal 0.96 0.92 0.88 0.85
"Stop-controlled approaches are not considered in determining the number of approaches with right-turn lanes.
CMF,,-Lighting
The base condition for lighting is the absence of intersection lighting. The CMF for lighted intersections is adapted
from the work of Elvik and Vaa (2), as:
CMF41 = 1-0.38 x p"1 (10-24)
Where:
CMF41 crash modification factor for the effect of lighting on total crashes; and
p "1 = proportion of total crashes for unlighted intersections that occur at night.
This CMF applies to total intersection crashes. Table 10-15 presents default values for the nighttime crash proportion
P"r HSM users are encouraged to replace the estimates in Table 10-15 with locally derived values.
Table 10-15, Nighttime Crash Proportions for Unlighted Intersections

Proportion of Crashes that Occur at Night
Intersection Type
3ST 0.260
4ST 0.244
4SG 0.286
Note: Based on HSIS data for California (2002-2006)
10.8. CALIBRATION OF THE SPFS TO LOCAL CONDITIONS

In Step 10 of the predictive method, presented in Section 10.4, the predictive model is calibrated to local state or
geographic conditions. Crash frequencies, even for nominally similar roadway segments or intersections, can vary
widely from one jurisdiction to another. Geographic regions differ markedly in climate, anintal population, driver
populations, crash reporting threshold, and crash reporting practices. These variations may result in some jurisdictions
.l...
experiencing a different nwnber of reported traffic crashes on rural two-lane, two-way roads than others. Calibration
factors are included in the methodology to allow highway agencies to adjust the SPFs to match actual local conditions.
The calibration factors for roadway segments and intersections (defined as C, and c, respectively) will have values greater
than 1.0 for roadways that, on average, experience more crashes than the roadways used in the development of the SPFs.
The calibration factors for roadways that experience fewer crashes on average than the roadways used in the development
of the SPFs will have values less than 1.0. The calibration procedures are presented in Appendix A to Part C.
Calibration factors provide one method of incorporating local data to improve estimated crash frequencies for indi-
vidual agencies or locations. Several other default values used in the predictive method, such as collision type distri-
bution, can also be replaced with locally derived values. The derivation of values for these parameters is addressed in
the calibration procedure in Appendix A to Part C.
10.9. LIMITATIONS OF PREDICTIVE METHOD IN CHAPTER 10

This section discusses limitations of the specific predictive models and the application of the predictive method in
Chapter 10.
Where rural two-lane, two-way roads intersect access-controlled facilities (i.e., freeways), the grade-separated
interchange facility, including the two-lane road within the interchange area, cannot be addressed with the predictive
method for rural two-lane, two-way roads.
The SPFs developed for Chapter 10 do not include signalized three-leg intersection models. Such intersections are
occasionally found on rural two-lane, two-way roads.
10.10. APPLICATION OF CHAPTER 10 PREDICTIVE METHOD

The predictive method presented in Chapter 10 applies to rural two-lane, two-way roads. The predictive method
is applied to a rural two-lane, two-way facility by following the 18 steps presented in Section I 0.4. Appendix I OA
provides a series of worksheets for applying the predictive method and the predictive models detailed in this chapter.
All computations within these worksheets are conducted with values expressed to three decimal places. This level of
precision is needed for consistency in computations. In the last stage of computations, rounding the final estimate of
expected average crash frequency to one decimal place is appropriate.
10.11. SUMMARY
The predictive method can be used to estimate the expected average crash frequency for a series of contiguous sites
(entire rural two-lane, two-way facility), or a single individual site. A rural two-lane, two-way facility is defined in Sec-
tion 10.3, and consists of a two-lane, two-way undivided road which does not have access control and is outside of cities
or towns with a population greater than 5,000 persons. Two-lane, two-way undivided roads that have occasional added
lanes to provide additional passing opportunities can also be addressed with the Chapter I 0 predictive method.
The predictive method for rural two-lane, two-way roads is applied by following the 18 steps of the predictive
method presented in Section 10.4. Predictive models, developed for rural two-lane, two-way facilities, are applied
in Steps 9, 10, and II of the method. These predictive models have been developed to estimate the predicted aver-
age crash frequency of an individual site which is an intersection or homogenous roadway segment. The facility is
divided into these individual sites in Step 5 of the predictive method.
Each predictive model in Chapter I 0 consists of a safety performance function (SPF), crash modification fac-
tors (CMFs), and a calibration factor. The SPF is selected in Step 9 and is used to estimate the predicted aver-
age crash frequency for a site with base conditions. The estimate can be for either total crashes or organized by
crash-severity or collision-type distribution. In order to account for differences between the base conditions and
the specific conditions of the site, CMFs are applied in Step 10, which adjust the prediction to account for the
geometric design and traffic control features of the site. Calibration factors are also used to adjust the prediction
to local conditions in the jurisdiction where the site is located. The process for determining calibration factors for
the predictive models is described in Part C, Appendix A.l.
Section 10.12 presents six sample problems which detail the application of the predictive method. Appendix lOA
contains worksheets which can be used in the calculations for the predictive method steps.
10.12. SAMPLE PROBLEMS

In this section, six sample problems are presented using the predictive method for rural two-lane, two-way roads.
Sample Problems 1 and 2 illustrate how to calculate the predicted average crash frequency for rural two-lane road-
way segments. Sample Problem 3 illustrates how to calculate the predicted average crash frequency for a stop-con-
trolled intersection. Sample Problem 4 illustrates a similar calculation for a signalized intersection. Sample Problem
5 illustrates how to combine the results from Sample Problems I through 3 in a case where site-specific observed
crash data are available (i.e., using the site-specific EB Method). Sample Problem 6 illustrates how to combine the
results from Sample Problems 1 through 3 in a case where site-specific observed crash data are not available but
project-level observed crash data are available (i.e., using the project-level EB Method).
Table 10-16. List of Sample Problems in Chapter 10

Problem No. Page No. Description
10-35 Predicted average crash frequency for a tangent roadway segment

2 10-42 Predicted average crash frequency for a curved roadway segment
3 10-49 Predicted average crash frequency for a three-leg stop-controlled intersection
4 10-55 Predicted average crash frequency for a four-leg signalized intersection
5 10-60 Expected average crash frequency for a facility when site-specific observed crash data are available
6 10-62 Expected average crash frequency for a facility when site-specific observed crash data are not available
10.12.1. Sample Problem 1
The Site/Facility
A rural two-lane tangent roadway segment.
The Question
What is the predicted average crash frequency of the roadway segment for a particular year?
The Facts
• 1.5-mi length
• Tangent roadway segment
• 10,000 veh!day
• 2% grade
• 6 driveways per mi
• 10-ft lane width

• 4-ft gravel shoulder
• Roadside hazard rating = 4
.~. .
TT"
Assumptions
Collision type distributions used are the default values presented in Table I 0-4.
The calibration factor is assumed to be !.!0.
Results
Using the predictive method steps as outlined below, the predicted average crash frequency for the roadway segment
in Sample Problem I is determined to be 6.! crashes per year (rounded to one decimal place).
Steps
Step 1 through 8
To determine the predicted average crash frequency of the roadway segment in Sample Problem I, only Steps 9
through II are conducted. No other steps are necessary because only one roadway segment is analyzed for one year,
and the EB Method is not applied.
The SPF for a single roadway segment can be calculated from Equation I 0-6 as follows:
N sprrf ~ AADT x L x 365 x I IT' x e<..·312l
N .<prrf ~ 10 ' 000 x 1.5 x 365 x liT' x J-<l.3l2l ~ 4.008 crashes/year
frequency for base conditions to the site-specific geometric design and traffic control features.
Each CMF used in the calculation of the predicted average crash frequency of the roadway segment is calculated below:
Lane Width (CMF1)

CMF 1, can be calculated from Equation 10-11 as follows:
For a 10-ft lane width and AADT of!O,OOO, CMFro ~ 1.30 (see Table 10-8).
The proportion of related crashes, pro, is 0.574 (see discussion below Equation 10-11).
CMFJ, ~ (1.3- 1.0) X 0.574 + 1.0 ~ !.!7
Shoulder Width and Type (CMF2 )

CMF,, can be calculated from Equation 10-12, using values from Table 10-9, Table 10-10, and Table 10-4 as follows:
CMF2, ~ (CMFwro X CMF,ro- 1.0) X Pro+ 1.0
For 4-ft shoulders and AADT of I 0,000, CMFwro ~ !.!5 (see Table I 0-9).
For 4-ft gravel shoulders, CMF,ro ~ 1.01 (see Table 10-10).
The proportion of related crashes, pro, is 0.574 (see discussion below Equation 10-12).
CMF,, ~ (!.!5 X 1.01 - 1.0) X 0.574 + 1.0 ~ 1.09

CHAPTER 1D-PREDICTIVE METHOD FOR RURAL TWO-LANE, TWO-WAY ROADS 10-37
Horizontal Curves: Length, Radius, and Presence or Absence ofSpiral Transitions (CMF3)
Since the roadway segment in Sample Problem I is a tangent, CMF,, = 1.00 (i.e., the base condition for CMF 3, is
no curve).
Horizontal Curves: Superelevation (CMF4 )

Since the roadway segment in Sample Problem I is a tangent, and, therefore, has no superelevation, CMF 4, = 1.00.
Grade (CMF5)
From Table 10-11, for a two percent grade, CMF,, = 1.00
Driveway Density (CMF6)

The driveway density, DD, is 6 driveways per mile. CMF 6, can be calculated using Equation I 0-17 as follows:
CMF = 0.322 + DD X [0.05 - 0.005 x ln(AADT)]

6
' 0.322 + 5 X [0.05 - 0.005 x ln(AADT)]
= 0.322 + 6 x [0.05- 0.005 x In(IO,OOO)]
0.322 + 5 x [0.05- 0.005 x ln(!O,OOO)]
= 1.01
Centerline Rumble Strips (CMF7)

Since there are no centerline rumble strips in Sample Problem I, CMF 7, = 1.00 (i.e., the base condition for CMF,,
is no centerline rumble strips).
Passing Lanes (CMF,)

Since there are no passing lanes in Sample Problem I, CMF,, = 1.00 (i.e., the base condition for CMF,, is the
absence of a passing lane).
Two- Way Left-Turn Lanes (CMF9)

Since there are no two-way left-tum lanes in Sample Problem I, CMF 9, = 1.00 (i.e., the base condition for CMF,,
is the absence of a two-way left-tum lane).
Roadside Design (CMF10 )

The roadside hazard rating, RHR, in Sample Problem I is 4. CMF 10, can be calculated from Equation I 0-20 as follows:
e( --{).6869+ 0.066SxRHR)
CMFJo, = (--{).4865)
e
e( --{).6869 + 0.0668 x 4)
=
e(-0.4865)
= 1.07
Lighting (CMF11 )
Since there is no lighting in Sample Problem I, CMF 11, = 1.00 (i.e., the base condition for CMFw is the absence
of roadway lighting).
Automated Speed Enforcement (CMF12 )

Since there is no automated speed enforcement in Sample Problem I, CMF 12, = 1.00 (i.e., the base condition for
CMF J.?r is the absence of automated speed enforcement).
The combined CMF value for Sample Problem I is calculated below.
CMF<Omb = 1.17 X 1.09 X J.O] X 1.07 = 1.38

It is asswned a calibration factor, C,, of 1.10 has been determined for local conditions. See Part C, Appendix A.l
for further discussion on calibration of the predictive models.
Calculation ofPredicted Average Crash Frequency

The predicted average crash frequency is calculated using Equation I 0-2 based on the results obtained in Steps 9
through 11 as follows:
Npredictedrs = Nspfrs XC,. X (CMFlr X CMF2r X • •• X CMFJ;)
~ 4.008 x 1.10 x (1.38) ~ 6.084 crashes/year
WORKSHEETS
The step-by-step instructions above are provided to illustrate the predictive method for calculating the predicted
average crash frequency for a roadway segment. To apply the predictive method steps to multiple segments, a series
of five worksheets are provided for determining predicted average crash frequency. The five worksheets include:
• Worksheet SP JA (Corresponds to Worksheet 1A)-General Information and Input Data for Rural Two-Lane,
Two-Way Roadway Segments
• Worksheet SP 1B (Corresponds to Worksheet JB)-Crash Modification Factors for Rural Two-Lane, Two-Way
Roadway Segments
• Worksheet SP 1 C (Corresponds to Worksheet 1 C)-Roadway Segment Crashes for Rural Two-Lane, Two-Way
Roadway Segments
• Worksheet SP1D (Corresponds to Worksheet 1D)-Crashes by Severity Level and Collision Type for Rural
Two-Lane, Two-Way Roadway Segments
• Worksheet SP 1E (Corresponds to Worksheet 1E)-Summary Results for Rural Two-Laoe, Two-Way Roadway Segments
Details of these sample problem worksheets are provided below. Blank versions of corresponding worksheets are
provided in Chapter 10, Appendix 1OA.
Worksheet SP1A--Generallnformation and Input Data for Rural Two-Lane, Two-Way Roadway Segments
Worksheet SPlA is a summary of general information about the roadway segment, analysis, input data (i.e., "The
Facts"), and asswnptions for Sample Problem 1.
CHAPTER 1G--PREDICTIVE METHOD FOR RURAL TWO-LANE, TWO-WAY ROADS 10-39
Worksheet SP1A. General Information and Input Data for Rural Two-Lane, Two-Way Roadway Segments
General Information Location Information
Analyst ! Roadway
Agency or Company ! Roadway Section
Jurisdiction
Date Performed
i Analysis Year
Input Data Base Conditions I
Site Conditions
Length of segment, L (mi) - 1.5
AADT (veh/day) - 10,000
Laoe width (ft) 12 10
Shoulder width (ft) 6 ! 4
Shoulder type paved Gravel

Length of horizontal curve (mi) 0 not present
Radius of curvature (ft) 0 not present
Spiral transition curve not present not present

(present/not present)
Superelevation variance (ftlft) <0.01 not present
Grade(%) 0 2
Driveway density (driveways/mi) 5 6

Centerline rumble strips not present not present
Passing lanes not present not present
I
Two~way left-turn lane not present not present
Roadside hazard rating 3 4
(1-7 scale)
Segment lighting not present not present

Auto speed enforcement not present not present
Calibration factor, C, l.O 1.1
Worksheet SP1B-Crash Modification Factors for Rural Two-Lane, Two-Way Roadway Segments
In Step I 0 of the predictive method, crash modification factors are applied to account for the effects of site specific
geometric design and traffic control devices. Section 10.7 presents the tables and equations necessary for determin-
ing CMF values. Once the value for each CMF has been determined, all of the CMFs are multiplied together in
Column 13 ofWorksheet SPIB which indicates the combined CMF value.
Worksheet SP1 B. Crash Modification Factors for Rural Two-Lane, Two-Way Roadway Segments
CMF for , CMF for Shoulder I CMF for I CMF for I CMF for
Lane Width I Width and Type Horizontal Cunres ! Superelevation CMF for Grades Driveway Density
Worksheet SPJB continued

(7) ! (8) (9) (10) (11) (12) (13)
CMFfor
CMFfor CMFfor Automated
Centerline CMF for Two-Way CMF for C:MF for Speed
Rumble Strips 1
Passing Lanes Left-Turn Lane Roadside Design Lighting Enforcement Combined CMF
-~~!!.~ .. ~.~.!.?'...._,,_,,_,,_,,, -~~~.:~.b-"'"""""'""""

from Section from Section from Equation from Equation from Equation from Section (1)'(2)' ...
10.7.1 10.7.1 10-18 10-20 10-21 10.7.1 '(11)'(12)
---- -+ ------------ 1--------
1.00 1.00 1.00 1.07 1.00 1.00 1.38
Worksheet SP1C-Roadway Segment Crashes for Rural Two-Lane, Two-Way Roadway Segments
The SPF for the roadway segment in Sample Problem I is calculated using Equation 10-6 and entered into Column 2
ofWorksheet SPlC. The overdispersion parameter associated with the SPF can be entered into Column 3; however,
the overdispersion parameter is not needed for Sample Problem 1 (as the EB Method is not utilized). Column 4 of
the worksheet presents the default proportions for crash severity levels from Table 10-3. These proportions may be
used to separate the SPF (from Column 2) into components by crash severity level, as illustrated in Column 5. Col-
umn 6 represents the combined CMF (from Column 13 in Worksheet SPIB), and Column 7 represents the calibra-
tion factor. Column 8 calculates the predicted average crash frequency using the values in Column 5, the combined
CMF in Column 6, and the calibration factor in Column 7.
Worksheet SP1C. Roadway Segment Crashes for Rural Two-Lane, Two-Way Roadway Segments
(I) (2) I
(3) (4) (5) (6) (7) (8)
Predicted
Crash N'1'1,.by Average Crash
Crash Overdispersion Severity Severity Combined Calibration Frequency,
Severity Level N•r" Parameter, k Distribution Distribution CMFs Factor, C, Np.wlct«<"
from Equation from Equation from Table (2)total'(4) (13) from (5)'(6)'(7)
10-6 10-7 10-3 Worksheet
SP1B
! I
Total 4.008 !
0.16 ! 1.000 '
! 4.008 1.38 1.10 ! 6.084
Fatal and 0.321 1.287 1.38 1.10 1.954
injury (FI)
Property 0.679 2.721 1.38 1.10 4.131
damage only
(PDO)
CHAPTER 1~PREDICTIVE METHOD FOR RURAL TWO-LANE, TWO-WAY ROADS 10-41
Worksheet SP1 ll--Crashes by Severity Level and Collision for Rural Two-Lane, Two-Way Roadway Segments
Worksheet SPID presents the default proportions for collision type (from Table 10-4) by crash severity level as follows:
• Total crashes (Column 2)
• Fatal-and-injury crashes (Column 4)
• Property-damage-only crashes (Column 6)
Using the default proportions, the predicted average crash frequency by collision type is presented in Columns 3
(Total), 5 (Fatal and Injury, FI), and 7 (Property Damage Only, PDQ).
These proportions may be used to separate the predicted average crash frequency (from Column 8, Worksheet SPIC)
by crash severity and collision type.
Worksheet SP1 D. Crashes by Severity Level and Collision Type for Rural Two-Lane, Two-Way Roadway Segments
(I) (2) (3) (4) (5) (6) (7)
Proportion
of Collision N pR<IIC!Od ni (IOUOI) Proportion of Npm:llchld,.. (FI) Proportion of N P"'dlol«<"' (POD)
Type(rolal) I
(crashes/year) I Collision Type (FfJ i (crashes/year) : Collision TYPe rPDOJ I (crashes/year)
(8), 0,.1 from (8)F1from (S)PDO from
Collision Type from Table 10-4 I Worksheet SPIC from Table 10-4
I I I Worksheet SPIC I from Table 10-4 I Worksheet SPIC
Total I
1.000 I 6.084 I 1.000 I 1.954 1.000 4.131
!
(2)'(3)••, ' (4)'(5)" (6)'(7),00
SINGLE-VEHICLE
'
Collisron with 0.121 0.736 0.038 0.074 0.184 0.760
animal
- ------- -- --------
Collision with 0.002 0.012 0.004 0.008 0.001 0.004

bicycle
. _,_,_,,_,_,_,_,,_,_,_, ,_,,_,_ _,_,_,_, _,_,_,_,,_,_,,_,,_,_,_,,_ _,,_,_,,_,,_,_,,_,,_,_,_,_,,_,,_,,_
Collision with 0.003 O.Q18 0.007 0.014 0.001 0.004

pedestrian
Overturned 0.025 0.152 ' 0.037 0.072 O.Q15 0.062

Ran off road 0.521 3.170 0.545 1.065 0.505 2.086
Other single- 0.021 0.128 I
0.007 0.014 0.029 0.120
vehicle collision
'
Total single- 0.693 4.216 0.638 1.247 0.735 3.036

vehicle crashes
MULTIPLE-VEHICLE
Angle collision 0.085 0.517 0.100 0.195 0.072 0.297
~i~:: nj n:: mrm:: m:~nnnm~~mm ::.

collision ~
-0~38 1. '
········--- --- -· --·--··- ...
Other multiple-
vehicle collision
0.027 I 0.164
. -· _,_,_,_,_,_,_,_,_,_,_,_, __ ,_,_,,_,_,_,,_,_,_,_,_,_,_,_ ·- _,_,_,_,_,_,_,_,_,_,_,_,_,_,_,,_,_,_,_,_,_,_,
. 0.026
___
0 051 0.030 0.124
,_,_,_ .. _,,_,_._,_,_,_,_,_,_,_,,_,_,,_,_,,_,,_,_,,_,_,,_,_,_,_,_,_,_,_,_,_,_,:-·-"-"'-"""'-"'""'-"'_"_.,,_.. ,_.. _,_ .. _
Total multiple-
vehicle crashes
0.307
1
1.868
1
,
0.362 I' 0.707 ,
I
0.265 ·
I
1.095
• '1'
Worksheet SP1E-Summary Results or Rural Two-Lane, Two-Way Roadway Segments

Worksheet SPlE presents a summary of the results. Using the roadway segment length, the worksheet presents the
crash rate in miles per year (Column 5).
Worksheet SP1 E. Summary Results for Rural Two-Lane, Two-Way Roadway Segments
(I) (2) (3) (4) (5)
Crash Severity , Predicted Average Crash Roadway Segment Crash Rate

Crash Severity Level Distribution ! Frequency (crashes/year) Length (mi) (crashes/mi/year)
(4) from Worksheet SPIC (8) from Worksheet SP1C (3)1(4)
Total 1.000 6.084 1.5 4.1
Fatal and injury (FI) 0.321 1.954 1.5 1.3
Property damage only (PDO) 0.679 4.131 1.5 2.8
The Site/Facility
A rural two-lane curved roadway segment.
The Question
The Facts
• 0.1-mi length
• Curved roadway segment
• 8,000 veh/day
• 1% grade
• 1,2000-ft horizontal curve radius
• No spiral transition
• 0 driveways per mi
• Roadside hazard rating= 5
• 0.1-mi horizontal curve length
• 0.04 superelevation rate
Assumptions
Collision type distributions have been adapted to local experience. The percentage oftotal crashes representing
single-vehicle run-off-the-road and multiple-vehicle head-on, opposite-direction sideswipe, and same-direction
sideswipe crashes is 78 percent.
The calibration factor is assumed to be 1.1 0.
Design speed = 60 mph
Maximum superelevation rate, em.u = 6 percent
Results
in Sample Problem 2 is detennined to be 0.5 crashes per year (rounded to one decimal place).
Steps
Step 1 through 8
To determine the predicted average crash frequency of the roadway segment in Sample Problem 2, only Steps 9
The SPF for a single roadway segment can be calculated from Equation I 0-6 as follows:
Nspfrs =AADTx L x 365 x ]0-6 x e C-D 312l
= 8,000 x 0.1 x 365 x !0-6 x e C""· 312l = 0.214 crashes/year
Lane Width (CMF)

CMF ,, can be calculated from Equation 10-11 as follows:
CMF1, = (CMF,- 1.0) X Pro+ 1.0
For an 11-ft lane width and AADT of 8,000 veh!day, CMF" = 1.05 (see Table I 0-8)
The proportion of related crashes,p,, is 0.78 (see assumptions)
CMF1, = (1.05- 1.0) x 0.78 + 1.0 = 1.04
Shoulder Width and Type (CMF,)

CMF 2, can be calculated from Equation I 0-12, using values from Table 10-9, Table 10-10, and local data (p" = 0. 78)
as follows:
CMF2, = (CMFw, x CMF,,- 1.0) x P, + 1.0

For 2-ft shoulders andAADTof8,000 veh/day, CMFw, = 1.30 (see Table 10-9)
For 2-ft gravel shoulders, CMF,, = 1.01 (see Table 10-10)
The proportion of related crashes, P,•• is 0.78 (see assumptions)
CMF,, = (1.30 X 1.01- 1.0) X 0.78 + 1.0 = 1.24

Horizontal Curves: Length, Radius, and Presence or Absence of Spiral Transitions (CMF ) ,,
For a 0.1 mile horizontal curve with a 1,200 ft radius and no spiral transition, CMF can be calculated from Equa-
tion 10-13 as follows: ''
= 1.43
Horizontal Curves: Superelevation (CMF4)

CMF 4 , can be calculated from Equation 10-16 as follows:
CMF,, = 1.06 + 3 X (SV- 0.02)
For a roadway segment with an assumed design speed of60 mph and an assumed maximum superelevation (em.) of
six percent, AASHTO Green Book (1) provides for a 0.06 superelevation rate. Since the superelevation in Sample
Problem 2 is 0.04, the superelevation variance is 0.02 (0.06- 0.04).
CMF,, = 1.06 + 3 X (0.02- 0.02) = 1.06
Grade (CMF5)
From Table 10-11, for a one percent grade, CMF,, = 1.00.
Driveway Density (CMF6)

Since the driveway density, DD, in Sample Problem 2 is less than 5 driveways per mile, CMF6, = 1.00 (i.e., the base
condition for CMF6, is five driveways per mile. If driveway density is less than five driveways per mile, CMF6, is 1.00).
Centerline Rumble Strips (CMF7)

Since there are no centerline rumble strips in Sample Problem 2, CMF 7, = 1.00 (i.e., the base condition for CMF 7,
is no centerline rumble strips).
Passing Lanes (CMF,)

Since there are no passing lanes in Sample Problem 2, CMF 8 , = 1.00 (i.e., the base condition for CMF 8, is the
absence of a passing lane).
Two- Way Left-Turn Lanes (CMF9)

Since there are no two-way left-tum lanes in Sample Problem 2, CMF 9, = 1.00 (i.e., the base condition for CMF,,
is the absence of a two-way left-tum lane).
Roadside Design (CMF10)

The roadside hazard rating, RHR, is 5. Therefore, CMF 10, can be calculated from Equation 10-20 as follows:
e( -0.6869 + 0.0668xRHR)
CMFJo, = e(-0 .4865)
e( -0.6869 + 0.0668x 5)
=
e(-0.4865)
= 1.14
Lighting (CMF11 )
Since there is no lighting in Sample Problem 2, CMF11, = LOO (i.e., the base condition for CMF11 , is the absence of
roadway lighting).
Automated Speed Enforcement (CMF12 )

Since there is no automated speed enforcement in Sample Problem 2, CMF 12, = LOO (i.e., the base condition for
CMF 12, is the absence of automated speed enforcement).
The combined CMF value for Sample Problem 2 is calculated below.
CMF<Omb = 1.04 X L24 X 1.43 X LQ6 X 1.14 = 2.23
It is assumed that a calibration factor, C,, of 1.10 has been determined for local conditions. See Part C, AppendixA.l
Calculation of Predicted Average Crash Frequency

The predicted average crash frequency is calculated using Equation 10-2 based on the results obtained in Steps 9
= 0.214 x 1.10 x (2.23) = 0.525 crashes/year
WORKSHEETS
of five worksheets are provided for determining predicted average crash frequency. The five worksheets include:
• Worksheet SP2A (Corresponds to Worksheet !A)-General Information and Input Data for Rural Two-Lane,
Two-Way Roadway Segments
• Worksheet SP2B (Corresponds to Worksheet !B)-Crash Modification Factors for Rural Two-Lane, Two-Way
Roadway Segments
• Worksheet SP2C (Corresponds to Worksheet !C)-Roadway Segment Crashes for Rural Two-Lane, Two-Way
Roadway Segments
• Worksheet SP2D (Corresponds to Worksheet !D)-Crashes by Severity Level and Collision Type for Rural
Two-Lane, Two-Way Roadway Segments
• Worksheet SP2E (Corresponds to Worksheet !E)-Summary Results for Rural Two-Lane, Two-Way Roadway Segments
provided in Chapter lO,Appendix lOA.
Worksheet SP2A-Generallnformation and Input Data for Rural Two-Lane, Two-Way Roadway Segments
Worksheet SP2A is a summary of general information about the roadway segment, analysis, input data (i.e., "The
Facts"), and assumptions for Sample Problem 2.
Worksheet SP2A. General Information and Input Data for Rural Two-Lane, Two-Way Roadway Segments
General Information i' Location Information
::~=:_::jl:-::=-:-_::~=t~--~n= ~- -
1 Analysis Year
Input Data I Base Conditions 1 Site Conditions
~-e~~-~f_s:~:>~': ~-~~)- _______________________________ I__ ___ ____ =--_____ -----~- _________ -~ _1 ___________ _
::~J~~rt-
Shoulder width (ft)
---- --------------- --- - -------- :
1
------------ : . ---~~:---~~~==
Shoulder type
RaWusofc~ttrre(ft)
Spiral transition curve

Superelevation variance (ftlft) <0.01 0.02 (0.06--D.04)
Grade(%) 0
Driveway density (driveways/mi) 5 0
Centerline rumble strips not present not present
(presentfnot present)
Passing lanes not present not present
~~!-------------------------------------~~=+--=~-~~-~t--p-:-- -~~n~~
Segment lighting not present not present
Auto speed enforcement not present not present
Calibration factor, C, LO ]_]
Worksheet SP2B-Crash Modification Factors for Rural Two-Lane, Two-Way Roadway Segments
geometric design and traffic control devices. Section I 0.7 presents the tables and equations necessary for determin-
Column 13 of Worksheet SP2B which indicates the combined CMF value,
Worksheet SP2B. Crash Modification Factors for Rural Two-Lane, Two-Way Roadway Segments
(I) (2) (3) (4) (5) (6)
CMFfor CMF for Shoulder CMF for Horizontal i CMFfor CMFfor

Lane Width Width and Type I
Curves I Superelevation CMF for Grades Driveway Density
C:MF 1, CMF,, CMF,, C:MF4, CMF 5, CMF6,

from Equation 10-13
1.43
Worksheet SP2B continued
(7) (8) (9) (10) (11) (12) (13)
CMFfor
CMFfor CMFfor I
Automated
j
Centerline CMFfor Two-Way I CMFfor C:MF for Speed Combined
!
Rumble Strips Passing Lanes Left-Tum Lane Roadside Design Lighting Enforcement CMF
=. .
I
~. I ~. I ~.
......,_,______ -- _,_, ___ ,_,_,_______________,_,_----1---------·-·--.. --- ____ ,______ ,____ - ~. CMFI2r
- -··-
CMF ,_, -
__ __ _____
, ,
from Section from Section

1 from Equation from Equation from Equation from Section (1)*(2)* ...
10.7.1 10.7.1 10-18 10-20 10-21 10.7.1 *(11)'(12)
-- 1.00 ------- 1.00 T----J.o-o·---
1
114 · - - - · 1.00 ----

1.00
...... -------
2.23
Worksheet SP2C-Roadway Segment Crashes for Rural Two-Lane, Two-Way Roadway Segments
The SPF for the roadway segment in Sample Problem 2 is calculated using Equation 10·6 and entered into Column 2
of Worksheet SP2C. The overdispersion parameter associated with the SPF can be entered into Column 3; however,
the overdispersion parameter is not needed for Sample Problem 2. Column 4 of the worksheet presents the default
proportions for crash severity levels from Table 10-3 (as the EB Method is not utilized). These proportions may be
used to separate the SPF (from Column 2) into components by crash severity level, as illustrated in Column 5. Col·
umn 6 represents the combined CMF (from Column 13 in Worksheet SP2B), and Column 7 represents the calibra-
tion factor. Column 8 calculates the predicted average crash frequency using the values in Column 5, the combined
Worksheet SP2C. Roadway Segment Crashes for Rural Two-Lane, Two-Way Roadway Segments
(I) (2) (3) (4) (5) (6) (7) I (8)
Predicted
Crash N•pfn by Average Crash
Severity Level N., ,,.. Parameter, k Distribution Distribution CMFs Factor, C, N "'dktcd"'
from from Equation from Table (2)10\01"'(4) (13) from (5)'(6)*(7)

Equation 10-7 10-3 Worksheet
10-6 SP2B
- - - - - - -·- - -- _,_ ,_
Total 0.214 2.36 1.000 0.214 2.23 1.10 0.525
Fatal and - - 0.321 0.069 2.23 1.10 0.169

;njury (FI)
Property - - 0.679 0.145 2.23 1.10 0.356

damage only
(PDQ)
Worksheet SP2D--Crashes by Severity Level and Collision for Rural Two-Lane, Two-Way Roadway Segments
Worksheet SP2D presents the default proportions for collision type (from Table 10-3) by crash severity level as follows:

(Total), 5 (Fatal and Injury, FI), and 7 (Property Damage Only, PDO).
These proportions may be used to separate the predicted average crash frequency (from Column 8, Worksheet SP2C)
Worksheet SP2D. Crashes by Severity Level and Collision Type for Rural Two-Lane, Two-Way Roadway Segments
(I) (2) (3) (4) (5) (6) (7)
Proportion Proportion of
of Collision Npr<dlo<odn (lo~AI) I Proportion of Npftdlotedn(F/J Collision Type Np.,dlc1c'IIIS (PDO)
Collision Type Type(,or•ll (crashes/year) Collision Type rFIJ (crashes/year) (PDO)
(crashes/year)
1
from Table I 0-4 (8)to..J from I from Table 10-4 I (8)F1from I from Table 10-4 (8)PDO from
Worksheet SP2C Worksheet SP2C Worksheet SP2C
Total I 1.000 0.525 I 1.000 0.169 I 1.000 0.356
I <2>'<3>•• , 1 (4)'<5>" (6)'(7)'00

SINGLE-VEHICLE
Collision with 0.121 0.064 0.038 0.006 0.184 0.066
animal
Collision with 0.002 0.001 0.004 0.001 0.001 0.000

bicycle
Collision with 0.003 0.002 0.007 0.001 0.001 0.000

pedestrian
Overturned 0.025 0.013 0.037 0.006 O.DlS 0.005
Ran off road 0.521 0.274 0.545 0.092 0.505 0.180
---- +··---------- --1··················-------- -+----·---------+--·-··- --- ----
Other single- 0.021 0.011 0.007 0.001 0.029 0.010
vehicle collision
-i- ·-------- -- .....,.......... - -· ·---- -- .... -- .. ·----- ----+-- ···········--- ------- -·· · · · ·----- --····-·············-·
Total single- 0.693 0.364 0.638 0.108 0.735 0.262
vehicle crashes
MULTIPLE-VEHICLE
Angle collision 0.085 0.045 0.100 0.017 0.072 0.026
Head-on 0.016 0.008 0.034 0.006 0.003 0.001
collision
Rear-end 0.142 0.075

.. 1. 0.164 ___ +_____ 0.028 0.122
0.043
collision
Sideswipe
collision
··---
0.037 0.019
T --:-:-::-- --~-::::·------- 0.014
Other multiple- 0.027 0.014 0.011

vehiCle collision
Total multiple- 0.307 0.161 0.362 0.265 0.094
vehicle crashes
Worksheet SP2E-Summary Results for Rural Two-Lane, Two-Way Roadway Segments

Worksheet SP2E presents a summary of the results. Using the roadway segment length, the worksheet presents the
Worksheet SP2E. Summary Results for Rural Two-Lane, Two-Way Roadway Segments
(I) (2) (3) (4) (5)
Predicted Average
Cras h Severity Crash Frequency Roadway Segment Crash Rate
'
Crash Severity Level I Distribution (crashes/year) Length (mi) (crasbes/mi/year)
(4) from Worksheet SP2C I, (8) from Worksheet SP2C I (3)/(4)
I
'
Total LOOO 0.525 I 0.1 I
5.3
Property damage only 0.679 0.356 0.1 3.6
(PDQ) I I
10.12.3, Sample Problem 3
The Site/Facility
A three-leg stop-controlled intersection located on a rural two-lane roadway.
The Question
What is the predicted average crash frequency of the stop-controlled intersection for a particular year?
The Facts
• 3 legs
• Minor-road stop control
• No right-tum lanes on major road
• No lefr-turn lanes on major road

• 30-degree skew angle
• AADT of major road~ 8,000 veh/day
• AADT of minor road~ 1,000 veh/day
• Intersection lighting is present
Assumptions
• Collision type distributions used are the default values from Table I 0-6.
• The proportion of crashes that occur at night are not known, so the default proportion for nighttime crashes is assumed.
• The calibration factor is assumed to be 1.50.
Results
Using the predictive method steps as outlined below, the predicted average crash frequency for the intersection in
Sample Problem 3 is determined to be 2.9 crashes per year (rounded to one dedmal place).
Steps
Step 1 through 8
To determine the predicted average crash frequency of the intersection in Sample Problem 3, only Steps 9 through
II are conducted. No other steps are necessary because only one intersection is analyzed for one year, and the EB
Method is not applied.
The SPF for a single three-leg stop-controlled intersection can be calculated from Equation I 0-8 as follows:
N,pf3ST= exp[-9.86 + 0.79 X In(AADTm,) + 0.49 X In(AADTm 1J]
= exp[-9.86 + 0.79 x In(8,000) + 0.49 x In(!,OOO)] = 1.867 crashes/year
Step 10-Mnltiply the result obtained in Step 9 by the appropriate CMFs to adjust the estimated crash
Each CMF used in the calculation of the predicted average crash frequency of the intersection is calculated below:
Intersection Skew Angle (CMFn)

CMF 11 can be calculated from Equation I 0-22 as follows:
CMFJi = e (o.oo4 x $kewJ
The intersection skew angle for Sample Problem 3 is 30 degrees.
CMF 11 =e<0·004 x 30>= 1.13
Intersection Left-Turn Lanes (CMF,)

Since no left-turn lanes are present in Sample Problem 3, CMF21 = 1.00 (i.e., the base condition for CMF21 is the
absence ofleft-turn lanes on the intersection approaches).
Intersection Right-Turn Lanes (CMF3)

Since no right-turn lanes are present, CMF 3, = 1.00 (i.e., the base condition for CMF 31 is the absence of right-turn
lanes on the intersection approaches).
Lighting (CMF41 )
CMF 41 can be calculated from Equation I 0-24 using Table 10-15.
From Table 10-15, for a three-leg stop-controlled intersection, the proportion of total crashes that occur at night (see
assumption), P,,, is 0.26.
CMF,, = I - 0.38 X 0.26 = 0.90
CMFoomb = 1.13 X 0.90 = 1.02

CHAPTER 1()-PREDICTIVE METHOD FOR RURAL TWO-LANE, TWO-WAY ROADS 10-51
It is assumed that a calibration factor, C., of 1.50 has been detennined for local conditions. See Part C, Appendix A.l

The predicted average crash frequency is calculated using Equation I 0-3 based on the results obtained in Steps 9
tbrough 11 as follows:
= 1.867 x 1.50 x (1.02) = 2.857 crashes/year
WORKSHEETS
The step-by-step instructions above are the predictive method for calculating the predicted average crash frequency
for an intersection. To apply the predictive method steps to multiple intersections, a series of five worksheets are
provided for determining predicted average crash frequency. The five worksheets include:
• Worksheet SP3A (Corresponds to Worksheet 2A)-General Information and Input Data for Rural Two-Lane,
Two-Way Road Intersections
• Worksheet SP3B (Corresponds to Worksheet 2B)-Crash Modification Factors for Rural Two-Lane, Two-Way Road
Intersections
• Worksheet SP3C (Corresponds to Worksheet 2C)-Intersection Crashes for Rural Two-Lane, Two-Way Road
Intersections
• Worksheet SP3D (Corresponds to Worksheet 2D)-Crashes by Severity Level and Collision Type for Rural
Two-Lane, Two-Way Road Intersections
• Worksheet SP3E (Corresponds to Worksheet 2E)-Summary Results for Rural Two-Lane, Two-Way Road Intersections
provided in Chapter 10, Appendix 1OA.
Worksheet SP3A--Generallnformation and Input Data for Rural Two-Lane, Two-Way Road Intersections
Worksheet SP3A is a summary of general information about the intersection, analysis, input data (i.e., "The Facts"),
and assumptions for Sample Problem 3.
Worksheet SP3A. General Information and Input Data for Rural Two-Lane, Two-Way Road Intersections
General Information 1
Location Information
Analyst Roadway
Agency or Company Intersection
Date Performed
Input Data Base Conditions Site Conditions

Intersection type 3ST
(3ST, 4ST, 4SG)
(veblday) 8,000
(veblday) 1,000
Intersection skew angle (degrees) 0 30
Number of signalized or 0 0
Wlcontrolled approaches with a
left-tum lane (0, I, 2, 3, 4)
Number of signalized or 0 0
uncontrolled approaches with a
right-tum lane (0. I, 2, 3, 4)
Intersection lighting not present present
---·-------------··-····------·····-----········--·C·-···---····-·----------------+-----------------··----
Calibration factor, C, 1.0 !.50
Worksheet SP3B-Crash Modification Factors for Rural Two-Lane, Two-Way Road Intersections
In Step 10 of the predictive method, crash modification factors are applied to account for the effects of site specific
geometric design and traffic control devices. Section I 0. 7 presents the tables and equations necessary for determin-
Column 5 of Worksheet SP3B which indicates the combined CMF value.
Worksheet SP3B. Crash Modification Factors for Rural Two-Lane, Two-Way Road Intersections
(!) (2) (3) (4) (5)
CMF for Intersection C:MFfor

Skew Angle CMF for Left-Turn Lanes Lanes
Right~ Turn CMF for Lighting Combined CMF
CMF 11 C1v!F2, C:MF,1 CMF41 CMF
"•'
from Equations from Table 10-13 from Table 10-14 from Equation 10-24 (I )*(2)*(3)*( 4)
10-22 or 10-23
1.13 !.00 1.00 0.90 1.02
Worksheet SP3C-Intersection Crashes for Rural Two-Lane, Two-Way Road Intersections

The SPF for the intersection in Sample Problem 3 is calculated using Equation 10-8 and entered into Column 2 of
Worksheet SP3C. The overdispersion parameter associated with the SPF can be entered into Column 3; however, the
overdispersion parameter is not needed for Sample Problem 3 (as the EB Method is not utilized). Column 4 of the
worksheet presents the default proportions for crash severity levels from Table 10-5. These proportions may be used
to separate the SPF (from Column 2) into components by crash severity level, as illustrated in Column 5. Column 6
represents the combined CMF (from Column 13 in Worksheet SP3B), and Column 7 represents the calibration fac-
tor. Column 8 calculates the predicted average crash frequency using the values in Column 5, the combined CMF in
Column 6, and the calibration factor in Column 7.
Worksheet SP3C. Intersection Crashes for Rural Two-Lane, Two-Way Road Intersections
(1) (2) (3) (4) (5) (6) (7) (8)
Predicted
I Crash N,<pf 3ST, 4ST or 4SG Average Crash
Crash Overdispersion Severity by Severity Combined Calibration Frequency,
Severity Level I N,,of JS'T. 4ST or 4SG I
Parameter, k Distribution Distribution CMFs Factor, C1 I' Nplft!lct«<lm
from Equations from Section from Table (2),,,'(4) from (5) of (5)'(6)'(7)
10-8, 10-9, 10.6.2 10-5 Worksheet
or 10-10 SPJB
Total 1.867 0.54 1.000 1.867 1.02 I

1.50 2.857
I
Fatal and - - 0.415 0.775 1.02 1.50 1.186
injury (FI) ''
-"'"""""'"""''"""""""'"""-"'"
_,_,_ _,_- -- ------ - ~-- . -. -··-- ---L. __ -- ----- _
,,_,_,, _ .,,_,, _ . - - -· ,_,,_,,_,,_,_,_,_,_,_,_,_ -
Property - - 0.585 1.092 1.02 1.50 1.671

damage only
(PDQ) I '
I
'
Worksheet SP3D-Crashes by Severity Level and Collision for Rural Two-Lane, Two-Way Road Intersections
(Total), 5 (Fatal and Injury, Fl), and 7 (Property Damage Only, PDO).
Worksheet SP3D. Crashes by Severity Level and Collision Type for Rural Two-Lane, Two-Way Road Intersections
w I w w oo '
~ ~ 1
'
ro
I Proportion ' Proportion !
'
of Collision NprodJcJod/nt(IOtol) I Proportion of I N pmU«od /nt (PI) of Collision N p"'dlt(od lnt (PDO)
Collision Type Type(to!Bl) I
(crashes/year) '
Collision TyperFIJ I (crashes/year) Type(PDO! I
(crashes/year)
I
from Table 10-6 (8)10rat from from Table 10-6 (S)F, from from Table 10-6 (S)PDO from
I I
Worksheet SP3C I Worksheet SP3C I Worksheet SP3C
Total 1.000 I
2.857 1.000 I
!.186 1.000 1.671
'
I I (2)'(3).,., i I (4)'(5)n I (6)'(7),00
SINGLE-VEHICLE
Collision with 0.019 0.054 0.008 0.009 0.026 0.043
animal
Collision with 0.001 0.003 0.001 0.001 0.001 0.002

bicycle
---.
Collision with 0.001 0.003 0.001 0.001

o.oo1 --. · r· 0.002
pedestrian
Overturned 0.013 0.037 0.007 0.012

- ·-· t· .. -
Ran off road 0.244 0.697 0.247 0.413
Other single- 0.016 0.046 0.011 0.020 0.033
vehicle collision
- ""'""'""'""'""'"""""'""""'""""-'- -·-"""'""'""'""'"""""""'"""""""
Total single- 0.294 0.840 0.283 0.336 0.302 0.505

vehicle crashes i
MULTIPLE·VEIDCLE
Angle collision 0.237 0.677 0.275 0.326 0.210 0.351
---·· -- . -- - --+-·-----····-···- . - -- -+-- ---··-··-···- -··- -·+--- --------·· --+ ----- ----- -. -- ----- --··-·-··-····
Head-on collision 0.052 0.149 0.081 0.096 0.032 0.053
Rear-end 0.278 0.794 0.260 0.308 0.292 0.488

collision
. . . :.:. cc::=.:.::. ______ -+·---··----------;----- --------·--+------ --- --- -~----·-··-···-- -------+-------·--· ----1----- -- ---· -- - .
Sideswipe 0.097 0.277 0.051 0.060 0.131 0.219
collision
-·· · · · · -· ·-· ·-· -· ·-· ·-· -·-· -- -·-·-···-···-··-··-···-···-··-··-··----- +----- . --------+--·-···--- --- - ---+---·-······--· --- - - -+ ---- -- --- ---- ---- --- --- ----- -
Other multiple- 0.042 0.120 0.050 0.059 0.033 0.055
vehicle collision
Total multiple- 0.706

-------.. ----+-----------.. --------.---.
2.017 I 0.717 0.850 0.698 1.166
vehicle crashes
Worksheet SP3E-Summary Results for Rural Two-Lane, Two-Way Road Intersections

Worksheet SP3E presents a summary of the results.
Worksheet SP3E. Summary Results for Rural Two-Lane, Two-Way Road Intersections
(I) (2) (3)
Predicted Average Crash Frequency
Crash Severity Level Crash Severity Distribution (crashes/year)
(4) from Worksheet SP3C (8) from Worksheet SP3C
Total 1.000 2.857

Fatal and injury (Fl) 0.415 1.186
Property damage only (PDQ) 0.585 1.671

A four-leg signalized intersection located on a rural two-lane roadway.
The Question
What is the predicted average crash frequency of the signalized intersection for a particular year?
The Facts
• 4legs
• 1 right-tum lane on one approach
• Signalized intersection
• 90-degree intersection angle
• No lighting present
• AADT of major road= I 0,000 vehiday
• AADT of minor road= 2,000 vehiday
• !left-tum lane on each of two approaches
Assumptions
• Collision type distributions used are the default values from Table 10-6.
Results
Sample Problem 4 is determined to be 5.7 crashes per year (rounded to one decimal place).
Steps
Step 1 through 8
11 are conducted. No other steps are necessary because only one intersection is analyzed for one year, and the EB
The SPF for a signalized intersection can be calculated from Equation I 0-10 as follows:
N,pf<SG = exp[-5.!3 + 0.60 x In(AADT=;l + 0.20 x In(AADTm,)]
= exp[-5.13 + 0.60 x In(!O,OOO) + 0.20 x In(2,000)] = 6.796 crashes/year

Step 10-Multiply the result obtained in Step 9 by the appropriate CMFs to adjust the estimated crash frequency
for base conditions to the site specific geometric design and traffic control features.
Intersection Skew Angle (CMF)

The CMF for skew angle at four-leg signalized intersections is 1.00 for all cases.
Intersection Left-Turn Lanes (CMF2)

From Table 10-13 for a signalized intersection with left-tum lanes on two approaches, CMF _,,. ~ 0.67.
Intersection Right-Tum Lanes (CMF3i)

From Table 10-14 for a signalized intersection with a right-tum lane on one approach, CMF" ~ 0.96.
Lighting (CMF,)
Since there is no intersection lighting present in Sample Problem 4, CMF" ~ 1.00 (i.e., the base condition for CMF41
is the absence of intersection lighting).
CMFoomb ~ 0.67 X 0.96 ~ 0.64
It is assumed that a calibration factor, C,, of 1.30 has been determined for local conditions. See Part C, Appendix A.l
Calculation of Predicted Average Crash Frequency

The predicted average crash frequency is calculated using the results obtained in Steps 9 through II as follows:
~ 6.796 x 1.30 x (0.64) ~ 5.654 crashes/year
WORKSHEETS
for an intersection. To apply the predictive method steps to multiple intersections, a series of five worksheets are
provided for determining predicted average crash frequency. The five worksheets include:
• Worksheet SP4A (Corresponds to Worksheet 2A)--Genera1 Information and Input Data for Rural Two-Lane,
• Worksheet SP4B (Corresponds to Worksheet 2B)-Crash Modification Factors for Rural Two-Lane, Two-Way Road
Intersections
• WOrksheet SP4C (Corresponds to WOrksheet 2C)-Intersection Crashes for Rural Two-Lane, Two-Way Road Intersections
• Worksheet SP4D (Corresponds to Worksheet 2D)-Crashes by Severity Level and Collision for Rural Two-Lane,
• WOrksheet SP4E (Corresponds to WOrksheet 2E)-Summary Results for Rural Two-Lane, Two-Way Road Intersections
provided in Chapter 10, Appendix lOA.
Worksheet SP4A--General Information and Input Data for Rural Two-Lane, Two-Way Road Intersections
Worksheet SP4A is a swnmary of general information about the intersection, analysis, input data (i.e., "The Facts"),
Worksheet SP4A. General Information and Input Data for Rural Two-Lane, Two-Way Road Intersections
General Information I Location Information
Number of signalized or
uncontrolled approaches with a 0 2
left-tum lane (0, I, 2, 3, 4)
Number of signalized or
------·r· --.--------:----.-------
uncontrolled approaches with a
right-tum lane (0. I, 2, 3. 4)
______ 1-----
Intersection lighting
(present/not present) __ ________ n_o~~:~s~:~-- -· ______________n_o~ ~:e~enr______ _.
Calibration factor, C1 1 1.0 I 1.3
Worksheet SP4B-Crash Modification Factors for Rural Two-Lane, Two-Way Road Intersections
Worksheet SP4B. Crash Modification Factors for Rural Two-Lane, Two-Way Road Intersections
(1) (2) (3) (4) (5)
CMF for Intersection CMFfor CMFfor
Skew Angle Left-Turn Lanes Right- Turn Lanes CMF for Lighting Combined CMF
-- -·-"~"~'~'~!................
from Equations from Table I0-13 from Table I0-14 from Equation 10-24 (I )'(2)'(3)'(4)
10-22 orl0·23
1.00 0.67 0.96 1.00 0.64
Worksheet SP4C-Intersection Crashes for Rural Two-Lane, Two-Way Road Intersections

The SPF the intersection in Sample Problem 4 is calculated using Equation I 0-8 and entered into Column 2 of
Worksheet SP4C. The overdispersion parameter associated with the SPF can be entered into Column 3; however, the
worksheet presents the default proportions for crash severity levels from Table 10-5. These proportions may be used
to separate the SPF (from Column 2) into components by crash severity level, as illustrated in Column 5. Column 6
represents the combined CMF (from Column 13 in Worksheet SP4B), and Column 7 represents the calibration fac-
tor. Column 8 calculates the predicted average crash frequency using the values in Column 5, the combined CMF in
Worksheet SP4C. Intersection Crashes for Rural Two-Lane, Two-Way Road Intersections
(8)
'
'
I . Predicted
' !
Average
Crash N,,pfJST, JST, o• 4SG ' Crash
'
' I
Severity Level N.<J)( JST, 4ST. "' JSG
Parameter, k , Distribution Distribution '
I CMFs !
Factor, C1 N l"'dlctodlnl
from Equations from Section from Table (2)1olnt"'(4) I

from (5) of (5)*(6)'(7)
10-8. 10-9. 10.6.2 I' 10-5 I Worksheet
I
or 10-10 I
I
I SP4B
Total
Fatal and
;njury (FI)
I
6.796
-
o.n
-
1.ooo
0.340
6.796
2.311
--··--I·--·--·--·--·--_,,_,_,_____,___,___ ,,_, ___,___ ,, ______________,_______
o.64
0.64
uo
1.30 l
1 5.654
1.923
i -------
Property - ( - 0.660 4.485 0.64 1.30 . 3.732

damage only
(PDQ)
Worksheet SP4D---Crashes by Severity Level and Collision for Rural Two-Lane, Two-Way Road Intersections
(Total), 5 (Fatal and Injury, FI), and 7 (Property Damage Only, PDO).
Worksheet SP4D. Crashes by Severity Level and Collision Type for Rural Two-Lane, Two-Way Road Intersections
(1) (2) (3) (4) (5) (6) (7)
I Proportion
I Proportion of , Npffillorod In/ (toto I) Proportion of Npftdl«od /nt (FI) of Collision N pffille!od im (PDO)
Collision Type Collision Type 1,.,.0 ; (crashes/year) Collision Type (FJ! : (crashes/year) Type(PDOI (crashes/year)
from Table 10-6 (8),=1 from from Table 10-6 (8)FI from from Table 10-6 (8)PDO from
I Worksheet SP4C Worksheet SP4C Worksheet SP4C
Total 1.000 5.654 1.000 1.923 1.000 3.732

I
(2)'(3),~, (4)'(5)p; (6)'(7),DD
I
SINGLE-VEIDCLE
Collision with 0.002 0.011 0.000 0.000 0.003 0.011
animal
-·- ·- ·-·- .... _.. _,_,_,_,_ -··-"-"'""'-"'_"_"_"_"'-"'_"_"'-"'- -
,_,_, .......................,... _,_,,_,_,_ _,_,_,_,_,_,_,_,_,_,_,_,_ ·- - - ..-.............................................
Collision with 0.001 0.006 0.001 0.002 0.001 0.004
bicycle
-·---·----·---·-------··-- --~-~-- ~--· --
______ ,,,_
Collision with 0.001 0.006 0.001 0.002 0.001 0.004

pedestrian
Overturned 0.003 0.017 0.003 0.006 0.003 0.011

Ran off road 0.064 0.362 0.032 0.062 0.081 0.302
- - -· -' ...,... _,_
Other single- 0.005 0.028 0.003 0.006 0.018 0.067
vehicle collision
Total single- 0.076 0.430 0.040 0.077 0.107 0.399
vehicle crashes
MULTIPLE-VEHICLE
Angle collision I 0.274 !
1.549 0.336 0.646 0.242 0.903
!
Head-on collision 0.054 0.305 0.080 0.154 0.040 0.149
- -·-·- --- -··-· ·-·· --- "_,_,_,_,_
Rear-end collision 0.426 2.409 0.403 0.775 0.438 1.635
_,_,_,_,,_,_,_,,_,,_,,_,,_, __ .. - _,_,_,_ -- -·-·-·---
Sideswipe 0.118 0.667 0.051 0.098 0.153 0.571
collision
'
Other multiple- 0.052 0.294 0.090 0.173 0.020 0,075
vehicle
- - •.. .•....•
,. ,_, ____
collision__ ___ , , ,,_
-----·-..·-·-- ------~-------- - - - - - - - - ----
Total multiple- 0.924 5.224 0.960 1.846 0.893 I 3.333
vehicle crashes
I
Worksheet SP4E-Summary Results for Rural Two-Lane, Two-Way Road Intersections

Worksheet SP4E presents a swnmary of the results.
Worksheet SP4E. Swnmary Results for Rural Two-Lane, Two-Way Road Intersections
(1) (2) (3)
Crash Severity Level Crash Severity Distribution Predicted Average Crash Frequency (crashes/year)
(4) from Worksheet SP4C (8) from Worksheet SP4C
Total 1.000 5.654
Fatal and injury (FI) 0.340 1.923
Property damage only (PDQ) 0.660 3.732
.. L.'
The Project
A project of interest consists of three sites: a rural two-lane tangent segment, a rural two-lane curved segment, and a
three-leg intersection with minor-road stop control. (This project is a compilation of roadway segments and intersec-
tions from Sample Problems I, 2, and 3.)
The Question
What is the expected average crash frequency of the project for a particular year incorporating both the predicted
average crash frequencies from Sample Problems I, 2, and 3 and the observed crash frequencies using the site-
specific EB Method?
The Facts
• 2 roadway segments (2U tangent segment, 2U curved segment)

• I intersection (3ST intersection)
• 15 observed crashes (2U tangent segment: 10 crashes; 2U curved segment: 2 crashes; 3ST intersection: 3 crashes)
Outline of Solution
To calculate the expected average crash frequency, site-specific observed crash frequencies are combined with
predicted average crash frequencies for the project using the site-specific EB Method (i.e., observed crashes are
assigned to specific intersections or roadway segments) presented in Section A.2.4 of Part C, Appendix A.
Results
The expected average crash frequency for the project is 12.3 crashes per year (rouoded to one decimal place).
WORKSHEETS
To apply the site-specific EB Method to multiple roadway segments and intersections on a rural two-lane, two-way
road combined, two worksheets are provided for determining the expected average crash frequency. The two work-
sheets include:
• Worksheet SP5A (Corresponds to Worksheet 3A)-Predicted and Observed Crashes by Severity and Site Type
Using the Site-Specific EB Method for Rural Two-Lane, Two-Way Roads and Multilane Highways
• Worksheet SP5B (Corresponds to Worksheet 3B)-Site-Specific EB Method Suouoary Results for Rural Two-Lane,
Two-Way Roads and Multilane Highways
provided in Chapter I 0, Appendix I OA.
Worksheets SPSA-Predicted and Observed Crashes by Severity and Site Type Using the Site-
Specific EB Method for Rural Two-Lane, Two-Way Roads and Multilane Highways
The predicted average crash frequencies by severity type determined in Sample Problems I through 3 are entered
into Coluoms 2 through 4 of Worksheet SP5A. Column 5 presents the observed crash frequencies by site type, and
Coluom 6 presents the overdispersion parameters. The expected average crash frequency is calculated by applying
the site-specific EB Method which considers both the predicted model estimate and observed crash frequencies for
each roadway segment and intersection. Equation A-5 from Part C, Appendix A is used to calculate the weighted
adjustment and entered into Coluom 7. The expected average crash frequency is calculated using Equation A-4 and
entered into Column 8. Detailed calculation of Coluoms 7 and 8 are provided below.
Worksheet SPSA, Predicted and Observed Crashes by Severity and Site Type Using the Site-Specific EB Method
for Rural Two-Lane, Two-Way Roads and Multilane Highways
(I) (2) (3) I (4) (5) (6) (7) (8)
!
Expected
Weighted average crash
Adjustment, frequency,
1 Predicted Average Crash Frequency (crashes/year) ! w Nu M
Obsenred Equation
Crashes, A-5 from Equation A-4
Noboorwd Overdispersion Part C, from Part C,
Site Type Np.....tlct..:l.(l<ltal) I
N "'dlcl<d(PJ) N "'dlcl«l(PDOJ (crashes/year) 1
Parameter, k Appendix A Appendix A
ROADWAY SEGMENTS
Segment 1
·---·----
Segment 2
INTERSECTIONS
6.084
0.525
1.954
0.169 -+-
I
4.131
0.356
I 10
-----·-·-
2
0.16
2.36
_, __ ____
0.507
,
0.447
!
8.015
1.341
Intersection 1 2.857 1.186 1.671 I 3 0.54 I 0.393 I 2.944

Combined 9.466 3.309 I 6.158 15 - : - 12.300
(Sum of i
Column) I I
Column 7-Weighted Adjustment

The weighted adjustment, w, to be placed on the predictive model estimate is calculated using Equation A-5 from
Part C, Appendix A as follows:
w = -----,--:_____ _----;-
! +k X[ L
all study
Npredicted ]
years
Segment I
I
w= 0.507
I+ 0.16 x ( 6.084)
Segment 2
I
w= =0.447
I+ 2.36 x (0.525)
Intersection I
I
w= =0.393
I+ 0.54 X (2.857)
Column 8-ExpectedAverage Crash Frequency

The estimate of expected average crash frequency, N expecle,, is calculated using Equation A -4 from Part C, Appendix
A as follows:
Nexpected =wxNpredicted +(1-w)xNobserved
•• L...
Segment I
N~"''"" = 0.507 X 6.084 +(I- 0.507) X 10 = 8.015
Segment 2
N~P""' = 0.447 X 0.525 +(I- 0. 447) X 2 = 1.341
Intersection I
N~"''"" = 0.393 X 2.857 +(I- 0. 393) X 3 = 2.944
Worksheet SPSB-Site-Specific EB Method Summary Results for Rural Two-Lane, Two-Way Roads
and Multilane Highways
Worksheet SP5B presents a summary of the results. The expected average crash frequency by severity level is calcu-
lated by applying the proportion of predicted average crash frequency by severity level to the total expected average
crash frequency (Column 3).
Worksheet SPSB. Site-Specific EB Method Summary Results for Rural Two-Lane, Two-Way Roads and
Multilane Highways
(1) (2) (3)
Crash Severity Level
Total
1- -·~~2~.~k-~.~~"~?.~~"~~:.:~."~~~-~""""'"
9.466
1.-..--· _,_,~~)~~~.~~.~-~~:~~~.:.:~-.~~5-~- " " "'" " "'" "'12.3
Fatal and injury (FI) (3)"'mb from Worksheet SPSA

3.309 4.3
- - T -, - .. _,_ ............................................................,_,,_,_,,_,,, ...................... ,_, .......... ,_,,_ - ·- ·- - -+ - · - - - - - - ,._ - - - - - - - - · - ·-"'""""""'""""""""""'-"'""""""'""''""""""""'"""""""""
Propeny damage only (PDO) 1- ___ _(4l.,,,[r_oll1'\;'~rksheetSP5~ __ . _--1- _.. _.. . !~)-"~-'!~)!~(2},~---- ___ _
1 6.rss s.o
The Project
A project of interest consists oftlrree sites: a rural two-lane tangent segment; a rural two-lane curved segment; and a
three-leg intersection with minor-road stop control. (This project is a compilation of roadway segments and intersec-
tions from Sample Problems 1, 2, and 3.)
The Question
average crash frequencies from Sample Problems I, 2, and 3 and the observed crash frequencies using the project-
level EB Method?
The Facts
• 2 roadway segments (2U tangent segment, 2U curved segment)

• 15 observed crashes (but no information is available to attribute specific crashes to specific sites within the project)
Outline of Solution
Observed crash frequencies for the project as a whole are combined with predicted average crash frequencies for the
project as a whole using the project-level EB Method (i.e., observed crash data for individual roadway segments and
intersections are not available, but observed crashes are assigned to a facility as a whole) presented in Section A.2.5
of Part C, Appendix A.
Results
The expected average crash frequency for the project is 11.7 crashes per year (rounded to one decimal place).
WORKSHEETS
To apply the project-level EB Method to multiple roadway segments and intersections on a rural two-lane, two-way
road combined, two worksheets are provided for determining the expected average crash frequency. The two work-
sheets include:
Using the Project-Level EB Method for Rural Two-Lane, Two-Way Roads and Multilane Highways
• Worksheet SP6B (Corresponds to Worksheet 4B)-Project-Level EB Method Summary Results for Rural
Two-Lane, Two-Way Roads and Multilane Highways
provided in Chapter I 0, Appendix I OA.
Worksheets SP6A-Predicted and Observed Crashes by Severity and Site Type Using the Project-
Level EB Method for Rural Two-Lane, Two-Way Roads and Multilane Highways
The predicted average crash frequencies by severity type determined in Sample Problems I through 3 are entered in
Columns 2 through 4 of Worksheet SP6A. Column 5 presents the total observed crash frequencies combined for all
sites, and Column 6 presents the overdispersion parameters. The expected average crash frequency is calculated by
applying the project-level EB Method which considers both the predicted model estimate for each roadway seg-
ment and intersection and the project observed crashes. Column 7 calculates NwD and Column 8 N. 1 • Equations A-1 0
through A-14 from Part C, Appendix A are used to calculate the expected average crash frequency of combined sites.
The results obtained from each equation are presented in Columns 9 through 14. Section A.2.5 in Part C, Appendix
A defines all the variables used in this worksheet. Detailed calculations of Columns 9 through 13 are provided below.
•. L. .•
10-64
r
'
HIGHWAY SAFETY MANUAL
Worksheet SP6A. Predicted and Observed Crashes by Severity and Site Type Using the Project-Level EB Method
(1) (2) (3) (4) (5) (6)
Predicted Average Crash Frequency (crashes/year)
Observed Crashes, Overdispersion
Site Type N "'dl«od (total) N ....dktffi(FJJ N "'dlcted (PDO) Nob••,..,.,d (crashes/year) Parameter, k
ROADWAY SEGMENTS
Segment 1 6.084 -·-- - I ____ -
0.16
Segment 2 0.525 I
0.169 0.356 2.36
INTERSECTIONS
Intersection 1 2.857 1.186 1.671 0.54
Combined (Sum of Column) 9.466 3.309 6.158 15
Worksheet SP6A continued
I (7) ! (8) I (9) (10) I (11) (12) I (13)
~dl«od...O Npro<ll<t<d wl I w, I N, w, N, I N••pe<:todlcnmb

Equation Equation !
'
'
A-8 A-9 Equation Equation Equation Equation Equation
Site Type (6)'(2)' sqrt((6)'(2)) A-10 A-ll I
A-12 I'
A-13 A-14
ROADWAY SEGMENTS
Segment 1
Segment 2
INTERSECTIONS
----+---------+-~~,----~
s.n2
0.651
o.987
1.113
I
+-----
Intersection 1 4.408 1.242
Combined (Sum of Column) i 10.981 3.342 0.463 0.739 10.910 11.674
Note: NP<•dlaed wo = Predicted number of total crashes assuming that crash frequencies are statistically independent
5 5 5 4 4
NpredictedwO = LkrmjN~j + LkrsjN;sj + Lkrq;N,241. + LkimjNi~j + LkisjN~j (A-8)

J=l J=l J=l j=l j=l
Np..dlmo .,.1 = Predicted number of total crashes assuming that crash frequencies are perfectly correlated
5 5 5 4 4
Npnxlictodwl = L,.~k,mjN,mj + L,.~k,.,jN,sj + L,.~k,-djNnij + L,.~kimjNimj + L,.~k~1 N~1 (A-9)

i=l i=I i=l 1=l i=l
Column 9-w0
The weight placed on predicted crash frequency under the assumption that crashes frequencies for different roadway
elements are statistically independent, w0 , is calculated using Equation A-1 0 from Part C, Appendix A as follows:
I
I + ----:c:N-'p"-r_.edi::"'::''::d:_:w::.:O:__
N predicted (total)
10.981
I+--
9.466
= 0.463
Column IO-N,
The expected crash frequency based on the assumption that different roadway elements are statistically independent,
N 0, is calculated using EquationA-l! from Part C, Appendix A as follows:
N0 ~ w0 X Nprcdicted(lotul) + (1 - w 0) X Nobserved(rot:l.l)
~ 0.463 X 9.466 + (1- 0.463) X ]5 ~ 12.438
Column 11-w1
elements are perfectly correlated, w" is calculated using Equation A-12 from Part C, Appendix A as follows:
WJ ~ --~-=---
l + __N..!Pe:"~di..,·c..,te00
d_::w"-I-
Npredicted (total)
~-_:_-
I
3.342
1+ - -
9.466
~ 0.739
Column 12-N1
The expected crash frequency based on the assumption that different roadway elements are perfectly correlated, N"
is calculated using Equation A-13 from Part C, Appendix A as follows:
N 1 ~ w 1 X Npn:dicted(torol) + (1 - w 1) X Nobserved(tol:!l)
~ 0.739 X 9.466 + (1- 0.739) X 15 ~ 10.9!0
Column 13-N i!XJN!~

,ewconr
"' ,
The expected average crash frequency based of combined sites, N expecteUJcom
.. ,, is calculated using Equation A-14 from
No+NI
NexpectedJcomb
2
12.438+10.910
2
~11.674
Worksheet SPGB-Project-Level EB Method Summary Results for Rural Two-Lane, Two-Way Roads
crash frequency (Column 3) .
.• ~. j
Worksheet SP6B. Project-Level EB Method Summary Results for Rural Two-Lane, Two-Way Roads and
Multilane Highways
(I) (2) (3)
Total (2tmb from Worksheet SP6A (13)comb from Worksheet SP6A

9.466 11.7
Fatal and injury (Fl) "_.<_32~~~f~o-~ -~~:~~:e_t ~~.~.~ _ (3).~ '(2>/(2)_."' --
- - - - - - - - - - - - - - - - - - - - - ---- 330~----
Property damage only (PDO) ~- ____ (4)~._, from ~:~he_et SP6~- ______
---------!- __________
4.!
(3)_.~'(~~"'(2)'~-------- __
1
10.13. REFERENCES
( I) AASHTO. A Policy on Geometric Design ofHighways and Streets. American Association of State and
Highway Transportation Officials, Washington, DC, 2004.
( 2) Elvik, R. and T. Vaa. The Handbook ofRoad Safety Measures. Elsevier Science, Burlington, MA, 2004.
( 3) FHWA. Interactive Highway Safety Design Model. Federal Highway Administration, U.S. Department of
Transportation, Washington, DC. Available from http://www.tfhrc.gov/safety/ihsdm/ihsdm.htm.
( 4) Griffin, L. I. and K. K. Mak. The Benefits to Be Achievedfrom Widening Rural, Two-Lane Farm-to-Market Roads
in Texas, Report No. IAC(86-87)- 1039, Texas Transportation Institute, College Station, TX, Aprill987.
( 5) Harwood, D. W, F. M. Council, E. Hauer, W E. Hughes, and A. Vogt. Prediction of the Expected Safety
Performance ofRural Two-Lane Highways, Report No. FHWA-RD-99-207. Federal Highway Administration,
U.S. Department ofTransportation, Washington, DC, December 2000.
( 6) Harwood, D. Wand A. D. St. John. Passing Lanes and Other Operational Improvements on Two-Lane High-
ways. Report No. FHWA/RD-85/028, Federal Highway Administration, U.S. Department ofTransportation,
Washington, DC, July 1984.
( 7) Hauer, E. Two- fiily Left-Turn Lanes: Review and Interpretation ofPublished Literature, unpublished, 1999.
( 8) Miaou, S-P. Vertical Grade Analysis Summary, unpublished, May 1998.
( 9) Muskaug, R. Accident Rates on National Roads, Institute of Transport Economics, Oslo, Norway, 1985.
( I 0) Nette!blad, P. Traffic Safety Effects ofPassing (Climbing) Lanes: An Accident Analysis Based on Data for
1972-1977, Medde!ande TU 1979-5, Swedish National Road Administration, BorHinge, Sweden, 1979.
( II) Rinde, E. A. Accident Rates vs. Shoulder Width, Report No. CA-DOT-TR-3147-1-77-01, California Depart-
ment of Transportation, Sacramento, CA, 1977.
( 12) Srinivasan, R., F. M. Council, and D. L. Harkey. Calibration Factors for HSM Part C Predictive Models. Unpub-
lished memorandum prepared as part of the Federal Highway Administration Highway Safety Information System
project. Highway Safety Research Center, University of North Carolina, Chapel Hill, NC, October, 2008.
( 13) Vogt, A. Crash Models for Rural Intersections: 4-Lane by 2-Lane Stop-Controlled and 2-Lane by 2-Lane
Signalized, Report No. FHWA-RD-99-128, Federal Highway Administration, October 1999.
( 14) Vogt, A, and l G. Bared. Accident Models for Two-Lane Rural Roads: Segments and Intersections, Report No.
FHWA-RD-98-133, Federal Highway Administration, Washington, DC, October 1998.
( 15) Vogt, A, and l G. Bared, Accident Models for Two-Lane Rural Segments and Intersection. In Transportation
Research Record 1635. TRB, National Research Council, Washington, DC, 1998.
( 16) Zegeer, C. V., R. C. Deen, and l G. Mayes. Effect of Lane and Shoulder Width on Accident Reduction on
Rural, Two-Lane Roads. In Transportation Research Record 806. TRB, National Research Board, Washington,
DC, 1981.
( 17) Zegeer, C. V., D. W Reinfurt, l Hummer, L. Herf, and W Hunter. Safety Effects of Cross-Section Design for
Two-Lane Roads. In Transportation Research Record 1195. TRB, National Research Council, Washington,
DC, 1988.
( 18) Zegeer, C. V., J. R. Stewart, F M. Council, D. W Reinfurt, and E. Hamilton Safety Effects of Geometric
Improvements on Horizontal Curves. Transportation Research Record 1356. TRB, National Research Board,
Washington, DC, 1992.
( 19) Zegeer, C., R. Stewart, D. Reinfurt, F Council, T. Neuman, E. Hamilton, T. Miller, and W Hunter. Cost-
Effective Geometric Improvements for Safety Upgrading of Horizontal Curves, Report No. FHWA-R0-90-021,
Federal Highway Administration, U.S. Department ofTransportation, Washington, DC, October 1991.
10A.1. APPENDIX 10A-WORKSHEETS FOR PREDICTIVE METHOD FOR RURAL

TWO-LANE, TWO-WAY ROADS
Worksheet 1A. General Information and Input Data for Rural Two-Lane, Two-Way Roadway Segments
General Information \ Location Information

Length of segment, L (mi)
AADT (veh!day)
Lane width (ft)
Shoulder width (ft)
Length of horizontal curve (mi) 0

Radius of curvature (ft) 0
-_-:._: -_:-_r:-~-=-:-,:n-~:'-~-v:-u-~rv~_':_c~"(-~-':-~-v_n_:_ot_.pc.r_es_en_t")_____. .-.. .-..--.·------.-..-.. _.._. _-..ll---.--._-_-_no___~- ':-~;- -~-en_t__-.,_.-____-,- -~-..--- . . -.. . ----~~------. -..-.. . ~_-.._-_-.. .-.. . -.. -=~--. .-. .-_
. _-_--
Grade(%) 0
Driveway density (driveways/mile) 5
Centerline rumble strips (present/not present) not present
Passing lanes (present/not present) not present
Two-way left-turn lane (present/not present):___ _ _ _ _ _ ---1---=no:::t~P::_"::'_:_'":::'

Roadside hazard rating (1-7 scale) 3
r---~
not present
--------------------------------------
Calibration factor, C, LO
Worksheet 1 B. Crash Modification Factors for Rural Two-Lane, Two-Way Roadway Segments
(I) (2) (3) (4) (5) (6)
CMF for I CMF for Shoulder I CMF for CMF for I CMF for Grades I CMF for
Lane Width 1 Width and Type Horizontal Cunres Superelevation 1 Driveway Density
Worksheet 1B continued
(7) ! (8) I
(9) (10) (11) (12) (13)
CMFfor
Centerline
CMFfor
Passing Lanes
I
~=~:~ I Roa~:~ ~:sign! j' i:ti~; I Auto~::o;peed I Combined CMF
Rumble Strips I
Lefc-Turn Lane ' Enforcement
Worksheet 1 C. Roadway Segment Crashes for Rural Two-Lane, Two-Way Roadway Segments
I I
(I) I
(2) I (3) (4) '
(5) I (6) I (7) (8)
'
I I I Predicted
Crash N_,p1 ,.. by I Average Crash
'
Severity Level I N_,.,,.. I
Parameter, k Distribution I
Distribution I
CMFs I
Factor, C, N prodkl<d rs
from I from from (2) •• ,'(4) (!3) from (5)*(6)*(7)

I Worksheet lB I
I Equation 10-6 Equation 10-7 I Table 10-3 I
I
Total i 1.000 I I
I I '
Fatal and '
- - 0.321 ' '
injury (FI)
_,,_,_,_,,_,,_,,_,,_,,_
Property
- - -· _,,_,,_,,_,_,,_,_ ·-·-
--1---- _,_,_,, ...... . - ---
damage only - - 0.679

'
(PDO) i
I i
'
Worksheet 1 D. Crashes by Severity Level and Collision Type for Rural Two-Lane, Two-Way Roadway Segments
w I w w oo ~ oo m
Proportion
Proportion of Np.,dlcl..t"' (toed)
Proportion of Np.,.dk1<d"' (FI) of Collision N predicted r>' (PDO)
Collision Type1totron (crashes/year) 1
Collision Type (FI! (crashes/year) TypefPDO! (crashes/year)
I
(S)total from (S)FI from (S)PDO from
Collision Type from Table 10-4 Worksheet 1C from Table 10-4 Worksheet 1C from Table 10-4 Worksheet 1 C
--------l--
Total 1.000
(2)'(3)•• ,
1.000
(4)'(5)"
1.000
--·---------
(6)'(7J,oo
SINGLE-VEIDCLE
Collision with 0.121 O.DJS 0.184
animal
- _..
, .. ,_,,_,,_,,_,,_
Collision with 0.002 0.004 0.001

bicycle
Collision with 0.003 0.007 0.001

pedestrian
Overturned 0.025 0.037 O.Q15

Ran off road 0.521 0.545 0.505
_,,_,,_ ---
-·- ,_, ............ ... ... _,,_,,_,,_,,_,,_,,_'-
Other single- 0.021 0.007 0.029

vehicle collision
Total single- 0.693 0.638 0.735

vehicle crashes
MULTIPLE-VEHICLE
-:!~:;;lh~IO~- -1- --- -~-~;~---- -1-------.------ -1---- -~~;;---- -1-------------- +---- -~ ~~;---- -1-----. ---------
;~~~;:-- - -1-- --01~2----1---------- -----~-----o.164-----l--- ----- -----r-----ol22 r- ------- --
~:; l :;
Total mulnple-
vehicle crashes
0.307
:mm mni m:::
1 1
0 362
~1
mu: n""::mn: nm • 0.265
Worksheet 1 E. Summary Results for Rural Two-Lane, Two-Way Roadway Segments

(I) (2) (3) I (4) (5)
Crash Severity Predicted Average Crash I Crash Rate

Distribution Frequency (crashes/year) (crashes/mi/year)
Roadway Segment
Crash Severity Level (4) from Worksheet lC (8) from Worksheet 1C Length (mi) (3)/(4)
I ' ,
Total
. ------------- ---------c-- ---- ---------------~--------- ----~------- ---- -- --
Fatal and injury (FI)
Property damage only (PDQ) 1 I , ,

Worksheet 2A, General Information and Input Data for Rural Two-Lane, Two-Way Road Intersections
Analyst Roadway
Jurisdiction
Date Performed
Analysis Year
Intersection type (3ST, 4ST, 4SG) I

f,j\_[)~e_•(vehlday) __ _____ ------------- _ _ _ __ _
AADTm,, (vehlday)
Intersection skew angle (degrees) 0
Nwnber of signalized or uncontrolled approaches with a left-tum lane 0
-_:~-----
(0, 1,2, 3, 4)
Number of signalized or uncontrolled approaches with a right-tum

lane (0, 1, 2, 3, 4)
0
--:-_j __- _____
" ........ ,_,,_,_,,_,,_ - - - _,_,_,,_,,_,,_,,_,,_,,_,,_,,_,,_,,_,,_,,_,,_,,_,,_,,_,,_,- - - - ..
_,_,,_,,_,,_,,_ ,, ...............
Intersection lighting (present/not present) not present
-----+-----__:_---~~------
Calibration factor, C 1 1.0
Worksheet 2B. Crash Modification Factors for Rural Two-Lane, Two-Way Road Intersections
(I) (2) (3) (4) (5)
CMF for Intersection C:MF for CMFfor CMF for Lighting Combined CMF
Skew Angle Left-Turn Lanes Right-Turn Lanes
CMF 21 C:NlF 41 CMFcomb
from Equations from Table 10-13 from Table 10-14 from Equation 10-24 (1 )'(2)'(3)'(4)
10-22 or10-23
.. - - L... _-
Worksheet 2C. Intersection Crashes for Rural Two-Lane, Two-Way Road Intersections
(!) (2) (3) (4) (5) (6) (7) (8)
Predicted
Crash N•pf3ST, .JST 0' JSG Average Crash
Severity Level N spf 3ST. JST or .JSG Parameter, k Distnbution Distnbution CMFs Factor, C1 Npl"<dkl<d lnt
'
, from Equations from Section from Table (2)... '(4) from (5) of (5)'(6)'(7)
10-8, 10-9, or 10.6.2 10-5 Worksheet 2B
10-10
Total
Fatal and
injury (FI)
Property
damage only
(PDO)
Worksheet 20. Crashes by Severity Level and Collision Type for Rural Two-Lane, Two-Way Road Intersections
(1) (2) I (3) (4) (5) (6) (7)
I I
Proportion of
' Collision N P"'dlctod lnr (<otal) Proportion of Npftdloted lm (FI) Proportion of N pmllc<ed lm (PDO)
Collision Type I l)'pe rtotall I
(crashes/year) , Collision Type (Ff) (crashes/year) Collision Type (PDQ) I
(crashes/year)
I from Table I 0-61 (8)10..1 from from Table 10-6 (8)F1from from Table 10-6 (8)PDO from
!
Worksheet 2C I
Worksheet 2C I
Worksheet 2C
Total ' 1.000 1.000 1.000
I I
(2)'(3),,., (4)'(5), i (6)'(7),00
SINGLE-VEHICLE
Collision with
. I I
animal
Collision with
bicycle
Collision with
pedestrian
Overturned
Ran off road

-------f.----·-------+-------
Other single-
vehicle collision
Total single-
vehicle crashes
MULTIPLE-VEHICLE
-~gl~ c~llr~:~n- _1
1
________ _
Head-on
colhston
----------
Rear-end
'
collision
Sideswipe
collision
Other multiple-
vehicle collision
--
Total multiple-
vehicle crashes
Worksheet 2E. Summary Results for Rural Two-Lane, Two-Way Road Intersections
(I) (2) (3)
Crash Severity Level Crash Severity Distribution I Predicted Average Crash Frequency (crashes/year)
Total
~--------~------------------_, _______________
Property damage only (PDQ)
Worksheet 3A. Predicted and Observed Crashes by Severity and Site Type Using the Site-Specific EB Method for
Rural Two-Lane, Two-Way Roads and Multilane Highways
(1) (2) (3) (4) (5) (6) (7) (8)
I I
Expected Average
Predicted Average Crash Frequency Weighted Crash Frequency,
(crashes/year) Adjustment, w N "ffiffi
Observed
Crashes, Equation A-5 Equation A-4
Noboo,.,.,.j Overdispersion from Part C, from Part C,
Sitel)'pe N ftdlctod (<OIBI) N R<ll<t<d (Ff) N "'dlcl<d(PDOJ
(crashes/year) Parameter, k Appendix A Appendix A
ROADWAY SEGMENTS
Segment 1 I
I
Segment 2 !
I
Segment 3 !
Segment 4 I
Segment 5 I
Segment 6 I I
Segment 7 I
Segment 8 I
lNTERSECTIONS
Intersection 1
Intersection 2
Intersection 3
Intersection 4
Intersection 5 I I
'
Intersection 6
Intersection 7
Intersection 8
Combined I - -
(Sum of Column) I
Worksheet 38. Site-Specific EB Method Swnmary Results for Rural Two-Lane, Two-Way Roads and
Multilane Highways
(1) (2) (3)
Total -----'(2~)'"',_"'"''-fr_o_m_W._o_rk_sh_e_e_t_JA
_ _ _ _~---'(-'8)co-'"'''-fr_om Worksheet 3A
Worksheet 4A. Predicted and Observed Crashes by Severity and Site Type Using the Project-Level EB Method for
Rural Two-Lane, Two-Way Roads and Multilane Highways
(I) (2) (3) (4) (5) (6) (7)
Observed Crashes,
Nobocrv<d Overdispersion Equation A-8
Site Type N mllotcd (<otol) N ,...dlcrod (Fli N ~lctcd (PDOI
(crashes/year) Parameter, k (6)'(2)'
ROADWAY SEGMENTS
Segment 1
Segment2
Segment 3
·----+------+------·----~--- -----------------------------
Segment 4
Segment 5
--~-:~:.::::.:.:.: . :. .---------------1----- --- ---- -1 -- -------------1----------- -;·-- --- ------------7-- --- ----- -- -I -- ----------
Segment 6
---- --------- --- - - --- - - -- 4---- ---------·f··-···-···-- - -- -- - - -- - - -·-···f··-···---- --- ---··-···-···f- - -- - - --- - - -·
Segment 7
Segment 8
11\"TERSECTIONS
Intersection l T
Intersection 2
-- --- -- --- .. --- . -- . --·--- --- ------------ ----- ---c ----- -- ----- -1---- ---- -·-- ------ --1- - --- ------- ---+--- ----- ----
Intersection 3
Intersection 4
-=:::.:~_:.:::_· __ -------·-----------------:1---------------L--- ------ ---1 ---------------- -1--- --------- - +------------
Intersection 5
Intersection 6
Intersection 7
Intersection 8 I
Combined (Sum of Column)
Worksheet 4A continued on next page

Worksheet 4A continued
(8) (9) (10) (11) (12) (13)
N, N e:rpe«odlcomb
Equation A-9 Equation
Site Type sqrt((6)*(2)) ! EquationA-10 I EquationA-11 ! EquationA-121 EquationA-13 A-14
ROADWAY SEGMENTS
Segment 1 .
Segment 5
Segment 6
Segment 7
Segment 8
INTERSECTIONS
Intersection 1
Intersection 2
Intersection 3
Intersection 4
- - - - - - - - -·- --·- _... - ... " I
Intersection 5
Worksheet 4B. Project-Level EB Method Summary Results for Rural Two-Lane, Two-Way Roads and
Multilane Highways
(1) (2) (3)
Crash Severity Level N,rn~tctec~

I
Total (2)comb from Worksheet 4A from Worksheet 4A
_,,_,_,,_ - - -' -· _,,_,,_,,""""""""'"""""'-""""'-
- -j
Fatal and injury (FI) (3)comb from Worksheet 4A (3).,"' 1"'(2)F/(2)totnl
----~~-------------+------~~~~-------
Property damage only (PDO) (4)oonrb from Worksheet 4A (3)10,. 1*(2)PDJ(2)10,. 1

,_,,_,,_,,_,,_,,_,,_,,_,,_,_,,_,,_,,_,,_,,_,_,_,_,_,_,_ - ,_,_,_,_,_,_,,_,,_,_,_,_,,_ ,_,_,_,,_,,_,,_,_,,_,,_,_,,_,,_,_,_, __ ..
,_,_,,_ ,_,_,_ ·- - -·-
I
Chapter 11-Predictive Method
"~ ,_;y ,/,
for Rural Multilane Highways ,>~:;.,/ ·: \
11.1. INTRODUCTION
This chapter presents for the predictive method for rural multilane highways. A general introduction to the Highway
Safety Manual (HSM) predictive method is provided in the Part C-Introduction and Applications Guidance.
The predictive method for rural multilane highways provides a structured methodology to estimate the expected
average crash frequency, crash severity. and collision types for a rural multilane highway facility with known
characteristics. All types of crashes involving vehicles of all types, bicycles, and pedestrians are included, with the
exception of crashes between bicycles and pedestrians. The predictive method can be applied to existing sites, design
alternatives to existing sites, new sites, or for alternative traffic volume projections. An estimate can be made for
crash frequency in a period of time that occurred in the past (i.e., what did or would have occurred) or in the future
(i.e., what is expected to occur). The development of the predictive models in Chapter 11 is documented in Lord et
al. (5). The CMFs used in the predictive models have been reviewed and updated by Harkey et al. (3) and in related
work by Srinivasan et al. (6). The SPF coefficients, default collision type distributions, and default nighttime crash
proportions have been adjusted to a consistent basis by Srinivasan et al. (7).
This chapter presents the following information about the predictive method for rural multilane highways:
• The definitions of the facility types included in Chapter II and site types for which predictive models have been
developed for Chapter II.
• Details for dividing a rural multilane facility into individual sites, consisting of intersections and roadway
segments.
• Safety performance functions (SPFs) for rural multilane highways.

• Guidance for application of the Chapter II predictive method and limitations of the predictive method specific to
Chapter 11.
• Sample problems illustrating the application of the Chapter II predictive method for rural multilane highways.

The predictive method provides an 18-step procedure to estimate the "expected average crash frequency," Nexpected
(by total crashes, crash severity, or collision type), of a roadway network, facility, or site. In the predictive method
11-1
the roadway is divided into individual sites, which are homogenous roadway segments and intersections. A facility
consists of a contiguous set of individual intersections and roadway segments, referred to as "sites." Different facility
types are determined by surrounding land use, roadway cross-section, and degree of access. For each facility type, a
number of different site types may exist, such as divided and undivided roadway segments, and signalized and unsig-
nalized intersections. A roadway network consists of a number of contiguous facilities.
of all sites used as the estimate for an entire facility or network. The estimate is for a given time period of interest (in
years) during which the geometric design and traffic control features are unchanged and traffic volumes are known
or forecasted. The estimate relies on estimates made using predictive models which are combined with observed
crash data using the Empirical Bayes (EB) Method.
The predictive models used in Chapter 11 to determine the predicted average crash frequency, Npre,.!cte,, are of the
general form shown in Equation 11-1.
Npredicled = N.I'P/X X ( CMFJ:r; X CMF2:t X ... X CMFy) X c;r; (!!-!)

Where:
N predicted ~ predicted average crash frequency for a specific year on site type x;
N,Pf• ~ predicted average crash frequency determined for base conditions of the SPF developed for site type x;
CMFY• ~ crash modification factors specific to site type x and specific geometric design and traffic control features
y; and
C, ~ calibration factor to adjust SPF for local conditions for site type x.
11.3. RURAL MULTILANE HIGHWAYS-DEFINITIONS AND PREDIGIVE MODELS IN CHAPTER 11

This section provides the definitions of the facility and site types and the predictive models for each the site types
in Section 11.4.
11.3.1. Definition of Chapter 11 Facility and Site Types

Chapter 11 applies to rural multilane highway facilities. The term "multilane" refers to facilities with four through
lanes. Rural multilane highway facilities may have occasional grade-separated interchanges, but these are not to be
the primary form of access and egress. The predictive method does not apply to any section of a multilane highway
within the limits of an interchange which has free-flow ramp terminals on the multilane highway of interest. Facili-
ties with six or more lanes are not covered in Chapter 11
The terms "highway" and "road" are used interchangeably in this chapter and apply to all rural multilane facilities
independent of official state or local highway designation.
ClassifYing an area as urban, suburban, or rural is subject to the roadway characteristics, surrounding population and
Highway Administration (FHWA) guidelines which classifY "urban" areas as places inside urban boundaries where
the population is greater than 5,000 persons. "Rural" areas are defined as places outside urban areas which have a
population less than 5,000 persons. The HSM uses the term "suburban" to refer to outlying portions of an urban
area; the predictive method does not distinguish between urban and suburban portions of a developed area.
Table 11-1 identifies the specific site types on rural multilane highways for which predictive models have been devel-
oped for estimating expected average crash frequency, severity, and collision type. The four-leg signalized intersection
models do not have base conditions and, therefore, can be used only for generalized predictions of crash frequencies.
CHAPTER 11-PREDICTIVE METHOD FOR RURAL MULTILANE HIGHWAYS 11-3
No predictive models are available for roadway segments with more than four lanes or for other intersection types such
as all-way stop-controlled intersections, yield-controlled intersections, or uncontrolled intersections.
Table 11-1. Rural Multilane Highway Site Type with SPFs in Chapter II
Roadway Segmen~ Rural four-lane undivided segments (4U)

Rural four-lane divided segments (4D)
Intersections Unsignalized three-leg (Stop control on minor-road approaches) (3ST)

Unsignalized four-leg (Stop control on minor-road approaches) (4ST)
Signalized four-leg (4SG)'
• The four-leg signalized intersection models do not have base conditions and, therefore, can be used only for generalized predictions of
crash frequency.

• Undivided four-lane roadway segment (4U)-a roadway consisting of four lanes with a continuous cross-section
which provides two directions of travel in which the lanes are not physically separated by either distance or a
barrier. While multilane roadways whose opposing lanes are separated by a flush median (i.e., a painted median)
are considered undivided facilities, not divided facilities, the predictive models in Chapter II do not address rural
multilane highways with flush separators.
• Divided four-lane roadway segment (4D)-Divided highways are non-freeway facilities (i.e., facilities without
full control of access) that have the lanes in the two directions of travel separated by a raised, depressed, or flush
median which is not designed to be traversed by a vehicle; this may include raised or depressed medians with or
without a physical median barrier, or flush medians with physical median barriers.
• Three-leg intersection with stop control (3ST)-an intersection of a rural multilane highway (i.e., four lane divided or
undivided roadway) and a minor road. A stop sign is provided on the minor-road approach to the intersection only.
• Four-leg intersection with stop control (4ST)-an intersection of a rural multilane highway (i.e., four lane divided or
undivided roadway) and two minor roads. A stop sign is provided on both minor-road approaches to the intersection.
• Four-leg signalized intersection (4SG)-an intersection of a rural multilane highway (i.e., four lane divided or un-
divided roadway) and two other rural roads which may be two lane or four lane rural highways. Signalized control
is provided at the intersection by traffic lights.
11.3.2. Predictive Models for Rural Multilane Roadway Segments

The predictive models can be used to estimate total crashes (i.e., all crash severities and collision types) or can be used
to estimate the expected average frequency of specific crash severity types or specific collision types. The predictive
model for an individual roadway segment or intersection combines a SPF with CMFs and a calibration factor.
The predictive models for roadway segments estimate the predicted average crash frequency of non-intersection-
related crashes. In other words, the roadway segment predictive models estimate crashes that would occur regardless
of the presence of an intersection.
The predictive models for undivided roadway segments, divided roadway segments and intersections are presented in
Equations 11-2, 11-3, and 11-4 below.
For undivided roadway segments the predictive model is:
(I 1-2)
For divided roadway segments the predictive model is:
Npredicted rs = N~pfrd XC,. X (CMF1rd X CMF2rd X ..• X CMF5,.) (11-3)

Where:
Npredictcd rs = predictive model estimate of expected average crash frequency for an individual roadway
segment for the selected year;
Nspfrv expected average crash frequency for an undivided roadway segment with base conditions;
c, calibration factor for roadway segments of a specific type developed for a particular jurisdiction
or geographical area;
CMF 1 ~ ••• CMF,~ = crash modification factors for undivided roadway segments;
expected average crash frequency for a divided roadway segment with base conditions; and
CMFM .. CMF,,. = crash modification factors for divided roadway segments.
11.3.3. Predictive Models for Rural Multilane Highway Intersections

The predictive models for intersections estimate the predicted average crash frequency of crashes within the limits
of an intersection, or crashes that occur on the intersection legs, and are a result of the presence of the intersection
(i.e., intersection-related crashes).
For all intersection types in Chapter II the predictive model is:
Npredicted inr = Nspfinl X C, X ( CMFJi X CMF2i X ''. X CMF4) (11-4)
Where:
Npredictc.d In/
= predicted average crash frequency for an individual intersection for the selected year;
= predicted average crash frequency for an intersection with base conditions;

CMF1r .. CMF41 = crash modification factors for intersections-however, these CMFs are only applicable to three-
and four-leg stop-controlled intersections. No CMFs are available for four-leg signalized intersec-
tions; and
c, calibration factor for intersections of a specific type developed for use for a particular jurisdiction
of geographical area.
The SPFs for rural multilane highways are presented in Section 11.6. The associated CMFs for each of the SPFs are
presented in Section 11.7, and summarized in Table il-10. Only the specific CMFs associated with each SPF are appli-
cable to that SPF (as these CMFs have base conditions which are identical the base conditions of the SPF). The calibra-
tion factors, C, and C1, are determined in Part C, Appendix A.!.!. Due to continual change in the crash frequency and
severity distributions with time, the value of the calibration factors may change for the selected year of the study period.
11.4. PREDICTIVE METHOD FOR RURAL MULTILANE HIGHWAYS

The predictive method for rural multilane highways is shown in Figure Il-l. Applying the predictive method yields
an estimate of the expected average crash frequency (and/or crash severity and collision types) for a rural multilane
highway facility. The components of the predictive models in Chapter II are determined and applied in Steps 9, 10,
and II of the predictive method. Further information needed to apply each step is provided in the following sections
and in Part C, Appendix A.
There are 18 steps in the predictive method. In some situations, certain steps will not be needed because the data is
not available or the step is not applicable to the situation at hand. In other situations, steps may be repeated if an es-
timate is desired for several sites or for a period of several years. In addition, the predictive method can be repeated
as necessary to undertake crash estimation for each alternative design, traffic volume scenario or proposed treatment
option (within the same period to allow for comparison).
The following explains the details of each step of the method as applied to rural multilane highways.
Step 1 I Define roadway limits and facility type
+
Step 2 I Define the period of study. I
~
Step 3
I Determine AADT and availability of crash data for
every year in the period of interest.
Step 4
.
Determine geometric conditions
~
Step 5
I Divide roadway Into individual roadway
. segments and i11tersectlons.
I
_±
Step 6 I Assign observed crashes to Individual sites (if applicable). I
_±
Step 7 ----->! Select a roadway segment or intersection. I
_±
Step 8 c+l Select first or next year of the evaluation period.
I
_±
Step 10
+
Apply CMFs.
+
Step 11 I Apply a calibration factor
I
~
YES
Step 12 1
Step 13 I Apply site-specific EB method {if applicable).

J
Step 14
Step 1S
YES
<$> 1
Apply project-level EB method (if applicable).

I
Step 16 I Sum all sites and years.
Is there
an alternative
design, treatment, YES
Step 17
or forecast AADT to
be evaluated?
Step 18 I Compare and evaluate results

I

11-6 HiGHWAY SAFETY MANUAL
Step 1-Define the limits ofthe roadway and facility types in the study network, facility, or site for which the
intersection or a homogeneous roadway segment. Sites may consist of a number of types, such as signalized and
unsignalized intersections. The definitions of a rural multilane highway, an intersection and roadway segments, and
the specific site types included in Chapter II are provided in Section 11.3.
The predictive method can be undertaken for an existing roadway, a design alternative for an existing, or a new road-
way (which may be either unconstructed or yet to experience enough traffic to have observed crash data).
The limits of the roadway of interest will depend on the nature of the study. The study may be limited to only one
specific site or a group of contiguous sites. Alternatively, the predictive method can be applied to a very long cor-
ridor for the purposes of network screening (determining which sites require upgrading to reduce crashes) which is
discussed in Chapter 4, Network Screening.

The predictive method can be undertaken for either a past or future period measured in years. Years of interest will
be determined by the availability of observed or forecast average annual daily traffic (AADT) volumes, observed
crash data, and geometric design data. Whether the predictive method is used for a past or future period depends
upon the purpose of the study. The period of study may be:
• A past period (based on observedAADTs) for:
design features, traffic control features, and traffic volumes are known.

• An existing roadway network, facility, or site for which alternative geometric design or traffic control features
• A new roadway network, facility, or site that does not currently exist, but is proposed for construction during
some future period.
Step 3-For the study period, determine the availability of annual average daily traffic volumes and, for an
existing roadway network, the availability of observed crash data to determine whether the EB Method is
applicable.
The SPFs used in Step 9 (and some CMFs in Step 10), include AADT volumes (vehicles per day) as a variable. For
a past period, the AADT may be determined by automated recording or estimated from a sample survey. For a future
period, the AADT may be a forecast estimate based on appropriate land use planning and traffic volume forecasting
models, or based on the assumption that current traffic volumes will remain relatively constant.
For each roadway segment, the AADT is the average daily two-way, 24-hour traffic volume on that roadway segment
in each year of the period to be evaluated selected in Step 8.
For each intersection, two values are required in each predictive model. These are the AADT of the major street,
AADTmOJ., and the two-way AADT of the minor street, AADTmm..
In Chapter II, AADT ma1. and AADT mm. are determined as follows: if the AADTs on the two major-road legs of an in-
tersection differ, the lar£!er
'-'
of the two AADT values are used for AADT moJ.. For a three-leg intersection, the AADT of
the minor-road leg is used for AADT m;,· For a four-leg intersection, the larger of the AADTs for the two minor-road
legs should be used for AADT m;,· If a highway agency lacks data on the entering traffic volumes, but has two-way
AADT data for the major and minor-road legs of the intersection, these may be used as a substitute for the entering
volume data. \Vhere needed, AADT 1a1al can be estimated as the sum of AADTmaj and AADTmin·
an estimate of AADT for each year of the evaluation period is interpolated or extrapolated, as appropriate. If there is
no established procedure for doing this, the following may be applied within the predictive method to estimate the
AADTs for years for which data are not available.
• If AADT data are available for only a single year, that same value is assumed to apply to all years of the before period
• If two or more years of AADT data are available, the AADTs for intervening years are computed by interpolation.
that first year.
If the EB Method is used (discussed below), AADT data are needed for each year of the period for which observed
crash frequency data are available. If the EB Method will not be used, AADT for the appropriate time period-past,
present, or future--determined in Step 2 are used.

Where an existiug site or alternative conditions to an existing site are being considered, the EB Method is used. The
EB Method is only applicable when reliable observed crash data are available for the specific study roadway net-
work, facility, or site. Observed data may be obtained directly from the jurisdiction's crash report system. At least
If observed crash data are not available, then Steps 6, 13, and 15 of the predictive method are not conducted. In this
case, the estimate of expected average crash frequency is limited to using a predictive model (i.e., the predicted aver-
age crash frequency).
Step 4-Determine geometric design features, traffic control features, and site characteristics for all sites in
the study network.
In order to determine the relevant data needs and to avoid unnecessary data collection, it is necessary to understand
the base conditions of the SPFs in Step 9 and the CMFs in Step I 0. The base conditions are defined in Section 11.6.1
and 11.6.2 for roadway segments and in Section 11.6.3 for intersections.
The following geometric design and traffic control features are used to select a SPF and to determine whether the
site specific conditions vary from the base conditions and, therefore, whether a CMF is applicable:
• Length of roadway segment (miles)
• Presence of median and median width (feet) (for divided roadway segments)
• Sideslope (for undivided roadway segments)

• Shoulder widths (feet)
• Presence oflighting
• Presence of automated speed enforcement
For each intersection in the study area, the following geometric design and traffic control features are identified:
• Type of traffic control (minor-road stop or signalized)
• Intersection skew angle (stop-controlled intersections)
• Presence ofleft-turn and right-turn lanes (stop-controlled intersections)
• Presence or absence of lighting (stop-controlled intersections)
Step 5-Divide the roadway network or facility nnder consideration into individnal homogenous roadway
segments and intersections, which are referred to as sites-
Using the information from Step I and Step 4, the roadway is divided into individuarsites, consisting of individual
individual intersections and homogenous roadway segments for use with the Chapter II predictive models are
provided in Section 11.5. When dividing roadway facilities into small homogenous roadway segments, limiting the
segment length to a minimum ofO.IO miles will minimize calculation efforts and not affect results.
Step 6-Assign observed crashes to the individnal sites (if applicable).

way segments or intersections are presented in Section A.2.3 of Appendix A to Part C.
are assigned to the intersection and used in the EB Method together with the predicted average crash frequency for
the intersection. Crashes that occur between intersections and are not related to the presence of an intersection are
assigned to the roadway segment on which they occur; such crashes are used in the EB Method together with the
Step 7-Select the first or next individual site in the study network If there are no more sites to be evaluated,
proceed to Step 15.
In Step 5, the roadway network within the study limits has been divided into a number of individual homogenous
sites (intersections and roadway segments).
number of crashes expected to occur over ail sites during the period of interest. If a crash frequency is desired (crash-
es per year), the total can be divided by the number of years in the period of interest.
Step S---For the selected site, select the first or next year in the period of interest. If there are no more years to
be evaluated for that site, proceed to Step 14.
segment or intersection because SPFs and some CMFs (e.g., lane and shoulder widths) are dependent onAADT,
Steps 9 through 13, described below, are repeated for each year of the evaluation period as part of the evaluation
of any particular roadway segment or intersection. The predictive models in Chapter 11 follow the general form
shown in Equation 11-1. Each predictive model consists of a SPF, which is adjusted to site specific conditions using
CMFs (in Step I 0) and adjusted to local jurisdiction conditions (in Step II) using a calibration factor (C). The SPFs,
CMFs and calibration factor obtained in Steps 9, I 0, and II are applied to calculate the predictive model estimate
of predicted average crash frequency for the selected year of the selected site. The SPFs available for rural multilane
highways are presented in Section 11.6
The SPF (which is a statistical regression model based on observed crash data for a set of similar sites) determines
the predicted average crash frequency for a site with the base conditions (i.e., a specific set of geometric design and
traffic control features). The base conditions for each SPF are specified in Section 11.6. A detailed explanation and
overview of the SPFs in Part Cis provided in Section C.6.3 of the Part C-Introduction and Applications Guidance.
The SPFs (and base conditions) developed for Chapter 11 are summarized in Table 11-2 in Section 11.6. For the
selected site, determine the appropriate SPF for the site type (intersection or roadway segment) and geometric and
traffic control features (undivided roadway, divided roadway, stop-controlled intersection, signalized intersection).
The SPF for the selected site is calculated using the AADT determined in Step 3 (or AADT.,1 and AADT ml' for inter-
sections) for the selected year.
Each SPF determined in Step 9 is provided with default distributions of crash severity and collision type (presented
in Section 11.6). These default distributions can benefit from being updated based on local data as part of the cali-
bration process presented in Part C, Appendix A. 1.1.
Step 10-Multiply the result obtained in Step 9 by the appropriate CMFs to adjust base conditions to site
specific geometric conditions and traffic control features.
In order to account for differences between the base conditions (Section 11.6) and the site specific conditions, CMFs
are used to adjust the SPF estimate. An overview ofCMFs and gnidance for their use is provided in Section C.6.4 of
the Part C-Introduction and Applications Guidance, including the limitations of current knowledge related to the
effects of simultaneous application of multiple CMFs. In using multiple CMFs, engineering judgment is required to
assess the interrelationships and/or independence of individual elements or treatments being considered for imple-
mentation within the same project.
All CMFs used in Chapter II have the same base conditions as the SPFs used in Chapter II (i.e., when the specific
site has the same condition as the SPF base condition, the CMF value for that condition is 1.00). Only the CMFs pre-
sented in Section 11.7 may be used as part of the Chapter 11 predictive method. Table 11-10 indicates which CMFs
are applicable to the SPFs in Section 11.6.
periods in the data sets. Calibration of the SPFs to local conditions will account for differences in the data set. A
calibration factor (C, for roadway segments or C1 for intersections) is applied to each SPF in the predictive method.
An overview of the use of calibration factors is provided in the Part C-Introduction and Applications Guidance,
Section C.6.5. Detailed guidance for the development of calibration factors is included in Part C, Appendix A. I.!.
Steps 9, 10, and II together implement the predictive models in Equations 11-2, 11-3, and 11-4 to determine pre-
dicted average crash frequency.
Step 12-If there is another year to be evaluated in the study period for the selected site, return to Step 8.

the Chapter II predictive model estimate of predicted average crash frequency, N'""""'' with the observed crash fre-
quency of the specific site, N'"''~"· This provides a more statistically reliable estimate of the expected average crash
frequency of the selected site.
In order to apply the site-specific EB Method, overdispersion parameter, k, for the SPF is used. This is in addition to
the material in Part C, Appendix A.2.4. The overdispersion parameter provides an indication of the statistical reliabil-
ity of the SPF. The closer the overdispersion parameter is to zero, the more statistically reliable the SPF. This param-
eter is used in the site-specific EB Method to provide a weighting toN""'"""' and N,..,~,,- Overdispersion parameters
are provided for each SPF in Section 11.6.
Apply the site-specific EB Method to a fUture time period, if appropriate.
The estimated expected average crash frequency obtained above applies to the time period in the past for which the
observed crash data were obtained. Section A.2.6 in Appendix A to Part C provides a method to convert the estimate
of expected average crash frequency for a past time period to a future time period.
Step 14--Ifthere is another site to be evaluated, return to Step 7, otherwise, proceed to Step 15.
This step creates a loop through Steps 7 to 13 that is repeated for each roadway segment or intersection within the
facility.
Step 15-Apply the project level EB Method (if the site specific EB Method is not applicable).
This step is only applicable to existing conditions when observed crash data are available but cannot be accurately
assigoed to specific sites (e.g., the crash report may identify crashes as occurring between two intersections, but is
provided in Part C, AppendixA.2.5.
Step 16-Sum all sites and years in the study to estimate total crash frequency.
lated using Equation 11-5:
Ntotal = Lall
N rs + Lall
Nint
segments (11-5)
Where:
N"~' ~ total expected nwnber of crashes within the limits of a rural two-lane, two-way road facility for the
period of interest. Or, the swn of the expected average crash frequency for each year for each site within
the defined roadway limits within the study period;
N~ ~ expected average crash frequency for a roadway segment using the predictive method for one specific
year; and
N,", ~ expected average crash frequency for an intersection using the predictive method for one specific year.
Equation 11-5 represents the total expected nwnber of crashes estimated to occur during the study period. Equation
11-6 is used to estimate the total expected average crash frequency within the network or facility limits during the
study period.
N Ntotai
total average =--
n (11-6)
Where:
Ntob.t averuse = total expected average crash frequency estimated to occur within the defined network or facility limits
n ~ nwnber of years in the study period.

Steps 3 through 16 of the predictive method are repeated as appropriate for the same roadway limits but for alterna-
tive conditions, treatments, periods of interest, or forecast AADTs.
Step 18---Evaluate and compare results.

provided with each SPF in Section 11.6. These default distributions can benefit from being updated based on local
data as part of the calibration process presented in Part C, Appendix A.!.

necessary to apply the predictive method steps on rural multilane roads. Detail regarding the procedure for determin-
ing a calibration factor to apply in Step II is provided in Part C, Appendix A. I. Detail regarding the EB Method,
which is applied in Steps 6, 13, and 15, is provided in Part C, Appendix A.2.
In Step 5 of the predictive method, the roadway within the defined roadway limits is divided into individual sites,
which are homogenous roadway segments and intersections. A facility consists of a contiguous set of individual
intersections and roadway segments, referred to as "sites." A roadway network consists of a nwnber of contiguous
facilities. Predictive models have been developed to estimate crash frequencies separately for roadway segments and
intersections. The definitions of roadway segments and intersections presented below are the same as those for used
in the FHWA Interactive Highway Safety Design Model (IHSDM) (2).
Roadway segments begin at the center of an intersection and end at either the center of the next intersection or where there
is a change from one homogeneous roadway segment to another homogenous segment. The roadway segment model es-
timates the frequency of roadway-segment-related crashes which occur in Region B in Figure 11-2. When a roadway seg-
ment begins or ends at an intersection, the length of the roadway segment is measured from the center of the intersection.
..l..
TT
Chapter II provides predictive models for stop-controlled (three- and foUl'-leg) and signalized (four-leg) intersec-
tions. The intersection models estimate the predicted average frequency of crashes that occur within the curbline
limits of an intersection (Region A of Figure 11-2) and intersection-related crashes that occur on the intersection legs
(Region B in Figure 11-2).
Segment Length
A All crashes that occur within th1s region are classif1ed as intersection crashes.
B Crashes 1n this region may be segment or intersection related, depending on

Figure 11-2. Definition of Segments and Intersections
with respect to characteristics such as traffic volumes, key roadway design characteristics, and traffic control fea-
tures. Figure 11-2 shows the segment length, L, for a single homogenous roadway segment occurring between two
intersections. However, it is likely that several homogenous roadway segments will occur ben:veen two intersections.
A new (unique) homogeneous segment begins at the center of an intersection or where there is a change in at least
one of the following characteristics of the roadway:
• Average annual daily traffic (vehicles per day)
• Presence of median and median width (feet)
The following rounded median widths are recommended before determining "homogeneous" segments:
Measured Median Width Rounded Median Width
1 ftto 14ft 10ft
15ftto24ft 20ft
25 ft to 34ft 30ft
35 ft to 44ft 40ft
45ftto54ft 50ft
55ftto64ft 60ft
65 ftto 74ft 70ft
75ftto84ft 80ft
85 ftto 94ft 90ft
95ft or more 100ft
• Sideslope (for undivided roadway segments)
• Shoulder type
• Shoulder width (feet)
For shoulder widths measures to a 0.1-ft level of precision or similar, the following rounded paved shoulder widths
are recommended before determining "homogeneous" segments:
Measured Shoulder Width Rounded Shoulder Width
0.5 ft or less Oft
7.6 ft or more 8ft or more
For lane widths measured to a 0.1-ft level of precision or similar, the following rounded lane widths are recommend-
ed before determining "homogeneous" segments:
Measured Lane Width Rounded Lane Width
9.2 ft or less 9ft or less

9.3 ftto 9.7 ft 9.5 ft
9.8 :ft to 10.2 ft 10ft
10.3 ft to 10.7 ft 10,5 ft
10.8 ft to 11.2 ft 11 ft
ll.Jfttoll.7ft 11.5ft
ll.Sftormore 12ft or more
• Presence of lighting
In addition, each individual intersection is treated as a separate site for which the intersection-related crashes are
estimated using the predictive method.
There is no minimum roadway segment length, L, for application of the predictive models for roadway segments.
However, as a practical matter, when dividing roadway facilities into small homogenous roadway segments, limiting
the segment length to a minimum of 0.10 miles will minimize calculation efforts and not affect results.
and intersections. Observed crashes that occur between intersections are classified as either intersection-related
or roadway-segment related. The methodology for assignment of crashes to roadway segments and intersections
for use in the site-specific EB Method is presented in Section A.2.3 in Appendix A to Part C.

In Step 9 of the predictive method, the appropriate safety performance functions (SPFs) are used to predict average
crash frequency for the selected year for specific base conditions. SPFs are regression models for estimating the pre-
dicted average crash frequency of individual roadway segments or intersections. Each SPF in the predictive method was
developed with observed crash data for a set of similar sites. The SPFs, like all regression models, estimate the value of a
dependent variable as a fimction of a set of independent variables. In the SPFs developed for the HSM, the dependent vari-
able estimated is the predicted average crash frequency for a roadway segment or intersection under base conditions, and
the independent variables are the AADTs of the roadway segment or intersection legs (and, for roadway segments,
the length of the roadway segment).
The predicted crash frequencies for base conditions are calculated from the predictive method in Equations 11-2,
11-3, and 11-4. A detailed discussion ofSPFs and their use in the HSM is presented in Chapter 3, Fundamentals,
Section 3.5.2 and the Part C-lntroduction and Applications Guidance, Section C.6.3.
Each SPF also has an associated overdispersion parameter, k. The overdispersion parameter provides an indication
of the statistical reliability of the SPF. The closer the overdispersion parameter is to zero, the more statistically reli-
able the SPF. This parameter is used in the EB Method discussed in Part C, Appendix A. The SPFs in Chapter II are
summarized in Table 11-2.
Table 11-2. Safety Performance Functions included in Chapter II

Chapter 11 SPFs for Rural Multilane Highways SPF Equations and Exhibits
Undivided rural four-lane roadway segments Equations 11-7 and 11-8, Table 11-3, Figure 11-3
Divided roadway segments Equations 11-9 and 11-10, Tables 11-4 and 11-5
Three- and four-leg stop-controlled intersections Equation 11-11 , Table 11-7
Four-leg signalized intersections Equations 11-11 and 11-12, Tables 11-7 and 11-8
11.6.1. Safety Performance Functions for Undivided Roadway Segments

The predictive model for estimatiog predicted average crash frequency on a particular undivided rural multilane roadway
segment was presented in Equation 11-2. The effect of traffic volume (AADT) on crash frequency is incorporated through
the SPF, while the effects of geometric design and traffic control features are incorporated through the CMFs.
The base conditions of the SPF for undivided roadway segments on rural multilane highways are:
• Shoulder width 6 feet
• Shoulder type Paved
• Sideslopes 1V: 7H or flatter
• Lighting None

The SPF for undivided roadway segments on a rural multilane highway is shown in Equation 11-7 and presented
graphically in Figure 11-3:
N = eCo + b )( ln(AADT) + /n(L)) (11-7)

spfro
Where:
N spfro ~ base total expected average crash frequency for a roadway segment;
AADT ~ annual average daily traffic (vehicles per day) on roadway segment;
L ~ length of roadway segment (miles); and
a, b = regression coefficients.
Guidance on the estimation of traffic volumes for roadway segments for use in the SPFs is presented in Step 3 of the
predictive method described in Section 11.4. The SPFs for undivided roadway segments on rural multilane highways
are applicable to the AADT range from zero to 33,200 vehicles per day. Application to sites with AADTs substan-
tially outside this range may not provide accurate results.
The value of the overdispersion parameter associated with Nspfru is detem:t.IDed as a function of segment length. The
closer the overdispersion parameter is to zero, the more statistically reliable the SPF. The value is determined as:
k~ I
e(o+ln(L))
(11-8)
Where:
k ~ overdispersion parameter associated with the roadway segment;
c = a regression coefficient used to determine the overdispersion parameter.
Table 11-3 presents the values of the coefficients used for applying Equations 11-7 and 11-8 to determine the SPF
for expected average crash frequency by total crashes, fatal-and-injury crashes, and fatal, injury and possible injury
crashes.
Table 11-3. SPF Coefficients for Total and Fatal-and-Injury Crashes on Undivided Roadway Segments (for use in
Equations 11-7 and 11-8)
Crash Severity Level a b c
4-lane total -9.653 1.176 1.675
4-lane fatal and injury -9.410 1.094 1.796
4-lane fatal and injury" -8.577 0.938 2.003
a Using the KABCO scale, these include only KAB crashes. Crashes with severity level C (possible injury) are not included
.l..
'T
15r----------------------------------------------------------
fatal and injury crashes

(el<Ciudlng only possible injuries)
o~~===:J 5,000 10,000 15,000 20,000

AADT (veh/day)
30,000 35,000
Figure 11-3. Graphical Form of the SPF for Undivided Roadway Segments (from Equation 11-7 and Table 11-3)
The default proportions in Table 11-3 are used to break down the crash frequencies from Equation 11-7 into specific
collision types. To do so. the user multiplies the crash frequency for a specific severity level from Equation 11-7 by
the appropriate collision type proportion for that severity level from Table 11-4 to estimate the number of crashes for
that collision type. Table 11-4 is intended to separate the predicted frequencies for total crashes (all severity levels
combined), fatal-and-injury crashes, and fatal-and-injury crashes (with possible injuries excluded) into components
by collision type. Table 11-4 cannot be used to separate predicted total crash frequencies into components by severity
level. Ratios for PDO crashes are provided for application where the user has access to predictive models for that
severity level. The default collision type proportions shown in Table 11-4 may be updated with local data.
There are a variety of factors that may affect the distribution of crashes among crash types and severity levels. To
account for potential differences in these factors between jurisdictions, it is recommended that the values in Table
11-4 be updated with local data. The values for total, fatal-and-injury, and fatal-and-injury (with possible injuries
excluded) crashes in this exhibit are used in the worksheets described in Appendix llA.
Undivided Roadway Segments
Proportion of Crashes by Collision Type and Crash Severity Level
Severity Level
Collision Type Total Fatal and Injury Fatal and Injury" PDO
Head-on 0.009 0.029 0.043 0.001
Sideswipe 0.098 0.048 0.044 0.120
Rear-end 0.246 0.305 0.217 0.220
Angle 0.356 0.352 0.348 0.358

Single 0.238 0.238 0.304 0.237
Other 0.053 0.028 0.044 0.064
a Using the KABCO scale, these include only KAB crashes. Crashes with severity level C (possible injury) are not included.
Appendix JIB presents alternative SPFs that can be applied to predict crash frequencies for selected collision types
for undivided roadway segments on rural multilane highways. Use of these alternative models may be considered
when estimates are needed for a specific collision type rather than for all crash types combined. It should be noted
that the alternative SPFs in Appendix liB do not address all potential collision types of interest and there is no as-
surance that the estimates for individual collision types would sum to the estimate for all collision types combined
provided by the models in Table II-3.
11.6.2. Safety Performance Functions for Divided Roadway Segments

The predictive model for estimating predicted average crash frequency on a particular divided rural multilane road-
way segment was presented in Equation 11-3 in Section I 1.3. The effect of traffic volume (AADT) on crash frequen-
cy is incorporated through the SPF, while the effects of geometric design and traffic control features are incorporated
through the CMFs. The SPF for divided rural multilane highway segments is presented in this section. Divided rural
multilane highway roadway segments are defined in Section 11.3.
Some divided highways have two roadways, built at different times, with independent alignments and distinctly
different roadway characteristics, separated by a wide median. In this situation, it may be appropriate to apply the
divided highway methodology twice, separately for the characteristics of each roadway but using the combined traf-
fic volume, and then average the predicted crash frequencies.
The base conditions for the SPF for divided roadway segments on rural multilane highways are:
• Right shoulder width 8 feet
• Median width 30 feet

• Lighting None
The SPF for expected average crash frequency for divided roadway segments on rural multilane highways is shown
in Equation 11-9 and presented graphically in Figure 11-4:
N = e(a + b ~ ln(AADT) + Jn(L)) (11-9)

spfrd
\Vhere:
N,Pf" ~ base total number of roadway segment crashes per year;

AADT ~ annual average daily traffic (vehicles/day) on roadway segment;
predictive method described in Section 11.4. The SPFs for undivided roadway segments on rural multilane highways
are applicable to the AADT range from zero to 89,300 vehicles per day. Application to sites with AADTs substan-
tially outside this range may not provide reliable results.
The value of the overdispersion parameter is determined as a function of segment length as:
I
k
e(c+In(L))
(11-10)
Where:
k overdispersion parameter associated with the roadway s~gment;
L length of roadway segment (mi); and
c a regression coefficient used to determine the overdispersion parameter.
Table 11-5 presents the values for the coefficients used in applying Equations 11-9 and 11-10.
Table 11-5. SPF Coefficients for Total and Fatal-and-Injury Crashes on Divided Roadway Segments (for use in
Equations 11-9 and 11-10)
Severity Level a b c
4-lane total -9.025 1.049 1.549
4-lane fatal and injury -8.837 0.958 1.687
4-lane fatal and injury" -8.505 0.874 1.740
• Using the KABCO scale, these include only KAB crashes. Crashes with severity level C (possible injury) are not included.
20~---------------------------------------------------------,
"" 15
" ~
~
~
c
,.,.•
10
.,..!:
.s•
~
~
•>
<
"C
-~
•
"C
~
0 10,000 20,000 30,000 40,000 50,000 60,000 70,000 80,000 90,000

AADT (veh/day)
Figure 11-4. Graphical Form ofSPF for Rural Multilane Divided Roadway Segments (from Equation 11-9 and
Table 11-5)
The default proportions in Table 11-5 are used to break down the crash frequencies from Equation 11-9 into specific
collision types. To do so, the user multiplies the crash frequency for a specific severity level from Equation 11-9 by
the appropriate collision type proportion for that severity level from Table 11-6 to estimate the number of crashes for
that collision type. Table 11-6 is intended to separate the predicted frequencies for total crashes (all severity levels
combined), fatal-and-injury crashes, and fatal-and-injury crashes (with possible injuries excluded) into components
by collision type. Table 11-6 cannot be used to separate predicted total crash frequencies into components by sever-
ity level. Ratios for property-damage-only (PDO) crashes are provided for application where the user has access to
predictive models for that severity level. The default collision type proportions shown in Table 11-6 may be updated
with local data.
•:L ••
T ...
Divided Roadway Segments
Proportion of Crashes by Collision Type and Crash Severity Level
Severity Level
Collision Type Total Fatal and Injury Fatal and Injury" PDO
Head~on 0.006 0.013 0.018 0.002

Sideswipe 0.043 0.027 0.022 0.053
Rear~end 0.116 0.163 0.114 0.088

Angle 0.043 0.048 0.045 0.041
Single 0.768 0.727 0.778 0.792
Other 0.024 0.022 0.023 0.024
• Using the KABCO scale, these include only KAB crashes, Crashes with severity level C (possible injury) are not included.
11.6.3. Safety Performance Functions for Intersections

The predictive model for estimating predicted average crash frequency at particular rural multilane intersection was
presented in Equation 11-4. The effect of traffic volume (AADT) on crash frequency is incorporated through the
SPF, while the effects of geometric design and traffic control features are incorporated through the CMFs. The SPFs
for rural multilane highway intersection are presented in this section. Three- and four-leg stop-controlled intersec-
tions and four-leg signalized rural multilane highway intersections are defined in Section 11.3.
SPFs have been developed for three types of intersections on rural multilane highways. These models can be used for
intersections located on both divided and undivided rural four-lane highways. The three types of intersections are:
• Three-leg intersections with minor-road stop control (3ST)
• Four-leg intersections with minor-road stop control (4ST)
The SPFs for four-leg signalized intersections (4SG) on rural multilane highways have no specific base conditions
and, therefore, can only be applied for generalized predictions. No CMFs are provided for 4SG intersections and
predictions of average crash frequency cannot be made for intersections with specific geometric design and traffic
control features.
Models for three-leg signalized intersections on rural multilane roads are not available.
The SPFs for three- and four-leg stop-controlled intersections (3ST and 4ST) on rural multilane highways are
applicable to the following base conditions:
• Intersection skew angle
• Intersection left-tum lanes 0, except on stop-controlled approaches
• Intersection right-tum lanes 0, except on stop-controlled approaches
• Lighting None
The SPFs for crash frequency have two alternative functional forms, shown in Equations 11-11 and 11-12, and
presented graphically in Figures 11-5, 11-6, and 11-7 (for total crashes only):
Nsp_,.mt = exp[a + b x JntAADT

1 \• mo./I+ c x In(AADTm1.)] (11-11)
or
N,p['" = exp[a + d x In(AADT,.,)J (11-12)
Where:
N,,1 ," = SPF estimate of intersection-related expected average crash frequency for base conditions;
AADTm,J = AADT (vehicles per day) for major-road approaches;

AADT.,, = AADT (vehicles per day) for minor-road approaches;
AADT""'" = AADT (vehicles per day) for minor and major-roads combined approaches; and
a, b, c, d = regression coefficients.
The functional form shown in Equation 11-11 is used for most site types and crash severity levels; the functional
form shown in Equation 11-12 is used for only one specific combination of site type and facility type-four-leg
signalized intersections for fatal-and-injury crashes (excluding possible injuries)-as shown in Table 11-8.
Guidance on the estimation of traffic volumes for the major- and minor-road legs for use in the SPFs is presented in
Step 3 of the predictive method described in Section 11.4. The intersection SPFs for rural multilane highways are
applicable to the following AADT ranges:
3ST: AADT=J 0 to 78,300 vehicles per day and

AADTm'' 0 to 23,000 vehicles per day
4ST: AADTm'J 0 to 78,300 vehicles per day and

AADTm,, 0 to 7,400 vehicles. per day
4SG: AADT=! 0 to 43,500 vehicles per day and

AADT mi, 0 to 18,500 vehicles per day
Application to sites with AADTs substantially outside these ranges may not provide reliable results.
Table 11-7 presents the values of the coefficients a, b, and c used in applying Equation 11-11 for stop-controlled
intersections along with the overdispersion parameter and the base conditions.
Table 11-8 presents the values of the coefficients a, b, c, and d used in applying Equations 11-11 and 11-12 for four-
leg signalized intersections along with the overdispersion parameter. Coefficients a, b, and c are provided for total
crashes and are applied to the SPF shown in Equation II-II. Coefficients a and dare provided for injury crashes and
are applied to the SPF shown in Equation 11-12. SPFs for three-leg signalized intersections on rural multilane roads
are not currently available.
Iffeasible, separate calibration of the models in Tables 11-7 and 11-8 for application to intersections on divided and
undivided roadway segments is preferable. Calibration procedures are presented in Appendix A to Part C.
Table 11-7. SPF Coefficients for Three- and Four-Leg Intersections with Minor-Road Stop Control for Total and
Fatal-and-Injury Crashes (for use in Equation 11-11)
Intersection Type/ Overd.ispersion Parameter
Severity Level a b c (Fixed k)"
4STTotal -10.008 0.848 0.448 0.494
4ST Fatal and injury -11.554 0.888 0.525 0.742
4ST Fatal and injuryb -10.734 0.828 0.412 0.655
3STTota1 -12.526 1.204 0.236 0.460
3ST Fatal and injury -12.664 1.107 0.272 0.569
3ST Fatal and injuryb -11.989 1.013 0.228 0.566
• This value should be used directly as the overdispersion parameter; no further computation is required.
bUsing the KABCO scale, these include only KAB crashes. Crashes with severity level C (possible injury) are not included.
Table 11-8. SPF Coefficients for Four-Leg Signalized Intersections for Total and Fatal-and-Injury Crashes
(for use in Equations 11-11 and 11-12)
Intersection Type/ Overdispersion Parameter
Severity Level a b c d (Fixed k)"
4SGTotal -7.182 0.722 0.337 0.277
4SG Fatal and injury -Q.393 0.638 0.232 0.218
4SG Fatal and injuryb -12.011 1.279 0.566
a This value should be used directly as the overdispersion parameter; no further computation is required.
0 Using the KABCO scale, these include only KAB crashes. Crashes with severity level C (possible injury) are not included,
35,--------------------------------------------------------,
0 10,000 20,000 30,000 40,000 50,000 60,000 70,000 80,000
Figure 11-5. Graphical Form of SPF for Three-Leg Stop-Controlled Intersections--for Total Crashes Only
(from Equation 11-11 and Table 11-7)
50 ~-------------------------7CAADTm'""7,400
. 45
40
~ 35
,•
0
~ 30
.~
~
u
25
& 20
~
•>
< 15
~
-~
~
10
~
~
0
0 10,000 20,000 30,000 40,000 50,000 60,000 70,000 80,000
AADT maJ (veh/day)
Figure 11-6. Graphical Form of SPF for Four-Leg Stop-Controlled Intersections-for Total Crashes Only
(from Equation 11-11 aod Table 11-7)
50
45
40
11 35
0
,•
.sI 30
25
~ 20
••
<
>
15
~
:!!•
~
10
£ 5
0
0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000
AADT maJ (veh/day)
Figure 11-7. Graphical Form of SPF for Four-leg Signalized Intersections-for Total Crashes Only
(from Equation 11-11 aod Table 11-7)
The default proportions in Table 11-9 are used to break down the crash frequencies from Equation 11-11 into specif-
ic collision types. To do so the user multiplies the predicted average frequency for a specific crash severity level from
Equation 11-11 by the appropriate collision type proportion for that crash severity level from Table I 1-9 to estimate
the predicted average crash frequency for that collision type. Table 11-9 separates the predicted frequencies for total
crashes (all severity levels combined), fatal-aod-injury crashes, aod fatal-aod-injury crashes (with possible injuries
excluded) into components by collision type. Table 11-9 cannot be used to separate predicted total crash frequen-
cies into components by crash severity level. Ratios for PDO crashes are provided for application where the user has
access to predictive models for that crash severity level. The default collision type proportions shown in Table 11-9
may be updated with local data.
There are a variety of factors that may affect the distribution of crashes among crash types and crash severity levels.
To account for potential differences in these factors between jurisdictions, it is recommended that the values in Table
11-9 be updated with local data. The values for total, fatal-and-injury, and fatal-and-injury (excluding crashes involv-
ing only possible injuries) in this exhibit are used in the worksheets described in Appendix II A.
Table 11-9. Default Distribution oflntersection Crashes by Collision Type and Crash Severity
Proportion of Crashes by Severity Level
Three-Leg Intersections with .Minor-Road Stop Control Four-Leg Intersections with Minor-Road Stop Control
Collision
Type Fatal and Fatal and Fatal and Fatal and
Total Injury Injury" PDQ Total Injury Injury" PDQ
Head-on 0.029 0.043 0.052 0.020 0.016 0.018 0.023 0.015
Sideswipe 0.133 0.058 0.057 0.179 0.107 0.042 0.040 0.156
Rear-end 0.289 0.247 0.142 0.315 0.228 0.213 0.108 0.240
Angle 0.263 0.369 0.381 0.198 0.395 0.534 0.571 0.292
Single 0.234 0.219 0.284 0.244 0.202 0.148 0.199 0.243
Other 0.052 0.064 0.084 0.044 0.051 0.046 0.059 0.055

Three-Leg Signalized Intersections Four-Leg Signalized Intersections
Collision
Type Fatal and Fatal and Fatal and Fatal and
Total Injury Injury" PDQ Total Injury Injury" PDQ
Head-on 0.054 0.083 0.093 0.034
Sideswipe 0.106 0.047 0.039 0.147
Rear-end 0.492 0.472 0.314 0.505
Angle 0.256 0.315 0.407 0.215
Single 0.062 0.041 0.078 0.077
Other 0.030 0.041 0.069 0.023
~ Using the KABCO scale, these include only KAB crashes. Crashes with severity level C {possible injury) are not included.
Appendix liB presents alternative SPFs that can be applied to predict crash frequencies for selected collision types
for intersections with minor-road stop control on rural multilane highways. Use of these alternative models may be
considered when safety predictions are needed for a specific collision type rather than for all crash types combined.
Care must be exercised in using the alternative SPFs in Appendix liB because they do not address all potential
collision types of interest and because there is no assurance that the safety predictions for individual collision types
would sum to the predictions for all collision types combined provided by the models in Table 11-7.
11.7. CRASH MODIFICATION FACTORS

In Step I 0 of the predictive method shown in Section 11.4, crash modification factors are applied to the selected
safety performance function, which was selected in Step 9. SPFs provided in Chapter II are presented in Section
11.6. A general overview of crash modification factors (CMFs) is presented in Chapter 3, Section 3.5.3. The Part
C-Introduction and Applications Guidance provides further discussion on the relationship of CMFs to the pre-
dictive method. This section provides details of the specific CMFs applicable to the safety performance functions
presented in Section 11.6.
Crash modification factors (CMFs) are used to adjust the SPF estimate of expected average crash frequency for
Chapter II shown in Equation Il-l, The CMF for the SPF base condition of each geometric design or traffic control
feature has a value of 1.00. Any feature associated with higher average crash frequency than the SPF base condition
has a CMF with a value greater than 1.00; any feature associated with lower average crash frequency than the SPF
base condition has a CMF with a value less than 1.00.
The CMFs in Chapter 11 were determined from a comprehensive literature review by an expert panel (5). They rep-
resent the collective judgment of the expert panel concerning the effects of each geometric design and traffic control
feature of interest. Others were derived by modeling data assembled for developing the predictive models rural mul-
tilane roads. The CMFs used in Chapter 11 are consistent with the CMFs in the Part D-Crash Modification Factors,
although they have, in some cases, been expressed in a different form to be applicable to the base conditions. The
CMFs presented in Chapter II, and the specific SPFs to which they apply, are summarized in Table 11-10.
Table 11-10. Summary ofCMFs in Chapter 11 and the Corresponding SPFs

Applicable SPF CMF CMF Description CMF Equations and Exhibits
Lane Width on
C:MF11"11 Equation 11-13, Table 11-11 and Figure 11-8
Undivided Segments
Shoulder Width and Equation 11-14, Figure 11-9,
CMF21"11 Tables 11-12 and 11-13
Shoulder Type
Undivided Roadway Segment SPF
CMF3ru Sideslopes Table 11-14
C:MF4,., Lighting Equation 11-15, Table 11-15
CMFj-,., Automated Speed Enforcement See text
CMFJ,... Lane Width on Divided Segments Equation 11-16, Table 11-16, Figure 11-10
Right Shoulder Width on
CMF2,... Table 11-17
Divided Roadway Segment
Divided Roadway Segment SPF
CMF]rd Median Width Table 11-18
C:MF4,... Lighting Equation 11-17, Table 11-19

CMF 5ro Automated Speed Enforcement See text
CMF 11 Intersection Angle Tables ll-20, 11-21
Three- and Four-Leg C:MFlt Left-Turn Lane on Major Road Tables 11-20, 11-21
Stop-Controlled Intersection SPFs CT~·tfFJI Right-Turn Lane on Major Road Tables 11-20, 11-21
C:MF4t Lighting Tables 11-20, 11-21
11.7.1. Crash Modification Factors for Undivided Roadway Segments

The CMFs for geometric design and traffic control features of undivided roadway segments are presented below.
These CMFs are applicable to the SPF presented in Section 11.6.1 for undivided roadway segments on rural multi-
Jane highways. Each of the CMFs applies to all of the crash severity levels shown in Table 11-3.
CMFJru-Lane Width
The CMF for lane width on undivided segments is based on the work of Harkey eta!. (3) and is determined as fol-
lows:
(11-13)
Where:
CMF1ro ~ crash modification factor for total crashes;
CMFliA ~ crash modification factor for related crashes (run-off-the-road, head-on, and sideswipe),
from Table 11-11; and ·
pRA ~ proportion of total crashes constituted by related crashes (default is 0.27).
CMFRA is determined from Table 11-11 based on the applicable lane width and traffic volume range. The relation-
ships shown in Table 11-11 are illustrated in Figure 11-8. This effect represents 75 percent of the effect oflane width
on rural two-lane roads shown in Chapter 10, Predictive Method for Rural Two-Lane, Two-Way Roads. The default
value of pRA for use in Equation 11-13 is 0.27, which indicates that run-off-the-road, head-on, and sideswipe crashes
typically represent 27 percent of total crashes. This default value may be updated based on local data. The SPF base
condition for the lane width is 12 ft. Where the lane widths on a roadway vary, the CMF is determined separately for
the lane width in each direction of travel and the resulting CMFs are then averaged.
For lane widths with 0.5-ft increments that are not depicted specifically in Table 11-11 or in Figure 11-8, a CMF value
can be interpolated using either of these exhibits since there is a linear transition between the various AADT effects.
Table 11-11. CMFliA for Collision Types Related to Lane Width

Average Annual Daily Traffic (AADT) (vehicles per day)
Lane Width <400 400 to 2000 >2000
9ft or less 1.04 1.04 + 2.13 X 1o-'(AADT- 400) 1.38
10ft 1.02 1.02+ 1.31 X 1o-'(AADT-400) 1.23
11ft 1.01 1.01 + 1.88 X 10-'(AADT-400) 1.04
12ft or more 1.00 1.00 1.00
140
This factor •PPII"'" to slngiii!-Vf!hlde run·off.tk.,.road a"d 1.38 9-ft lana•
muldpla-vahlcle head-on, opposite dlrel;tlon 1lda•wlpe and
<am....:llractlon sideswipe crashes.
135
1.30
1.25
1.20
1.15
1.10
1.05 ,...
·-~
1.00
400 800 1,200 1,600 2,000 2.400
AADT (veh/day)
Figure 11-8. CMF" for Lane Width on Undivided Segments
CMF2 -Shoulder W"ulth

The CMF for shoulder width on undivided segments is based on the work of Harkey eta!. (3) and is determined as
follows:
(11-14)
Where:
crash modification factor for total crashes;
CMFWRA ~ crash modification factor for related crashes based on shoulder width from Table 11-12;
CMFTlU ~ crash modification factor for related crashes based on shoulder type from Table 11-13; and
p" ~ proportion of total crashes constituted by related crashes (default is 0.27).
CMF wRA is determined from Table 11-12 based on the applicable shoulder width and traffic volume range. The
relationships shown in Table 11-12 are illustrated in Figure 11-9. The default value ofp" for use in Equation 11-14
is 0.27, which indicates that run-off-the-road, head-on, and sideswipe crashes typically represent 27 percent of total
crashes. This default value may be updated based on local data. The SPF base condition for shoulder width is 6 ft.
Table 11-12. CMF for Collision Types Related to Shoulder Width (CMF WRA)
Annual Average Daily Traffic (AADT) (vehicles per day)
Shoulder Width <400 400 to 2000 > 2000
Oft 1.10 1.10 + 2.5 X !04 (AADT- 400) 1.50
2ft 1.07 1.07 + 1.43 X 104 (AADT -400) 1.30
4ft 1.02 1.02 + 8.125 X Jo-'(AADT- 400) 1.15
6ft 1.00 1.00 1.00
8ft or more 0.98 0.98-6.875 X 10-'(AADT- 400) 0.87
1.60
1.50 0-ftShoulders
1.50 This factor appllet to slngle-V<!hl~la run-off-ttle-road and
multiple-vehicle head·on, opposite-direction sideswipe,
1.40
1.30 2-ft Shoulders

£130-
v
~•
§ 1 20
1.15 4-ftShoulders
"•
·":s
0
1.10
1,10
::;;
-s 1.00
u
e 0.98
0.90 - o.87 8-ft Shoulders
0.80
o • ~a a·~,_ •-•a ,_ ~ ~ ,_
AADT (veh/day)
Figure 11-9. CMF WRA for Shoulder Width on Undivided Segments
CMF TRA is detennined from Table 11-13 based on the applicable shoulder type and shoulder width.
Table 11-13. CMF for Collision Types Related to Shoulder Type and Shoulder Width (CMF TRA)
Shoulder Width (ft)
Shoulder
Type 0 I 2 3 4 6 8
Paved 1.00 1.00 1.00 1.00 1.00 1.00 1.00
Gravel 1.00 1.00 1.01 1.01 1.01 1.02 1.02
Composite 1.00 1.01 1.02 1.02 1.03 1.04 1.06

Turf 1.00 1.01 1.03 1.04 1.05 1.08 1.11
If the shoulder types and/or widths for the two directions of a roadway segment differ, the CMF is determined sepa-
rately for the shoulder type and width in each direction of travel and the resulting CMFs are then averaged.
CMF3ro -Sides/opes
A CMF for the sideslope for undivided roadway segments of rural multilane highways has been developed by Har-
key et a!. (3) from the work of Zegeer et a!. (8). The CMF is presented in Table 11-14. The base conditions are for a
sideslope of 1:7 or flatter.
Table 11-14. CMF for Sideslope on Undivided Roadway Segments (CMF 3J

1:2 or Steeper 1:3 1:4 1:5 1:6 1:7 or Flatter
1.18 1.15 1.12 1.09 1.05 1.00
CMF4ro-Lighting
The SPF base condition for lighting of roadway segments is the absence oflighting. The CMF for lighted roadway
segments is determined, based on the work ofElvik and Vaa (1), as:
CMF4ro ~ 1- [(J- 0.72 X p 1""- 0.83 X PpJ X P,_] (11-15)

Where:
CMF4 ~ ~ crash modification factor for the effect oflighting on total crashes;
~ proportion of total nighttime crashes for unlighted roadway segments that involve a fatality or injury;
~ proportion of total nighttime crashes for unlighted roadway segments that involve property damage only;
and
~ proportion of total crashes for unlighted roadway segments that occur at night.
This CMF applies to total roadway segment crashes. Table ll-15 presents default values for the nighttime crash propor-
tions p 1nr, ppnr, and pnr. HSM users are encouraged to replace the estimates in Table ll-15 with locally derived values.

Proportion of Crashes that
Roadway Type Proportion of Total Night~ Time Crashes by Severity Level Occur at Night
Fatal and Injury p1n, PDOp, P.,
4U 0.361 0.639 0.255
CMF,,.-Automated Speed Enforcement

Automated speed enforcement systems use video or photographic identification in conjunction with radar or lasers
to detect speeding drivers. These systems automatically record vehicle identification information without the need
for police officers at the scene. The SPF base condition for automated speed enforcement is that it is absent. Chapter
17, Road Networks presents a CMF of 0.83 for the reduction of all types of injury crashes from implementation of
automated speed enforcement. This CMF applies to roadway segments with fixed camera sites where the camera
is always present or where drivers have no way of knowing whether the camera is present or not. Fatal-and-injury
crashes constitute 31 percent of total crashes on rural two-lane highway segments. No information is available on the
effect of automated speed enforcement on noninjury crashes. With the conservative assumption that automated speed
enforcement has no effect on noninjury crashes, the value of CMF5 ~ for automated speed enforcement would be 0.95
based on the injury crash proportion.
11.7.2. Crash Modification Factors for Divided Roadway Segments

The CMFs for geometric design and traffic control features of divided roadway segments for rural multilane high-
ways are presented below. Each of the CMFs applies to all of the crash severity levels shown in Table ll-5.
CMF1._-Lane Width on Divided Roadway Segments

The CMF for lane width on divided segments is based on the work of Harkey eta!. (3) and is determined as follows:
CMF1ro ~ (CMFRA- !.0) x pRA + !.0 (ll-16)
\Vhere:
CMF1ro crash modification factor for total crashes;
CMFRA crash modification factor for related crashes (run-off-the-road, head-on, and sideswipe),
from Table ll-16; and
pRA ~ proportion of total crashes constituted by related crashes (default is 0.50).
CMFRA is determined from Table 11-16 based on the applicable lane width and traffic volume range. The relation-
ships shown in Table 11-16 are illustrated in Figure 11-10. This effect represents 50 percent of the effect oflane
width on rural two-lane roads shown in Chapter I 0. The default value ofpRA for use in Equation 11-16 is 0.50, which
indicates that run-off-the-road, head-on, and sideswipe crashes typically represent 50 percent of total crashes. This
default value may be updated based on local data. The SPF base condition for lane width is 12ft. Where the lane
widths on a roadway vary, the CMF is determined separately for the lane width in each direction of travel and the
resulting CMFs are then averaged.
Table 11-16. CMF for Collision Types Related to Lane Width (CMFR)
Annual Average Daily Traffic (AADT) (vehicles/day)
Lane Width <400 400 to 2000 >2000
9ft 1.03 1.03 + 1.38 X 1o-'(AADT- 400) 1.25

10ft 1.01 1.01 +8.75 X 1Q-'(AADT-400) 1.15
11ft 1.01 1.01 + 1.25 X 1Q-'(AADT-400) 1.03
12ft 1.00 1.00 1.00
1.30
The factor applies to single-vehicle run-off-the-road

1.27 and multiple-vehic.Je heacl-on, opposite-direction
sideswipe, and same-direction sideswipe crashes. 1.25 9-ftLanes
1.24
1.21
"
~
~• 118
= 1.15 10-ftlanes
-~ 115
"',;
0 1.12
::;
-5i 1.09
5
1.06
1.03 1.03 11-ft lanes

1.03
1.01
1.00 12-ft Lanes
1.00
0 400 800 1,200 1,600 2,000 2,400
AADT (veh/day)
Figure 11-10. CMFRA for Lane Width on Divided Roadway Segments
CMF2"'-Right Shoulder Width on Divided Roadway Segments

The CMF for right shoulder width on divided roadway segments was developed by Lord et al. ( 5) and is presented in
Table 11-17. The SPF base condition for the right shoulder width variable is 8 ft. If the shoulder widths for the two
directions of travel differ, the CMF is based on the average of the shoulder widths. The safety effects of shoulder
widths wider than 8 ft are unknown, but it is recommended that a CMF of 1.00 be used in this case.
The effects of unpaved right shoulders on divided roadway segments and ofleft (median) shoulders of any width or
material are unknown. No CMFs are available for these cases.
Table 11-17. CMF for Right Shoulder Width on Divided Roadway Segments (CMF,,)
Average Shoulder Width (ft)
0 2 4 6 8 or more
1.18 1.13 1.09 1.04 1.00
Note: This CMF applies to paved shoulders only.
CMF3_,-Median Width
A CMF for median widths on divided roadway segments of rural multilane highways is presented in Table 11-18
based on the work of Harkey et al. (3). The median width of a divided highway is measured between the inside edges
of the through travel lanes in the opposing direction of travel; thus, inside shoulder and turning lanes are included
in the median width. The base condition for this CMF is a median width of 30 ft. The CMF applies to total crashes,
but represents the effect of median width in reducing cross-median collisions; the CMF assumes that nonintersec-
tion collision types other than cross-median collisions are not affected by median width. The CMF in Table 11-18
has been adapted from the CMF in Table 13-9 based on the estimate by Harkey et al. (3) that cross-median collisions
represent 12.2 percent of crashes on multilane divided highways.
This CMF applies only to traversable medians without traffic barriers. The effect of traffic barriers on safety would be expect-
ed to be a function of the barrier type and offse~ rather than the median width; however, the effects of these factors on safety
have not been quantified Until better infonnation is available, a CMF value of 1.00 is used for medians with traffic barriers.
Table 11-18. CMFs for Median Width on Divided Roadway Segments without a Median Barrier (CMF 3, )
Median Width (ft) CMF
10 1.04
20 !.02
30 1.00
40 0.99
50 0.97
60 0.96
70 0.96
80 0.95
90 0.94
100 0.94
Note: This CMF applies only to medians without traffic barriers.
CMF4_,-Lighting
The SPF base condition for lighting is the absence of roadway segment lighting. The CMF for lighted roadway seg-
ments is determined, based on the work ofElvik and Vaa (!),as:
CMF4rd = I -[(I- 0.72 X p.mr - 0.83 X p pnrl' X p nrl (11-17)
Where:
CMF,., = crash modification factor for the effect oflighting on total crashes;
P;,. = proportion of total nighttime crashes for unlighted roadway segments that involve a fatality or injury;
pP" = proportion of total nighttime crashes for unlighted roadway segments that involve property damage
only; and
p" = proportion of total crashes for unlighted roadway segments that occur at night.
.1..
This CMF applies to total roadway segment crashes. Table 11-19 presents default values for the nighttime crash propor-
tions P;,~ Pp,~ and P,,· HSM users are encouraged to replace the estimates in Table 11-19 with locally derived values.

Proportion of Crashes that
Proportion of Total Nighttime Crashes by Severity Level Occur at Night
Roadway Type Fatality and Injury p1,., PDOp, P.,

4D 0.323 0.677 0.426
CMF,,,-Automated Speed Enforcement

Automated speed enforcement systems use video or photographic identification in conjunction with radar or lasers to
detect speeding drivers. These systems automatically record vehicle identification information without the need for
police officers at the scene. The SPF base condition for automated speed enforcement is that it is absent. Chapter 17
presents a CMF of 0.83 for the reduction of all types of fatal-and-injury crashes from implementation of automated
speed enforcement. This CMF applies to roadway segments with fixed camera sites where the camera is always pres-
ent or where drivers have no way of knowing whether the camera is present or not. Fatal-and-injury crashes con-
stitute 37 percent of total crashes on rural multilane divided highway segments. No information is available on the
effect of automated speed enforcement on noninjury crashes. With the conservative assumption that automated speed
enforcement has no effect on noninjury crashes, the value of CMF sro for automated speed enforcement would be 0.94
based on the injury crash proportion.

The effects of individual geometric design and traffic control features of intersections are represented in the safety
prediction procedure by CMFs. The equations and exhibits relating to CMFs for stop-controlled intersections are
summarized in Tables 11-20 and 11-21 and presented below. Except where separate CMFs by crash severity level are
shown, each of the CMFs applies to all of the crash severity levels shown in Table 11-7. As noted earlier, CMFs are
not available for signalized intersections.
Table 11-20. CMFs for Three-Leg Intersections with Minor-Road Stop Control (3ST)
CMFs Total Fatal and Injury
Intersection Angle Equation ll~18 Equation 11-19

Left-Tum Lane on Major Road Table 11-22 Table 11-22
Right-Tum Lane on Major Road Table 11-23 Table ll-23
Lighting Equation 11-22 Equation 11-22
Table 11-21- CMFs for Four-Leg Intersection with Minor-Road Stop Control (4ST)
CMFs Total Fatal and Injury
Intersection Angle Equation 11-20 Equation 11-21
Left-TurnLane on Major Road Table 11-22 Table 11-22

Right-Turn Lane on Major Road Table 11-23 Table 11-23
Lighting Equation 11-22 Equation 11-22
CMFu-Intersection Skew Angle

The SPF base condition for intersection skew angle is 0 degrees of skew (i.e., an intersection angle of 90 degrees).
Reducing the skew angle of three- or four-leg stop-controlled intersections on rural multilane highways reduces
total intersection crashes, as shown below. The skew angle is the deviation from an intersection angle of 90 degrees.
Skew carries a positive or negative sign that indicates whether the minor road intersects the major road at an acute or
obtuse angle, respectively.
Illustration of Intersection Skew Angle
SKEW·/
ANGLE
Three-Leg Intersections with Stop-Control on the Minor Approach

The CMF for total crashes for intersection skew angle at three-leg intersections with stop-control on the minor ap-
proach is:
CMFli = 0.016 x skew + LO

(0.98 + 0.16 x skew) (11-18)
and the CMF for fatal-and-injury crashes is:
CMF], = 0.017 x skew + LO

(0.52 + 0.17 x skew) (11-19)
Where:
CMF11 = crash modification factor for the effect of intersection skew on total crashes; and
skew = intersection skew angle (in degrees); the absolute value of the difference between 90 degrees and the
Four-Leg Intersections with Stop-Control on the Minor Approaches

The CMF for total crashes for intersection angle at four-leg intersection with stop-control on the minor approaches is:
CMF]; = 0.053 x skew + l.O

(1.43 + 0.53 x skew) (11-20)
The CMF for fatal-and-injury crashes is:
CMFJ, = 0.048 x skew + l.O

(0.72 + 0.48 x skew) (11-21)
CMF21-lntersection Left-Turn Lanes

The SPF base condition for intersection left-tum lanes is the absence ofleft-tum lanes on all of the intersection ap-
proaches. The CMFs for presence of left-tum lanes are presented in Table 11-22 for total crashes and injury crashes.
These CMFs apply only on uncontrolled major-road approaches to stop-controlled intersections. The CMFs for
installation of left-tum lanes on multiple approaches to an intersection are equal to the corresponding CMF for in-
stallation of a left-tum lane on one approach raised to a power equal to the number of approaches with left-tum lanes
(i.e., the CMFs are multiplicative, and Equation 3-7 can be used). There is no indication of any effect of providing
a left-tum lane on an approach controlled by a stop sign, so the presence of a left-tum lane on a stop-controlled ap-
proach is not considered in applying Table 11-22. The CMFs for installation ofleft-tum lanes are based on research
by Harwood et al. (4) and are consistent with the CMFs presented in Chapter 14, Intersections. A CMF of 1.00 is
used when no left-tum lanes are present.
Table 11-22. Crash Modification Factors (CMF,) for Installation of Left-Turn Lanes on Intersection Approaches
Number ofNon-Stop-ControlledApproaches
with Left-Turn Lanes•
Intersection Type Crash Severity Level One Approach Two Approaches
Three-leg minor-road stop Total 0.56

control" Fatal and Injury 0.45
Four-leg minoHoad stop controlb Total 0.72 0.52
-------------------------------------------------
Fatal and Injury 0.65 0.42
• Stop-controlled approaches are not considered in determining the number of approaches with !eft-turn lanes
b Stop signs present on minor-road approaches only.
CMF31-lntersection Right-Turn Lanes

The SPF base condition for intersection right-tum lanes is the absence of right-tum lanes on the intersection ap-
proaches. The CMFs for the presence of right-tum lanes are based on research by Harwood et al. (4) and are con-
sistent with the CMFs in Chapter 14. These CMFs apply to installation of right-tum lanes on any approach to a
signalized intersection, but only on uncontrolled major-road approaches to stop-controlled intersections. The CMFs
for installation of right-tum lanes on multiple approaches to an intersection are equal to the corresponding CMF for
installation of a right-tum lane on one approach raised to a power equal to the number of approaches with right-tum
lanes (i.e., the CMFs are multiplicative, and Equation 3-7 can be used). There is no indication of any safety effect
for providing a right-tum lane on an approach controlled by a stop sign, so the presence of a right-tum lane on a
stop-controlled approach is not considered in applying Table 11-23. The CMFs for presence of right-tum lanes are
presented in Table 11-23 for total crashes and injury crashes. A CMF value of 1.00 is used when no right-tum lanes
are present. This CMF applies only to right-tum lanes that are identified by marking or signing. The CMF is not ap-
plicable to long tapers, flares, or paved shoulders that may be used informally by right-tum traffic.
Table 11-23. Crash Modification Factors (CMF3 ) for Installation of Right-Turn Lanes on Intersections Approaches
Number ofNon-Stop-ControlledApproaches
with Right- Turn Lanes•
Intersection Type Crash Severity Level One Approach Two Approaches
Three-leg minor-road Total 0.86

stop controJb Fatal and Injury 0.77
Total 0.86 0.74

Four-leg minor-road stop controJb
Fatal and Injury 0.77 0.59
~Stop-controlled approaches are not considered in determining the number of approaches with right-turn lanes.
CMF41-Lighting
The SPF base condition for lighting is the absence of intersection lighting. The CMF for lighted intersections is
adapted from the work of Elvik and Vaa ( 1), as:
CMF, ~ 1.0-0.38 x P,; (11-22)
Where:
CMF,; ~ crash modification factor for the effect oflighting on total crashes; and
p ,; ~ proportion of total crashes for unlighted intersections that occur at night.
This CMF applies to total intersections crashes (not including vehicle-pedestrian and vehicle-bicycle collisions).
Table 11-24 presents default values for the nighttime crash proportion, P,;- HSM users are encouraged to replace the
estimates in Table 11-24 with locally derived values.
Table 11-24. Default Nighttime Crash Proportions for Unlighted Intersections

Intersection Type Proportion of Crashes that Occur at Night, p,;
JST 0.276
4ST 0.273
11.8. CALIBRATION TO LOCAL CONDITIONS

widely from one jurisdiction to another. Geographic regions differ markedly in climate, animal population, driver
populations, crash-reporting threshold, and crash-reporting practices. These variations may result in some jurisdic-
tions experiencing a different number of traffic crashes on rural multilane highways than others. Calibration factors
are included in the methodology to allow highway agencies to adjust the SPFs to match actual local conditions.
The calibration factors for roadway segments and intersections (defined below as C, and C;, respectively) will have
values greater than 1.0 for roadways that, on average, experience more crashes than the roadways used in the devel-
opment of the SPFs. The calibration factors for roadways that experience fewer crashes on average than the road-
ways used in the development of the SPFs will have values less than 1.0. The calibration procedures are presented in
vidual agencies or locations. Several other default values used in the methodology, such as collision type distribu-
tion, can also be replaced with locally derived values. The derivation of values for these parameters is addressed in
11.9. LIMITATIONS OF PREDICTIVE METHODS IN CHAPTER 11

This section discusses limitations of the specific predictive models and the application ofthe predictive method in
Chapter II.
Where rural multilane highways intersect access-controlled facilities (i.e., freeways), the grade-separated interchange
facility, including the rural multilane road witbin the interchange area, cannot be addressed with the predictive
method for rural multilane highways.
The SPFs developed for Chapter II do not include signalized three-leg intersection models. Such intersections may
be found on rural multilane highways.
CMFs have not been developed for the SPF for four-leg signalized intersections on rural multilane highways.
11.10. APPLICATION OF CHAPTER 11, PREDICTIVE METHOD

The predictive method presented in Chapter II applies to rural multilane highways. The predictive method is applied
to a rural multilane highway facility by following the 18 steps presented in Section 11.4. Worksheets are presented
in Appendix II A for applying calculations in the predictive method steps specific to Chapter II. All computations
of crash frequencies within these worksheets are conducted with values expressed to three decimal places. This level
of precision is needed only for consistency in computations. In the last stage of computations, rounding the final
estimates of expected average crash frequency be to one decimal place is appropriate.
11.11. SUMMARY
The predictive method can be used to estimate the expected average crash frequency for an entire rural multilane
highway facility, a single individual site, or series of contiguous sites. A rural multilane highway facility is defined in
Section 11.3, and consists of a four-lane highway facility which does not have access control and is outside of cities
or towns with a population greater than 5,000 persons.
The predictive method for rural multilane highways is applied by following the 18 steps of the predictive method
presented in Section 11.4. Predictive models, developed for rural multilane highway facilities, are applied in Steps
9, 10, and II of the method. These predictive models have been developed to estimate the predicted average crash
frequency of an individual intersection or homogenous roadway segment. The facility is divided into these individual
sites in Step 5 of the predictive method.
Each predictive model in Chapter II consists of a safety performance fimction (SPF), crash modification factors (CMFs),
and a calibration factor. The SPF is selected in Step 9 and is used to estimate the predicted average crash frequency for a
site with base conditions. This estimate can be for either total crashes or organized by crash-severity or collision-type dis-
tribution. In order to account for differences between the base conditions and the specific conditions of the site, CMFs are
applied in Step I 0, which adjust the prediction to account for the geometric design and traffic control features of the site.
Calibration factors are also used to adjust the prediction to local conditions in the jurisdiction where the site is located. The
process for determining calibration factors for the predictive models is described in Part C, Appendix A.!.
Where observed data are available, the EB Method is applied to improve the reliability of the estimate. The EB
Method can be applied at the site-specific level or at the project-specific leveL It may also be applied to a future time
period if site conditions will not change in the future period. The EB Method is described in Part C, Appendix A.2.
Section 11.12 presents six sample problems which detail the application of the predictive method. Appendix !lA
contains worksheets which can be used in the calculations for the predictive method steps.

In this section, six sample problems are presented using the predictive method for rural multilane highways. Sample
Problem 1 illustrates how to calculate the predicted average crash frequency for a divided rural four-lane highway
segment. Sample Problem 2 illustrates how to calculate the predicted average crash frequency for an undivided rural
four-lane highway segment. Sample Problem 3 illustrates how to calculate the predicted average crash frequency for a
three-leg stop-controlled intersection. Sample Problem 4 illustrates how to combine the results from Sample Problems
1 through 3 in a case where site-specific observed crash data are available (i.e., using the site-specific EB Method).
Sample Problem 5 illustrates how to combine the results from Sample Problems 1 through 3 in a case where site-spe-
cific observed crash data are not available (i.e .• using project level EB Method). Sample Problem 6 applies the Project
Estimation Method I, presented in Section C.7 of the Part C-lntroduction and Applications Guidance, to determine
the effectiveness of a proposed upgrade from a rural two-lane roadway to a rural four-lane highway.

11-37 Predicted average crash frequency for a divided roadv,ray segment
2 11-43 Pre~cted average crash frequency for an undivided roadway segment
Expected average crash frequency for a facility when site-specific observed crash
4 11-54
frequencies are available
Expected average crash frequency for a facility when site-specific observed crash
5 11-56
frequencies are not available
Expected average crash frequency and the crash reduction for a proposed rural four-
6 11-<50
lane highway facility that will replace an existing rural two-lane roadway
The Site/Facility
A rural four-lane divided highway segment.
The Question
The Facts
• 1.5-mi length
• 10,000 veh/day
• 6-ft paved right shoulder
• 20-ft traversable median
• No roadway lighting
• No automated enforcement
11-38 HIGHWAY SAFElY MANUAL
Assumptions
Collision type distributions are the defaults values presented in Table 11-6.
Results
in Sample Problem I is determined to be 3.3 crashes per year (rounded to one decimal place).
Steps
Step 1 through 8
The SPF for a divided roadway segment is calculated from Equation 11-9 and Table 11-5 as follows:
N = era. + b "' In.(AADTJ + In.(LJJ
spjrd
= e(-9.025 + t.049 x /n(lo.ooo) + In(l.S)) = 2.835 crashes/year
Step tO-Multiply the result obtained in Step 9 by the appropriate CMFs to adjust base conditions to site
specific geometric conditions and traffic control features.
Each CMF used in the calculation of the predicted average crash frequency of the roadway segment is calculated
below:
Lane Width (CMF,j

Since the roadway segment in Sample Problem I has 12-ft lanes, CMF 1ni = 1.00 (i.e., the base condition for CMF1,d
is 12-ft lane width).
Shoulder Width and Type (CMF ,j

From Table 11-17, for 6-ft paved shoulders, CMF,ni = 1.04.
Median Width (CMF3 j

From Table 11-18, for a traversable median width of 20 ft, CMF 3ni = 1.02.
Lighting (CMF4 j
Since there is no lighting in Sample Problem I, CMF 4ni = 1.00 (i.e., the base condition for CMF 4ni is absence of
roadway lighting).
Automated Speed Enforcement (CMF5j

Since there is no automated speed enforcement in Sample Problem I, CMF 5"' = 1.00 (i.e., the base condition for
CMF '"'is the absence of automated speed enforcement).
The combined CMF value for Sample Problem I is calculated below.
= 1.06
It is assumed in Sample Problem I that a calibration factor, C~ of 1.10 has been determined for local conditions.
See Part C, Appendix A.l for further discussion on calibration of the predictive models.

Npredictcd rs = ~pfrd X cr X ( CMFlrd X CMF2rd X ..• X CMF5rd)

= 2.835 X 1.10 X (1.06)
= 3.305 crashes/year
WORKSHEETS
of five worksheets are provided for determining the predicted average crash frequency. The five worksheets include:
• Worksheet SP !A (Corresponds to Worksheet !A)-General Information and Input Data for Rural Multilane
Roadway Segments
• Worksheet SP IE (Corresponds to Worksheet IE (a))-Crash Modification Factors for Rural Multilane Divided
Roadway Segments
• Worksheet SP I C (Corresponds to Worksheet I C (a))-Roadway Segment Crashes for Rural Multilane Divided
Roadway Segments
• Worksheet SP JD (Corresponds to Worksheet JD (a))--Crashes by Severity Level and Collision Type for Rural
Multilane Divided Roadway Segments
• Worksheet SP IE (Corresponds to Worksheet I E)-Summary Results for Rural Multilane Roadway Segments
Details of these sample problem worksheets are provided below. Blank versions of the corresponding worksheets
are provided in Chapter 11, Appendix llA.
Worksheet SP1A-Generallnformation and Input Data for Rural Multilane Roadway Segments
Worksheet SPlA is a summary of general information about the roadway segment, analysis, input data
(i.e., "The Facts") and assumptions for Sample Problem 1.
Worksheet SP1A. General Information and Input Data for Rural Multilane Roadway Segments
Analyst I Highway
Agency or Company Roadway Section
l .!:z:i:~=~~~·-·-··-··- -· ----··-·- --··-··-··-·1- ----------- ------- ----- --

Date Performed
Analysis Year .
Roadway type (divided/undivided) - divided
-------------------------------------------------·-t··------------·-----------1·-------------------------
Length of
__segment, L (mi)
_,_,_,_,,_,_,_, _,_,_,_,_,_ __ __ ---·-·-·-
,,_,_,_,,_,_,_,_,_,_,,_,_,_,_,_,_,_,_,_ ..
- ___ _,_ __
.. _, ___ ____ 1.5
,_,_,_,_,_,_, ,_,_,_,_,_,_,_,,_,,_, .. .. ,_,_,_,_,_,_,_, .. .. .. _,_,_,,_, ,_,_,_,,,_,,_, ,_,_,,_, ,_,_,_,_,_,_,_,_,_,,_,_,_,_,_"
AADT (vehlday) I - 10,000

Lane width (ft) 12 1
12
Shoulder
___ width
_,_,_,_,_, ___ ..(ft)-right
___
,_ , __
shoulder
..... -·-·-·-·-
,_,,_ ,
width for divided ......
.. -·-·- _,_,_, ___ ,_,
,_,_, _____ ,,_ ,_,,_, __,_,_ .. ,_,_,
I
_,_,_,_ ,_.. _,_,_,_,_,_,_,,_, ___8 ,_, __
....,_,_ . ,_,,_,,_ .. __ ,_
1
,_,_,_,_, __,_,_,_ __ 6
.. _,, ,_,_,_,_ .. ,_,_, __ ,_,,_,_,_, __
Shoulder type-right shoulder type for divided paved paved
--------·------·---------------------------------·----------------------------1.-------------------------
Median -width divided only 30 20
Sideslopes-for undivided only 1:7 or :flatter N/A
. ----- ----- -··- . -- -- -·- . --- -·· · .. -- --··- -... ------ .. ··-·-- . --+- ... ---- -- --- -- -- --·- --- . -_,_ -"""" -. --- .....:.:.:.:. :. . . "" ---- ----
Lighting (present/not present) not present not present
Auto speed enforcement (present/not present) not present not present
Calibration factor, C, LO L1
Worksheet SP1B-Crash Modification Factors for Rural Multilane Divided Roadway Segments
In Step I 0 ofthe predictive method, crash modification factors are applied to account for the effects of site specific
geometric design and traffic control devices. Section I L7 presents the tables and equations necessary for determin-
ing the CMF values. Once the value for each CMF has been determioed, all of the CMFs multiplied together io
Colwnn 6 ofWorksheet SPIB which iodicates the combined CMF value.
Worksheet SP1 B. Crash Modification Factors for Rural Multilane Divided Roadway Segments
(I) (2) (3) (4) (5) (6)
CMFfor
Lane Width
CMF for Right
_______ ____ __ _____,_____
Shoulder Width
,
'I
. ,_,
CMF for
Median Width
i
rl
• • I CMF for Auto
C:MF for Lighting I Speed Enforcement
-··-----·--..------·-;·-----·-..
.
Combmed CMF
.. --- -----.. ---·---·- .. -~---
I -·-·~
CMF2,... CMF1.,.. CMF4,... CMF5,... CMFcomh

from Equation 11-16 from Table 11-17 from Table 11-18 from Equation 11-17 from Section 11.7.2 (1)*'(2)'*'(3)"'(4)"'(5)
LOO L04 L02 LOO LOO L06
Worksheet SP1C-Roadway Segment Crashes for Rural Multilane Divided Roadway Segments
The SPF for the roadway segment in Sample Problem I is calculated usiog the coefficients found io Table 11-5 (Col-
wnn 2), which are entered into Equation 11-9 (Colwnn 3). The overdispersion parameter associated with the SPF can
be calculated usiog Equation Il-l 0 and entered iota Colwnn 4; however, the overdispersion parameter is not needed
for Sample Problem I (as the EB Method is not utilized). Colwnn 5 represents the combioed CMF (from Colwnn 6
in Worksheet SPIB), and Column 6 represents the calibration factor. Column 7 calculates predicted average crash
frequency using the values in Colwnn 4, the combioed CMF io Colwnn 5, and the calibration factor io Colwnn 6.
Worksheet SP1 C. Roadway Segment Crashes for Rural Multilane Divided Roadway Segments
(I) (2) (3) (4) (5) (6) (7)
Predicted Average
Crash Overdispersion Combined
SPF Coefficients Crash Frequency,
Severity Parameter, k CMFs
Calibration Npmllo,.dn
Level Factor, C,
from Table 11~5 from from Equation (6) from
----------
T b
-~---~----·-t·--'-·
T Equation 11-9 11-10 Worksheet SPIB
----------- -----------·- ~-------- -----------
(3)*(5)*(6)
Total -9.025 1 L049 , 1.549 2.835 0.142 L06 1.10 3.306

Fatal and
--8.837 0.958 1.687 L480 0.123 L06 1.10 1.726
injury (Fl)
Fatal and
injury" -8.505 0.874 1.740 0.952 0.117 L06 1.10 LllO
(F1')
Property
damage
only 1.580
(PDO)
Worksheet SP1 O-Crashes by Severity Level and Collision Type for Rural Multilane Divided
Roadway Segments
Worksheet SPlD presents the default proportions for collision type (from Table ll-6) by crash severity level as fol-
lows:
• Fatal-and-injury crashes, not including "possible injury" crashes (i.e., on a KABCO injury scale, only KAB
crashes) (Column 6)
(Total), 5 (Fatal and Injury, Fl), 7 (Fatal and Injury, not including "possible injury"), and 9 (Property Damage Only,
PDO).
These proportions may be used to separate the predicted average crash frequency (from Column 7, Worksheet SPlC)
Worksheet SP1 D. Crashes by Severity Level and Collision Type for Rural Multilane Divided Roadway Segments
(1) (2) (3) (4) (5) I (6) I (7) (8) (9)
Collision I Proportion
of Collision
I N,~ldffi"'""'
(crashes/
Proportion
of Collision
N predl«ad r.r (FI)
(crashes/
Npn!dkl<d rs(Ff')
(crashes/
Proportion
of Collision NP"'"I!ctM"' (PDO)
Type
Type (<o<ol\ year) Type (FI! year) Type rFfJ year) Type(PDO)
(7) 1.,.1 from (7)FI from (7)Ffl from (7)PDO from

from Table Worksheet from Table Worksheet from Table Worksheet from Table Worksheet
+
11-6 SP1C 11-6 SP1C 11-6 SP1C IH SPIC
Total
- -.. .......-.. -·--
,.- "
1.000
- - - -.... --
1.
3.306 I 1.000 '
-·
1.726 1.000 1.110
- f--
1.000 _, ______________
I 1.580
i
(2)*(3)toUJI (4)'(5)n (6)'(7)" (8)*(9)PDO
~~~020
Head-on
0.013 0.022 I O.Ql8 0.020 ' 0.002 0.003
collision
--
Sideswipe I
-
--~---
0.043 I 0.142 0.027 0.047 0.022 0.024 0.053 0 084
collision
- _,,_,, ..................... _,,_,,_ -·-·- ---- - - --------
Rear-end ' 0.116 0.383 0.163 0.281 0.114 0.127 0.088 0.139
collision
Angle
0 043 0 142 0 048 0 083 0 045 0 050 0 041 0 065
collision '
--
Single-
vehicle 0.768 2.539 0.727 1.255 0.778 0.864 0.792 1.251
collision
..... _,,_,_" _,,_ - - ·-
Other
0.024 0.079 0.022 O.Q38 0.023 0.026 0.024 O.Q38
collision ' I i
Worksheet SP1 E-Summary Results for Rural Multilane Roadway Segments

Worksheet SPlE presents a swnmary of the results. Using the roadway segment length. the worksheet presents the
Worksheet SP1 E. Swnmary Results for Rural Multilane Roadway Segments

(1) (2) (3) (4)
Predicted Average Crash
Frequency (crashes/year) Crash Rate (crashes/mi!year)
Crash Severity Level (7) from Worksheet SPIC Roadway Segment Length (mi) (2)/(3)
Total 3.306 1.5 2.2
Fatal and injury (FI) L 726 1.5 L2

- - - _,_,, _,_,,_,, ............. _,,_,_,_, _,_ - ,_,_,_,_ .. ,_ _,,_,,_,,_,,_,,_,_,_,_,_,_,_,_,_,_,_,_,_,,_,,_,_,,_,,_,,_,_
Fatal and injury' (Fl') L II 0 1.5 0.7
Property damage only (PDO) 1.580 1.5 Ll
The Site/Facility
A rural four-lane undivided highway segment.
The Question
The Facts
• 0.1-mi length
• 8,000 vehlday
• Sideslope of 1:6
• Roadside lighting present

• Automated enforcement present
Assumptions
Collision type distributions have been adapted to local experience. The percentage oftotal crashes representing
single-vehicle run-off-the-road and multiple-vehicle head-on, opposite-direction sideswipe, and same-direction
sideswipe crashes is 33 percent.
The proportion of crashes that occur at night are not known, so the default proportions for nighttime crashes will be used.
Results
in Sample Problem 2 is determined to be 0.3 crashes per year (rounded to one decimal place).
Steps
Step 1 through 8
through 11 are conducted. No other steps are necessary because only one roadway segment is analyzed for one year,
The SPF for an undivided roadway segment is calculated from Equation 11-7 and Table 11-3 as follows:
N = efa + b X Jn(AADT) + ln(L)j
spfru
= eC-9.653 + 1.176 x In(S,OOO)+ In(O !JJ = 0.250 crashes/year

Step tO-Multiply the result obtained in Step 9 by the appropriate CMFs to adjust base conditions to site
specific geometric conditions and traffic control features-
Each Cl\fF used in the calculation of the predicted average crash frequency of the roadway segment is calculated below:
Lane Width (CMF1, )

CMF 1rn can be calculated from Equation 11-13 as follows:
For 11-ft lane width andAADT of8,000, CMF RA= 1.04 (see Table 11-11).
The proportion of related crashes, pRA, is 0.33 (from local experience, see assumptions).
CMF1ro = (1.04- 1.0) X 0.33 + 1.0 = 1.01

Shoulder Width and Type (CMF2, )
CMF,ro can be calculated from Equation 11-14 as follows:
For 2-ft shoulders andAADT of8,000, CMFwRA = 1.30 (see Table 11-12).
For 2-ft gravel shoulders, CMF 7RA = 1.01 (see Table 11-13).
The proportion of related crashes, pRA, is 0.33 (from local experience, see assumptions).
CMF2ro = (1.30 X !.OJ- J.O) X 0.33 + 1.0 = 1.10
Sides/opes (CMF3, )
From Table 11-14, for a sideslope of 1:6, CMF,ro = 1.05.
Lighting (CMF4 , )
CMF,ro can be calculated from Equation 11-15 as follows:
CMF4rv =l-[(l-0.72xp.mr -0.83xppn.r )xp,,.J1
Local values for nighttime crashes proportions are not known. The default nighttime crash proportions used are
PInr=0.36l,ppn.r=0.639,andpnr=0.255(seeTable 11-15).
CMF4ro =)-((I- 0.72 X 0.361-0.83 X 0.639) X 0.255] = 0.95
Automated Speed Enforcement (CMF5, )

For an undivided roadway segment with automated speed enforcement, CMF 5 ~~ 0.95 (see Section 11.7.1).
CMF"mb ~ 1.04 X 1.02 X 1.05 X 0.95 X 0.95 ~ 1.05
It is assumed in Sample Problem 2 that a calibration factor, C,, of 1.10 bas been determined for local conditions.
See Part C, Appendix A. I for further discussion on calibration of the predictive models.

through II as follows:
~ 0.250 X 1.10 X (1.05)
~ 0.289 crashes/year
WORKSHEETS
The step-by-step instroctions above are provided to illustrate the predictive method for calculating the predicted
of five worksheets are provided for determining the predicted average crash frequency. The five worksheets include:
• Worksheet SP2A (Corresponds to Worksheet JA)-General Information and Input Data for Rural Multilane
Roadway Segments
• Worksheet SP2B (Corresponds to Worksheet JB (b))-Crash Modification Factors for Rural Multilane Undivided
Roadway Segments
• Worksheet SP2C (Corresponds to Worksheet JC (b))-Roadway Segment Crashes for Rural Multilane Undivided
Roadway Segments
• Worksheet SP2D (Corresponds to Worksheet JD (b))-Crashes by Severity Level and Collision Type for Rural
Multilane Undivided Roadway Segments
• Worksheet SP2E (Corresponds to Worksheet I E)-Summary Results for Rural Multilane Roadway Segments
Details of these sample problem worksheets are provided below. Blank versions of the corresponding worksheets
are provided in Chapter II, Appendix II A.
Worksheet SP2A-General Information and Input Data for Rural Multilane Roadway Segments
Worksheet SP2A is a summary of general information about the roadway segment, analysis, input data
(i.e., "The Facts") and assumptions for Sample Problem 2.
Worksheet SP2A. General Information and Input Data for Rural Multilane Roadway Segments
Analyst H1ghv.ray
Date Performed Jurisdiction
Analysis Year
Roadway type (divided/undivided) undivided
Length
___ _"_____
of segment, L (mi)
":____:_'__:_ ___________________ ----------------------- - - - - - - -0.1- - - - - - -
AADT (vehlday) 8,000
Lane width (ft) 12 II
Shoulder width (ft)-right shoulder width for divided 6 2
Shoulder type-right shoulder type for divided paved gravel
Median width (ft)----for divided only 30 N/A
- + - - - - - - - - - - - 1 - - - ------------------
Sideslopes-for undivided only 1:7 or flatter
Lighting (present/not present) not present present
Auto speed enforcement (present/not present) not present present
Calibration factor, C, 1.0 ]_]
Worksheet SP2B-Crash Modification Factors for Rural Multilane Undivided Roadway Segments
ing the CMF values. Once the value for each CMF has been determined, all of the CMFs multiplied together in
Colunm 6 ofWorksheet SP2B which indicates the combined CMF value.
Worksheet SP2B. Crash Modification Factors for Rural Multilane Undivided Roadway Segments
(I) (2) (3) (4) (5) I
(6)
I CMFfor
CMFfor CMFfor CMFfor '' '
CMF for Lighting Automated Speed ' Combined CMF
Lane Width Shoulder Width Sideslopes
I I Enforcement
--~~!r!J I
CMF2111 CMF;,.,
I
CMF 4111
CM:F~':"-·-· CMFcomb
'
from Equation 11-13 from Equation 11-14 from Table 11-14 I
from Equation 11-15 from Section 11.7.1 (I )'(2)'(3)'(4)'(5)
i
1.01 1.10 LOS ! 0.95 0.95 1.05
Worksheet SP2C-Roadway Segment Crashes for Rural Multilane Undivided Roadway Segments
The SPF for the roadway segment in Sample Problem 2 is calculated using the coefficients found in Table 11-3
(Column 2), which are entered into Equation 11-7 (Colunm 3). The overdispersion parameter associated with the
SPF can be calculated using Equation 11-8 and entered into Colunm 4; however, the overdispersion parameter is not
needed for Sample Problem 2 (as the EB Method is not utilized). Colunm 5 represents the combined CMF (from
Colunm 6 in Worksheet SP2B), and Colunm 6 represents the calibration factor. Colunm 7 calculates the predicted
average crash frequency using the values in Column 4, the combined CMF in Column 5, and the calibration factor in
Column6.
Worksheet SP2C. Roadway Segment Crashes for Rural Multilane Undivided Roadway Segments
{1) (2) (3) (4) I {5) ! <6) (7)
Predicted
Average
Overdispersion Combined
SPF Coefficients N,p[ru Crash
Crash Parameter, k CMFs
Calibration Frequency,
Severity
Factor, C, NpN<IIO!Odni
Level
from Table 11-3 from (6) from
from Equation
Equation Worksheet (3)*{5)*(6)
I
a b c 11-8
11-7 SP2B
Total -9.653 1.176 !.675 I o.25o 1.873 ..-- ------:---------
I 1.05 ..- 1.10__
____
I 0.289
--..-··-··--- ---
l
-- - - - ----- ...... ] ·- ------ -----"-""'"'""'-- - -----·-·- -""---
, ,,_
-- --
Fatal and
-9.410 1.094 1.796 0.153 1.660 1.05 1.10 0.177
injury (FI)
Fatal and
-8.577 0.938 2.003 0.086 1.349 1.05 1.10 0.099
injury' (Fl') I I
- -- 1--
Proprty
e I I
(7\oud-{7)Fl
damage only
(PDQ) 1·· 0.112
• Us'mg the KABCO scale, these include only KAB crashes. Crashes with severity level C (possible injury) are not included.
Worksheet SP2D-Crashes by Severity Level and Collision Type for Rural Multilane Undivided
Roadway Segments
Worksheet SP2D presents the default proportions for collision type (from Table 11-4) by crash severity level as fol-
lows:
• Fatal-and-injury crashes, not including "possible-injury" crashes (i.e., on a KABCO injury scale, only KAB
crashes) (Column 6)
(Total), 5 (Fatal and Injury, Fl), 7 (Fatal and Injury, not including "possible injury''), and 9 (Property Damage Only, PDO).
11·48 HIGHWAY SAFETY MANUAL
Worksheet SP2D. Crashes by Severity Level and Collision Type for Rural Multilane Undivided Roadway Segments
Proportion I Nprodlct<d,.. (total) Proportion Npredlctedr..-(FI) Proportion N prodlc<od ,.. (PI") Proportion j N pmllcted ,.. (}'DO)
of Collision : (crashes/ of Collision (crashes/ of Collision (crashes/ of Collision (crashes/
Type (<otall I year) Type (FI) year) Type(Ff'! year) Type (PDOJ year)
(7),oto1 from (7)F1 from (1)FI" from (7)PDO from

Collision from Table Worksheet from Table Worksheet from Table Worksheet from Table Worksheet
Type 11-4 SP2C 11-4 SP2C 11-4 SP2C 11-4 SP2C
Total 1.000 0.289 1.000 0.177 1.000 0.099 1.000 0.112
Head~on
0.009 0.003 0.029 0.005 0.043 0.004 0.001 0.000
collision
Sideswipe
0.098 0.028 0.048 0.008 0.044 0.004 0.120 0.013
collision
Rear-end
0.246 0.071 0.305 0.054 0.217 0.021 0.220 0.025
collision
.......: ............... : ................ ·!······················································! ···--··················:· . . ............................................... ....... .............. .................................... ··································· ................ .
Angle
0.356 0.103 0.352 0.062 0.348 0.034 0.358 0.040
collision ,_,_,_, __ ,,_,,_,_,_,,_,_,_,_,_
+·······························!·······················! ........... ············"·····································
Single-
vehicle 0.238 0.069 0.238 0.042 0.304 0.030 0.237 0.027
collision
Other
0.053 0.015 0.028 0.005 0.044 0.004 0,064 0.007
collision
Worksheet SP2E-Summary Results for Rural Multilane Roadway Segments

Worksheet SP2E presents a summary of the results. Using the roadway segment length, the worksheet presents the
Worksheet SP2E. Summary Results for Rural Multilane Roadway Segments

(1) (2) (3) (4)
Crash Severity Level Roadway Segment Length (mi) Crash Rate (crasheslmilyear)
Frequency (crashes/year)
(7) from Worksheet SP2C (2)/(3)
The Site/Facility
A three-leg stop-controlled intersection located on a rural four-lane highway.
The Question
What is the predicted average crash frequency of the stop-controlled intersection for a particular year?
The Facts
• 3 legs
• Minor-road stop control
• 0 right-tum lanes on major road
• 1 left-tum lane on major road
• 30-degree skew angle
• AADT of major road= 8,000 veh/day
• AADT of minor road= 1,000 veh/day
• Calibration factor= 1.50
• Intersection lighting is present
Assumptions
• Collision type distributions are the default values from Table 11-9.
Results
Sample Problem 3 is determined to be 0.8 crashes per year (rounded to one decimal place).
Steps
Step 1 through 8
11 are conducted. No other steps are necessary because only one intersection is analyzed for one year, and the EB
The SPF for a three-leg intersection with minor-road stop control is calculated from Equation 11-11 and Table 11-7
as follows:
N,pf;,, = exp[a + b x !n(AADTm,) + C X !n{AADTm)]

= exp[-12.526 + 1.204 x /n(S,OOO) + 0.236 x /n(1,000)] = 0.928 crashes/year
specific geometric conditions and traffic control features
Intersection Skew Angle (CMF1)

CMF 11 can be calculated from Equation 11-18 as follows:
0.016xskew +l.O
(0.98 + 0.16 x skew)
The intersection skew angle for Sample Problem 3 is 30 degrees.
CMF!i 0.016x30 +l.O=l.0 8

(0.98+0.16x30)
Intersection Left-Turn Lanes (CMF)

From Table 11-22, for a left-tum lane on one non-stop-controlled approach at a three-leg stop-controlled intersec-
tion, CMF21 = 0.56.

Since no right-tum lanes are present, CMF31 = 1.00 (i.e., the base condition for CMF" is the absence of right-tum
Lighting (CMF4)
CMF 41 can be calculated from Equation 11-22 as follows:
CMF4i = 1.0-0.38 X p"'

From Table 11-24, for intersection lighting at a three-leg stop-controlled intersection, p"' = 0.276.
CMF41 = 1.0-0.38 X 0.276 = 0.90
CMFoomb = 1.08 X 0.56 X 0.90 = 0.54
It is assumed that a calibration factor, C,, of 1.50 has been determined for local conditions. See Part C, Appendix A. I

= 0.928 x 1.50 x (0.54) = 0.752 crashes/year

WORKSHEETS
for an intersection. To apply the predictive method steps, a series of five worksheets are provided for determining
the predicted average crash frequency. The five worksheets include:
• Worksheet SP3A (Corresponds to Worksheet 2A)-General Information and Input Data for Rural Multilane
Highway Intersections
• Worksheet SP3B (Corresponds to Worksheet 2B)--Crash Modification Factors for Rural Multilane Highway
Intersections
• Worksheet SP3C (Corresponds to Worksheet 2C)-Intersection Crashes for Rural Multilane Highway Intersections
• Worksheet SP3D (Corresponds to Worksheet 2D)--Crashes by Severity Level and Collision Type for Rural
Multilane Highway Intersections
• Worksheet SP3E (Corresponds to Worksheet 2E)-Surnmary Results for Rural Multilane Highway Intersections
Details of these sample problem worksheets are provided below. Blank versions of the corresponding worksheets are
provided in Chapter II, Appendix IIA.
Worksheet SP3A-Generallnformation and Input Data for Rural Multilane Highway Intersections
Worksheet SP3A is a summary of general information about the intersection, analysis, input data (i.e., "The Facts")
Worksheet SP3A. General Information and Input Data for Rural Multilane Highway Intersections
General Information 1 Location Information
Analyst Highway
, Analysis Year
Intersection type (3ST, 4ST, 4SG) 3ST
AADT•< (veh/day) 8,000

AADT•'" (veh/day) 1,000
Intersection skew angle (degrees) 0 30
Number of signalized or uncontrolled approaches 0
with a left-tum lane (0, 1, 2, 3, 4)
-'-'----------------··········-·----
Number of signalized or uncontrolled approaches 0 0
with a right-tum lane (0, 1, 2, 3, 4)
Intersection lighting (present/not present) not present present
Calibration factor, C1 1.0 1.5
Worksheet SP3B-Crash Modification Factors for Rural Multilane Highway Intersections

ing the CMF values. Once the value for each CMF has been determined, all of the CMFs are multiplied together in
Column 6 ofWorksheet SP3B which indicates the combined CMF value.
11-52 HIGHWAYSAFETYMANUAL
Worksheet SP3B. Crash Modification Factors for Rural Multilane Highway Intersections
C:MF for
Intersection CMFfor CMFfor
Skew Angle Left-Turn Lanes Right- Turn Lanes CMF for Lighting Combined CMF
CMF 11 CMF, CMFj1 CMF41 CMFc.,•b
from Equations
Crash 11-18 or 11-20 and from
Severity Level 11-19 or 11-21 from Table 11-22 from Table 11-23 Equation 11-22 (1)*(2)*(3)*(4)
Total 1.08 0.56 1.00 0.90 0.54
Fatal and injury (FI) 1.09 0.45 1.00 0.90 0.44
Worksheet SP3C-Intersection Crashes for Rural Multilane Highway Intersections

The SPF for the intersection in Sample Problem 3 is calculated using the coefficients shown in Table 11-7 (Col-
umn 2), which are entered into Equation 11-11 (Column 3). The overdispersion parameter associated with the SPF
is also found in Table 11-7 and entered into Column 4; however, the overdispersion parameter is not needed for
Sample Problem 3 (as the EB Method is not utilized). Column 5 represents the combined CMF (from Column 6 in
Worksheet SP3B), and Column 6 represents the calibration factor. Column 7 calculates the predicted average crash
frequency using the values in Column 3, the combined CMF in Column 5, and the calibration factor in Column 6.
Worksheet SP3C Intersection Crashes for Rural Multilane Highway Intersections

(1) (2) (3) (4) (5) (6) (7)
Predicted
Average Crash
•II Overdispersion I Combined Frequency,
SPF Coefficients Parameter, k i CMFs Npredle!Od/nt
from Tables 11-7 or 11-8 from

I
Crash Equation from (6) of
Severity 11-11 or from Tables I Worksheet I Calibration
Level a b c 1 11-12 j 11-7 or 11-8 SP3B Factor, C, (3)*(5)*(6)
-:::~:~---~-~:-::!~-'I" ---;-~~---j--!::~:---~---~~;!---~-----::;:-:- ---- ~----:~:---l--- ;~!-- --1- ---~~:~-----

------- --------------------- ---~- .. ----. -1·---------. -1-----------------. --------------------------- .. ---
:~a.;~!') j -II 989 I 0!3 0.228 0 270 0.566 I 0.44 l !.50 I. 0.!78
(?J~~;;~ )!c __
7
=fonly . , ___
Worksheet SP3D---<:rashes by Severity Level and Collision Type for Rural Multilane Highway Intersections

• Fatal-and-injury crashes, not including "possible-injury" crashes (i.e., on a KABCO injury scale, only KAB
crashes) (Column 6)

Using the default proportions, the predicted average crash frequency by collision type in Columns 3 (Total),
5 (Fatal and Injury, FI), 7 (Fatal and Iojury, not including "possible injury"), and 9 (Property Damage Only, PDO).
Worksheet SP3D. Crashes by Severity Level and Collision Type for Rural Multilane Highway Iotersections
(I) (2) (3) (4) (5) (6) (7) (8) (9)
Np,...dl«odlm
Proportion Proportion Proportion Proportion
Collision of Collision
(tntll)
(crashes/ of Collision
Npft<llctodlm(FJ)
(crashes/ of Collision I N'"""" '"' (F,,
(crashes/ I
of Collision
int(PDQ)
(crashes/
Type Type rto,.L) year) Type (FI/ year) Type (FI"! year) Type (PDO) year)
(7)totoL from (7}F1 from (7)Ft' from (7)p00 from

from Table Worksheet from Table Worksheet from Table Worksheet from Table Worksheet
11-9 SP3C 11-9 SP3C 11-9 1 SP3C 11-9 SP3C
~~;-1-;-~~~~~~~~~~~~ ~0~-
1. 000 I 0.466
--- ---- __(~l~_(i~~:~-
Sideswipe
collision
0.133
1
0.0~.057
17
I C
0.010 0.179 0.083
Rear-end
0.147
collision
Angle
0.092
collision
Single-
vehicle 0.219 0.063 0,114
collision
Other
0.052 0.039 0.064 O.Dl8 0.044 0.021
collision
Worksheet SP3E-Summary Results for Rural Multilane Highway Intersections

Worksheet SP3E presents a summary of the results.
Worksheet SP3E. Summary Results for Rural Multilane Highway Intersections

(1) - (2)
Crash Severity Level (7) from Worksheet SP3C
Total 0.752
Fatal and (FI) 0.286
Fatal and injury• (FI") 0.178
Property damage only (PDO) 0.466
• Using the KABCO scale, these include only KAB crashes. Crashes with severity level C (possible injury) are not included .
.
;~. .
'.,,.
The Project
A project of interest consists of three sites: a rural four-lane divided highway segment, a rural four-lane undivided
highway segment, and a three-leg intersection with minor-road stop control. (This project is a compilation of road-
way segments and intersections from Sample Problems I, 2, and 3.)
The Question
crash frequencies from Sample Problems I, 2, and 3 and the observed crash frequencies using the site-specific
EBMethod?
The Facts
• 2 roadway segments (4D segment, 4U segment)
• 9 observed crashes (4D segment: 4 crashes; 4U segment: 2 crashes; 3ST intersection: 3 crashes)
Outline of Solution
predicted average crash frequencies for the project using the site-specific EB Method (i.e., observed crashes are as-
signed to specific intersections or roadway segments) presented in SectionA.2.4 of Part C, Appendix A.
Results
WORKSHEETS
To apply the site-specific EB Method to multiple roadways segments and intersections on a rural multilane highway
combined, two worksheets are provided for determining the expected average crash frequency. The two worksheets
include:
• Worksheet SP4A (Corresponds to lfOrksheet 3A)-Predicted and Observed Crashes by Severity and Site Type
Using the Site-Specific EB Method for Rural Two-Lane, Two-Way Roads and Multilane Highways
• lfOrksheet SP4B (Corresponds to Worksheet 3B)-Site-Specific EB Method Summary Results for Rural Two-Lane,
Two-Way Roads and Multilane Highways
provided in Chapter II, Appendix !lA.
Worksheets SP4A-Predicted and Observed Crashes by Severity and Site Type Using the
Site-Specific EB Method for Rural Two-Lane, Two-Way Roads and Multilane Highways
The predicted average crash frequencies by severity type determined in Sample Problems I through 3 are entered into
Columns 2 through 4 of Worksheet SP4A. Column 5 presents the observed crash frequencies by site type, and Column
6 the overdispersion parameter. The expected average crash frequency is calculated by applying the site-specific EB
Method which considers both the predicted model estimate and observed crash frequencies for each roadway segment
and intersection. Equation A-5 from Part C, Appendix A is used to calculate the weighted adjustment and entered into
Column 7. The expected average crash frequency is calculated using Equation A-4 and entered into Column 8.
Worksheet SP4A. Predicted and Observed Crashes by Severity and Site Type Using the Site-Specific EB Method
for Rural Two-Lane. Two-Way Roads and Multilane Highways
(1) (2) (3) (4) (5) (6) (7) (8)
Expected
Average
Weighted Crash
Predicted Average Crash Frequency Adjustment, Frequency,
(crashes/year) w N.,.. oc<od
Observed
Crashes, Equation A-5 Equation A-4
No~od Overdispersion from Parte, from Part C,
Site Type I N ftdlo«d (10101\ N "'dieted (F[J I N mllctod (PDO) I
(crashes/year) Parameter, k Appendix A Appendix A
Roadway Segments
-~~~;~---1-----~;;;----l-----~i;;----I----~;;;---+-----;------I-----;:~~;----+----~:~;----!----~:~;:-----
Intersections
Intersection 1 0.752 0.286 0.466 3 0.460 0.743 1.330
--------- --- f--- ------ - --+------ . ····--
Combined
(Sum of 4.347 2.189 2.158 9 5.747
Co1umo)
Column 7-WeightedAdjustment
The weighted adjustment, w, to be placed on the predictive model estimate is calculated using EquationA-5 from
1
w
l+kx[ L
all study
Np<edicted;
years
Segment 1
w 0.681
1+ 0.142x (3.306)
Segment2
I
w= 0.649
I+ 1.873 X ( 0.289)
Intersection I
I
w 0.743
I+ 0.460 X ( 0.752)
Column 8-ExpectedAverage Crash Frequency

The estimate of expected average crash frequency, N expectecl, is calculated using Equation A-4 from Part C ' Appendix
A as follows:
Nexpected = w x N predicted +(I- w) x Nobserved
Segment 1: N~,_, = 0.681 x 3.306 + (1- 0.681) x 4 = 3.527

Segment 2: N~"''"' = 0.649 x 0.289 + (!- 0.649) x 2 = 0.890
Intersection 1: N~''"'' = 0.743 x 0.752 +(I- 0.743) x 3 = 1.330
Worksheet SP4B-Site-Specific EB Method Summary Results for Rural Two-Lane, Two-Way Roads
Worksheet SP4B. Site-Specific EB Method Summary Results for Rural Two-Lane, Two-Way Roads
(1) (2) (3)
Crash
"_,_,_ . _,_ . . _Severity
. _. . . . _. _. _._,_Level N mllowd
. _. _. _. . _. . _._ . _. _,_ . _. _. _._ . _. _,_,_,_ . . _._ . . _.I_. . _._,_,_,_,_,_ . _,_ . _,_ . _,_,_,_r_ . _,_,_._,_,_,_,_,_,_ . _. _. _,_,_, _. _....I _. . _. _. _,_,_,_ . _,_ . . ___ . . _.,,_,_,_,_._
N.,.p. .....
_. _M. _. _. . _,_,_ . _. _. . _. . _. __,_,_. _. _
, (2).,,.,b from Worksheet SP4A (S)canrb from Worksheet SP4A
Total
4.347 5.7
~
(3).".' from Worksheet 3A I . (3),,.,*(2M(2).~
Fatal and iojury (FI)
2.189 . 2.9
_,,_,_,,_,,_,,_ _,_,_,_,_,,_,_,_,,_,_,_,_,_,_,,_,,_,,_,,_,,_,,_,_, __ ,,_,_,_,,_,,_,,_,,_,_,_,_,_,_,,_,,_,,_,_,_,_,_,_,_,_,_,,_,,_,,_,_ .. _,_,_,_,,_,_,,_,,_,_,_,_,_,_,_,,_,,_,,_
(4\o•• from Worksheet SP4A I (3\,,*(2),DJ(2),,.1

Property damage only (PDQ) r---------------;_;58-------------------------------;~-----------------
11.12.5. Sample ProblemS
The Project
A project of interest consists of three sites: a rural four-lane divided highway segment, a rural four-lane undivided
highway segment, and a three-leg intersection with minor-road stop control. (This project is a compilation of road-
way segments and intersections from Sample Problems I, 2, and 3.)
The Question
What is the expected average crash frequency of the project for a particular year incorporating both the predicted crash
frequencies from Sample Problems 1, 2, and 3 and the observed crash frequencies using the project-level EB Method?
The Facts
• 2 roadway segments (4D segment, 4U segment)
• 9 observed crashes (hut no information is available to attribute specific crashes to specific sites within the project)
Outline of Solution
intersections are not available, but observed crashes are assigned to a facility as a whole) presented in SectionA.2.5
Results
WORKSHEETS
To apply the project-level EB Method to multiple roadway segments and intersections on a rural multilane highway
combined, two worksheets are provided for determining the expected average crash frequency. The two worksheets
include:
Using the Project-Level EB Method for Rural Two-Lane, Two-Way Roads and Multilane Highways
• Worksheet SP5A (Corresponds to Worksheet 4B)-Project-Level Summary Results for Rural Two-Lane, Two-Way
Roads and Multilane Highways
provided in Chapter 11, Appendix llA.
Worksheets SPSA-Predicted and Observed Crashes by Severity and Site Type Using the Project-
Level EB Method for Rural Two-Lane, Two-Way Roads and Multilane Highways
The predicted average crash frequencies by severity type determined in Sample Problems I through 3 are entered
in Colwnns 2 through 4 of Worksheet SP5A. Colwnn 5 presents the observed crash frequencies by site type, and
Colwnn 6 the overdispersion parameter. The expected average crash frequency is calculated by applying the project-
level EB Method which considers both the predicted model estimate for each roadway segment and intersection and
the project observed crashes. Colwnn 7 calculates N,., and Colwnn 8 Nw1• Equations A-10 throughA-14 from Part C,
Appendix A are used to calculate the expected average crash frequency of combined sites. The results obtained from
each equation are presented in Colwnns 9 through 14. Section A.2.5 in Part C, Appendix A defines all the variables
used in this worksheet.
.1..
Worksheet SPSA. Predicted and Observed Crashes by Severity and Site Type Using the Project-Level EB Method
(I) (2) (3) (4) (5) (6) (7)
Predicted Average Crash Frequency (crashes/year) N,..
Observed Crashes,
Nob<;orvod I Overdispersion Equation
Site Type N mll«od(!olol) N mllc!ci (FI) N n-lllmd.(PDO! (crashes/year) Parameter, k A-8 (6)* (2)'
Roadway Segments
Segment 1
Segment 2
Intersections
Intersection 1 0.752 _J -----~:~~o ____ I ___ 026~- _
Combined
4.347 9 1.968
(sum of co!wnn) I
Note: Npredlcted wo = Predicted number of total crashes assuming that crash frequencies are statistically independent
Worksheet SP5A. Continued

(I) (8) (9) (10) (11) (12) (13)
w, N, w, N, N expected/Comb
Equation A-9
Site Type sqrt((6)*(2)) EquationA-10 EquationA-11 EquationA-12 EquationA-13 EquationA-14
Roadway Segments
Intersections
Intersection 1 0.588
Combined
2.009 0.688 5.799 0.684 5.817 5.808
(Sum of Column)
Note: Nprodlcted wo = Predicted number of total crashes assuming that crash frequencies are statistically independent
5 5 5 4 4
NpredictedwO:::;: LkrmjN~j + LkrsjN~j + Lkr41·N;dj + LkmyNi~j + LkisjN~j
J=l )=l )=1 )=I )=1 (A-8)
Nprodlctod .,.
1
= Predicted number of total crashes assuming that crash frequencies are perfectly correlated
5 5 5 4 4
Npredictedwl:::;: L~krmjNrmj + L~krsjNrsj + L~krdjNrdj + L~kim}Nim} + L~kisjNisj
}=1 }=I J=l J=l }=1 (A-9)
Column 9-w,
elements are statistically independent, w,, is calculated using Equation A-10 from Part C, Appendix A as follows:
I
wo = ----,-,.-------
1+ N predicted wO
N predicted (total)
1
=-,...,-,,-
] + 1.968
4.347
= 0.688
Column 1O-N0
N 0, is calculated using Equation A-ll from Part C, Appendix A as follows:
N 0 =w0 xNpredicted (total) +(1-w)xN
0 observed (total)
= 0.688 X 4.347 + (J- 0.688) X 9 = 5.799

Column JJ-w1
I
I + N predicted w1
Npredicted (total)
1
2.009
I +--
4.347
= 0.684
Column 12-N1
The expected crash frequency based on the assumption that different roadway elements are perfectly correlated, N 1,
is calculated using Equation A-13 from Part C, Appendix A as follows:
= 0.684 X 4.347 + (J- 0.684) X 9 = 5.817

Column 13-Napeeted/comh
The expected average crash frequency based of combined sites, N •. ,, is calculated using Equation A-14 from
c:q~ectewcom
No+Nl
Nexpected!comb = 2
5.799 + 5.817
= .::..:..:..:..=--:-=-:..::..::..:.
2
= 5.808
.II...
Worksheet SPSB-Project-Level EB Method Summary Results for Rural Two-Lane, Two-Way Roads
Worksheet SPSB. Project-Level EB Method Summary Results for Rural Two-Lane, Two-Way Roads
(1) (2) (3)
(2)co,.b from Worksheet SP5A (13)comb from Worksheet SP5A

Total
4.347 5.8

2.189 2.9
(4tm, from Worksheet SP5A (3).,. '(2),,J(2).,.
_,_,_,_,_,_,_,_,_,_._,_, __ ,_,_,_,_,_,_,_,_,_,_,,_,_,_,_,_,_,_,_,_,.J.,_,_,_,_,_,_,_,,_,_,,_,,_,,_,,_,_,_, __ ,_ .. ,_,_,_,_,,_,_,,_,,_,_,_,_,_,_,_,_,_,
~158 ~9
The Project
An existing rural two-lane roadway is proposed for widening to a four-lane highway facility. One portion of the
project is planned as a four-lane divided highway, while another portion is planned as a four-lane undivided highway.
There is one three-leg stop-controlled intersection located within the project limits.
The Question
What is the expected average crash frequency of the proposed rural four-lane highway facility for a particular year,
and what crash reduction is expected in comparison to the existing rural two-lane highway facility?
The Facts
• Existing rural two-lane roadway facility with two roadway segments and one intersection equivalent to the facili-
ties in Chapter IO's Sample Problems I, 2, and 3.
• Proposed rural four-lane highway facility with two roadway segments and one intersection equivalent to the facili-
ties in Sample Problems I, 2, and 3 presented in this chapter.
Outline of Solution
Sample Problem 6 applies the Project Estimation Method I presented in Section C.7 of the Part C-Introduction
and Applications Guidance (i.e., the expected average crash frequency for existing conditions is compared to the
predicted average crash frequency of proposed conditions). The expected average crash frequency for the existing
. rural two-lane roadway can be represented by the results from applying the site-specific EB Method in Chapter 1O's
Sample Problem 5. The predicted average crash frequency for the proposed four-lane facility can be determined from
the results of Sample Problems I, 2, and 3 in this chapter. In this case, Sample Problems I through 3 are considered
to represent a proposed facility rather than an existing facility; therefore, there is no observed crash frequency data,
and the EB Method is not applicable.
Results
The predicted average crash frequency for the proposed four-lane facility project is 4.4 crashes per year, and the
predicted crash reduction from the project is 8.1 crashes per year. Table 11-26 presents a summary of the results.
Table 11-26. Summary of Results for Sample Problem 6

Expected Average Crash Predicted Average Crash Predicted Crash Reduction
Frequency for the Existing Frequency for the Proposed from Project Implementation
Site Condition (crashes/year)" Condition (crashes/year)b (crashes/year)
Segment 1 8.2 3.3 4.9
Segment 2 1.4 0.3 1.1
Intersection 1 2.9 0.8 2.1
Total 12.5 4.4 8.1
• From Sample Problems 5 in Chapter 10

b From Sample Problems 1 through 3 in Chapter 11
11.13. REFERENCES
( I) Elvik, R. and T. Vaa. The Handbook of Road Safety Measures. Elsevier Science, Burlington, MA, 2004.
( 2) FHWA. Interactive Highway Safety Design Model. Federal Highway Administration, U.S. Department of
Transportation, Washington, DC. Available from http://www.tfhrc.gov/safety/ihsdm/ihsdm.htrn.
( 3) Harkey, D.L., S. Raghavan, B. Jongdea, F.M. Council, K. Eccles, N. Lefler, F. Gross, B. Persaud, C. Lyon, E.
Hauer, and J. Bonneson. National Cooperative Highway Research Program Report 617: Crash Reduction
Factors for Traffic Engineering and ITS Improvement. NCHRP, Transportation Research Board, Washington,
DC,2008.
( 4) Harwood, D.W, E.R.K. Rabbani, K.R. Richard, H.W McGee, and G.L. Gittings. National Cooperative High-
way Research Program Report 486: Systemwide Impact ofSafety and Traffic Operations Design Decisions for
3R Projects. NCHRP, Transportation Research Board, Washington, DC, 2003.
( 5) Lord, D., S.R. Geedipally, B.N.Persaud, S.P.Washington, I. van Schalkwyk, J.N. Ivan, C. Lyon, and T. Jonsson.
National Cooperative Highway Research Program Document 126: Methodology for Estimating the Safety
Performance of Multilane Rural Highways. (Web Only). NCHRP, Transportation Research Board, Washing-
ton, DC, 2008.
( 6) Srinivasan, R., C. V. Zegeer, F. M. Council, D. L. Harkey, and D. J. Torbic. Updates to the Highway Safety
Manual Part D CMFs. Unpublished memorandum prepared as part of the FHWA Highway Safety Informa-
tion System Project. Highway Safety Research Center, University ofNorth Carolina, Chapel Hill, NC, July
2008.
( 7) Srinivasan, R., F. M. Council, and D. L. Harkey. Calibration Factors for HSM Part C Predictive Models.
Unpublished memorandum prepared as part of the FHWA Highway Safety Information System Project.
Highway Safety Research Center, University of North Carolina, Chapel Hill, NC, October 2008.
( 8) Zegeer, C. V., D. W Reinfurt, W W Hunter, J. Hummer, R. Stewart, and L. Herf. Accident Effects of Side-
slope and Other Roadside Features on Two-Lane Roads. Transportation Research Record 1195, TRB, National
Research Council, Washington, DC, 1988. pp. 33-47.
,;L.
APPENDIX 11A-WORKSHEETS FOR APPLYING THE PREDICTIVE METHOD FOR RURAL

MULTILANE ROADS
Worksheet 1A. General Information and Input Data for Rural Multilane Roadway Segments
Analyst Highway
Agency or Company
--·- ·-·- ,_,_,_, ___
,_,_,_,_,_,_,,_,_,_,,_,,_ ,_,_,_,_,_,_,_,_,_,,_,_,_,_, __ Roadway Section
__ ,_,_,_,_,_,,_,_,_,,_,_,_,,_,_,_,_ ,_,_,_,_,_,_,_,_
,_,_,_,_,_,,_,_ ,_,,_,,_,,_,,_, _,_,_,,_,_,,_,_,_,,_,_,,_,_,_,_,_,_,_,_,_,_,,_,_,_,_,_,_,_,_,,_,_
Analysis Year
Roadway type (divided/undivided)
Length of segment, L (mi)
AADT (veh!day)
Lane width (ft) 12
Shoulder width (:ft)-right shoulder width for divided 8.
Shoulder type-right shoulder type for divided paved_
Median width (ft)-for divided only 30
Sideslopes-for undivided only 1:7 or flatter

Lighting (present/not present) not 'present
Auto speed enforcement (present/not present) not present
Calibration factor, C, LO
Worksheet 1B (a). Crash Modification Factors for Rural Multilane Divided Roadway Segments
(1) (2) (3) (4) I
(5) I
(6)
C:MF for CMF for Right CMF for I CMF for Auto I
Lane Width Shoulder Width Median Width CMF for Lighting Speed Enforcement Combined CMF
Worksheet 1B (b). Crash Modification Factors for Rural Multilane Undivided Roadway Segments
(1) (2) I (3) I (4) (5) (6)
C:MF for CMFfor CMFfor CMFforAuto

Lane Width Shoulder Width Sideslopes CMF for Lighting Speed Enforcement Combined CMF
- -- - - -~~,~
- -- - - -- - - - - ----
from Equation 11~13 from Equation 11~14
~ -~~2> ---~~'"-
-r---from Table
----- ------ ~-~"'-- ---- _,_,_,_,,_,_~~~~~
11~14
from Equation from Section 11.7.1 11~15
. -·-"-"_"_"_ ,_,_, __ ,_,,_,_~-~~.~f_,_,_ ,_,_,_
(1)'(2)'(3)'(4)'(5)
~ ....... ... ..
~ ~ ·~"'""""""""'-"'"""~"-"-'- ,_ .. ,_ .. ,_, _,_,_,_,,_,,_,,_,_,,_,,_,,_,,_,,_,,_,,_,,_,_, ,_,_,_,_ .. ,_,_,_,_,,_,_,,_,,_,,_,_,,_,_,T"_"_"'""'-"'_"_"'-"'-"'-'- ,_,_,_,_,_,_..,,. """-"'""'-"'_"_,_,,_,,_,_,_,_,,_,_ ,_ ·- ·- - _,_,,_,,_,,_,,_,_,_ - - - ·- - - _,
Worksheet 1 C (a). Roadway Segment Crashes for Rural Multilane Divided Roadway Segments
(1) (2) (3) (4) (5) (6) (7)
Predicted
Average Crash
Overdispersion Combined Frequency,
SPF Coefficients N,pfrJ Parameter, k CMFs Np...,d!OIO<I,...
Crash from Table 11-S from (6) from

Severity Equation from Equation Worksheet Calibration
Level • b c 11-9 11-10 !B (a) Factor, C, (3)'(5)'(6)
Total -9.025 1.049 1.549

---------·--- --------- ---------+------ +-·--- -------+-----·--- -----+-------·--- ---1------·---· -+- ------------
Fatal and
-8.837 0.958 1.687
injury (FI)
Fatal and
-8.505 0.874 1.740
InJury' (FI')
Property
damage only
(PDO) I I
Worksheet 1C (b). Roadway Segment Crashes for Rural Multilane Undivided Roadway Segments
(1) (2) (3) (4) (5) (6) (7)
Predicted Average
Overdispersion Combined Crash Frequency,
SPF Coefficients N,,,. Parameter, k CMFs N"""'kt<d"'
Crash from Table 11-3 from (6) from
Severity Equation from Equation Worksheet Calibration
Level • b c 11-7 11-8 1B(b) Factor, C, (3)'(5)'(6)
Total -9.653 1.176 1.675
Fatal and
-9.410 1.094 1.796
injury (FI)
--'--'-' -·'···-··-·-\-- -- -·-+- -- ----1-- ------ --- -- ----- -·-·- ---- -- ----- --. ---·-·-·-- ----- - -- - ------ --- -- ---- -. -. ---.
Fatal and
-8.577 0.938 2.003
injury' (Fl')
Property
damage only
(PDO)
• Using the KABCO scale, these include only KAB crashes. Crashes with severity level C (possible injury) are not included .
.,1,.
Worksheet 1D (a). Crashes by Severity Level and Collision Type for Rural Multilane Divided Roadway Segments
(I) (2) (3) (4) (5) (6) (7) (8) (9)
Proportion N pr<dl«odn(ma&l) I
Proportion Np,dk1«1 "'(FI) Proportion Nprtdlcrod n1 (Ffl) Proportion
of Collision (crashes/ of Collision (crashes/ of Collision (crashes/ of Collision Npr<dlc«:d"'
Type (rotal) year) ,
1 Type rn; year) Type rFP; year) Type fPDo; (PIJO)
(7)totol from (7)n from (1)Ff' from (1)Poo from

Type 11-6 lC (a) 11·6 lC (a) 11-6 lC (a) 11-6 lC (a)
Total
1.000 ___ I _ - r 1000 1.000 i 1.000
__ (2)*(3\,.,- ---- -- --- -- _(4)_*_(5)_,, -- _________ :t_(6):(7)"

- -· -·- T -
(8)'(9)MO
_,_,_,_, .......................................... _,_,_ - _,_,_,_
Head-on
0.006 0.013 0.018 0.002
collision
------- -·-
Sideswipe
0.043 0.027 0.022 0.053
collision
1
--------
---------f------l-------r------+------+------~-----+-------
Rear-end
0.116 0.163 0.114 0.088
collision
Angle
0.043 0.048 0.045 0.041
collision
Single-
vehicle 0.768 0.727 0.778 0.792
collision
---------
Other
0.024 0.022 0.023 0.024
collision
• Using the KABCO scale, these include only KAB crashes. Crashes with severity level C {possible injury) are not included.
Worksheet 1D (b). Crashes by Severity Level and Collision Type for Rural Multilane Undivided Roadway Segments
w w w oo ~ oo m oo ~
Proportion ' Npr<dlO!Odr.r(IO!Ol)
Proportion Np.,dla«<.,(l'[) Proportion N pmllotod ,... (Ffl) Proportion Np=llotedrs(PDQ)
of Collision (crashes/ of Collision (crashes/ of Collision (crashes/ of Collision (crashes/
Type (1o1on year) Type tFIJ year) Type tFflJ year) Type rPDOI year)
(7) 101 oJ from (7)F, from (7)Ffl from from
(1)PDO
Type 11-4 lC (b) 11-4 lC (b) 11-4 lC (b) 11-4 IC (b)
Total 1.000 1.000 ___ !_~oo___ -~------ ___ ___ !.~~o__ __
------·---- __(6)*\1)"-- -·- -- (S)*(?)'29_
Head·on
0.009 0.029 0043 0001
collision -· _,_,_,_,,, . ,_. ,,. ,_,,_,_,___ ,,_,_,,_,,_,_,_,_,_ - - _,_,_,_,,_,,_,,_,,_,,_,t,_"_, ___,_,___,___,_ - _,_, ___,......................T"_,,_, _________,_,__ -·- - _,_,_,,............,. ,_,,_ -·-·- _,_,_, _,_,_,_,_
Sides'Nipe
0.098 0.048 0.044 0.120
collision
Rear-end
0.246 0.305 0.217 0.220
collision I
~~~~~n +-:-::-:-:-+----------------+---::-:-::--r-----t--·-:-· -:- :- - _- ----~1 i----_-_. . ._. . u ; =--.. -_· __

collision
-------L----------------------~-----------r
1
Other
0.053 0.028 0.044 0.064
collision
• Using the KABCO scale, these include only KAB crashes. Crashes with severity level C {possible injury) are not included.
Worksheet 1 E. Summary Results for Rural Multilane Roadway Segments

(I) (2) (3) (4)

Frequency (crashes/year) I Crash Rate (crashes/mi/year)
Crash Severity Level (7) from Worksheet IC (a) or (b) Roadway Segment Length (mi) I (2)/(3)
Total

Fatal and injury" (Fl')
-- . -- ----- --- ----- -- ------ --- -- ------ ----- --_L_ ------ ---------------------------------- ---
Worksheet 2A. General Information and Input Data for Rural Multilane Highway Intersections
General Information I Local Information
-~~.~~~:. _. _,_,_,_ _,_,_,_,,_,_,,_,,_,_ ·-·-..·-- _,- -·-,_,_,_,_,,_,_,,_,,_,_,,_,_,_,_ ·-~~~-~~~-. ,_,_,_,_,,_,,_,_,_,,_,_,_,_,_,_,,_,,_,_,,_,,_,,_,_j_,_,_,_,_,_,_,__,_,_,_,_,,_,,_,,_,_,_,_,_,_ ,_,_,_,_,_,_,_,_,,_

Agency or Company
_,_,_,_,_·- ..-·-·-·-- _..
_,,_,_,,_,_,_,_,,_,_,_,,_,_,_,_,_,_,_,_,_,_,_,,_, __ Intersection
_,_,_,_,,_,,_,,_,_,_, __
,_,_,_,_,_,_,_,,_,_,_,_,_,_,_,,_,,_,_,_,_,_,_,
-~ __,_ -- -·--· --·-
,_,_,_,_,_,,_,,_,_,,_,_,_,_,,_,_,_,_ ,_,_,_,_,_,,_,,_,,_,,_,,
'
Analysis Year

Intersection type (3ST, 4ST, 4SG)
AADT•< (vehlday)
AADT m (veh/day)
--- --~"---'···-- --···-'·'···--- ·-- -~~ --------- ------ - -. -- ---- -+- -- ---- --------- ---- ---- -- ---1 -- -- --- - -- ---- --- --- - -- -- - - --
Intersection skew angle (degrees) 0
~------- - - - - - - - - - ' - - ....•. 0 ..• _,_ - - - - - - - - - - - - - - - - - - -'---------~~--- - - - - - - - - -- - -~ ----+----------- --------- ----- ----
with a left-tum lane (0, 1, 2, 3, 4)
with a right-tum lane (0, L 2, 3, 4) j
Worksheet 2B. Crash Modification Factors for Rural Multilane Highway Intersections
(!) (2) (3) (4) (5) (6)
I
CMFfor I
Intersection CMFfor CMFfor I
Skew Angle Left-Turn Lanes Right-Tum Lanes C:MF for Lighting I
CMF 11 CMF21 CMFJI CMF41 Combined CMF

from Equations I
Crash 11-18 or 11-20 and from Equation I
Severity Level 11-19 or 11-21 from Table 11-22 from Table 11-23 11-22 1
(1)"'(2)*(3)*(4)
.ll,.
!"'"
Worksheet 2C Intersection Crashes for Rural Multilane Highway Intersections

CIJ C2J I (3J c•J C5J (6) (7)
Predicted
·~ Average Crash
Overdispersion Combined Calibration Frequency,
SPF Coefficients Parameter, k CMFs Factor N redl01od lnr
Crash from Table 11~7 or 11~8 from from (6) of I I

Severity 1 Equation from Table Worksheet
Level a b j c 11~11 or 11-12 11~7 or 11~8 2B C, (3)'(5)'(6)
Total
Fatal and
m)ury (Fl)
Fatal and
injury" (FI")
Property
damage only
(PDO)
Worksheet 20- Crashes by Severity Level and Collision Type for Rural Multilane Highway Intersections
(!) (2) (3) (4) (5) (6) (7) (8) (9)
!
Npredlot.d Nprec~lctcd
Proportion lnr(l<ltal)
Proportion 1 Npn.-dlored/m(F/) Proportion Np,.,dlctcd In/ (FI')
Proportion int(PDO)
of Collision (crashes/ of Collision (crashes/ of Collision (crashes/ of Collision (crashes/
Type rtotal) year) l)rpe (FI) year) Type fFf'J year) Type (PDO) year)
(7)total
from (7)F1 from (7)FI' from (7)Pno from
Collision from Worksheet from Worksheet from Worksheet from Worksheet
Type Table 11~9 2C Table 11~9 2C Table 11~9 2C Table 11~9 2C
Total 1.000 1.000 1.000 1.000
(2)'(3) •• (4)'(5), (6)'(7),_ (8)'(9),DD
Head-on •
collision
Sideswipe
collision
Rear-end
collision
Angle
. ·····················
II ' • . . I
collision
.... ----+-----· --- . ----- ---- -- _j __ -- - - - . . - - - - - - - - -- - . · ; . . . - - . ----·--- - - - -- -- --1- ---- ---- ..
r-1 . ·
Single-
vehicle
collision
Other
collision
Worksheet 2E. Summary Results for Rural Multilane Highway Intersections

(!) (2)
Crash Severity Level (7) from Worksheet 2C
. . !o_tal_______________________________________________ :_______________ -···-···-··------ _____ ---·· -· -· -·- ________________ _

_,,_,_,_,_,,_,,_,_,,_,,_,_,_,_,_,_,_,_,,_,,_,_,,_,_,_,_,_,_,_,,_,_,_,_,_ .. ,_,,_,_,_,_,_,_,_,_,,_,_,_,_,_,_,_,_,_,_,_ ,_,_,_,_,_,,_,_,_,,_,_,,_,_,_,_,_,_,_,_,,_,,_,_,_,_,_,_,_,_,_,_,_,_,,_,,_,,_,,_,_,_,_._ .. _,_,, ___ ,_,,_,_,,_,_,_,_,_,_
Fatal and injury" (FI') 1

---------------------------------------------------1·---------------------------------------------------
Worksheet 3A. Predicted and Observed Crashes by Severity and Site Type Using the Site-Specific EB Method
!
(!) (2) I (3) (4) (5) (6) I (7) (8)
I
'
' Weighted Expected Average
Predicted Average Crash Frequency Adjustment, Crash Frequency,
(crashes/year) w N.,. ec1ed
Observed
'
Crashes, Equation A-5 !
Noboorwd Overd.ispersion from PartC, Equation A-4 from

Site Type
N """""'""' I N,.,.,,• .,,. ~redlctcd (PDO! (crashes/year) Parameter, k I' Appendix A Part C,AppendixA
Roadway Segmen[s
ln[ersections
Intersection 1
Intersection 2
t=--_:___
Intersection 3
Intersection 4
I··
Intersection 5 I__ - - - - - - - - - . . . . -
- - - - - - - - - - - - - ___ l_
Intersection 6
--·· -··················- ----
IntersectiOn 7
·--------. ---------------- ··------------- ___ l _____________ [_ __ _
Intersecnons__ J ________ -------------~------- _________ ----------------------------------------------------
'
Combmed I
(Sum of I'
Column)
.Li
Worksheet 38. Site-Specific EB Method SlllD111lU)' Results

(I) (2) (3)
Crash Severity Level Npmll«od
(2)~omh from Worksheet 3A (B).,.,b from Worksheet 3A

Total
(3)"'""b from Worksheet 3A (3),, <;o) If?)

(4).,mb from Worksheet 3A

Worksheet 4A. Predicted and Observed Crashes by Severity and Site Type Using the Project-Level EB Method
(I) (2) (3) (4) (5) I (6) (7)
N~
Observed Crashes,
No~>t;e,...d Overdispersion Equation
Site Type , N (crashes/year) Parameter, k , A-8 (6)* (2)'
I "'dkted(!ol.ol) N "'dlc!O<I(PDOJ
Roadway Segments
Segment 6
Segment 7
Segment 8
---~-·
·------~--- =t---
Intersections
Intersection 1
Intersection 2
Intersection 3
Intersection 4
Intersection 5
Intersection 6
Intersection 7
Intersection 8
Combined
(Sum of
Colunm)
Worksheet 4A continued on next page.

Worksheet 4A. Continued

(1) (8) (9) (10) (11) (12) (13)
Site Type w, N, w, N,
Equation A-9
sqrt((6)*(2)) EquationA-10 Equation A-ll EquationA-12 Equation A-13 Equation A-14
Roadway Segments
l-- --- -=---

I I
Segment 1 1
- - - -· - .. - ... - ...........,_,_,,_,_,- T - ,_,_,,_,,_,,_,,_,_,_,_,_,_,_ _,_,_,,_,i"'-"'_"_"_"_'- ,_. ,_,_,,_,,_,__,,_,,_,1!-.. -.. -..-·- ·-"-"-""""""'-"'-"'-"'_"_';-""- -

Segment2 1
·---'- · - - - - - - - - - - - - - - - - - -
Segment 3
Segment4
~!¥~···r;·~••J••••~····n·•J••••••~··••••• J••• ~n·•·J••·~~

Intersections
Intersection 1
Intersection 2
Intersection 3
Intersection 4
Intersection 5
Intersection 6
Intersection 7
Intersection 8
Combined (Sum
of Column)
Worksheet 48. Project-Level EB Method Summary Results

(1) (2) (3)
Crash Severity Level Npftdlc<ed [ N.,.po<1N

_,_,_,_,_,_,_ - · - · - ,_,_,,_,_,_,_,,_,,_,_,,_,_,,_,,_,,_,,_,,_,,_,_,_,_,_,_,_,_,,L., _._,_,_,_,_,_,_,_,_,,_,,_,_,,_,,_,,_,,_,_,_,_,_,_,_,_,_,_,_,_,,_,_,_,_,_,_' _,,_,_,_,,_,_,_,_,_,_,_,_,_,_,_,_, __ , _,_,,_,,_,,_,,_,_,,_,,_,,_,,_,,_,,_,,_,,_,_,,_,,
Total ,! . _. _,_,_,_ . . _. _,_(2)..,mb

. ,_. _,_,_ . from Worksheet 4A __ . _,_,_,_,_,_1,_
,_. ,_. ,_. _,_,_,_,_,_,_,_,_,_,_, ! (13).,.,mb
. . _. _. ,_. ,_. _,_,_,_,_ . ,_. _,_,_,from Worksheet. ,_ .4A
__ . _,_,_,_,_,_,_,_ ,_. ,_. _........ _. _,_,_, __ ,_
--- ----1 -"". ""--- ---- ----- -·-- "----- """-"-"'" -----
I '
Fatal and injury (FI) ~ (3)ro•o from Wo;ksheet 4A --·--- "--('=3").,..,,_'('::2'!.!)':!./('=2")'='"'''-------
Property damage only (PDO)
APPENDIX 118-PREDICTIVE MODELS FOR SELECTED COLLISION TYPES

The main text of this chapter presents predictive models for crashes by severity level. Tables with crash proportions
by collision type are also presented to allow estimates for crash frequencies by collision type to be derived from
the crash predictions for specific severity levels. Safety prediction models are also available for some, but not all,
collision types. These safety prediction models are presented in this appendix for application by HSM users, where
appropriate. Users should generally expect that a more accurate safety prediction for a specific collision type can
.,l, i
be obtained using a model developed specifically for that collision type than using a model for all collision types
combined and multiplying the result by the proportion of that specific collision type of interest. However. prediction
models are available only for selected collision types. And such models must be used with caution by HSM users
because the results of a series of collision models for individual collision types will not necessarily sum to the pre-
dicted crash frequency for all collision types combined. In other words, when predicted crash frequencies for several
collision types are used together, some adjustment of those predicted crash frequencies may be required to assure
that their sum is consistent with results from the models presented in the main text of this chapter.
11 8.1 Undivided Roadway Segments

Table IlB-I summarizes the values for the coefficients used in prediction models that apply Equation 11-4 for estimating
crash frequencies by collision type for undivided roadway segments. Two specific collision types are addressed: single-
vehicle and opposite-direction collisions without turning movements (SvOdn) and same-direction collisions without
turning movements (SDN). These models are assumed to apply for base conditions represented as the average value of the
variables in a jurisdiction. There are no CMFs for use with these models; the crash predictions provided by these models
are assumed to apply to average conditions for these variables for which CMFs are provided in Section 11.7.
Table 11 B-1. SPFs for Selected Collision Types on Four-Lane Undivided Roadway Segments
(Based on Equation 11-4)
Overdispersion Parameter
Severity LeveVCollision Type a b (Fixed k)•
Total-SvOd.n -5.345 0.696 0.777

Fatal and Injury-SvOdn -7.224 0.821 0.946
Fatal and Injuryb-SvOdn -7.244 0.790 0.962
Total-SON -14.962 1.621 0.525
Fatal and Injury-SON -12.361 1.282 0.218
Fatal and Jnjury"-SDN -14.980 1.442 0.514
Note: SvOdn-Single Vehicle and Opposite Direction without Turning Movements Crashes {Note: These two crash types were modeled together)
SDN-Same Direction without Turning Movement (Note: This is a subset of all rear·end collisions)
b Excluding crashes involving only possible injuries.
Divided Roadway Segments

No models by collision type are available for divided roadway segments on rural multilane highways.
Stop-Controlled Intersections
Table llB-2 summarizes the values for the coefficients used in prediction models that apply Equation 11-4 for
estimating crash frequencies by collision type for stop-controlled intersections on rural multilane highways. Four
specific collision types are addressed:
• Single-vehicle collisions
• Intersecting direction collisions (angle and left-tum-through collisions)
• Opposing-direction collisions (head-on collisions)
• Same-direction collisions (rear-end collisions)
Table l!B-2 presents values for the coefficients a, b, c, and d used in applying Equations 11-11 and 11-12 for predicting
crashes by collision type for three- and four-leg intersections with minor-leg stop-control. The intersection types and
severity levels for which values are shown for coefficients a, b, and care addressed with the SPF shown in Equation JI-
ll. The intersection types and severity levels for which values are shown for coefficients a and d are addressed with the
SPF shown in Equation 11-12. The models presented in this exhibit were developed for intersections without specific
base conditions. Thus, when using these models for predicting crash frequencies, no CMFs should be used, and it is as-
sumed that the predictions apply to typical or average conditions for the CMFs presented in Section 11.7.
Table 11 B-2- Collision Type Models for Stop-Controlled Intersections without Specific Base Conditions
(Based on Equations 11-11 and 11-12)
Intersection Type/Severity
Level/Collision Type a b c d Overdispersion Parameter (Fixed k) 1
4ST Total Single Vehicle -9.999 0.950 0.452
4ST Fatal and Injury

-10.259 0.884 0.651
Single Vehicle
4ST Fatal and Injuryb
-9.964 0.800 1.010
Single Vehicle
4ST Total Int. Direction -7.095 0.458 0.462 1.520

-7.807 0.467 0.505 1.479
Int. Direction
-7.538 0.441 0.420 1.506
Int. Direction
4ST Total Opp. Direction -8.539 0.436 0.570 1.068

10.274 0.465 0.529 1.453
Opp. Direction
-10.058 0.497 0.547 1.426
Opp. Direction
4ST Total Same Direction -11.460 0.971 0.291 0.803
-11.602 0.932 0.246 0.910
Same Direction
-13.223 1.032 0.184 1.283
Same Direction
3ST Total Single Vehicle -10.986 1.035 0.641

-10.835 0.934 0.741
Single Vehicle
-11.608 0.952 0.838
Single Vehicle
3ST Total Int. Direction -10.187 0.671 0.529 1.184

-11.171 0.749 0.487 1.360
Int. Direction
-12.084 0.442 0.796 1.5375
Int. Direction
3ST Total Opp. Dtrection -13.808 1.043 0.425 1.571
-14.387 1.055 0.432 1.629
Opp. Direction
-15.475 0.417 1.105 1.943
Opp. Direction
3ST Total Same Direction -15.457 1.381 0.306 0.829
-14.838 1.278 0.227 0.754
Same Direction
-14.736 1.199 0.147 0.654
Same Direction
Note: Int. Direction = Intersecting Direction (angle and left-turn-through crashes)

Opp, Direction= Opposing Direction {head-on)
b Excluding crashes involving only possible injuries.
Signalized Intersections
No models by collision type are available for signalized intersections on rural multilane highways.
Chapter 12-Predictive Method
for Urban and Suburban Arterials
12.1.1NTRODUCTION
This chapter presents the predictive method for urban and suburban arterial facilities. A general introduction to the
Highway Safety Manual (HSM) predictive method is provided in the Part C-Introduction and Applications Guidance.
The predictive method for urban or suburban arterial facilities provides a structured methodology to estimate the
expected average crash frequency, crash severity, and collision types for facilities with known characteristics. All
types of crashes involving vehicles of all types, bicycles, and pedestrians are included, with the exception of crashes
between bicycles and pedestrians. The predictive method can be applied to existing sites, design alternatives to exist-
ing sites, new sites, or for alternative traffic volume projections. An estimate can be made for crash frequency in a
period of time that occurred in the past (i.e., what did or would have occurred) or in the future (i.e., what is expected
to occur). The development of the SPFs in Chapter 12 is documented by Harwood eta!. (8, 9). The CMFs used in
this chapter have been reviewed and updated by Harkey et al. (6) and in related work by Srinivasan et al. (13). The
SPF coefficients, default collision type distributions, and default nighttime crash proportions have been adjusted to a
consistent basis by Srinivasan et al. (14).
This chapter presents the following information about the predictive method for urban and suburban arterial facilities:
• The definitions of the facility types included in Chapter 12, and site types for which predictive models have been
developed for Chapter 12.
• Details for dividing an urban or suburban arterial facility into individual sites, consisting of intersections and
roadway segments.
• Safety performance functions (SPFs) for urban and suburban arterials.

• Guidance for applying the Chapter 12 predictive method, and limitations of the predictive method specific to
Chapter 12.
• Sample problems illustrating the application of the Chapter 12 predictive method for urban and suburban arterials.

The predictive method provides an 18-step procedure to estimate the "expected average crash frequency" '
N expected
(by total crashes, crash severity, or collision type) of a roadway network, facility, or site. In the predictive method,
the roadway is divided into individual sites, which are homogenous roadway segments and intersections. A facility
12·1
consists of a contiguous set of individual intersections and roadway segments referred to as "sites." Different facility
types are determined by surrounding land use, roadway cross-section, and degree of access. For each facility type, a
number of different site types may exist, such as divided and undivided roadway segments and signalized and unsig-
nalized intersections. A roadway netvvork consists of a number of contiguous facilities.
of all sites used as the estimate for an entire facility or network. The estimate is for a given time period of interest (in
years) during which the geometric design and traffic control features are unchanged and traffic volumes are known
or forecasted. The estimate relies on estimates made using predictive models which are combined with observed
crash data using the Empirical Bayes (EB) Method.
The predictive models used within the Chapter 12 predictive method are described in detail in Section 12.3.
The predictive models used in Chapter 12 to predict average crash frequency, N,.diaod' are of the general form shown
in Equation 12-1.
(12-1)
Where:
N predicred = predicted averae:e

....., crash frequency for a specific year on site type x;
N,Pf• = predicted average crash frequency determined for base conditions of the SPF developed for site type x;
N~"'"' = predicted average number of vehicle-pedestrian collisions per year for site type x;
N""" = predicted average number of vehicle-bicycle collisions per year for site type x;
CMF"" = crash modification factors specific to site type x and specific geometric design and traffic control features
y;and
ex= calibration factor to adjust SPF for local conditions for site type X.
The predictive models in Chapter 12 provide estimates of the crash severity and collision type distributions for road-
way segments and intersections. The SPFs in Chapter 12 address two general crash severity levels: fatal-and-injury
and property-damage-only crashes. Fatal-and-injury crashes include crashes involving all levels of injury severity in-
cluding fatalities, incapacitating injuries, nonincapacitating injuries, and possible injuries. The relative proportions of
crashes for the two severity levels are determined from separate SPFs for each severity level. The default estimates
of the crash severity and crash type distributions are provided with the SPFs for roadway segments and intersections
in Section 12.6.
12.3. URBAN AND SUBURBAN ARTERIALS-DEFINITIONS AND PREDiaiVE MODELS IN CHAPTER 12

This section provides the definitions of the facility and site types and the predictive models for each of the site types
in Section 12.4.
12.3.1. Definition of Chapter 12 Facility Types

The predictive method in Chapter 12 addresses the following urban and suburban arterial facilities: two- and four-
lane undivided facilities, four-lane divided facilities, and three- and five-lane facilities with center two-way left-tum
lanes. Divided arterials are nonfreeway facilities (i.e., facilities without full control of access) that have lanes in the
two directions of travel separated by a raised or depressed median. Such facilities may have occasional grade-sepa-
rated interchanges, but these are not the primary form of access. The predictive models do not apply to any section
of an arterial within the limits of an interchange which has free-flow ramp terminals on the arterial of interest. Ar-
terials with a flush separator (i.e., a painted median) between the lanes in the two directions of travel are considered
CHAPTER 12- PREDICTIVE METHOD FOR URBAN AND SUBURBAN ARTERIALS 12-3
undivided facilities, not divided facilities, Separate prediction models are provided for arterials with a flush separator
that serves as a center two-way left-tum lane. Chapter 12 does not address arterial facilities with six or more lanes.
The terms "highway" and "road" are used interchangeably in this chapter and apply to all urban and suburban arteri-
als independent of official state or local highway designation.
Classifying an area as urban, suburban, or rural is subject to the roadway characteristics, surrounding population and
Highway Administration (FHWA) guidelines which classify "urban" areas as places inside urban boundaries where
the population is greater than 5,000 persons. "Rural" areas are defined as places outside urban areas where the
population is less than 5,000 persons. The HSM uses the term "suburban" to refer to outlying portions of an urban
area; the predictive method does not distinguish between urban and suburban portions of a developed area. The term
"arterial" refers to facilities the meet the FHWA definition of "roads serving major traffic movements (high-speed,
high volume) for travel between major points" (5).
Table 12-1 identifies the specific site types on urban and suburban arterial highways that have predictive models.
In Chapter 12, separate SPFs are used for each individual site to predict multiple-vehicle nondriveway collisions,
single-vehicle collisions, driveway-related collisions, vehicle-pedestrian collisions, and vehicle-bicycle collisions
for both roadway segments and intersections. These are combined to predict the total average crash frequency at an
individual site.
Table 12-1. Urban and Suburban Arterial Site Type SPFs included in Chapter 12
Roadway Segments Two-lane undivided arterials (2U)
Three-lane arterials including a center two-way left-turn.lane (TWLTL) (3T)
Four-lane undivided arterials (4U)
Four-lane divided arterials (i.e., including a raised or depressed median) (4D)

Five-lane arterials including a centerT\VLTL (5T)
Intersections Unsignalized three-leg intersection (stop control on minor-road approaches) (3ST)

Signalized three-leg intersections (3SG)
Unsignalized four-leg intersection (stop control on minor-road approaches) (4ST)
Signalized four-leg intersection (4SG)

• Two-lane undivided arterial (2U)-a roadway consisting of two lanes with a continuous cross-section providing
two directions of travel in which the lanes are not physically separated by either distance or a barrier.
• Three-lane arterials (3 T)-a roadway consisting of three lanes with a continuous cross-section providing two
directions of travel in which center lane is a two-way left-tum lane (TWLTL).
• Four-lane undivided arterials (4U)-a roadway consisting of four lanes with a continuous cross-section providing
two directions of travel in which the lanes are not physically separated by either distance or a barrier.
• Four-lane divided arterials (i.e., including a raised or depressed median) (4D)-a roadway consisting of two lanes
with a continuous cross-section providing two directions of travel in which the lanes are physically separated by
either distance or a barrier.
• Five-lane arterials including a center TWLTL (5T)-a roadway consisting of five lanes with a continuous cross-
section providing two directions of travel in which the center lane is a two-way left-tum lane (TWLTL).
• Three-leg intersection with stop control (3ST)-an intersection of a urban or suburban arterial and a minor road.
A stop sign is provided on the minor road approach to the intersection only.
• Three-leg signalized intersection (3SG)-an intersection of a urban or suburban arterial and one minor road.
Signalized control is provided at the intersection by traffic lights.
• Four-leg intersection with stop control (4ST)-an intersection of a urban or suburban arterial and two minor roads.
A stop sign is provided on both the minor road approaches to the intersection.
• Four-leg signalized intersection (4SG)-an intersection of a urban or suburban arterial and two minor roads.
Signalized control is provided at the intersection by traffic lights.
12.3.2. Predictive Models for Urban and Suburban Arterial Roadway Segments
The predictive models can be used to estimate total average crashes (i.e., all crash severities and collision types) or
can be used to predict average frequency of specific crash severity types or specific collision types. The predictive
model for an individual roadway segment or intersection combines the SPF, CMFs, and a calibration factor. Chapter
12 contains separate predictive models for roadway segments and for intersections.
The predictive models for roadway segments estimate the predicted average crash frequency of non-intersection-
related crashes. Non-intersection-related crashes may include crashes that occur within the limits of an intersection
but are not related to the intersection. The roadway segment predictive models estimate crashes that would occur
regardless of the presence of the intersection.
The predictive models for roadway segments are presented in Equations 12-2 and 12-3 below.
N prediC1ed rs =Cx(N+N
r br pedr
+N'
biiw.,J
(12-2)
N,, = N,Pf" X (CMF,, X CMF,, X ••• X CMFJ (12-3)
Where:
= predicted average crash frequency of an individual roadway segment for the selected year;
= predicted average crash frequency of an individual roadway segment (excluding vehicle-
pedestrian and vehicle-bicycle collisions);
Nspfrs = predicted total average crash frequency of an individual roadway segment for base conditions
(excluding vehicle-pedestrian and vehicle-bicycle collisions);
Npedr = predicted average crash frequency of vehicle-pedestrian collisions for an individual roadway
segment;
= predicted average crash frequency of vehicle-bicycle collisions for an individual roadway segment;
CMF,, ... CMF, = crash modification factors for roadway segments; and
c. = calibration factor for roadway segments of a specific type developed for use for a particular
geographical area.
Equation 12-2 shows that roadway segment crash frequency is estimated as the sum of three components: N,,, NP''''
and N",.,. The following equation shows that the SPF portion ofN,,, designated as N,P1,, is further separated into
three components by collision type shown in Equation 12-4:
Nspfrs =Nbrmv +Nbrsv +Nbni>vy (12-4)
Where:
N,,., = predicted average crash frequency of multiple-vehicle nondriveway collisions for base conditions;
N 6"" = predicted average crash frequency of single-vehicle crashes for base conditions; and
N,,,.., = predicted average crash frequency of multiple-vehicle driveway-related collisions.
Thus, the SPFs and adjustment factors are applied to determine five components: N 6rmv, N 6rsv, N 6r, wy, Npe, r, and N 61""'
•.• ,
which together provide a prediction of total average crash frequency for a roadway segment.
Equations 12-2 through 12-4 are applied to estimate roadway segment crash frequencies for all crash severity levels
combined (i.e., total crashes) or for fatal-and-injury or property-damage-only crashes.
12.3.3. Predictive Models for Urban and Suburban Arterial Intersections

The predictive models for intersections estimate the predicted total average crash frequency including those crashes
that occur within the limits of an intersection and are a result of the presence of the intersection. The predictive
model for an urban or suburban arterial intersection is given by:
N predicted In/ =Cx(N

i bi
+Npedi +Nblkei) (12-5)
N,, = N,pfi" X ( CMFII X CMF, X .. • X CMF,) (12-6)
'Where:
= predicted average crash frequency of an intersection for the selected year;
= predicted average crash frequency of an intersection (excluding vehicle-pedestrian and

vehicle-bicycle collisions);
= predicted total average crash frequency of intersection-related crashes for base conditions
(excluding vehicle-pedestrian and vehicle-bicycle collisions);
= predicted average crash frequency of vehicle-pedestrian collisions;
= predicted average crash frequency of vehicle-bicycle collisions;
CMF11 ••• CMF61 = crash modification factors for intersections; and
c, = calibration factor for intersections developed for use for a particular geographical area.
The CMFs shown in Equation 12-6 do not apply to vehicle-pedestrian and vehicle-bicycle collisions. A separate set
ofCMFs that apply to vehicle-pedestrian collisions at signalized intersections is presented in Section 12.7.
Equation 12-5 shows that the intersection crash frequency is estimated as the sum of three components: N", N,."'
and N",.,. The following equation shows that the SPF portion ofN61 , designated as N,Pi'"' is further separated into
two components by collision type:
Nspflnt ~Nblmv +Nblsv (12-7)
Where:
N,,m, ~ predicted average number of multiple-vehicle collisions for base conditions; and
N,., ~ predicted average number of single-vehicle collisions for base conditions.
Thus, the SPFs and adjustment factors are applied to determine four components oftotii.l intersection average crash
frequency: N bimv' N bm>' Npedl' and Nblkei'
The SPFs for urban and suburban arterial highways are presented in Section 12.6. The associated CMFs for each of
the SPFs are presented in Section 12.7 and summarized in Table 12-18. Only the specific CMFs associated with each
SPF are applicable to that SPF (as these CMFs have base conditions which are identical to the base conditions of
the SPF). The calibration factors, C, and C1, are determined in the Part C, Appendix A. I.!. Due to continual change
in the crash frequency and severity distributions with time, the value of the calibration factors may change for the
selected year of the study period.
12.4. PREDICTIVE METHOD STEPS FOR URBAN AND SUBURBAN ARTERIALS

The predictive method for urban and suburban arterials is shown in Figure 12-1. Applying the predictive method
yields an estimate of the expected average crash frequency (and/or crash severity and collision types) for an urban
or suburban arterial facility. The components of the predictive models in Chapter 12 are determined and applied in
Steps 9, 10, and II of the predictive method. The information to apply each step is provided in the following sections
and in the Part C, Appendix A. In some situations, certain steps will not require any action. For example, a new facil-
ity will not have observed crash data and therefore steps relating to the EB Method require no action.
There are 18 steps in the predictive method. In some situations certain steps will not be needed because data is not
available or the step is not applicable to the situation at hand. In other situations, steps may be repeated if an esti-
mate is desired for several sites or for a period of several years. In addition, the predictive method can be repeated as
necessary to undertake crash estimation for each alternative design, traffic volume scenario, or proposed treatment
option (within the same period to allow for comparison).
The following explains the details of each step of the method as applied to urban and suburban arterials.
Step 1 Define roadway limits and facility type.
Step 2 Define the penod of study.
Determine AADT and availability of crash dat:J for

Step 3 every year in the period of interest.
Step 4 Determine geometric conditions.
Divide roadway into individual roadway

Step 5 segments and intersections.
Step 6 Assign observed crashes to individual sites (if cpplicable).
Step 7 Select a roadway segment or intersection.
Step 8 Select first or next year of the evaluation period.
Step 10 Apply CMFs.
Step 11 Apply a calibration factor.
YES
Step 12
Step 13
YES
Step 14
Step 15
Step 16 Sum all sites and years.
Is there
an alternative
design, treatment, YES
Step 17
or forecast AADT to
be evaluated?
Step 18 Compare and evaluate results.

Step !-Define the limits of the roadway and facility types in the study network, facility, or site for which the
intersection or a homogeneous roadway segment. Sites may consist of a number of types, such as signalized and
unsignalized intersections. The definitions of urban and suburban arterials, intersections, and roadway segments and
the specific site types included in Chapter 12 are provided in Section 12·3.
The predictive method can be undertaken for an existing roadway, a design alternative for an existing roadway, or a
new roadway (which may be either unconstructed or yet to experience enough traffic to have observed crash data).
The limits of the roadway of interest will depend on the nature of the study. The study may be limited to only one
specific site or a group of contiguous sites. Alternatively, the predictive method can be applied to a very long cor-
ridor for the purposes of network screening which is discussed in Chapter 4.

The predictive method can be undertaken for either a past period or a future period. All periods are measured in
years. Years of interest will be determined by the availability of observed or forecast average annual daily traffic
(AADT) volumes, observed crash data, and geometric design data. Whether the predictive method is used for a past
or future period depends upon the purpose of the study. The period of study may be:
• A past period (based on observed AADTs) for:
design features, traffic control features and traffic volumes are known.

• An existing roadway network, facility, or site for which alternative geometric design or traffic control features
• A new roadway network, facility, or site that does not currently exist but is proposed for construction during
some future period.
Step 3-For the study period, determine the availability of annual average daily traffic volumes, pedestrian
crossing volumes, and, for an existing roadway network, the availability of observed crash data (to determine
whether the EB Method is applicable).

The SPFs used in Step 9 (and some CMFs in Step I 0) include AADT volumes (vehicles per day) as a variable. For a
past period the AADT may be determined by an automated recording or estimated by a sample survey. For a future
period, the AADT may be a forecast estimate based on appropriate land use planning and traffic volume forecasting
models or based on the assumption that current traffic volumes will remain relatively constant.
For each roadway segment, the AADT is the average daily two-way 24-hour traffic volume on that roadway segment
in each year of the period to be evaluated selected in Step 8.
For each intersection, two values are required in each predictive model. These are: the two-way AADT of the major
street (AADT=) and the two-way AADT of the minor street (AADTm)·
AADT mo;. andAADT mn1 are determined as follows: if the AADTs on the two major-road legs of an intersection differ,
the larger of the two AADT values is used for the intersection. If the AADTs on the two minor road legs of a four-
leg intersection differ, the larger of the AADTs for the two minor road legs is used. For a three-leg intersection, the
AADT of the single minor road leg is used. If AADTs are available for every roadway segment along a facility, the
major-road AADTs for intersection legs can be determined without additional data.
an estimate of AADT for each year of the evaluation period is interpolated or extrapolated, as appropriate. If there is
not an established procedure for doing this, the following may be applied within the predictive method to estimate
the AADTs for years for which data are not available.
• IfAADT data are available for only a single year, that same value is assumed to apply to all years of the before period.
• If two or more years ofAADT data are available, the AADTs for intervening years are computed by interpolation.
• The AADTs for years before the first year for which data are available are assumed to be equal to the AADT for that
first year.
lfthe EB Method is used (discussed below), AADT data are needed for each year of the period for which observed
crash frequency data are available. If the EB Method will not be used, AADT data for the appropriate time period-
past, present, or future-determined in Step 2 are used.
For signalized intersections, the pedestrian volumes crossing each intersection leg are determined for each year of the
period to be evaluated. The pedestrian crossing volumes for each leg of the intersection are then summed to determine
the total pedestrian crossing volwne for the intersection. Where pedestrian volwne connts are not available, they may be
estimated using the guidance presented in Table 12-15. Where pedestrian volume counts are not available for each year,
they may be interpolated or extrapolated in the same manner as explained above for AADT data.

Where an existing site or alternative conditions for an existing site are being considered, the EB Method is used.
The EB Method is only applicable when reliable observed crash data are available for the specific study roadway
network, facility, or site. Observed data may be obtained directly from the jurisdiction's crash report system. At least
If observed crash frequency data are not available, then Steps 6, 13, and 15 of the predictive method are not
conducted. In this case the estimate of expected average crash frequency is limited to using a predictive model
(i.e., the predictive average crash frequency).
Step 4--Determine geometric design features, traffic control features, and site characteristics for all sites in
the study network.
In order to determine the relevant data needs and avoid unnecessary collection of data, it is necessary to understand
the base conditions and CMFs in Step 9 and Step 10. The base conditions are defined in Section 12.6.1 for roadway
segments and in Section 12.6.2 for intersections.
The following geometric design and traffic control features are used to determine whether the site specific conditions
vary from the base conditions and, therefore, whether a CMF is applicable:
• Length of roadway segment (miles)
• Number of through lanes

• Presence/type of median (undivided, divided by raised or depressed median, centerTWLTL)
• Presence/type of on-street parking (parallel vs. angle; one side vs. both sides of street)
• Number of driveways for each driveway type (major commercial, minor commercial; major industrial/institutional;
minor industrial/institutional; major residential; minor residential; other)
• Roadside fixed object density (fixed objects/mile, only obstacles 4-in or more in diameter that do not have a break-
away design are counted)
• Average offset to roadside fixed objects from edge of traveled way (feet)
• Presence/absence of roadway lighting

• Speed category (based on actual traffic speed or posted speed limit)

• For all intersections within the study area, the following geometric and traffic control features are identified:

• Type of traffic control (minor-road stop or signal)
• Number of approaches with intersection left-tum lane (all approaches, 0, I, 2, 3, or 4 for signalized intersection;
only major approaches, 0, 1, or 2, for stop-controlled intersections)
• Number of major-road approaches with left-tum signal phasing (0, 1, or 2) (signalized intersections only) and type
of left-tum signal phasing (permissive, protected/permissive, permissive/protected, or protected)
• Number of approaches with intersection right tum lane (all approaches, 0, 1, 2, 3, or 4 for signalized intersection;
only major approaches, 0, I, or 2, for stop-controlled intersections)
• Number of approaches with right-tum-on-red operation prohibited (0, 1, 2, 3, or 4) (signalized intersections only)
• Presence/absence of intersection lighting

• Maximum number of traffic lanes to be crossed by a pedestrian in any crossing maneuver at the intersection con-
sidering the presence of refuge islands (for signalized intersections only)
• Proportions of nighttime crashes for unlighted intersections (by total, fatal, injury, and property damage only)
For signalized intersections, land use and demographic data used in the estimation of vehicle-pedestrian collisions
include:
• Number of bus stops within 1,000 feet of the intersection
• Presence of schools within 1,000 feet of the intersection
• Number of alcohol sales establishments within 1,000 feet of the intersection
• Presence of red light camera

• Number of approaches on which right-tum-on-red is allowed
Step 5-Divide the roadway network or facility into individual homogenous roadway segments and
intersections which are referred to as sites.
Using the information from Step I and Step 4, the roadway is divided into individual sites, consisting of individual
individual intersections and homogenous roadway segments for use with the Chapter 12 predictive models are
provided in Section 12.5. When dividing roadway facilities into small homogenous roadway segments, limiting the
segment length to a minimum of 0.10 miles will decrease data collection and management efforts.
Step 6-Assign observed crashes to the individual sites (if applicable).

way segments or intersections are presented in Section A.2.3 of the Appendix A to Part C.
Crashes that occur at an intersection or on an intersection leg, and are related to the presence of an intersection, are
assigned to the intersection and used in the EB Method together with the predicted average crash frequency for the
intersection. Crashes that occur between intersections, and are not related to the presence of an intersection, are
assigned to the roadway segment on which they occur. Such crashes are used in the EB Method together with the
Step 7-Select the first or next individual site in the study network. If there are no more sites to be evaluated,
proceed to Step 15.
In Step 5 the roadway network within the study limits has been divided into a number of individual homogenous
sites (intersections and roadway segments).
number of crashes expected to occur over all sites during the period of interest. If a crash frequency is desired, the
total can be divided by the number of years in the period of interest.
Step 8-For the selected site, select the first or next year in the period of interest. If there are no more years to
be evaluated for that site, proceed to Step 14
segment or intersection because SPFs and some CMFs (e.g., lane and shoulder widths) are dependent on AADT,
Steps 9 through 13, described below, are repeated for each year of the evaluation period as part of the evaluation of
any particular roadway segment or intersection. The predictive models in Chapter 12 follow the general form shown in
Equation 12-1. Each predictive model consists of a SPF, which is adjusted to site specific conditions using CMFs (in
Step 10) and adjusted to local jurisdiction conditions (in Step 11) using a calibration factor (C). The SPFs, CMFs, and
calibration factor obtained in Steps 9, 10, and 11 are applied to calculate the predicted average crash frequency for the
selected year of the selected site. The SPFs available for urban and suburban arterials are presented in Section 12.6
The SPF (which is a regression model based on observed crash data for a set of similar sites) determines the pre-
dicted average crash frequency for a site with the same base conditions (i.e., a specific set of geometric design and
traffic control features). The base conditions for each SPF are specified in Section 12.6. A detailed explanation and
overview of the SPFs are provided in Section C.6.3 of the Part C-Introduction and Applications Guidance.
The SPFs developed for Chapter 12 are summarized in Table 12-2 in Section 12.6. For the selected site, determine
the appropriate SPF for the site type (intersection or roadway segment) and the geometric and traffic control
features (undivided roadway, divided roadway, stop-controlled intersection, signalized intersection). The SPF for
the selected site is calculated using the AADT determined in Step 3 (AADTmaj and AADTmin for intersections)
for the selected year.
Each SPF determined in Step 9 is provided with default distributions of crash severity and collision type (presented
in Section 12.6). These default distributions can benefit from being updated based on local data as part of the cali-
bration process presented in Part C, Appendix A.l.l.
specific geometric design and traffic control features.
In order to account for differences between the base conditions (Section 12.6) and the specific conditions of the site,
CMFs are used to adjust the SPF estimate. An overview of CMFs and guidance for their use is provided in Section
C.6.4 of the Part C-Introduction and Applications Guidance, including the limitations of current knowledge related
to the effects of simultaneous application of multiple CMFs. In using multiple CMFs, engineering judgment is re-
quired to assess the interrelationships and/or independence of individual elements or treatments being considered for
implementation within the same project.
All CMFs used in Chapter 12 have the same base conditions as the SPFs used in Chapter 12 (i.e., when the specific
site has the same condition as the SPF base condition, the CMF value for that condition is 1.00). Only the CMFs pre-
sented in Section 12.7 may be used as part of the Chapter 12 predictive method. Table 12-18 indicates which CMFs
are applicable to the SPFs in Section 12.6.
The CMFs for roadway segments are those described in Section 12.7.1. These CMFs are applied as shown in
Equation 12-3.
The CMFs for intersections are those described in Section 12.7.2, which apply to both signalized and stop-controlled
intersections, and in Section 12.7.3, which apply to signalized intersections only. These CMFs are applied as shown
in Equations 12-6 and 12-28.
In Chapter 12, the multiple- aod single-vehicle base crashes determined in Step 9 and the CMFs values calculated in
Step 10 are then used to estimate the vehicle-pedestrian and vehicle-bicycle base crashes for roadway segments and
intersections (present in Section 12.6.1 and 12.6.2 respectively).
periods. Calibration to local conditions will account for these differences. A calibration factor (Cr for roadway seg-
ments or Ci for intersections) is applied to each SPF in the predictive method. An overview of the use of calibration
factors is provided in the Part C-Introduction and Applications Guidance, Section C.6.5. Detailed guidance for the
development of calibration factors is included in Part C, Appendix A.l.1.
Steps 9, 10, and 11 together implement the predictive models in Equations 12-2 througb 12-7 to determine predicted
average crash frequency.
Step 12-lf there is another year to be evaluated in the study period for the selected site, return to Step 8.
This step creates a loop througb Steps 8 to 12 that is repeated for each year of the evaluation period for the selected site.

the Chapter 12 predictive model estimate of predicted average crash frequency, N'"';""' with the observed crash fre-
quency of the specific site, Nob•m•'· This provides a more statistically reliable estimate of the expected average crash
frequency of the selected site.
In order to apply the site-specific EB Method, overdispersion parameter, k, for the SPF is also used. This is in
addition to the material in Part C, Appendix A.2.4. The overdispersion parameter provides an indication of the
statistical reliability of the SPF. The closer the overdispersion parameter is to zero, the more statistically reliable
the SPF. This parameter is used in the site-specific EB Method to provide a weighting to Npredicted and Nobserved·
Overdispersion parameters are provided for each SPF in Section 12.6.
The estimated expected average crash frequency obtained above applies to the time period in the past for which the
observed crash data were obtained. Section A.2.6 in Appendix A to Part C provides a method to convert the estimate of
expected average crash frequency for a past time period to a future time period. In doing this, consideration is given to
significant changes in geometric or roadway characteristics cause by the treatments considered for future time period.
Step 14--If there is another site to be evaluated, return to 7, otherwise, proceed to Step 15.
This step creates a loop through Steps 7 to 13 that is repeated for each roadway segment or intersection within the
facility.
Step 15-Apply the project level EB Method (ifthe site-specific EB Method is not applicable).
This step is only applicable to existing conditions when observed crash data are available, but cannot be accurately
assigned to specific sites (e.g., the crash report may identify crashes as occurring between two intersections, but is
provided in Part C, Appendix A.2.5.
Step 16--Sum all sites and years in the study to estimate total crash frequency. .
lated using Equation 12-8:
(12-8)
segments
Where:
N,"1 ~total expected number of crashes within the limits of an urban or suburban arterial for the period of interest.
Or, the sum of the expected average crash frequency for each year for each site within the defined roadway
limits within the study period;
~s = expected average crash frequency for a roadway segment using the predictive method for one specific year;
and
N;, ~ expected average crash frequency for an intersection using the predictive method for one specific year.
Equation 12-8 represents the total expected number of crashes estimated to occur during the study period. Equation
12-9 is used to estimate the total expected average crash frequency within the network or facility limits during the
study period.
Ntotal
Ntotal. aveage = -- (12-9)
n
Where:
N,ot:J.J overage = total expected average crash frequency estimated to occur within the defined netv.rork or facility limits
n = number of years in the study period.

Steps 3 through 16 of the predictive method are repeated as appropriate for the same roadway limits but for alterna-
tive conditions, treatments, periods of interest, or forecast AADTs.

provided with each SPF in Section 12.6. These default distributions can benefit from being updated based on local
data as part of the calibration process presented in Part C, Appendix A.l.l.

necessary to apply the predictive method steps. Detail regarding the procedure for determining a calibration factor to
apply in Step II is provided in the Part C, Appendix A. I. Detail regarding the EB Method, which is applied in Steps
6, 13, and 15, is provided in the Part C, Appendix A.2.
In Step 5 of the predictive method, the roadway within the defined limits is divided into individual sites, which are
homogenous roadway segments and intersections. A facility consists of a contiguous set of individual intersections
and roadway segments, referred to as "sites." A roadway network consists of a number of contiguous facilities.
Predictive models have been developed to estimate crash frequencies separately for roadway segments and
intersections. The definitions of roadway segments and intersections presented below are the same as those used in
the FHWA Interactive Highway Safety Design Model (IHSDM) (4).
Roadway segments begin at the center of an intersection and end at either the center of the next intersection or where
there is a change from one homogeneous roadway segment to another homogenous segment. The roadway segment
model estimates the frequency of roadway-segment-related crashes which occur in Region Bin Figure 12-2. When a
roadway segment begins or ends at an intersection, the length of the roadway segment is measured from the center of
the intersection.
Chapter 12 provides predictive models for stop-controlled (three- and four-leg) and signalized (three- and four-leg)
intersections. The intersection models estimate the predicted average frequency of crashes that occur within the
limits of an intersection (Region A of Figure 12-2) and intersection-related crashes that occur on the intersection legs
(Region B in Figure 12-2).
Segment Length
A All crashes that occur w1thin this region are class1fied as intersection crashes.
8 Crashes in this region may be segment or intersection related, depending on

Figure 12-2. Definition of Roadway Segments and Intersections
with respect to characteristics such as traffic volumes and key roadway design characteristics and traffic control
features. Figure 12-2 shows the segment length, L, for a single homogenous roadway segment occurring between
two intersections. However, several homogenous roadway segments can occur between two intersections. A new
(unique) homogeneous segment begins at the center of each intersection and where there is a change in at least one
of the following characteristics of the roadway:
• Annual average daily traffic volume (AADT) (vehicles/day)

• Number of through lanes
• Presence/type of median
The following rounded widths for medians without barriers are recommended before determining "homogeneous"
segments:
Measured Median Width Rounded Median Width
1ftto14ft 10ft
15ft to 24ft 20ft
25 ftto 34ft 30ft
35ftto44ft 40ft
45ftto54ft 50ft
55ftto64ft 60ft
65ftto74ft 70ft
75ftto84ft 80ft
85ftto94ft 90ft
95 ft or more 100ft
• Presence/type of on-street parking
• Roadside fixed object density
• Presence oflighting
• Speed category (based on actual traffic speed or posted speed limit)

In addition, each individual intersection is treated as a separate site for which the intersection-related crashes are
estimated using the predictive method.
There is no minimum roadway segment length, L, for application of the predictive models for roadway segments.
When dividing roadway facilities into small homogenous roadway segments, limiting the segment length to a mini-
mum of 0.10 miles will minimize calculation efforts and not affect results.
and intersections. Observed crashes that occur benveen intersections are classified as either intersection-related or
roadway-segment related. The methodology for assigning crashes to roadway segments and intersections for use
in the site-specific EB Method is presented in Section A.2.3 in Appendix A to Part C. In applying the EB Method
for urban and suburban arterials, whenever the predicted average crash frequency for a specific roadway segment
during the multiyear study period is less than Ilk (the inverse of the overdispersion parameter for the relevant
SPF), consideration should be given to combining adjacent roadway segments and applying the project-level EB
Method. This guideline for the minimum crash frequency for a roadway segment applies only to Chapter 12 which
uses fixed-value overdispersion parameters. It is not needed in Chapter 10 or Chapter II which use length-depen-
dent overdispersion parameters.

In Step 9 of the predictive method, the appropriate safety performance functions (SPFs) are used to predict crash
frequencies for specific base conditions. SPFs are regression models for estimating the predicted average crash
frequency of individual roadway segments or intersections. Each SPF in the predictive method was developed with
observed crash data for a set of similar sites. The SPFs, like all regression models, estimates the value of a dependent
variable as a function of a set of independent variables. In the SPFs developed for the HSM, the dependent variable
estimated is the predicted average crash frequency for a roadway segment or intersection under base conditions, and
the independent variables are the AADTs of the roadway segment or intersection legs (and, for roadway segments,
the length of the roadway segment).
The predicted crash frequencies for base conditions obtained with the SPFs are used in the predictive models in
Equations 12-2 through 12-7. A detailed discussion ofSPFs and their use in the HSM is presented in Chapter 3,
Section 3.5.2 and the Part C-lntroduction and Applications Guidance, Section C.6.3.
Each SPF also has an associated overdispersion parameter, k. The overdispersion parameter provides an indication of
the statistical reliability of the SPF. The closer the overdispersion parameter is to zero, the more statistically reliable
the SPF. This parameter is used in the EB Method discussed in the Part C, Appendix A. The SPFs in Chapter 12 are
summarized in Table 12-2.
Table 12-2. Safety Performance Functions included in Chapter 12

Chapter 12 SPFs for Urban
and Suburban Arterials SPF Components by Collision Type SPF Equations, Tables, and Figures
Roadway segments multiple-vehicle nondriveway collisions Equation 12-10, 12-11, 12-12, Figure 12-3, Tables 12-3, 12-4
single-vehicle crashes Equations 12-13, 12-14, 12-15,

Figure 12-4, Tables 12-5, 12-6
multiple-vehicle driveway-related collisions Equations 12-16, 12-17, 12-18, Figures 12-5, 12-6, 12-7, 12-8,
12-9, Table 12-7
vehicle-pedestrian collisions Equation 12-19, Table 12-8
vehicle-bicycle collisions Equation 12-20, Table 12-9
Intersections multiple-vehicle collisions Equations 12-21, 12-22, 12-23, Figures 12-10.12-11, 12-12.
12-13, Tables 12-10, 12-11
single-vehicle crashes Equations 12-24, 12-25, 12-26, 12-27, Figures 12-14. 12-15,
12-16, 12-17, Tables 12-12, 12-13
vehicle-pedestrian collisions Equations 12-28, 12-29, 12-30,
Tables 12-14, 12-15, 12-16
vehicle-bicycle collisions Equation 12-31, Table 12-17
12.6.1. Safety Performance Functions for Urban and Suburban Arterial Roadway Segments
The predictive model for predicting average crash frequency on a particular urban or suburban arterial roadway
segment was presented in Equation 12-2. The effect of traffic volume (AADT) on crash frequency is incorporated
through the SPF, while the effects of geometric design and traffic control features are incorporated through the
CMFs. The SPF for urban and suburban arterial roadway segments is presented in this section. Urban and suburban
arterial roadway segments are defined in Section 12.3.
SPFs and adjustment factors are provided for five types of roadway segments on urban and suburban arterials:
• Two-lane undivided arterials (2U)
• Three-lane arterials including a center two-way left-tum lane (TWLTL) (3T)
• Four-lane undivided arterials (4U)
• Four-lane divided arterials (i.e., including a raised or depressed median) (4D)
• Five-lane arterials including a center TWLTL (5T)
predictive method described in Section 12.4. The SPFs for roadway segments on urban and suburban arterials are
applicable to the following AADT ranges:
• 2U: 0 to 32,600 vehicles per day
• 3T : 0 to 32,900 vehicles per day

• 4U: 0 to 40,100 vehicles per day
• 4D: 0 to 66.000 vehicles per day
• 5T: 0 to 53,800 vehicles per day
Application to sites with AADTs substantially outside these ranges may not provide reliable results.
Other types of roadway segments may be found on urban and suburban arterials but are not addressed by the
predictive model in Chapter 12.
The procedure addresses five types of collisions. The corresponding equations, tables, and figures are indicated in
Table 12-2 above:
• multiple-vehicle nondriveway collisions
• single-vehicle crashes
• multiple-vehicle driveway-related collisions
• vehicle-pedestrian collisions
• vehicle-bicycle collisions
The predictive model for estimating average crash frequency on roadway segments is shown in Equations 12-2
through 12-4. The effect of traffic volume on predicted crash frequency is incorporated through the SPFs, while the
effects of geometric design and traffic control features are incorporated through the CMFs. SPFs are provided for
multiple-vehicle nondriveway collisions and single-vehicle crashes. Adjustment factors are provided for multi-vehi-
cle driveway-related, vehicle-pedestrian, and vehicle-bicycle collisions.
Multiple-Vehicle Nondriveway Collisions

The SPF for multiple-vehicle nondriveway collisions is applied as follows:
N,,m, ~ exp(a + b x ln(AADT) + In(L)) (12-10)
Where:
AADT average armual daily traffic volume (vehicles/day) on roadway segment;
L length of roadway segment (rni); and

Table 12-3 presents the values of the coefficients a and bused in applying Equation 12-10. The overdispersion
parameter, k, is also presented in Table 12-3.
Table 12-3- SPF Coefficients for Multiple-Vehicle Nondriveway Collisions on Roadway Segments
Coefficients Used in Equation 12-10
Intercept AADT Overdispersion Parameter
Road Type (a) (b) (k)
Total crashes
2U -15.22 1.68 0.84
3T -12.40 1.41 0.66
4U -11.63 1.33 1.01
4D -12.34 1.36 1.32
5T -9.70 1.17 0.81
Fatal-and~injury crashes
2U -16.22 1.66 0.65
JT -16.45 1.69 0.59
4U -12.08 1.25 0.99
4D -12.76 1.28 1.31
5T -10.47 1.12 0.62
Property-damage-only crashes
2U -15.62 1.69 0.87
JT -11.95 1.33 0.59
4U -12.53 1.38 1.08
4D -12.81 1.38 1.34
5T -9.97 1.17 0.88
25
~
:1 20
~
"-
~
"..,. 4D
"~ 15
~
.J:
~
~
..
~
u
..~"'
~
~
10
.,..
~
u
'5 5
~
"-
0 10,000 20,000 30,000 40,000 50,000 60,000 70,000
AADT (veh/day)
Figure 12-3_ Graphical Form of the SPF for Multiple Vehicle Nondriveway collisions (from Equation 12-10 and Table 12-3)
Equation 12-10 is first applied to determine N,rmv using the coefficients for total crashes in Table 12-3. N brmv is then
divided into components by severity level, N,,m~Fn for fatal-and-injury crashes and N,,m~PDO) for property-damage-
only crashes. These preliminary values ofN,,m~m and N,,m,POOl' designated as N' ,,m~m and N' ''m"(PDO> in Equation
12-11, are determined with Equation 12-10 using the coefficients for fatal-and-injury and property-damage-only
crashes, respectively, in Table 12-3. The following adjustments are then made to assure that N,rmv(r.., and N, rmv(PD0 ) 1
sum to Nbrmv:
Nbrmv(FI) = Nbrmv(total) •
N~rmv(FI)
•
J (12-11)
[
Nbrmv(FI) + Nbrmv(PDO)
Nbrmv(PDO) =Nbrmv(tor.al) -Nbrmv{FJ) (12-12)
The proportions in Table 12-4 are used to separate N,,m"(F" and N,,m,(PDO) into components by collision type.
Table 12-4. Distribution ofMu1tiple-Vehicle Nondriveway Collisions for Roadway Segments by Manner of
Collision Type
Proportion of Crashes by Severity Level for Specific Road Types
2U 3T 4U 4D ST
Collision Type FI PDO FI PDO FI PDO FI PDO FI PDO
Rear-end collision 0.730 0.778 0.845 0.842 0.511 0.506 0.832 0.662 0.846 0.651
Head-on collision 0.068 0.004 0.034 0.020 0.077 0.004 0.020 0.007 0.021 0.004
Angle collision 0.085 0.079 0.069 0.020 0.181 0.130 0.040 0.036 0.050 0.059
Sideswipe.
O.Ql5 0.031 0.001 O.o78 0.093 0.249 0.050 0.223 0.061 0.248
same direction
Sideswipe,
0.073 0.055 0.017 0.020 0.082 0.031 0.010 0.001 0.004 0.009
opposite direction
Other multiple-vehicle
0.029 0.053 0.034 0.020 0.056 0.080 0.048 0.071 O.Ql8 0.029
collisions
Source: HSIS data for Washington (2002-2006)
Single-Vehicle Crashes
SPFs for single-vehicle crashes for roadway segments are applied as follows:
Nbrsv = exp(a + b x In(AADJ) + Jn(L)) (12-13)
Table 12-5 presents the values of the coefficients and factors used in Equation 12-13 for each roadway type.
Equation 12-13 is first applied to determine N,,~ using the coefficients for total crashes in Table 12-5. N,m is then
divided into components by severity level; N,n"F" for fatal-and-injury crashes and N,m(PDO) for property-damage-
only crashes. Preliminary values ofNbr.n>(F!J and Nbrsv(PDOJ' designated as N' brsv(Ff) and N' brsv(PDOJ in Equation 12-14,
are determined with Equation 12-13 using the coefficients for fatal-and-injury and property-damage-only crashes,
respectively, in Table 12-5. The following adjustments are then made to assure that Nbm(Ff) and Nbr)'>'(PDO) sum to Nbrsv:
Nbrsv(FI) = Nbrsv(toW) [ N'

N~rsv(FI)
N'
J (12-14)
brsv(FI) + brsv(PDO)
N bm(PDO) = N brsv(rotnl) -Nbrs~(Ff) (12-15)
The proportions in Table 12-6 are used to separate N,m<Fn and N,,,PDO) into components by crash type.
Table 12-5. SPF Coefficients for Single-Vehicle Crashes on Roadway Segments

Intercept AADT Overdispersion Parameter

Road Type (a) (b) (k)
Total crashes
2U -5.47 0.56 0.81

3T -5.74 0.54 1.37
4U -7.99 0.81 0.91
4D -5.05 0.47 0.86
5T -4.82 0.54 0.52
Fatal-and-injury crashes
2U -3.96 0.23 0.50

3T --6.37 0.47 1.06
4U -7.37 0.61 0.54
4D -8.71 0.66 0.28
5T -4.43 0.35 0.36
Property-damage-only crashes
2U -6.51 0.64 0.87
3T -6.29 0.56 1.93
4U -8.50 0.84 0.97
4D -5.04 0.45 1.06
5T -5.83 0.61 0.55
3
ST
.!!!
~
~
""->.
u
"":s 2
~
""
i!!
-"
~
!!!
u
""' 40
~
"'
">
<(
-o
~
"
u
'5
i!!
"-
0
0 10,000 20,000 30,000 40,000 50,000 60,000 70,000
AADT (veh/day)
Figure 12-4. Graphical Form of the SPF for Single-Vehicle Crashes (from Equation 12-13 and Table 12-5)
Table 12-6. Distribution of Single-Vehicle Crashes for Roadway Segments by Collision Type
2U 3T 4U 4D ST
Collision Type FI PDO FI PDO FI PDO FI PDO FI PDO

Collision with animal 0.026 0.066 0.001 0.001 0.001 0.001 0.001 0.063 0.016 0.049
Collision with fixed object 0.723 0.759 0.688 0.963 0.612 0.809 0.500 0.813 0.398 0.768
Collision \\lith other object 0.010 0.013 0.001 0.001 0.020 0.029 0.028 0.016 0.005 0.061
Other single-vehicle collision 0.241 0.162 0.310 0.035 0.367 0.161 0.471 0.108 0.581 0.122
Multiple-Vehicle Driveway-Related Collisions

The model presented above for multiple-vehicle collisions addressed only collisions that are not related to drive-
ways. Driveway-related collisions also generally involve multiple vehicles, but are addressed separately because the
frequency of driveway-related collisions on a roadway segment depends on the number and type of driveways. Only
unsignalized driveways are considered; signalized driveways are analyzed as signalized intersections.
The total number of multiple-vehicle driveway-related collisions within a roadway segment is determined as:
AADTJ(t)
Nb-~ = ~ n-xN-x
1 1 (
-- (12-16)
'""" LJ
all 15, 000
driveway
typoe
Where:
~ ~ Number of driveway-related collisions per driveway per year for driveway typej from Table 12-7;
n ~ number of driveways within roadway segment of driveway type j including all driveways on both sides of the
1
road; and
~ coefficient for traffic volume adjustment from Table 12-7.
The number of driveways of a specific type, np is the sum of the number of driveways of that type for both sides of the
road combined. The number of driveways is determined separately for each side of the road and then added together.
Seven specific driveway types have been considered in modeling. These are:
• Major commercial driveways
• Minor commercial driveways
• Major industrial/institutional driveways
• Minor industrial/institutional driveways
• Major residential driveways
• Minor residential driveways
• Other driveways
Major driveways are those that serve sites with 50 or more parking spaces. Minor driveways are those that serve
sites with less than 50 parking spaces. It is not intended that an exact count of the number of parking spaces be made
for each site. Driveways can be readily classified as major or minor from a quick review of aerial photographs that
show parking areas or through user judgment based on the character of the establishment served by the driveway.
Commercial driveways provide access to establishments that serve retail customers. Residential driveways serve
single- and multiple-family dwellings. Industrial/institutional driveways serve factories, warehouses, schools,
hospitals, churches, offices, public facilities, and other places of employment. Commercial sites with no restriction
on access along an entire property frontage are generally counted as two driveways.
Table 12-7. SPF Coefficients for Multiple-Vehicle Driveway Related Collisions

Coefficients for Specific Roadway Types
Driveway Type (j) 2U 3T 4U 4D ST

Number of Driveway-Related Collisions per Driveway per Year (N)
Major commercial 0.158 0.102 0.182 0.033 0.165
Minor commercial 0.050 0.032 0.058 0.011 0.053
Major industriaVinstitutional 0.172 0.110 0.198 0.036 0.181
Minor industriaVinstitutional 0.023 O.o!5 0.026 0.005 0.024
Major residential 0.083 0.053 0.096 O.Q18 0.087
Minor residential 0.016 0.010 O.o!S 0.003 0.016
Other 0.025 0.016 0.029 0.005 0.027
Regression Coefficient for AADT (t)
All driveways 1.000 1.000 1.172 1.106 1.172
Overdispersion Parameter (k)

All d.rivev.rays 0.81 1.10 0.81 1.39 0.10
Proportion of Fatal-and-Injury Crashes (fJ...)

All driveways 0.323 0.243 0.342 0.284 0.269
Proportion of Property-Damage-Only Crashes

All driveways 0.677 0.757 0.658 0.716 0.731
Note: Includes only unsignalized driveways; signalized driveways are analyzed as signalized intersections. Major driveways serve 50 or more
parking spaces; minor driveways serve less than 50 parking spaces.
0.4 , . - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Major lridustrialllnstitutional
Major ommercial
Major Residential
Minor ommercial
j....:;::;;;;;;;;;;;,;;;;;,;;;;;;;;..;;;;. .; .: ; :. ; . ,.,.,.,.; :;.;.;: : . : : :.: :~:h:ec~

0.1
Minor 1 dustrialllnstitutional
Minor esidential
0.0
0 5,000 10,000 15,000 20,000 25,000 30,000 35,000
AADT (veh/day)
Figure 12-5. Graphical Form of the SPF for Multiple Vehicle Driveway Related Collisions on Two-Lane Undivided
Arterials (2U) (from Equation 12-16 and Table 12-7)
0.3 - , - - - - - - - - - - - - - - - - - - - - - - - - - - - - ,
Major, Industrial/Institutional
Major Commercial
Major Residential
0.1
L
~~~~~~~~~~MlooCommocd•l Other
Mino lndustriaVtnstitutional
Mine Residential
0.0
0 5,000 10,000 15,000 20,000 25,000 30,000 35,000
AADT (veh/day)
Figure 12-6. Graphical Form of the SPF for Multiple Vehicle Driveway Related Collisions on Three-Lane
Undivided Arterials (3T) (from Equation 12-16 and Table 12-7)
0.7
Major lndustriaVInstitutional
"'
~
~
~
0.6
Major Commercial
.~
i5
~
~
c. 0.5
"'
u
~
.,.
~
~
0.4
.!:
.<:
~
Major Reside tial
b
~
0.3
'"~
~
> 0.2
<
"C
~
~
u
'6
~ 0.1
"-
0.0
0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000
AADT (veh/day)
Figure 12-7. Graphical Form of the SPF for Multiple Vehicle Driveway Related Collisions on Four-Lane Undivided
Arterials (4U) (from Equation 12-16 and Table 12-7)
0.2
Major Industrial/Institutional
"'
~
~
~
I
Majo 'Commercial
>
~
~
~
c.
"'
u
~
~
~
c-
~ 0.1
-"
~ Major Residential
~
u
~
"'
~
~
> Minor Commercial
<(
-a
~
t;
..
'5
~
0.0
0 10,000 20,000 30,000 40,000 50,000 60,000 70,000
AADT (veh/day)
Figure 12-8. Graphical Form of1he SPF for Multiple Vehicle Driveway Related Collisions on Four-Lane Divided
Arterials (4D) (from Equation 12-16 and Table 12-7)
0.9
» 0.8 + - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ' 7 ' - - ' 1 ' a j o r Industrial/Institutional

~
~
~
ajor Commercial
-~
5 0.7
~
~
c.
»
u 0.6
~
~
~
c-
~ OS
~
'u":!!' 0.4
~
~
"'
:!! 0.3
>
<(
-a
~
~ 0.2
u
..
'5
l::~~~~~~~~~~~~~~~~~~~~~~~::=:::=::::=::::=:~jther
~
0.1 inor lndustriaVInstitutional
inor Residential
0.0
0 5,000 10,000 15,000 20,000 25,000 3D DOD 35,000 40,000 45,000 50,000 55,000
AADT (veh/day)
Figure 12-9. Graphical Form of1he SPF for Multiple Vehicle Driveway Related Collisions on Five-Lane Arterials
Including a Center Two-Way Left-Tum Lane (from Equation 12-16 and Table 12-7)
Driveway-related collisions can be separated into components by severity level as follows:
Nbrdwy(FJ) = Nbrdwy(rotnl) X ~wy (12-17)
(12-18)
\\There:
~"" ~ proportion of driveway-related collisions !bat involve fatalities or injuries
The values ofN.j and f,"Y are showu in Table 12-7.
Vehicle-Pedestrian Collisions
The number of vehicle-pedestrian collisions per year for a roadway segment is estimated as:
Npcdr ~ N br xf.pedr (12-19)
'Where:
J;,,,, ~ pedestrian crash adjustment factor.
The value N,, used in Equation 12-19 is !bat determined witb Equation 12-3.
Table 12-8 presents the values off , for use in Equation 12-19. All vehicle-pedestrian collisions are considered to
P"
be fatal-and-injury crashes. The values off,,,, are likely to depend on the climate and tbe walking environment in
particular states or communities. HSM users are encouraged to replace tbe values in Table 12-8 with suitable values
for their own state or community through tbe calibration process (see Appendix A to Part C).
Table 12-8. Pedestrian Crash Adjustment Factor for Roadway Segments

Pedestrian Crash Adjustment Factor (fpedr)
Road Type Posted Speed 30 mph or Lower Posted Speed Greater than 30 mph
2U 0.036 0.005
3T 0.041 0.013
4U 0.022 0.009
40 0.067 0.019
5T 0.030 0.023
Note: These factors apply to the methodology for predicting total crashes (all severity levels combined).
All pedestrian collisions resulting from this adjustment factor are treated as fatal·and-injury crashes and
none as property-damage-only crashes.
Vehicle-Bicycle Collisions
The number of vehicle-bicycle collisions per year for a roadway segment is estimated as:
(12-20)
Where:
!;,,., ~ bicycle crash adjustment factor.
The value ofN,, used in Equation 12-20 is determined with Equation 12-3.
Table 12-9 presents the values of !biker for use in Equation 12-18. All vehicle-bicycle collisions are considered to
be fatal-and-injury crashes. The values of !biker are likely to depend on the climate and bicycling environment in
particular states or communities. HSM users are encouraged to replace the values in Table 12-9 with suitable values
for their own state or community through the calibration process (see Appendix A to Part C).
Table 12-9. Bicycle Crash Adjustment Factors for Roadway Segments

Bicycle Crash Adjustment Factor (fb 1u)
Road type Posted Speed 30 mph or Lower Posted Speed Greater than 30 mph
2U 0.018 0.004
3T 0.027 0.007
4U 0.011 0.002
4D 0.013 0.005
5T 0.050 0.012
Note: These factors apply to the methodology for predicting total crashes (all severity levels combined).
All bicycle collisions resulting from this adjustment factor are treated as fatal-and-injury crashes and none as
property-damage-only crashes.
12.6.2. Safety Performance Functions for Urban and Suburban Arterial Intersections
The predictive models for predicting the frequency of crashes related to an intersection is presented in Equations
12-5 through 12-7. The structure of the predictive models for intersections is similar to the predictive models for
roadway segments.
The effect of traffic volume on predicted crash frequency for intersections is incorporated through SPFs, while the
effect of geometric and traffic control features are incorporated through CMFs. Each of the SPFs for intersections
incorporates separate effects for the AADTs on the major- and minor-road legs, respectively.
SPFs and adjustment factors have been developed for four types of intersections on urban and suburban arterials.
These are:
• Three-leg intersections with stop control on the minor-road approach (3ST)
• Three-leg signalized intersections (3SG)
• Four-leg intersections with stop control on the minor-road approaches (4ST)
Other types of intersections may be found on urban and suburban arterials but are not addressed by the Chapter 12 SPFs.
The SPFs for each of the four intersection types identified above predict total crash frequency per year for crashes
that occur within the limits of the intersection. The SPFs and adjustment factors address the following four types of
collisions, (the corresponding equations, tables, and figures are indicated in Table 12-2):
• multiple-vehicle collisions
• single-vehicle crashes
• vehicle-pedestrian collisions
• vehicle-bicycle collisions
Guidance on the estimation of traffic volumes for the major and minor road legs for use in the SPFs is presented in
Step 3. The AADT(s) used in the SPF are the AADT(s) for the selected year of the evaluation period. The SPFs for
intersections are applicable to the following AADT ranges:
3ST Intersections AADT mqJ: 0 to 45,700 vehicles per day and AADT"''": 0 to 9,300 vehicles per day
4ST Intersections AADT ma;: 0 to 46,800 vehicles per day and AADT min: 0 to 5,900 vehicles per day
3SG Intersections AADT '"~1 : 0 to 58,100 vehicles per day and AADT min: 0 to 16,400 vehicles per day
4SG Intersecbons AADTmaJ: 0 to 67,700 vehicles per day and AADT'"'": 0 to 33,400 vehicles per day
4SG Intersections Pedestrian Models:
• AADT m,J' 80,200 vehicles per day
• AADTm1,: 49,100 vehicles per day

• PedVol: 34,200 pedestrians per day crossing all four legs combined
Application to sites with AADTs substantially outside this range may not provide reliable results.
Multiple-Vehicle Collisions
SPFs for multiple-vehicle intersection-related collisions are applied as follows:
Nbim" ~ exp(a + b X In(AADTm,) + c x In(AADTm,)) (12-21)
Where:
AADTm,J ~ average daily traffic volume (vehicles/day) for major road (both directions of travel combined);
AADTm,, ~ average daily traffic volume (vehicles/day) for minor road (both directions of travel combined); and
a, b, c = regression coefficients.
Table 12-10 presents the values of the coefficients a, b, and c used in applying Equation 12-21. The SPF
overdispersion parameter, k, is also presented in Table 12-10.
Equation 12-21 is first applied to determine N,.•mv using the coefficients for total crashes in Table 12-10. N,lmv
is then divided into components by crash severity level, Nbtmv(FI) for fatal-and-injury crashes and Nbimv(PDO) for
property-damage-only crashes. Preliminary values ofNbimv(FJ) and Nbimv(PDOJ' designated as N' bimv(F[) and N' bimv(PDOJ
in Equation 12-22, are determined with Equation 12-21 using the coefficients for fatal-and-injury and property-
damage-only crashes, respectively, in Table 12-10. The following adjustments are then made to assure that N,lmv(Ff)
and N bimv(PDO) sum to N bimv:
N
bimv(FI) ;::::
N
bimv(total) X
[
1
N~imv(FI)1
J (12-22)
Nbimv(FI) + Nbimv(PDO)
Nblmv(PDO) ~Nbimv(totnl) -Nbimv(F[) (12-23)
The proportions in Table 12-11 are used to separate Nbim"{Fn and Nblm"{PDOJ into components by manner of collision.
'
Table 12-10. SPF Coefficients for Multiple-Vehicle Collisions at Intersections
Intercept AADT,a; AADT,1" Overdispersion Parameter

Intersection Type (a) (b) (c) (k)
Total Crashes
3ST -13.36 1.11 0.41 0.80
3SG -12.13 1.11 0.26 0.33
4ST -8.90 0.82 0.25 0.40
4SG -10.99 1.07 0.23 0.39
Fatal-and-Injury Crashes
3ST -14.01 1.16 0.30 0.69
3SG -11.58 1.02 0.17 0.30
4ST -11.13 0.93 0.28 0.48
4SG -13.14 1.18 0.22 0.33
Property-Damage-Only Crashes
3ST -15.38 1.20 0.51 0.77
3SG -13.24 1.14 0.30 0.36
4ST -8.74 0.77 0.23 0.40
4SG -11.02 1.02 0.24 0.44
11
10
,., 9
""".,.::0 8
~
~ 7
.<::
~
:!! 6
u
""'
:!! 5
">
<(
~
4
~
"
"
'5 3
~
"-
2
0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000 50,000
AADT m•J (veh/day)
Figure 12-10. Graphical Form of the Intersection SPF for Multiple Vehicle Collisions on Three-Leg Intersections
with Minor-Road Stop Control (3ST) (from Equation 12-21 and Table 12-10)
14
13
12
,.,
u 11
c
""
r:T
10
~"
~
9
-"~
8
u"'
~
7
""'
"'">
~
6
.,..: 5
"
~
u
'C 4
...
!!!
3
~~
0
0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000 50,000 55,000 60,000
AADT m•i (veh/day)
Figure 12-11. Graphical Form of the Intersection SPF for Multiple Vehicle Collisions on Three-Leg Signalized
Intersections (3SG) (from Equation 12-21 and Table 12-10)
'
,.,
u
B mln=5,900
c AAD m•n=4,000
""r:T 7 mln=3,000
~"
~
-"
u
~
~
6
::dm'•"'·ooo m•n=1 ,000
" 5
"'~
..:">
., 4
"u
~
'C 3
...
~
0 S,OOO 10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000 50,000
Figure 12-12. Graphical Form of the Intersection SPF for Multiple Vehicle Collisions on Four-Leg Intersections
with Minor-Road Stop Control (4ST) (from Equation 12-21 and Table 12-10)
30
>.
.."
u
c
c-
25
DTmln"'20,000
DTmln=1 5,000
~
"- DTmln=10,000
-"
~
20
:!!
..
u
..~"'
:!! 15
"~u
-..
'0
"-
10
0 5,000 I 0,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000 50,000 55,000 60,000 65,000 70,000
AADTmaj (veh/day)
Figure 12-13. Graphical Form of the Intersection SPF for Multiple Vehicle Collisions on Four-Leg Signalized
Table 12-11. Distribution of Multiple-Vehicle Collisions for Intersections by Collision Type

Proportion of Crashes by Severity Level for Specific Intersections Types
3ST 3SG 4ST 4SG
Manner of Collision FI PDO FI PDO FI PDO FI PDO
Rear-end collision 0.421 0.440 0.549 0.546 0.338 0.374 0.450 0.483
Head-on collision 0.045 0.023 0.038 0.020 0.041 0.030 0.049 0.030
Angle collision 0.343 0.262 0.280 0.204 0.440 0.335 0.347 0.244
Sideswipe 0.126 0.040 0.076 0.032 0.121 0.044 0.099 0.032
Other multiple-vehicle 0.065 0.235 0.057 0.198 0.060 0.217 0.055 0.211
collisions
Source: HSIS data for California (2002-2006)
Single-Vehicle Crashes
SPFs for single-vehicle crashes are applied as follows:
(12-24)
Table 12-12 presents the values of the coefficients and factors used in Equation 12-24 for each roadway type. Equation
12-24 is first applied to determine N'"' using the coefficients for total crashes in Table 12-12. N,,,, is then divided into
components by severity level, Nb!"·<F~ for fatal-and-injury crashes and Nb!,,PDOJ for property-damage-only crashes.
Preliminary values ofNbisv(Fl) and Nbm<PDOl' designated as N 'bisv(Ff) and N 'bisv{PDOJ in Equation 12-25, are determined with
Equation !2-24 using the coefficients for fatal-and-injury and property-damage-only crashes, respectively, in Table 12-12.
The following adjustments are then made to assure that Nbis~CFJ) and Nbh-v(PDO) sum to Nbisv"
Nbisv(Fl) = Nbisv(tota1) X [ N'

N~isv(Fl)
N'
J (!2-25)
bisv(FI) + bisv(PDO)
(12-26)
Table 12-12. SPF Coefficients for Single-Vehicle Crashes at Intersections

Intercept AADTm,J AADTmln Overdispersion Parameter

Intersection Type (a) (b) (c) (k)
Total Crashes
3ST -6.81 0.16 0.51 1.14
3SG -9.02 0.42 0.40 0.36
4ST -5.33 0.33 0.12 0.65
4SG -10.21 0.68 0.27 0.36
Fatal-and-Injury Crashes
3ST
3SG -9.75 0.27 0.51 0.24

4ST
4SG -9.25 0.43 0.29 0.09
Property-Damage-Only Crashes
3ST -8.36 0.25 0.55 1.29

3SG -9.08 0.45 0.33 0.53
4ST -7.04 0.36 0.25 0.54
4SG -11.34 0.78 0.25 0.44
Note: Where no models are available, Equation 12-27 is used.

0.7
0.6
"'
u
~
~
0.5
~
"'"'"
-.. 0.4
-"'
.<:
~
u
0.3
~
~
"'
>
.,< 0.2 AADTm,n 1,000
~
"'
u
'5
!!!
0.. 0.1
0.0
0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000 50,000
Figure 12-14. Graphical Form of the Intersection SPF for Single-Vehicle Crashes on Three-Leg Intersections with
Minor-Road Stop Control (3ST) (from Equation 12-24 and Table 12-12)
0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000 50,000 55,000 60,000
Figure 12-15. Graphical Form of the Intersection SPF for Single-Vehicle Crashes on Three-Leg Signalized
0.5
AADTm•n:=S, 900
AADlim•n=4,000
AAD mln"'3,0QQ
AAD
,.,
u
0.4
n11n=2,000
c AAD mln=l,OOO
~
::0
""
~
~
~
.s=
~ 0.3
E
u
~
"'E
~
> 0.2
.,
<(
~
t:
'5
~
"- 0.1
0.0 +--+-+-+-i---+--H-+-b-0-+--t-f-+--+-+-f-+-H-+-+-+-+--+-f-+-+~-++-+-+--t-+-+-+-+-+-f-+-H--H
0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000 50,000
Figure 12-16. Graphical Form of the Intersection SPF for Single-Vehicle Crashes on Four-Leg Stop Controlled
Intersections (4ST) (from Equation 12:24 and Table 12-12)
1.2
,.,
u 1.0
I
c
~
::0
~
""
~
~
.s= 0.8
~
E
u
~
~
"'E 0.6
.,~
~
~ DTmon=1 ,000
u
'5 0.4
~
"-
0.2
0 10,000 20,000 30,000 40,000 50,000 60,000 70,000
Figure 12-17. Graphical Form of the Intersection SPF for Single-Vehicle Crashes on Four-Leg Signalized
The proportions in Table 12-13 are used to separate NM"(m and NM~(PDOJ into components by crash type.
Table 12-13. Distribution of Single-Vehicle Crashes for Intersection by Collision Type

JST JSG 4ST 4SG

Crash Type FI PDO Fl PDO FI PDO Fl PDO
Collision with parked vehicle 0.001 0.003 0.001 0.001 0.001 0.001 0.001 0.001
Collision with animal 0.003 O.Q18 0.001 0.003 0.001 0.026 0.002 0.002
Collision with fixed object 0.762 0.834 0.653 0.895 0.679 0.847 0.744 0.870
Collision with other object 0.090 0.092 0.091 0.069 0.089 0.070 0.072 0.070
Other single-vehicle collision 0.039 0.023 0.045 O.Q18 0.051 0.007 0.040 0.023
Noncollision 0.105 0.030 0.209 0.014 0.179 0.049 0.141 0.034
Source: HSIS data for California (2002-2006)
Since there are no models for fatal-and-injury crashes at three- and four-leg stop-controlled intersections in
Table 12-12, Equation 12-25 is replaced with the following equation in these cases:
(12-27)
Where:
J;," ~ proportion of fatal-and-injury crashes for combined sites.
The default value off,. in Equation 12-27 is 0.31 for 3ST and 0.28 for 4ST intersections. It is recommended that
these default values b~-~pdated based on locally available data.
SPFs for Vehicle-Pedestrian Collisions

Separate SPFs are provided for estimation of the number of vehicle-pedestrian collisions at signalized and unsignal-
ized intersections.
SPFs for Signalized Intersections

The number of vehicle-pedestrian collisions per year at a signalized intersection is estimated with a SPF and a set of
CMFs that apply specifically to vehicle-pedestrian collisions. The model for estimating vehicle-pedestrian collisions
at signalized intersections is:
(12-28)
Where:
~ predicted number of vehicle-pedestrian collisions per year for base conditions at signalized
intersections; and
CMF 1, ... CMF3, ~ crash modification factors for vehicle-pedestrian collisions at signalized intersections.
The SPF for vehicle-pedestrian collisions at signalized intersections is:
Npedba" ~ exp[a +bxln(AADT,01az)+cxln[AADTmi~

AADTmOJ
J+dxln(PedVol)+exn 10 nmJ (12-29)
Where:
AADT""' sum of the average daily traffic volumes (vehicles per day) for the major and minor roads
(= AADTm,J + AADTm1,);
PedVol sum of daily pedestrian volumes (pedestrians/day) crossing all intersection legs;
ntrmesx = maximum number of traffic lanes crossed by a pedestrian in any crossing maneuver at the intersection
considering the presence of refuge islands; and
a, b, c, d, e = regression coefficients.
Determination of values for AADT moJ and AADTmto is addressed in the discussion of Step 3. Only pedestrian crossing
maneuvers immediately adjacent to the intersection (e.g., at a marked crosswalk or along the extended path of any
sidewalk present) are considered in determining the pedestrian volumes. Table 12-14 presents the values of the
coefficients a, b, c, d, and e used in applying Equation 12-29.
The coefficient values in Table 12-14 are intended for estimating total vehicle-pedestrian collisions. All vehicle-pedestrian
collisions are considered to be fatal-and-injury crashes.
The application of Equation 12-29 requires data on the total pedestrian volumes crossing the intersection legs.
Reliable estimates will be obtained when the value ofPedVol in Equation 12-29 is based on actual pedestrian volume
counts. Where pedestrian volume counts are not available, they may be estimated using Table 12-15. Replacing the
values in Table 12-15 with locally derived values is encouraged.
The value ofn,,,, in Equation 12-29 represents the maximum number of traffic lanes that a pedestrian must cross
in any crossing maneuver at the intersection. Both through and turning lanes that are crossed by a pedestrian along
the crossing path are considered. If the crossing path is broken by an islaod that provides a suitable refuge for the
pedestrian so that the crossing may be accomplished in two (or more) stages, then the number oflanes crossed in
each stage is considered separately. To be considered as a suitable refuge, an island must be raised or depressed; a
flush or painted island is not treated as a refuge for purposes of determining the value of nlanesx.
Table 12-14. SPFs forVehicle-Pedestriao Collisions at Signalized Intersections

Coefficients used in Equation 12-29
Overdispersion
Intercept AADTtotal AADT,1/AADTmQJ PedVol 0 tone.cr Parameter
Intersection Type (a) (b) (c) (d) (e) (k)
Total crashes
JSG -6.60 0.05 0.24 0.41 0.09 0.52

4SG -9.53 0.40 0.26 0.45 0.04 0.24
Table 12-15. Estimates of Pedestrian Crossing Volumes Based on General Level of Pedestriao Activity
Estimate ofPedVol (pedestrians/day)
for Use in Equation 12-29
Genenl Level of Pedestrian Activity 3SG Intersections 4SG Intersections
Htgh 1.700 3.200
Medium-high 750 1.500
Medium 400 700
Medium-low 120 240
Low 20 50
SPFs for Stop-Controlled Intersections

The nwnber of vehicle-pedestrian collisions per year for a stop-controlled intersection is estimated as:
Npedi ~ N.xf
b1 pedi
(12-30)
Where:
/;edi = pedestrian crash adjustment factor.
The value ofN" used in Equation 12-30 is that determined with Equation 12-6.
Table 12-16 presents the values offpedi for use in Equation 12-30. All vehicle-pedestrian collisions are considered
to be fatal-and-injury crashes. The values off~"d; are likely to depend on the climate and walking environment in
Table 12-16. Pedestrian Crash Adjustroent Factors for Stop-Controlled Intersections

Intersection Type Pedestrian Crash Adjustment Factor (f )
JST 0.021
4ST 0.022
Note: These factors apply to the methodology for predicting total crashes
(all severity levels combined). All pedestrian collisions resulting from this
adjustment factor are treated as fatal-and-injury crashes and none as
Source: HSIS data for California {2002-2006)
Vehicle-Bicycle Collisions
The number of vehicle-bicycle collisions per year for an intersection is estimated as:
(12-31)
Where:
fw,; ~ bicycle crash adjustroent factor.
The value ofN" used in Equation 12-31 is determined with Equation 12-6.
Table 12-17 presents the values of f,;~re; for use in Equation 12-31. All vehicle-bicycle collisions are considered to
be fatal-and-injury crashes. The values of f,;~w;are likely to depend on the climate and bicycling environment in
Table 12-17. Bicycle Crash Adjustroent Factors for Intersections

Intersection Type Bicycle Crash Adjustment Factor (fh1k)
3ST 0.016
3SG 0.011
4ST 0.018
4SG 0.015
Note: These factors apply to the methodology for predicting total crashes
(all severity levels combined). All bicycle collisions resulting from this adjustment
factor are treated as fatal-and-injury crashes and none as property-damage-only
crashes. Source: HSIS data for California (2002-2006)
12.7. CRASH MODIFICATION FAGORS

In Step I 0 of the predictive method shown in Section 12.4, crash modification factors are applied to the selected safety
performance function (SPF), which was selected in Step 9. SPFs provided in Chapter 12 are presented in Section 12.6. A
general overview of crash modification factors (CMFs) is presented in Chapter 3, Section 3.5.3. The Part C-Introduction
and Applications Guidance provides fiuther discussion on the relationship of CMFs to the predictive method. This section
provides details of the specific CMFs applicable to the SPFs presented in Section 12.6.
Crash modification factors (CMFs) are used to adjust the SPF estimate of predicted average crash frequency for
Chapter 12 shown in Equation 12-1. The CMF for the SPF base condition of each geometric design or traffic control
feature has a value of 1.00. Any feature associated with higher crash frequency than the base condition has a CMF
with a value greater than 1.00; any feature associated with lower crash frequency than the base condition has a CMF
with a value less than 1.00.
The CMFs used in Chapter 12 are consistent with the CMFs in Part D, although they have, in some cases, been
expressed in a different form to be applicable to the base conditions of the SPFs. The CMFs presented in Chapter 12
and the specific SPFs which they apply to are summarized in Table 12-18.
Table 12-18. Summary ofCMFs in Chapter 12 and the Corresponding SPFs

Applicable SPF CMF CMF Description CMF Equations and Tables
CMF,r On-Street Parking Equation 12~32 and Table 12-19
CMF,, Roadside Fixed Objects Equation 12-33 and Tables 12-20 and 12-21
Roadway Segments CMFj, Median Width Table 12-22
CMF., Lighting Equation 12-34 and Table 12-23
CMF5, Automated Speed Enforcement See text
CMF 11 Intersection Left-Tum Lanes Table 12-24
CMF: 1 Intersection Left-Tum Signal Phasing Table 12-25
CMF 3 , Intersection Right-Tum Lanes Table 12-26

Multiple-Vehicle Collisions and
Single-Vehicle Crashes at Intersections
CMF 41 Right-Tum-on-Red Equation 12-35
C:MF,, Lighting Equation 12-36 and Table 12-27
CMF 61 Red-Ltght Cameras Equations 12-37, 12-38, 12-39
CMF,, Bus Stops Table 12-28
Vehicle-Pedestrian Collisions at
CMF 2, Schools Table 12-29
CMF 3, Alcohol Sales Establishments Table 12-30

12.7.1. Crash Modification Factors for Roadway Segments

The CMFs for geometric design and traffic control features of urban and suburban arterial roadway segments are
presented below. These CMFs are determined in Step 10 oftbe predictive method and used in Equation 12-3 to
adjust the SPF for urban and suburban arterial roadway segments to account for differences between the base condi-
tions and the local site conditions.
CMF 1,--0n-Street Parking

The CMF for on-street parking, where present, is based on research by Bonneson (1). The base condition is the
absence of on-street parking on a roadway segment. The CMF is determined as:
CMFlr = 1 +ppk X if.pk - 1.0) (12-32)
Where:
CMF1, = crash modification factor for the effect of on-street parking on total crashes;
f,, = factor from Table 12-19;
P,, = proportion of curb length with on-street parking= (0.5 L,jL); and
L,, sum of curb length with on-street parking for both sides of the road combined (miles); and
L length of roadway segment (miles).
This CMF applies to total roadway segment crashes.
The sum of curb length with on-street parking (L,,) can be determined from field measurements or video log
review to verify parking regulations. Estimates can be made by deducting from twice the roadway segment length
allowances for intersection widths, crosswalks, and driveway widths.
Table 12-19. Values off,, Used in Determining the Crash Modification Factor for On-Street Parking
'I)'pe of Parld.ng and Land Use
Parallel Parking Angle Parking
Commercial or Commercial or
Road Type Residential/Other Industrialllnstitutional Residential/Other IndustriaUinstitutional
2U 1.465 2.074 3.428 4.853
3T 1.465 2.074 3.428 4.853

4U 1.100 1.709 2.574 3.999
4D l.lOO 1.709 2.574 3.999
5T 1.100 1.709 2.574 3.999
CMF,,-Roadside Fixed Objects

The base condition is the absence of roadside fixed objects on a roadway segment. The CMF for roadside fixed ob-
jects, where present, has been adapted from the work of Zegeer and Cynecki (15) on predicting utility pole crashes.
The CMF is determined with the following equation:
(12-33)
Where:
CMF2, = crash modification factor for the effect of roadside fixed objects on total crashes;
•
J offset = fixed-object offset factor from Table 12-20;
D, = fixed-object density (fixed objects/mi) for both sides of the road combined; and
1
p, = fixed-object collisions as a proportion of total crashes from Table 12-21.
1
This CMF applies to total roadway segment crashes. If the computed value of CMF 2, is less than 1.00, it is set equal
to 1.00. This can only occur for very low fixed object densities.
In estimating the density of fixed objects (D ,), only point objects that are 4 inches or more in diameter and do not have
1
breakaway design are considered. Point objects that are within 70 ft of one another longitudinally along the road are
counted as a single object. Continuous objects that are not behind point objects are counted as one point object for each
70 ft oflength. The offset distance (0 ,) shown in Table 12-20 is an estimate of the average distance from the edge of
1
the traveled way to roadside objects over an extended roadway segment. If the average offset to fixed objects exceeds
30ft, use the value offoffset for 30ft. Only fixed objects on the roadside on the right side of the roadway in each
direction of travel are considered; fixed objects in the roadway median on divided arterials are not considered.
Table 12-20. Fixed-Object Offset Factor

Offset to Fixed Objects Fixed-Object Offset Factor
(0) (ft) (foffoec)
2 0.232
5 0.133
10 0.087
15 0.068
20 0.057
25 0.049
30 0.044
Table 12-21. Proportion of Fixed-Object Collisions

Proportion of Fixed--Object Collisions
Road Type (p)
2U 0.059
3T 0.034
4U 0.037
4D 0.036
5T 0.016
CMF,,-Median Width
A CMF for median widths on divided roadway segments of urban and suburban arterials is presented in Table 12-22
based on the work of Harkey et al. (6). The base condition for this CMF is a median width of 15ft. The CMF applies
to total crashes and represents the effect of median width in reducing cross-median collisions; the CMF assumes that
nonintersection collision types other than cross-median collusions are not affected by median width. The CMF in
Table 12-22 has been adapted from the CMF in Table 13-12 based on the estimate by Harkey et al. (6) that cross-
median collisions represent 12.0 percent of crashes on divided arterials.
This CMF applies only to traversable medians without traffic barriers; it is not applicable to medians serving as
TWLTLs (a CMF for TWLTLs is provided in Chapter 16). The effect of traffic barriers on safety would be expected
to be a function of barrier type and offset, rather than the median width; however, the effects of these factors on
safety have not been quantified. Until better information is available, a CMF value of 1.00 is used for medians with
traffic barriers. The value of this CMF is 1.00 for undivided facilities.
Table 12-22. CMFs for Median Widths on Divided Roadway Segments without a Median Barrier (CMF 3,)
Median Width (ft) CMF
10 1.01
15 1.00
20 0.99
30 0.98
40 0.97
50 0.96
60 0.95
70 0.94
80 0.93
90 0.93
100 0.92
CMF,,-Lighting
The base condition for lighting is the absence of roadway segment lighting (CMF,, ~ 1.00). The CMF for lighted
roadway segments is determined, based on the work ofElvik and Vaa (3), as:
CMF4r ~ l.O- (p nr x (l.O- 0.72 xp.mr -0.83 Xp pnr)) (12-34)
Where:
CMF4, crash modification factor for the effect of roadway segment lighting on total crashes;
P;,, ~ proportion of total nighttime crashes for unlighted roadway segments that involve a fatality or injury;
pP"' ~ proportion of total nighttime crashes for unlighted roadway segments that involve property daroage only; and
p"' ~ proportion of total crashes for unlighted roadway segments that occur at night.
CMF 4, applies to total roadway segment crashes. Table 12-23 presents default values for the nighttime crash propor-
tions p.mr, ppnr, and p nr . Replacement of the estimates in Table 12-23 with locally derived values is encoura2ed.
'-'
If
lighting installation increases the density of roadside fixed objects, the value of CMF2, is adjusted accordingly.

Proportion of Total Nighttime Crashes by Severity Level Proportion of Crashes that Occur at Night
Roadway Segment type Fatal and Injury p 1n, PDOp,,
2U · 0.424 0.576 0.316
3T 0.429 0.571 0.304
4U 0.517 0.483 0.365
4D 0.364 0.636 0.410
5T 0.432 0.568 0.274
CMF,,-Automated Speed Enforcement

Automated speed enforcement systems use video or photographic identification in conjunction with radar or lasers
to detect speeding drivers. These systems automatically record vehicle identification information without the need
for police officers at the scene. The base condition for automated speed enforcement is that it is absent. Chapter 17
presents a CMF of 0.83 for the reduction of all types of fatal-and-injury crashes from implementation of automated
speed enforcement. This CMF is assumed to apply to roadway segments between intersections with fixed camera
sites where the camera is always present or where drivers have no way of knowing whether the camera is present or
not. No information is available on the effect of automated speed enforcement on noninjury crashes. With the con-
servative assumption that automated speed enforcement has no effect on noninjury crashes, the value of the CMF for
automated speed enforcement would be 0.95.

The effects of individual geometric design and traffic control features of intersections are represented in the predic-
tive models by CMFs. CMF u through CMF 41 are applied to multiple-vehicle collisions and single-vehicle crashes
at intersections, but not to vehicle-pedestrian and vehicle-bicycle collisions. CMF 1, through CMF 3, are applied to
vehicle-pedestrian collisions at four-leg signalized intersections (4SG), but not to multiple-vehicle collisions and
single-vehicle crashes and not to other intersection types.
CMF u-Intersection Left-Thrn Lanes

The base condition for intersection left-tum lanes is the absence ofleft-tum lanes on the intersection approaches.
The CMFs for presence ofleft-tum lanes are presented in Table 12-24. These CMFs apply to installation ofleft-tum
lanes on any approach to a signalized intersection but only on uncontrolled major-road approaches to stop-controlled
intersections. The CMFs for installation ofleft-turn lanes on multiple approaches to an intersection are equal to
the corresponding CMF for installation of a left-tum lane on one approach raised to a power equal to the number
of approaches with left-tum lanes. There is no indication of any change in crash frequency for providing a left-tum
lane on an approach controlled by a stop sign, so the presence of a left-tum lane on a stop-controlled approach is not
considered in applying Table 12-24. The CMFs in the table apply to total intersection crashes (not including vehicle-
pedestrian and vehicle-bicycle collisions). The CMFs for installation ofleft-tum lanes are based on research by
Harwood eta!. (7). A CMF of 1.00 is always used when no left-tum lanes are present.
Table 12-24. Crash Modification Factor (CMF,) for Installation of Left-Tum Lanes on Intersection Approaches
Number of Approaches with Left-Turn Lanes•
Three-leg intersection MinoHoad stop controlb 0.67 0.45
Traffic signal 0.93 0.86 0.80
Four-leg_intersection MinoHoad stop controlb 0.73 0.53
Traffic signal 0.90 0.81 0.73 0.66
• Stop-<ontrolled approaches are not considered in determining the number of approaches with left-turn lanes.
CMF,;-Intersection Left-Thrn Signal Phasing

The CMF for left-tum signal phasing is based on the results of work by Hauer (10), as modified in a study by Lyon
eta!. (II). Types ofleft-turn signal phasing considered include permissive, protected, protected/permissive, and
permissive/protected. Protected/permissive operation is also referred to as a leading left-tum signal phase; permis-
sive/protected operation is also referred to as a lagging left-tum signal phase. The CMF values are presented in Table
12-25. The base condition for this CMF is permissive left-tum signal phasing. This CMF applies to total intersection
crashes (not including vehicle-pedestrian and vehicle-bicycle collisions) and is applicable only to signalized inter-
sections. A CMF value of 1.00 is always used for unsignalized intersections.
If several approaches to a signalized intersection have left-tum phasing, the values of CMF, for each approach are
multiplied together. ·'
Table 12-25. Crash Modification Factor (CMF 2) for Type of Left-Tum Signal Phasing
Type of Left-Tum Signal Phasing
Permissive 1.00
Protected/permissive or permissive/protected 0.99
Protected 0.94
Note: Use CMF 21 = 1.00 for all unsignalized intersections. If several

approaches to a signalized intersection have left-turn phasing,
the values of CMF2i for each approach are multiplied together.
CMF31-Intersection Right-Turn Lanes

The base condition for intersection right-tum lanes is the absence of right-tum lanes on the intersection approaches.
The CMFs for presence of right-tum lanes based on research by Harwood eta!. (7) are presented in Table 12-26.
These CMFs apply to installation of right-tum lanes on any approach to a signalized intersection, bnt only on un-
controlled major-road approaches to stop-controlled intersections. The CMFs for installation of right-tum lanes on
multiple approaches to an intersection are equal to the corresponding CMF for installation of a right-tum lane on one
approach raised to a power equal to the number of approaches with right-tum lanes. There is no indication of any
change in crash frequency for providing a right-tum lane on an approach controlled by a stop sign, so the presence of
a right-tum lane on a stop-controlled approach is not considered in applying Table 12-26.
The CMFs in Table 12-26 apply to total intersection crashes (not including vehicle-pedestrian and vehicle-bicycle
collisions). A CMF value of 1.00 is always used when no right-tum lanes are present. This CMF applies only to
right-tum lanes that are identified by marking or signing. The CMF is not applicable to long tapers, flares, or paved
shoulders that may be used informally by right-tum traffic.
Table 12-26. Crash Modification Factor (CMF31) for Installation of Right-Tum Lanes on Intersection Approaches
Number of Approaches with Right-Turn Lanes~
Intersection Type Type of Traffic Control One Approach Two Approaches Three Approaches Four Approaches
Three-leg intersection Minor-road stop controlb 0.86 0.74
Traffic signal 0.96 0.92
Four-leg intersection MinoNoad stop controJb 0.86 0.74
Traffic signal 0.96 0.92 0.88 0.85
~Stop-controlled approaches are not considered in determining the number of approaches with right-turn lanes.
CMF41-Right-Turn-on-Red
The CMF for prohibiting right-tum-on-red on one or more approaches to a signalized intersection has been derived
from a study by Clark (2) and from the CMFs for right-tum-on-red operation shown in Chapter 14. The base condi-
tion for CMF 4; is permitting a right-tum-on-red at all approaches to a signalized intersection. The CMF is deter-
mined as:
(12-35)
W'here:
CMF4 ; crash modification factor for the effect of prohibiting right turns on red on total crashes; and
number of signalized intersection approaches for which right-tum-on-red is prohibited.
This CMF applies to total intersection crashes (not including vehicle-pedestrian and vehicle-bicycle collisions) and
is applicable only to signalized intersections. A CMF value of 1.00 is used for unsignalized intersections.
CMF,-Lighting
The base condition for lighting is the absence of intersection lighting. The CMF for lighted intersections is adapted
from the work ofE!vik and Vaa (3), as:
CMF" = I - 0.38 x p,1 (12-36)
Where:
CMF51 = crash modification factor for the effect of intersection lighting on total crashes; and
p,1 = proportion of total crashes for unlighted intersections that occur at night.
This CMF applies to total intersection crashes (not including vehicle-pedestrian and vehicle-bicycle collisions).
Table 12-27 presents default values for the nighttime crash proportion, p,,. HSM users are encouraged to replace the
estimates in Table 12-27 with locally derived values.
Table 12-27. Nighttime Crash Proportions for Unlighted Intersections

Proportion of Crashes that Occur at Night
Intersection Type
3ST 0.238
4ST 0.229
3SG and4SG 0.235
CMF ,,-Red-Light Cameras

The base condition for red light cameras is their absence. The CMF for installation of a red light camera for enforce-
ment of red signal violations at a signalized intersection is based on an evaluation by Persaud eta!. (12). As shown in
Chapter 14, this study indicates a CMF for red light camera installation of0.74 for right-angle collisions and a CMF
of 1.18 for rear-end collisions. In other words, red light cameras would typically be expected to reduce right-angle
collisions and increase rear-end collisions. There is no evidence that red light camera installation affects other colli-
sion types. Therefore, a CMF for the effect of red light camera installation on total crashes can be computed with the
following equations:
CMF61 = J- Pro X (J- 0.74)- p~ X (1- 1.18) (12-37)
Promv(FJ) X ~imv(Fl) + Pramv(PDO) X ~imv(PDO)

Pro= (12-38)
( Nb,mv(FI) + N,;mv(PDO) + Nbi.v)
Prcmv(FI) X Nbimv(Fl) + Prcmv(PDO) X Nbimv(PDO)

Pre= (12-39)
( Nbimv(FJ) + Nbimv(PDO) + Nbisv)
Where:
CMF61 = crash modification factor for installation of red light cameras at signalized intersections;
pro = proportion of crashes that are multiple-vehicle, right-angle collisions;
pre = proportion of crashes that are multiple-vehicle, rear-end collisions;
p rom•Fn = proportion of multiple-vehicle fatal-and-injury crashes represented by right-angle collisions;
Prom,(PDOJ = proportion of multiple-vehicle property-damage-only crashes represented by right-angle collisions;
12-46 HIGHWAY SAFElY' MANUAL
P~m.;m ~ proportion of multiple-vehicle fatal-and-injury crashes represented by rear-end collisions; and
P~m.;Poo) ~ proportion of multiple-vehicle property-damage-only crashes represented by rear-end collisions.
The values ofN, 1 .,,F~ is available from Equation 12-22, the value ofN,1m,(PDOJ is available from Equation 12-23,
and the value ofNbisv is available from Equation 12-24. The values ofpramv(Fl)' Pramv(PDOJ' Premv(FI)' and Premv(PDO) can be
determined from data for the applicable intersection type in Table 12-11. The values in Table 12-11 may be updated
with data for a particular jurisdiction as part of the calibration process presented in Appendix A to Part C. The
data in Table 12-11, by definition, represent average values for a broad range of signalized intersections. Because
jurisdictions are likely to implement red-light cameras at intersections with higher than average proportions of
right-angle collisions, it is acceptable to replace the values in Table 12-11 with estimate based on data for a specific
intersection when determining the value of the red light camera CMF.
12.7.3. Crash Modification Factors for Vehicle-Pedestrian Collisions at Signalized Intersections

The CMFs for vehicle-pedestrian collisions at signalized intersections are presented below.
CMF1,-Bus Stops
The CMFs for the number of bus stops within 1,000 ft of the center of the intersection are presented in Table 12-28.
The base condition for bus stops is the absence of bus stops near the intersection. These CMFs apply to total vehicle-
pedestrian collisions and are based on research by Harwood et a!. (8).
Table 12-28. Crash Modification Factor (CMF 1,) for the Presence of Bus Stops near the Intersection
Number of Bus Stops within 1,000 ft of the Intersection CMF 1
0 1.00
1 or2 2.78
3 or more 4.15
In applying Table 12-28, multiple bus stops at the same intersection (i.e., bus stops in different intersection quadrants
or located some distance apart along the same intersection leg) are counted separately. Bus stops located at
adjacent intersections would also be counted as long as any portion of the bus stop is located within l ,000 ft of the
intersection being evaluated.
CMF,,-Scbools
The base condition for schools is the absence of a school near the intersection. The CMF for schools within l ,000
ft of the center of the intersection is presented in Table 12-29. A school may be counted if any portion of the school
grounds is within 1,000 ft of the intersection. Where one or more schools are located near the intersection, the value
of the CMF is independent of the number of schools present. This CMF applies to total vehicle-pedestrian collisions
and is based on research by Harwood et al. (8).
This CMF indicates that an intersection with a school nearby is likely to experience more vehicle-pedestrian col-
lisions than an intersection without schools even if the traffic and pedestrian volumes at the two intersections are
identical. Such increased crash frequencies indicate that school children are at higher risk than other pedestrians.
Table 12-29. Crash Modification Factor (CMF,) for the Presence of Schools near the Intersection
Presence of Schools within 1,000 ft of the Intersection CMF1
No school present l.OO
School present 1.35
CMF, -Alcohol Sales Establishments

The b:se condition for alcohol sales establishments is the absence of alcohol sales establishments near the inter-
section. The CMF for the number of alcohol sales establishments within I ,000 ft of the center of an intersection is
presented in Table 12-30. Any alcohol sales establishment wholly or partly within 1,000 ft of the intersection may be
counted. The CMF applies to total vehicle-pedestrian collisions and is based on research by Harwood et al. (8).
This CMF indicates that an intersection with alcohol sales establishments nearby is likely to experience more
vehicle-pedestrian collisions than an intersection without alcohol sales establishments even if the traffic and
pedestrian volumes at the two intersections are identical. This indicates the likelihood of higher risk behavior on
the part of either pedestrians or drivers near alcohol sales establishments. The CMF includes any alcohol sales
establishment which may include liquor stores, bars, restaurants, convenience stores, or grocery stores. Alcohol sales
establishments are counted if they are on any intersection leg or even on another street, as long as they are within
1,000 ft of the intersection being evaluated.
Table 12-30. Crash Modification Factor (CMF3,) for the Number of Alcohol Sales Establishments near the Intersection
Number of Alcohol Sales Establishments
within 1,000 ft of the Intersection eMF,
0 1.00
1-S !.12
9 or more !.56
12.8. CALIBRATION OF THE SPFS TO LOCAL CONDITIONS

widely from one jurisdiction to another. Geographic regions differ markedly in climate, animal population, driver
populations, crash reporting threshold, and crash reporting practices. These variations may result in some jurisdic-
tions experiencing a different number of reported traffic crashes on urban and suburban arterial highways than
others. Calibration factors are included in the methodology to allow highway agencies to adjust the SPFs to match
actual local conditions.
The calibration factors for roadway segments and intersections (defined below as C, and C 1, respectively) will have
values greater than 1.0 for roadways that, on average, experience more crashes than the roadways used in the devel-
opment of the SPFs. The calibration factors for roadways that experience fewer crashes on average than the road-
ways used in the development of the SPFs will have values less than 1.0. The calibration procedures are presented in
vidual agencies or locations. Several other default values used in the methodology, such as collision type distribu-
tion, can also be replaced with locally derived values. The derivation of values for these parameters is addressed in
12.9. INTERIM PREDICTIVE METHOD FOR ROUNDABOUTS

Sufficient research has not yet been conducted to form the basis for development of a predictive method for round-
abouts. Since many jurisdictions are planning projects to convert existing intersections into modern roundabouts,
an interim predictive method is presented here. This interim procedure is applicable to a location at which a modem
roundabout has been constructed or is being planned to replace an existing intersection with minor-road stop control
or an existing signalized intersection. The interim procedure is:
I. Apply the predictive method from Chapter 12 to estimate the crash frequency, N'"'' for the existing intersection.
2. Multiply N,, by the appropriate CMF from Chapter 12 for conversion on an existing intersection to a modem
roundabout. The applicable CMFs are:
• 0.56 for conversion of a two-way stop-contro!led intersection to a modern roundabout.

• 0.52 for conversion of a signalized intersection to a modem roundabout.
These CMFs are applicable to all crash severities and collision types for both one- and two-lane roundabouts in all
settings.
At present, there are no available SPFs to determine predicted average crash frequency of an existing or newly
constructed roundabout where no intersection currently exists.
12.10. LIMITATIONS OF PREDICTIVE METHOD IN CHAPTER 12

The limitations of the predictive method which apply generally across all of the Part C chapters are discussed in
Section C.l4 of the Part C-lntroduction and Applications Guidance chapter. This section discusses limitations
of the specific predictive models and the application of the predictive method in Chapter 12.
Where urban and suburban arterials intersect access-contro!led facilities (i.e., freeways), the grade-separated
interchange facility, including the arterial facility within the interchange area, cannot be addressed with the
predictive method for urban and suburban arterials.
12.11. APPLICATION OF CHAPTER 12 PREDICTIVE METHOD

The predictive method presented in Chapter 12 applies to urban and suburban arterials. The predictive method is
applied to by following the !8 steps presented in Section !2.4. Appendix !2A provides a series of worksheets for
applying the predictive method and the predictive models detailed in this chapter. All computations within these
worksheets are conducted with values expressed to three decimal places. This level of precision is needed for consis-
tency in computations. In the last stage of computation, rounding the final estimate expected average crash frequency
to one decimal place.
12.12. SUMMARY
The predictive method is used to estimate the expected average crash frequency for a series of contiguous sites
(entire urban or suburban arterial facility), or a single individual site. An urban or suburban facility is defined in
Section 12.3.
The predictive method for urban and suburban arterial highways is applied by following the 18 steps of the predic-
tive method presented in Section 12.4. Predictive models, developed for urban and suburban arterial facilities, are
applied in Steps 9, 10, and II of the method. These models have been developed to estimate the predicted average
crash frequency of an individual intersection or homogenous roadway segment. The facility is divided into these
individual sites in Step 5 of the predictive method.
Where observed data are available, the EB Method may be applied in Step 13 or 15 of the predictive method to
improve the reliability of the estimate. The EB Method can be applied at the site-specific level or at the project
specific level. It may also be applied to a future time period if site conditions will not change in the future period.
The EB Method is described in the Part C, Appendix A.2.
Each predictive model in Chapter 12 consists of a safety performance function (SPF), crash modification fac-
tors (CMFs), a calibration factor, and pedestrian and bicyclist factors. The SPF is selected in Step 9 and is used to
estimate the predicted average crash frequency for a site with base conditions. This estimate can be for either total
crashes or organized by crash-severity or collision-type distribution. In order to account for differences between the
base conditions of the SPF and the actu'!l conditions of the local site, CMFs are applied in Step 10 which adjust the
predicted number of crashes according the geometric conditions of the site.
In order to account for the differences in state or regional crash frequencies, the SPF is calibrated to the specific
state and or geographic region to which they apply. The process for determining calibration factors for the predictive
models is described in the Part C, Appendix A.l.
Section 12.13 presents six sample problems which detail the application of the predictive method. A series of template
worksheets have been developed to assist with applying the predictive method in Chapter 12. These worksheets are
utilized to solve the sample problems in Section 12.13, and Appendix 12A contains blank versions of the worksheets.

In this section, six sample problems are presented using the predictive method steps for urban and suburban arterials.
Sample Problems 1 and 2 illustrate how to calculate the predicted average crash frequency for urban and suburban
arterial roadway segments. Sample Problem 3 illustrates how to calculate the predicted average crash frequency for a
stop-controlled intersection. Sample Problem 4 illustrates a similar calculation for a signalized intersection. Sample
Problem 5 illustrates how to combine the results from Sample Problems 1 through 4 in a case where site-specific
observed crash data are available (i.e., using the site-specific EB Method). Sample Problem 6 illustrates how to com-
bine the results from Sample Problems 1 through 4 in a case where site-specific observed crash data are not available
(i.e., using the project-level EB Method).

12-49 Predicted average crash frequency for a three-lane TWLTL arterial roadway segment
2 12-63 Predicted average crash frequency for a four-lane divided arterial roadway segment
4 12-86 Predicted average crash frequency for a four-leg signalized intersection
5 12-97 Expected average crash frequency for a facility when site-specific observed crash data are available
6 12-101 Expected average crash frequency for a facility when site-specific observed crash data are not available
The Site/Facility
A three-lane urban arterial roadway segment with a center two-way left-turn lane (TWLTL).
The Question
The Facts
• 1.5-mi length
• 11,000 veh/day
• 1.0 mi of parallel on-street commercial parking on each side of street
• 30 driveways (10 minor commercial, 2 major residential, 15 minor residential,

3 minor industrial/institutional)
• I 0 roadside fixed objects per mile

• 6-ft offset to roadside fixed objects
• Lighting present
• 35-mph posted speed
Assumptions
Collision type distributions used are the default values presented in Tables 12-4 and 12-6 and Equations 12-19 and 12-20.
The calibration factor is assumed to be 1.00.
Results
in Sample Problem I is determined to be 7.0 crashes per year (rounded to one decimal place).
Steps
Step 1 through 8
For a three-lane urban arterial roadway segment with TWLTL, SPF values for multiple-vehicle nondriveway, single-
vehicle, multiple-vehicle driveway-related, vehicle-pedestrian, and vehicle-bicycle collisions are determined. The
calculations for vehicle-pedestrian and vehicle-bicycle collisions are shown in Step I 0 since the CMF values are
needed for these models.
Multiple- Vehicle Nondriveway Collisions

The SPF for multiple-vehicle nondriveway collisions for the roadway segment is calculated from Equation 12-10 and
Table 12-3 as follows:
N,,." ~ exp(a + b x In(AADT) + In(L))
N''""""'' ~ exp(-!2.40 + 1.41 x Jn(ll,OOO) + In(l.5))

N''""''~ ~ exp(-16.45 + 1.69 x In(ll,OOO) + In(l.5))

~ 0. 728 crashes/year
N,,m>'(PDD) ~ exp(-!!.95 + 1.33 x In(!! ,000) + In(!.5))

These initial values for fatal-and-injury (Fl) and property-damage-only (PDO) crashes are then adjusted using
Equations 12-11 and 12-12 to assure that they sum to the value for total crashes as follows:
12-51
CHAPTER 12- PREDICTIVE METHOD FOR URBAN AND SUBURBAN ARTERIALS
Nbrmv(FI) = Nbrmv(tota!) ( N'

N~rmv(FI)
N'
J
brmv(Fl) + brmv(PDO)
0 728
= 3.085( • )
0. 728 + 2.298
N brm~(PDO) =Nbrmv{!otnl) -Nbrmv{FI)
= 3.085-0.742
Single- Vehicle Crashes

The SFP for single-vehicle crashes for the roadway segments is calculated from Equation 12-13 and Table 12-5 as follows:
N,,.~ = exp(a + b x In(AADT) + In(L))

Nbm(cool) = exp(-5.74 + 0.54 X /n(ll,OOO) + In(l.5))
= 0. 734 crashes/year
Nbmlm = exp(-6.37 + 0.47 x In(ll,OOO) + In(1.5))

N,~.;PDO) = exp(-6.29 + 0.56 x In(ll,OOO) + In(1.5))

These initial values for fatal-and-injury (FI) and property-damage-only (PDO) crashes are then adjusted using
N~,v(FI) J
Nbrsv(FI) = Nbrsv(total) ( N' N'
brsv(FI) + brsv(PDO)
= 0.734x ( 0.204 )
0.204 + 0.510
N br:sv(PDO) = N b~:~·v{!ot:!l) -Nbrs~(FI)
= 0.734-0.210

The SPF for multiple-vehicle driveway-related collisions for the roadway segment is calculated from Equation 12-16
as follows:
Nbrdwy(total) = Iill l
nj X Nj X - - -
AADTJ(tJ
15,000
driveway
types
The number of driveways within the roadway segment, nl' for Sample Problem I is I 0 minor commercial, two major
residential, 15 minor residential, and three minor industrial/institutional.
The number of driveway-related collisions, N,) and the regression coefficient for AADT, t, for a three-lane arterial are
provided in Table 12-7.
II 000 )(l.O) (II 000 J(l.O)

Nbrdwy(total) = 10x0.032x ( - ' - +2x0.053x - ' -
15,000 15,000
+ J5 X 0.0J0 X ( I1,000)(l.O) + 3x 0.015 X ( l1, OOO)(l.O)

15,000 15,000
Driveway-related collisions can be separated into components by severity level using Equations 12-17 and 12-18 as
follows:
From Table 12-7, for a three-Jane arterial the proportion of driveway-related collisions that involve fatalities and
injuries, f,., = 0.243
= 0.455 X 0.243
Nbrrlwy(PDO) = Nbrdwy(total) - Nbrdwy(FJ)
= 0.455-0.111
On-Street Parking (CMF)

CMF 1, is calculated from Equation 12-32 as follows:
CMF), = I + Ppk X if;,- 1.0)

The proportion of curb length with on-street parking, ppk' is determined as follows:
Lpk
Ppk =0.5xL
Since 1.0 mile of on-street parking on each side of the road is provided, the sum of curb length with on-street park-
ing for both sides of the road combined, LP, = 2.
2
Ppk = 0.5x- = 0.66
1.5
From Table 12-19, fP, = 2.074.
1 + 0.66 X (2.074- 1.0)
1.71
Roadside Fixed Objects (CMF2 )

CMF 2 , is calculated from Equation 12-33 as follows:
From Table 12-20, for a roadside fixed object with a 6-ft offset, the fixed-object offset factor, f,.,.,, is interpolated as 0.124.
From Table 12-21, for a three-lane arterial the proportion of total crashes, p1, = 0.034.
CMF,, = 0.124 X 10 X 0.034 + (1.0- 0.034)
= 1.01
Median Width (CMF3 )

The value of CMF 3, is 1.00 for undivided facilities (see Section 12.7 .1 ). It is assumed that a roadway with TWLTL
is undivided.
Lighting (CMF4)
CMF 4, is calculated from Equation 12-34 as follows:
CMF,, = 1.0- (p,, X (1.0- 0.72 X P,,- 0.83 X PpJ)
For a three-lane arterial, p 1nr = 0.429, ppnr = 0.571, and p nr = 0.304 (see Table 12-23).
CMF,, = 1.0- (0.304 X (1.0- 0.72 X 0.429-0.83 X 0.571))
=0.93
Automated Speed Enforcement (CMF5)

Since there is no automated speed enforcement in Sample Problem 1, CMF5, = 1.00 (i.e., the base condition for
CMF ,, is the absent of automated speed enforcement).
CMF,omb = 1.71 X 1.01 X 0.93
= 1.61
Vehicle-Pedestrian and Vehicle-Bicycle Collisions

The predicted average crash frequency of an individual roadway segment (excluding vehicle-pedestrian and
vehicle-bicycle collisions) for SPF base conditions, N,,, is calculated first in order to determine vehicle-pedestrian
and vehicle-bicycle crashes. N,, is determined from Equation 12-3 as follows:
N,, = N,pfn X (CMF1, X CMF2, X •.• X CMFJ
From Equation 12-4, N,pfn can be calculated as follows:

N spfrs = Nbrm~ +Nbrw +Nbrdwy
= 3.085 + 0.734 + 0.455
The combined CMF value for Sample Problem I is 1.61.
N,, = 4.274 X (1.61)

The SPF for vehicle-pedestrian collisions for the roadway segment is calculated from Equation 12-19 as follows:
N pedr = N br xJ.pedr
From Table 12-8, for a posted speed greater than 30 mph on three-lane arterials the pedestrian crash adjustment factor,
f,.d, = 0.013.
Npedr = 6.881 X 0.013
0.089 crashes/year
The SPF for vehicle-bicycle collisions is calculated from Equation 12-20 as follows:
From Table 12-9, for a posted speed greater than 30 mph on three-lane arterials the bicycle crash adjustment factor,
!,,., = 0.007.
N,;.,, = 6.881 X 0.007
It is assumed in that a calibration factor, C,, of 1.00 has been determined for local conditions. See Part C, Appendix A.! for
further discussion on calibration of the predicted models.

N.
pre<l1cted rs
Cx(N
r br
+Npedr +N'
blkerl
1.00 X (6.88) + 0.089 + 0.048)
7.018 crashes/year
WORKSHEETS
The step-by-step instructions above are provided to illustrate the predictive method for calculating the predicted average
crash frequency for a roadway segment. To apply the predictive method steps to multiple segments, a series of 12 work-
sheets are provided for determining the predicted average crash frequency. The 12 worksheets include:
• Worksheet SP 1A (Corresponds to Worksheet 1A)-General Information and Input Data for Urban and Suburban
Arterial Roadway Segments
• Worksheet SP 1B (Corresponds to Worksheet 1B)-Crash Modification Factors for Urban and Suburban Arterial
Roadway Segments
• Worksheet SPIC (Corresponds to Worksheet !C)-Multiple-Vehicle Nondriveway Collisions by Severity Level for
Urban and Suburban Arterial Roadway Segments
• Worksheet SP I D (Corresponds to Worksheet I D)-Multiple-Vehicle Non driveway Collisions by Collision Type for
• Worksheet SP IE (Corresponds to Worksheet I E)-Single-Vehicle Crashes by Severity Level for Urban and
Suburban Arterial Roadway Segments
• Worksheet SP IF (Corresponds to Worksheet I F)-Single-Vehicle Crashes by Collision Type for Urban and
• Worksheet SP I G (Corresponds to Worksheet I G)-Multiple-Vehicle Driveway-Related Collisions by Driveway

Type for Urban and Suburban Arterial Roadway Segments
• Worksheet SP I H (Corresponds to Worksheet Il:\)-Multiple-Vehicle Driveway-Related Collisions by Severity

Level for Urban and Suburban Arterial Roadway Segments
• Worksheet SP II (Corresponds to Worksheet II)-Vehicle-Pedestrian Collisions for Urban and Suburban Arterial
Roadway Segments
• Worksheet SP JJ (Corresponds to Worksheet IJ)-Vehicle-Bicycle Collisions for Urban and Suburban Arterial
Roadway Segments
• Worksheet SP IK (Corresponds to Worksheet IK)----Crash Severity Distribution for Urban and Suburban Arterial
Roadway Segments
• Worksheet SP lL (Corresponds to Worksheet IL)-Summary Results for Urban and Suburban Arterial Roadway
Segments
provided in Appendix 12A.
Worksheet SP1A-Generallnformation and Input Data for Urban and Suburban Roadway Segments
Worksheet SPIA is a summary of general information about the roadway segment, analysis, input data
(i.e., "The Facts"), and assumptions for Sample Problem I.
Worksheet SP1 A. General Information and Input Data for Urban and Suburban Roadway Segments
Analyst Roadway
----- ------------ ------------------------- -
Analysis Year

Road type (2U, 3T, 4U, 4D, ST) I 3T
------------------------'------------- f - - - - - - - - - - - - - -
Length of segment. L (mi) 1.5
AADT (veh/day) 11,000
Type of on-street parking (none/paralleVangle) none parallel-commercial
Proportion of curb length with on-street parking 0.66
·-· -· ·-· ·' ---- ----- --- -"-- ---- ------'- -- -~-- ------- ----- ,.,.:. . . . . ------ ----- -- ---- ----1- ----- -- ---------------- ---- -
Median width (ft) I
15 not present
Lighting (presem/not present) not present present
Major commercial driveways (number) 0
Minor commercial driveways (number) 10
Major industrial/instrturional driveways (number) 0
Minor industriaUinstitutional driveways (number) 3
Major residential driveways (number) 2
Minor residential driveways (number) 15

----------~-------'----------------~------~--- -----------
Other driveways (number) 0
Speed Category intermediate or high speed (>30 mph)
Roadside fixed object density (fixed objects/mi) 10

Offset to roadside fixed objects (ft) not present 6
Calibration Factor, C, LO LO
Worksheet SP1 B. Crash Modification Factors for Urban and Suburban Roadway Segments
Column 6 of Worksheet SPlB which indicates the combined CMF value.
Worksheet SP1 B. Crash Modification Factors for Urban and Suburban Roadway Segments
(I) (2) (3) (4) (5) (6)
CMF for CMF for Roadside ' CMF for Auto I
On-Street Parking Fixed Objects CMF for Median Width CMF for Lighting Speed Enforcement Combined CMF
CMF 1, CMF,, CMF_lr CMF 4,. CMF_,, I

CMF mmh
from Equation 12-32 from Equation 12-33 from Table 12-22 ; from Equation 12-34 from Section 12.7.1 I (I )'(2)'(3 )'(4 )'( 5)
'
1.71 1.01 1.00 0.93 1.00 1.61
CHAPTER 12- PREDICTIVE METHOD FOR URBAN AND SUBURBAN ARTERIALS 12-S7
Worksheet SP1C-Multiple-Vehicle Nondriveway Collisions by Severity Level for Urban and

Suburban Roadway Segments
The SPF for multiple-vehicle nondriveway collisions along the roadway segment in Sample Problem I is calculated
using Equation 12-10 and entered into Column 4 of Worksheet SPIC. The coefficients for the SPF and the overdis-
persion parameter associated with the SPF are entered into Columns 2 and 3; however, the overdispersion parameter
is not needed for Sample Problem I (as the EB Method is not utilized). Column 5 of the worksheet presents the
proportions for crash severity levels calculated from the results in Column 4. These proportions are used to adjust
the initial SPF values (from Column 4) to assure that fatal-and-injury (FI) and property-damage-only (PDO) crashes
sum to the total crashes as illustrated in Column 6. Column 7 represents the combined CMF (from Column 6 in
Worksheet SPIB), and Column 8 represents the calibration factor. Column 9 calculates the predicted average crash
frequency of multiple-vehicle nondriveway crashes using the values in Column 6, the combined CMF in Column 7,
and the calibration factor in Column 8.
Worksheet SP1C. Multiple-Vehicle Nondriveway Collisions by Severity Level for Urban and Suburban
Roadway Segments
(1) (2) (3) (4)
SPF Coefficients Overdispersion Parameter, k Initial Nhrmv
from Table 12-3
Crash Severity Level a b from Table 12-3 from Equation 12-10
Total -12.40 1.41 0.66 3.085

Fatal and injury (FI) I -16.45 1.69 0.59 0.728
.. - - ...,_ -
r-
-"
Property damage only (PDQ) . -11.95 1.33 0.59 I 2.298
Worksheet SP I C. continued
(1) (5) (6) (7) (8) (9)
Combined Calibration
Adjusted N brm• CMFs Factor Predicted NMm•
(6) from
Crash Severity Level I Proportion ofTota1 Crashes (4),••,*(5) Worksheet SPIB c, i
(6)*(7)*(8)
Total I 1 ooo 3 085 1 61 1.00 I
4 967
I
Fatal and injury (PI) ~-'-(4":),~/_((4),+(4),00) 0.743 1.61 1.00 1.196
I
I
0.241
Property damage only (PDQ) i
(~~~.~=~.~2.t"' 2.342 1.61 f
I
1.00 ' 3.771
t· 0.759
Worksheet SP1 0-Multiple-Vehicle Nondriveway Collisions by Collision Type for Urban and
Worksheet SPID presents the default proportions for collision type (from Table 12-4) by crash severity level as follows:
Using the default proportions, the predicted average crash frequency for multiple-vehicle nondriveway crashes by
collision type is presented in Columns 3 (Fatal and Injury, Fl), 5 (Property Damage Only, PDO), and 6 (Total).
These proportions may be used to separate the predicted average crash frequency for multiple-vehicle nondriveway
crashes (from Column 9, Worksheet SP!C) into components by crash severity and collision type.
(I) (2) (3) (4) (5) I
(6)
Proportion of Predicted Nbrmvcm Proportion of I Predicted Na,mv(PDO) I Predicted Nbrnw (roml)
Collision Type IFII ' (crashes/year) Collision Type (PDO) (crashes/year) 1
(crashes/year)
I (9)F1from (9)p00 from (9),olo1 from

Collision Type from Table 12-4 Worksheet SPl C from Table 12-4 Worksheet SPlC Worksheet SPlC
I
Total 1.000 1.196 1.000 3.771 4.967
(2)*(3)" (4)*(5),00 (3)+(5)

Rear-end collision 0.845 1.011 0.842 3.175 4.186
Head-on collision 0.034 0.041 0.020 O.D75 0.116
Angle collision 0.069 0.083 0.020 I 0.075 0.158
Sideswipe, same direction 0.001 0.001 O.D78 1:
0.294 0.295
Sideswipe, opposite direction 0.017 0.020 0.020 0.075 0.095
Other multiple-vehicle collision 1

0.034 0.041 0.020 I O.D75 0.116
Worksheet SP1 E-Single-Vehicle Crashes by Severity Level for Urban and Suburban Roadway Segments
The SPF for single-vehicle crashes along the roadway segment in Sample Problem I is calculated using Equation
12-13 and entered into Column 4 of Worksheet SPlE. The coefficients for the SPF and the overdispersion param-
eter associated with the SPF are entered into Columns 2 and 3; however, the overdispersion parameter is not needed
for Sample Problem l (as the EB Method is not utilized). Column 5 of the worksheet presents the proportions for
crash severity levels calculated from the results in Column 4. These proportions are used to adjust the initial SPF
values (from Column 4) to assure that fatal-and-injury (FI) and property-damage-only (PDO) crashes sum to the
total crashes as illustrated in Column 6. Column 7 represents the combined CMF (from Column 6 in Worksheet
SPlB), and Column 8 represents the calibration factor. Column 9 calculates the predicted average crash frequency
of multiple-vehicle nondriveway crashes using the values in Column 6, the combined CMF in Column 7, and the
calibration factor in Column 8.
Worksheet SP1 E. Single-Vehicle Collisions by Severity Level for Urban and Suburban Roadway Segments
(I) I (2) ' (3) (4) (5) (6) (7) I (8) (9)
SPF Overdispersion Adjusted I Combined ! Calibration Predicted

Coefficients I
Parameter, k Initial Nb,.. Nh,.. CMFs Factor Nbr:w
'
from Table
Crash I from (6) from
12-5
Severity Equation Proportion of Worksheet I
Level a I b 1 from Table 12-5 12-13 Total Crashes '
'
(4),o,.t(5) SP!B I c, (6)'(7)'(8)
I
Total i -5.74 I o.54 1 1.37 0.734 1.000 0.734 1.61 1.00 1.182
I
Fatal and I (4M((4),+(4),0 , ) I
-6.37 ----
inJury 0.47 I
1.06 0.204 I 0.210 1.61 1.00 0.338
(FI) ' 0.286
- --- - . - -·-- ..........•.... I · . - - .. -···- - ··-- -····- . -···-·- -- .
Property
damage
. ~~!,~c:~.?2r.::.
-6.29 0.56 1.93 -1 0.524 1.61 1.00 0.844
only 0.714
(PDO) I
Worksheet SP1 F-Single-Vehicle Crashes by Collision Type for Urban and Suburban Roadway Segments
Worksheet SPlF presents the default proportions for collision type (from Table 12-5) by crash severity level as follows:
• Fatal-and-injury crashes (Colwnn 2)
• Property-damage-only crashes (Colwnn 4)
Using the default proportions, the predicted average crash frequency for single-vehicle crashes by collision type is
presented in 3 (Fatal and Injury, FI), 5 (Property Damage Only, PDO), and Colwnns 6 (Total).
These proportions may be used to separate the predicted average crash frequency for single-vehicle crashes
(from Colwnn 9, Worksheet SPlE) into components by crash severity and collision type.
Worksheet SP1 F. Single-Vehicle Collisions by Collision Type for Urban and Suburban Roadway Segments
(I) ! (2) (3) (4) (5) (6)
I
Proportion of
Proportion of Predicted Nb,.(F/J Collision Type Predicted Nhr:rv(J>DO) Predicted Nb.-s• twtall
Collision Type (F/) (crashes/year) (PDOl I
(crashes/year) 1
(crashes/year)
'
(9)F1from (9)PDO from (9),.m1 from

Collision Type from Table 12-6 Worksheet SPlE I from Table 12-6 I Worksheet SPIE Worksheet SPIE
Total 1.000 0.338 1.000 0.844 I 1.182
(2)'(3), I I (4)'(5),00 (3)+(5)
I
Collision with animal 0.001 0.000 0.001 0.001 0.001
----·-~---------- -
!
--------------
Collision with fixed object 0.688 0.233 0.963 0.813 1.046
..
Collision with other object 0.001 0.000 0.001 I 0.001 0.001
-- .,,_,,_ . - - - ,_.,, ....................... ,,_,,_ . -- """" - """""'-"'""""'- - _,_,_ ""
- ... -. -- """" - -"-"-"'""'" - """' ... -~-- - "'"" """- - -'-"'""'- - ...... ... - """"'"""" - ""'""'" -
Other single-vehicle collision i 0.310 '

'
0.105 i 0.035 0.030 0.135
Worksheet SP1G-Multiple-Vehicle Driveway-Related Collisions by Driveway Type for Urban and

Worksheet SPl G determines and presents the number of driveway-related multiple-vehicle collisions. The number of
driveways along both sides of the road is entered in Colwnn 2 by driveway type (Column 1). The associated number
of crashes per driveway per year by driveway type as found in Table 12-7 is entered in Column 3. Colwnn 4 contains
the regression coefficient for AADT also found in Table 12-7. The initial average crash frequency of multiple-vehicle
driveway-related crashes is calculated from Equation 12-16 and entered into Colwnn 5. The overdispersion param-
eter from Table 12-7 is entered into Column 6; however, the overdispersion parameter is not needed for Sample Prob-
lem 1 (as the EB Method is not utilized).
Worksheet SP1 G. Multiple-Vehicle Driveway-Related Collisions by Driveway Type for Urban and Suburban
Roadway Segments
(!) (2) (3) (4) (5) (6)
Crashes per Coefficient

Driveway per for Traffic Overdispersion
Year,NJ Adjustment, t Initial Nh,.......,. Parameter, k
from Table
Driveway Type i
.,
Number ofDriveways, I
from Table 12-71 12-7
Equation 12-16
n *N/(AADT/IS,OOO)t I from Table 12-7
Major commercial 0 0.102 1.000 0.000

--------
Minor commercial -I 10 0.032 1.000 0.235

Major industriaVinstitutional 0 0.110 1.000 0.000
Minor industriaVinstitutional 3 0.015 1.000 0.033
Major residential 2 0.053 1.000 0.078
Minor residential 15 0.010 1.000 0.110 --1
Other 0 0.016 1.000 0.000
--I
Total I 0.456 LIO
Worksheet SP1H-Multiple-Vehicle Driveway-Related Collisions by Severity Level for Urban and

The initial average crash frequency of multiple-vehicle driveway-related crashes from Column 5 of Worksheet SPlG
is entered in Column 2. This value is multiplied by the proportion of crashes by severity (Column 3) found in Table
12-7 and the adjusted value is entered into Column 4. Column 5 represents the combined CMF (from Column 6 in
Worksheet SPlB), and Column 6 represents the calibration factor. Column 7 calculates the predicted average crash
frequency of multiple-vehicle driveway-related crashes using the values in Column 4, the combined CMF in Column
5, and the calibration factor in Column 6.
Worksheet SP1 H. Multiple-Vehicle Driveway-Related Collisions by Severity Level for Urban and Suburban
Roadway Segments
(I) (2) (3) (4) (5) (6) (7)
Proportion of
Total Crashes
Initial Nhrd"'" (f,.,) Adjusted Nb..,..., Combined CMFs Predicted Nh,.....,.
(5\ooal from (6) from Calibration

Crash Severity Level Worksheet SPIG I from Table 12-7 ; (2),•• , '(3) Worksheet SPIB Factor, C, (4)'(5)'(6)
Total 0.456 I LOOO I 0.456 I 1.61 1.00 0.734
, __
:-:~~ ---1
--------------- -, -~- -- --- -- --
Property damage
0.757 1.61 LOO 0.555
only (PDQ)
Worksheet SP11-Vehicle-Pedestrian Collisions for Urban and Suburban Roadway Segments

The predicted average crash frequency of multiple-vehicle nondriveway, single-vehicle, and multiple-vehicle
driveway-related predicted crashes from Worksheets SPlC, SPlE, and SPlH are entered into Columns 2, 3, and 4,
respectively. These values are summed in Column 5. Column 6 contains the pedestrian crash adjustment factor
(see Table 12-8). Column 7 represents the calibration factor. The predicted average crash frequency of vehicle-pedes-
trian collisions (Column 8) is the product of Columns 5, 6, and 7. Since all vehicle-pedestrian crashes are assumed
to involve some level of injury, there are no property-damage-only crashes.
Worksheet SP11. Vehicle-Pedestrian Collisions for Urban and Suburban Roadway Segments
(1) I (2) I (3) (4) I (5) (6) (7) (8)
Predicted Predicted
I Predicted Nbrm• Predicted Nh.,., Nlm/wr ! Predicted N br 1
I
(7) from
I (9) from (9) from Worksheet from I Calibration I
Crash Severity Level Worksheet SPIC Worksheet SPlE SP1H (2)+(3)+(4) I Table 12-8 Factor, C, (5)*(6)*(7)
Total 4.967 1.182 0.734 6.883 o.olJ 1 1.oo o.os9
Worksheet SP1J-Vehicle-Bicycle Collisions for Urban and Suburban Roadway Segments

driveway-related predicted crashes from Worksheets SPlC, SPlE, and SPlH are entered into Columns 2, 3, and 4,
respectively. These values are summed in Column 5. Column 6 contains the bicycle crash adjustment factor
!(see Table 12-9). Column 7 represents the calibration factor. The predicted average crash frequency of vehicle-bicy-
cle collisions (Column 8) is the product of Columns 5, 6, and 7. Since all vehicle-bicycle collisions are assumed to
involve some level of injury, there are no property-damage-only crashes.
Worksheet SP1J. Vehicle-Bicycle Collisions for Urban and Suburban Roadway Segments
(1) (2) (3) (4) (5) (6) (7) (8)
Predicted Predicted
Nhrnn> Predicted Nb--. Nlmfw•• Predicted Nh, fb/kr Predicted Nb 1k..,.
(9) from (9) from (7) from
Worksheet Worksheet Worksheet I' from Calibration I
Crash Severity Level SPlC SP1E SPIH (2)+(3)+(4) , Table 12-9 1

Factor, C, (5)*(6)*(7)
I
Total 4.967 1.182 0.734 6.883 0.007 1.00 0.048
Fatal and injury 1.00 0.048
Worksheet SP1 K-Crash Severity Distribution for Urban and Suburban Roadway Segments
Worksheet SPlK provides a summary of all collision types by severity level. Values from Worksheets SP!C, SP!E,
SPlH, SPll, and SPIJ are presented and summed to provide the predicted average crash frequency for each severity
level as follows:

Worksheet SP1 K. Crash Severity Distribution for Urban and Suburban Roadway Segments
(!) I
(2) (3) (4)
Property Damage Only

I
Fatal and Injury (FI) (PDO) Total
I (3) from Worksheet I (6) from Worksheet
SP!D and SP!F; (7) I SP!D and SP!F; (7)
I from Worksheet SPlH; , (5) from Worksheet I from Worksheet SPlH;
and (8) from Worksheet I SP!D and SP!F; and (7) , and (8) from Worksheet
Collision Type I SPll and SP!J from Worksheet SPIH SPll and SP!J
'I
1
·I
MULTIPLE-VEHICLE
Rear-end collisions (from Worksheet SPlD) !.Oil 3.175 4.186
-1- - ---- .. -...... ,_,_-
0.041 0.075 0.116

----+----- -
collisions (from Worksheet SPlD) 0.083 0.075
SideS'.Vipe, same direction (from Worksheet SPlD) 0.001 0.294 0.295
Sideswipe, opposite direction (from Worksheet SPlD) 0.020 I 0.075 0.095
_Other multiple-vehicle collision (from Worksheet SP~D) ---ooo----~-----~-- 0.116
Subtotal 1.376 4.324 5.700

SINGLE-VEHICLE
Collision with pedestrian (from Worksheet SPli)
Collision with bicycle (from Worksheet SPlJ) 0.048 0.000 0.048

Subtotal 0.475 0.845 1.320
Total 1.851 5.169 7.020
Worksheet SP1 L-Summary Results for Urban and Suburban Roadway Segments
Worksheet SPlL presents a summary of the results. Using the roadway segment length and the AADT, the worksheet
presents the crash rate in miles per year (Column 4) and in million vehicle miles (Column 6).
Worksheet SP1 L Summary Results for Urban and Suburban Roadway Segments
(!) (2) (3) (4)
Predicted Average
Crash Frequency, Npro<ilcr..:~,. Crash Rate
(crashes/year) ( crashes/m.i/year)
Roadway Segment Length,
Crash Severity Level (Total) from Worksheet SPlK L (mi) (2)/(3)
The Highway
A four-lane divided urban arterial roadway segment.
The Question
The Facts
• 0.75-mi length
• 23,000 veh/day
• On-street parking not permitted

• 8 driveways (l major commercial, 4 minor commercial, 1 major residential, 1 minor residential,
I minor industrial/institutional)
• 20 roadside fixed objects per mile
• 12-ft offset to roadside fixed objects
• 40-ft median
• Lighting present
• 30-mph posted speed
Assumptions
Collision type distributions used are the default values presented in Tables 12-4 and 12-6 and Equations 12-19 and 12-20.
Results
in Sample Problem 2 is determined to be 3.4 crashes per year (rounded to one decimal place).
Steps
Step 1 through 8
For a four-lane divided urban arterial roadway segment, SPF values for multiple-vehicle nondriveway, single-vehicle,
multiple-vehicle driveway-related, vehicle-pedestrian, and vehicle-bicycle collisions are determined. The calculations
for total multiple-vehicle nondriveway, single-vehicle, and multiple-vehicle driveway-related collisions are presented
below. Detailed steps for calculating SPFs for fatal-and-injury (FI) and property-damage-only (PDO) crashes are
presented in Sample Problem I. The calculations for vehicle-pedestrian and vehicle-bicycle collisions are shown in
Step 10 since the CMF values are needed for these two models.

The SPF for multiple-vehicle nondriveway collisions for the roadway segment is calculated from Equation 12-10 and
Table 12-3 as follows:
N,,m" = exp(a + b x In(AADT) + In(L))
Nbrm>(loml) = exp(-12.34 + 1.36 X In(23,000) + In(0.75))


The SFP for single-vehicle crashes for the roadway segments is calculated from Equation 12-13 and Table 12-5 as
follows:
N,~, = exp(a + b x In(AADT) + In(L))
N,,,,~""'' = exp(-5.05 + 0.47 x In(23,000) + In(0.75))


The SPF for multiple-vehicle driveway-related collisions for the roadway segment is calculated from Equation 12-16
as follows:
Nbrdwy(total) =
· L nj X Nj X
(AADTJ(t)
---
all 15,000
driveway
types
The number of driveways within the roadway segment, n1, for Sample Problem 1 is one major commercial, four
minor commercial, one major residential, one minor residential, and one minor industriaVinstitutional.
The number of driveway-related collisions, N,I and the regression coefficient for AADT, t, for a four-lane divided
arterial, are provided in Table 12-7.
6 6 6
23, OOOJ(l.!0 ) [23, OOOJ(uo ) [ 23, OOOJ(uo )
Nbrdwy(total) =1x0.033x [ - - - +4xO.Ollx - - - +lx0.018x - - -
15,000 15,000 15,000
6 6
23 OOOJ(l.lo ) [23 OOOJ(l.!0 )
+lx0.003x - ' - - +lx0.005x - ' -
[ 15,000 15,000
The fatal-and-injury (FI) and property-damage-only (PDO) SPF values for multiple-vehicle nondriveway collisions,
single-vehicle crashes and multiple-vehicle driveway-related collisions can be determined by using the same
procedure presented in Sample Problem 1.
On-Street Parking (CMF1)

Since on-street parking is not permitted, CMF1, = 1.00 (i.e., the base condition for CMF1, is the absence of on-street parking).
Roadside Fixed Objects (CMF1)

CMF,, is calculated from Equation 12-33 as follows:
CMF2r =foffset xDfo Xpfo +(lO-p\
· jol
From Table 12-20, for a roadside fixed object with a 12-ft offset, the fixed-object offset factor, f,ff,"' is interpolated as
0.079.
From Table 12-21, for a four-lane divided arterial the proportion of total crashes, p1, = 0.036.
CMF,, = 0.079 X 20 X 0.036 + (1.0- 0.036)
= 1.02
Median Width (CMF,)

From Table 12-22, for a four-lane divided arterial with a 40-ft median, CMF 3, = 0.97.
Lighting (CMF4)
CMF 4, can be calculated from Equation 12-34 as follows:
CMF4r = 1.0-(pn~ x(l.0-0.72xp.mr -0.83xppn,.J')
For a four-lane divided arterial, p.mr = 0.364, ppnr = 0.636, and pnr = 0.410 (see Table 12-23).
CMF,, 1.0- (0.410 X (1.0- 0.72 X 0.364-0.83 X 0.636))
0.91
Automated Speed Enforcement (CMF,)

Since there is no automated speed enforcement in Sample Problem 2, CMF5, = 1.00 (i.e., the base condition for
CMF 5, is the absent of automated speed enforcement).
CMFcomb = 1.02 X 0.97 X 0.91

= 0.90

The predicted average crash frequency of an individual roadway segment (excluding vehicle-pedestrian and vehicle-
bicycle collisions) for SPF base conditions, Nb,' is calculated first in order to determine vehicle-pedestrian and
vehicle-bicycle crashes. Nb, is determined from Equation 12-3 as follows:
Nb, = N,pf~ X (CMFJ, X CMF,, X ••• X CMFJ
From Equation 12-4, N,Pf'·' can be calculated as follows:
= 2.804 + 0.539 + 0.165

The combined CMF value for Sample Problem 2 is 0.90.

Nb, = 3.508 X (0.90)
The SPF for vehicle-pedestrian collisions for the roadway segment is calculated from Equation 12-19 as follows:
Npedr = Nbr xfpedr
From Table 12-8, for a posted speed of 30 mph on four-lane divided arterials, the pedestrian crash adjustment factor
f'"' = 0.067.
Npedr = 3.157 X 0.067
From Table 12-9. for a posted speed of30 mph on four-lane divided arterials, the bicycle crash adjustment factor
fbi,., = 0.013.
N,,.,, = 3.157 X 0.013
Step 11-Multiply the result obtained in Step 10 by tbe appropriate calibration factor.
It is assumed in that a calibration factor, C,., of 1.00 has been determined for local conditions. See Part C, Appendix
A. I for further discussion on calibration of the predicted models.

J.00 X (3.157 + 0.2)2 + 0.04J)
= 3.410
WORKSHEETS
The step-by-step instructions above are provided to illustrate the predictive method for calculating the predicted average
crash frequency for a roadway segment. To apply the predictive method steps to multiple segments, a series of 12 work-
sheets are provided for determining the predicted average crash frequency. The 12 worksheets include:
• Worksheet SP2A (Corresponds to Worksheet I A)-General Information and Input Data for Urban and Suburban
Arterial Roadway Segments
• Worksheet SP2B (Corresponds to Worksheet I B)-Crash Modification Factors for Urban and Suburban Arterial
Roadway Segments
• Worksheet SP2C (Corresponds to Worksheet I C)-Multiple-Vehicle Nondriveway Collisions by Severity Level for
• Worksheet SP2D (Corresponds to Worksheet I D)-Multiple-Vehicle Nondriveway Collisions by Collision Type for
• Worksheet SP2E (Corresponds to Worksheet I E)-Single-Vehicle Crashes by Severity Level for Urban and
• Worksheet SP2F (Corresponds to Worksheet I F)-Single-Vehicle Crashes by Collision Type for Urban and
• Worksheet SP2G (Corresponds to Worksheet I G)-Multiple-Vehicle Driveway-Related Collisions by Driveway

Type for Urban and Suburban Arterial Roadway Segments
• Worksheet SP2H (Corresponds to Worksheet I H)-Multiple-Vehicle Driveway-Related Collisions by Severity
Level for Urban and Suburban Arterial Roadway Segments
• Worksheet SP21 (Corresponds to Worksheet 11)-Vehicle-Pedestrian Collisions for Urban and Suburban Arterial
Roadway Segments
• Worksheet SP2J (Corresponds to Worksheet J.J)-Vehicle-Bicycle Collisions for Urban and Suburban Arterial
Roadway Segments
• Worksheet SP2K (Corresponds to Worksheet JK)-Crash Severity Distribution for Urban and Suburban Arterial
Roadway Segments
• WOrksheet SP2L (Corresponds to WOrksheet JL)-Summary Results for Urban and Suburban Arterial Roadway Segments
Worksheet SP2A-Generallnformation and Input Data for Urban and Suburban Roadway Segments
Worksheet SP2A is a summary of general information about the roadway segment, analysis, input data (i.e., ''The Facts"),
and assumptions for Sample Problem 2
Worksheet SP2A. General Information and Input Data for Urban and Suburban Roadway Segments
Gi:neral Information Location Information
Analyst Roadway
Agency or Company , Roadway Section
Analysis Year
Length of segment, L (mi) 0.75

AADT (veh/day) 23,000
Type of on~street parking (none/paralleVangle) none None
Median width (ft) 15 40

Lighting (present/not present) present
Major commercial driveways (number)
Minor commercial driveways (number) I 4
---------------~---------=---------------- ------
Major industriaVinstitutional driveways (number) '
Minor industrial/institutional driveways (number)

Major residential driveways (number)
Minor residential driveways (number)
Other driveways (number)
Speed Category Low (30mph)
Roadside fixed object density (fixed objects/mi) not present 20
Offset to roadside fixed objects (ft) not 12

Calibration Factor, C, 1.0 1.0
Worksheet SP2B-Crash Modification Factors for Urban and Suburban Roadway Segments
Column 6 ofWorksheet SP2B which indicates the combined CMF value.
Worksheet SP2B. Crash Modification Factors for Urban and Suburban Roadway Segments
(1) (2) (3) (4) (5) (6)
CMFfor CMF for Roadside CMFfor CMF for Auto Speed
On-Street Parking Fixed Objects I
Median Width CMF for Lighting I Enforcement Combined CMF
CMF 1, CMF,,
- ·-
I
-- - I .~~Jr CMF4,
I CMFs, CMF
- .. c~.~~.~~ .. "
from Equation 12-32 from Equation 12-33 from Table 12-22 I from E~uation 12-34 from Section 12.7.1 (1 )'(2)'(3)'(4)'(5)
1.00 1.02 0.97 0.91 1.00 0.90
Worksheet SP2C-Multiple-Vehicle Nondriveway Collisions by Severity Level for Urban and

The SPF for multiple-vehicle nondriveway collisions along the roadway segment in Sample Problem 2 is calculated
using Equation 12-10 and entered into Column 4 ofWorksheet SP2C. The coefficients for the SPF and the overdis-
persion parameter associated with the SPF are entered into Columns 2 and 3; however, the overdispersion parameter
is not needed for Sample Problem 2 (as the EB Method is not utilized). Colunm 5 of the worksheet presents the
proportions for crash severity levels calculated from the results in Column 4. These proportions are used to adjust
the initial SPF values( from Column 4) to assure that fatal-and-injury (FI) and property-damage-only (PDO) crashes
sum to the total crashes as illustrated in Column 6. Column 7 represents the combined CMF (from Column 6 in
frequency of multiple-vehicle nondriveway crashes using the values in Co1unm 6, the combined CMF in Co1unm 7,
and the calibration factor in Column 8.
Worksheet SP2C. Multiple-Vehicle Nondriveway Collisions by Severity Level for Urban and Suburban Roadway Segments
(1) (2) (3) (4) (5) (6) (7) (8) (9)
1
SPF Overdispersion I Initial Adjusted Combined I Calibration Predicted

Coefficients Parameter, k Nbrmv Nb,.,., CMFs Factor 1
N brm•
!
from Table
Crash from (6) from
12-3
Severity from Table Equation Proportion of Worksheet
Level a b 12-3 12-10 Total Crashes
'
(4)(0101*(5) SP2B c, (6)'(7)*(8)
Total -12.34 1.36 1.32 2.804 1.000 2.804 0.90 1.00 2.524
Fatal and -12.76 1.28 ! 1.31 0.825 (4);/((4),+(4),00) 0.780 0.90 1.00 0.702
injury (FI)
0.278
Property -12.81 1.38 I
1.34 2.143 (5),o'III-(5)Ff 2.024 0.90 1.00 1.822
damage
I 0.722
only (PDO)
Worksheet SP2D-Multiple-Vehicle Nondriveway Collisions by Collision Type for Urban and

Using the default proportions, the predicted average crash frequency for multiple-vehicle nondriveway crashes by
collision type is presented in Columns 3 (Fatal and Injury, FI), 5 (Property Damage Only, PDO), and 6 (Total).
These proportions may be used to separate the predicted average crash frequency for multiple-vehicle nondriveway
crashes (from Column 9, Worksheet SP2C) into components by crash severity and collision type.
(1) (2) (3) (4) (5) (6)
Proportion of Predicted Nbrm•(FI) Proportion of Predicted Nhrmv(PDQ) I Predicted Nbrmv(t<uol)

Collision Type (Fl) I
(crashes/year) Collision Type (PDO) I
(crashes/year) 1 (crashes/year)
(9)n from (9) PDO from I (9)total from

Collision l)rpe from Table 12-4 I
Worksheet SP2C from Table 12-4 Worksheet SP2C I Worksheet SP2C
Total 1.000 0.702 1.000 1.822 2.524
(2)*(3)" (4)*(5)PDO (3)+(5)

Rear-end collision 0.662 1.206 1.790
Head-on collision 0.007 0.013 0.027
Angle collision 0.036 0.066 0.094
-···-··········· +···-········································ +··················· -
Sideswipe, same direction 0.050 0.035 0.223 0.406 0.441

Srdeswipe, opposite direction . . . .r··· 0.010 0.007 0.001 0.009
Other multi;l~-~~bi~·~~·"collision. -·· 0.048 0.034 0.071 0.163
Worksheet SP2E-Single-Vehicle Crashes by Severity Level for Urban and Suburban Roadway Segments
The SPF for single-vehicle crashes along the roadway segment in Sample Problem 2 is calculated using Equation
12-13 and entered into Column 4 of Worksheet SP2E. The coefficients for the SPF and the overdispersion param-
eter associated with the SPF are entered into Columns 2 and 3; however, the overdispersion parameter is not needed
for Sample Problem 2 (as the EB Method is not utilized). Column 5 of the worksheet presents the proportions for
crash severity levels calculated from the results in Column 4. These proportions are used to adjust the initial SPF
values (from Column 4) to assure that fatal-and-injury (FI) and property-damage-only (PDO) crashes sum to the
total crashes as illustrated in Column 6. Column 7 represents the combined CMF (from Column 6 in Worksheet
SP2B), and Column 8 represents the calibration factor. Column 9 calculates the predicted average crash frequency
of multiple-vehicle nondriveway crashes using the values in Column 6, the combined CMF in Column 7, and the
calibration factor in Column 8.
..L..
Worksheet SP2E, Single-Vehicle Collisions by Severity Level for Urban and Suburban Roadway Segments
(I) I (2) (3) (4) I (5) !
(6) (7) I (8) I (9)
I I I
I
SPF Overdispersion I Adjusted Combined I Calibration I Predicted
Coefficients I
Parameter, k I
Initial Nbr:<v N""' 1
C:MFs Factor ! N~m~.
from Table
Crash (6) from
12-5
Severity from Equation Proportion of Worksheet
Level a b from Table 12-5 12-13 I Total Crashes (4),..,*(5) SP2B c (6)'(7)'(8)
Total
I'
-5.05 I 0.47 I 0.86 0.539 1.000 0.539 0.90
I
-
I I 1.00 0.485
Fatal and (4V((4),+(4J,ool !
-8.71 0.66 • 0.28 0.094 0.094 ' 0.90 1.00 0.085

injury (FI) 1
0.174 i
'
Property (5)1oWI-(5)FI I
damage 1-5.04 0.45 1.06 0.446 0.445 0.90 1.00 D.401
only(PDO) 0.826
I I
Worksheet SP2F-Single-Vehicle Crashes by Collision Type for Urban and Suburban Roadway Segments
Worksheet SP2F presents the default proportions for collision type (from Table 12-5) by crash severity level as follows:
presented in 3 (Fatal and Injury, Fl), 5 (Property Damage Only, PDO), and Columns 6 (Total).
(from Column 9, Worksheet SP2E) into components by crash severity and collision type.
Worksheet SP2F. Single-Vehicle Collisions by Collision Type for Urban and Suburban Roadway Segments
(!) (2) (3) (4) (5) (6)
Proportion of Predicted N~Jm> (Ff) Proportion of 1 Predicted Nbr>v (PMJ Predicted N,...,.. (wtoiJ
Collision Type (F/) (crashes/year) Collision Type (PDOJ I (crashes/year) (crashes/year)
(9)F1 from (9)P00 from (9), 0 , 01 from
Collision Type from Table 12-6 Worksheet SP2E from Table 12-.6 Worksheet SP2E Worksheet SP2E
Total 1.000 0.085 1.000 D.401 0.485
(2)'(3)n (4)'(5),DD (3)+(5)
Collision with animal 0.001 0.000 0.063 0.025 : 0.025
--------------------·-··-·-··-·--·----·------------------·-··-··-······················-··.,···-··-···-··-··-··-··-··-·-··------ f----------·---· ·t-··-------------
Collision with fixed object 0.500 0.043 0.813 0.326 ; 0.369
Collision with other object 0.028 0.002 0.016 0.006 0.008
Other single~vehicle collision ~ .. 0.471 0.040 0.108 0.043 0.083
Worksheet SP2G-Multiple-Vehicle Driveway-Related Collisions by Driveway Type for Urban and

Worksheet SP2G determines and presents the number of multiple-vehicle driveway-relatedcollisions. The number of
driveways along both sides of the road is entered in Column 2 by driveway type (Column 1). The associated number
of crashes per driveway per year by driveway type as found in Table 12-7 is entered in Column 3. Column 4 contains
the regression coefficient for AADT also found in Table 12-7. The initial average crash frequency of multiple-vehicle
driveway-related crashes is calculated from Equation 12-16 and entered into Column 5. The overdispersion param-
eter from Table 12-7 is entered into Column 6; however, the overdispersion parameter is not needed for Sample Prob-
lem 2 (as the EB Method is not utilized).
Worksheet SP2G. Multiple-Vehicle Driveway-Related Collisions by Driveway Type for Urban and Suburban
Roadway Segments
(1) (2) (3) (4) (5) (6)

Year,N Adjustment, t Initial N brd>w Parameter, k
Number of from Table Equation 12-16

Driveway Type Driveways, n1 from Table 12-7 12-7 •/N/(AADT/15,000), from Table 12-7
Major commercial 0.033 1.106 0.053
Minor commercial 4 0.011 1.106 0.071
Major industrial/institutional 0 0.036 1.106 0.000
--~ ·--------------. - ···---·-·-----~----······--------·--·
Minor industriaVinstitutional I 0.005 1.106 0.008
Major residential o.ors I.I06 0.029

Minor residential 0.003 l.l06 0.005
Other 0 0.005 1.106 0.000
Total : 0.166 1.39
Worksheet SP2H-Multiple-Vehicle Driveway-Related Collisions by Severity Level for Urban and

The initial average crash frequency of multiple-vehicle driveway-related crashes from Column 5 of Worksheet SP2G
is entered in Column 2. This value is multiplied by the proportion of crashes by severity (Column 3) found in Table
12-7, and the adjusted value is entered into Column 4. Column 5 represents the combined CMF (from Column 6 in
frequency of multiple-vehicle driveway-related crashes using the values in Column 4, the combined CMF in Column
5, and the calibration factor in Column 6 .
•:L..
!"'"
Worksheet SP2H. Multiple-Vehicle Driveway-Related Collisions by Severity Level for Urbao aod Suburban
Roadway Segments
(!) (2) (3) (4) (5) (6) (7)
Proportion of Total Adjusted Combined

Initial N brd Crashes (f11...) Nlmiwv CMFs Predicted Nbn~,....
(6) from
(5),.,.1 from Worksheet Calibration
Crash Severity Level Worksheet SP2G from Table 12-7 (2),.,., *(3) SP2B Factor, C, (4)*(5)*(6)
Total 0.166 I 1.000 I
0.166 0.90 1.00 0.149
Fatal and inJury (FI) 0.284 0.047 0.90 1.00 0.042
Property damage only (PDQ) I 0.716 I
0.119 0.90 I 1.00 i
0.107
Worksheet SP21-Vehicle-Pedestrian Collisions for Urban and Suburban Roadway Segments

driveway-related predicted crashes from Worksheets SP2C, SP2E, and SP2H are entered into Columns 2, 3, and
4, respectively. These values are summed in Column 5. Column 6 contains the pedestrian crash adjustment factor
(see Table 12-8). Column 7 represents the calibration factor. The predicted average crash frequency of vehicle-
pedestrian collisions (Column 8) is the product of Columns 5, 6, aod 7. Since all vehicle-pedestrian crashes are
assumed to involve some level of injury, there are no property-damage-only crashes.
Worksheet SP21. Vehicle-Pedestriao Collisions

(1) (2) (3) (4) (5) (6) (7) (8)
Predicted Predicted Predicted
Predicted Nbrm• I Predicted N bm I Nlmiwy NM f tdr N
"'
I
Worksheet I Worksheet '
Worksheet from Table Calibration
I'
Crash Severity Level SP2C SP2E SP2H (2)+(3)+(4) 12-8 Factor, C, (5)*(6)*(7)
Total 2.524 0.485 0.149 3.158 0.067 1.000 0.212
Fatal and inJury (FI) 1.00 0.212
Worksheet SP2J-Vehicle-Bicycle Collisions for Urban and Suburban Roadway Segments

The predicted average crash frequency of multiple-vehicle nondriveway, single-vehicle, aod multiple-vehicle
driveway-related predicted crashes from Worksheets SP2C, SP2E, aod SP2H are entered into Columns 2, 3, aod 4,
respectively. These values are summed in Column 5. Column 6 contains the bicycle crash adjustment factor (see
Table 12-9). Column 7 represents the calibration factor. The predicted average crash frequency of vehicle-bicycle
collisions (Column 8) is the product of Columns 5, 6, aod 7. Since all vehicle-bicycle collisions are assumed to
involve some level of injury, there are no property-damage-only crashes.
Worksheet SP2J. Vehicle-Bicycle Collisions for Urbao aod Suburbao Roadway Segments
(1) (2) (3) I (4) (5) (6) (7) (8)
'
Predicted Predicted Predicted
Nb.,v 1 Predicted N"""" N,,.,. Predicted NM fblktr Nblk.-
Worksheet Worksheet Worksheet from Table Calibration
Crash Severity Level SP2C SP2E SP2H (2)+(3)+(4) 12-9 Factor, C, (5)*(6)*(7)
Total 2.524 0.485 0.149 3.158 0.013 1.00 0.041
Fatal and mjury 1.00 0.041

Worksheet SP2K-Crash Severity Distribution for Urban and Suburban Roadway Segments
Worksheet SP2K provides a sununary of all collision types by severity level. Values from Worksheets SP2C, SP2E,
SP2H, SP2I, and SP2J are presented and sununed to provide the predicted average crash frequency for each severity
level as follows:
Worksheet SP2K. Crash Severity Distribution for Urban and Suburban Roadway Segments
(I) (2) (3) (4)

Fatal and Injury (FI) (PDQ) Total
(3) from Worksheet (6) from Worksheet

SP2D and SP2F; (7) (5) from Worksheet SP2D and SP2F; (7)
from Worksheet SP2H; SP2D and SP2F; and from Worksheet SP2H;
and (8) from Worksheet (7) from Worksheet and (8) from Worksheet
Collision Type SP21 and SP2J SP2H SP21 and SP2J
MULTIPLE-VEHICLE
Rear-end collisions (from Worksheet SP2D) 0.584 1.206 1.790
Head-on collisions (from Worksheet SP2D) 0.014 0.013 0.027
Angle collisions (from Worksheet SP2D) 0.028 0.066 0.094
Sides\Vipe, same direction (from Worksheet SP2D) 0.035 0.406 0.441

··-------
Sideswipe, opposite direction (from Worksheet SP2D) 0.007 0.002 0.009
Driveway-related collisions (from Worksheet SP2H) 0.042 0.107 0.149

Other multiple~vehicle collision (from Worksheet SP2D) 0.034 0.129 0.163
Subtotal 0.744 1.929 2.673
SINGLE-VEHICLE
ColllSlon With anlmal (from Worksheet SP2F) 1 0.000 0.025 0.025
Colhston Wlth fixed objeC~ (fr~m-Workshee; sP2F) •t 0.043 0.326 0.369
j_ :~L . . ..
Colltston wtth other obJeCt (from Worksheet SP2F) 0.002 0.008
-- ---------------- -~--- - ---
Other smgle~vehicle colhston (from Worksheet SP2F) 1
0.040 0.083
Collision with pedestrian (from Worksheet SP21) 0.212 0.212
.
-c·;n·i-~i~;·~fu.bi~;~J·; (fr~~-w~;k~h~·;~·-s~;n . -..-·-·-..-..-.. -.. -.. .
~ ~r·_ . _. _. _. _. _. . -..-·
_·o·:;·~
0.000 0.041
Subtotal / 0.338 0.400 0.738

Total I 1.082 2.329 3.411
!""
Worksheet SP2L-Summary Results for Urban and Suburban Roadway Segments

Worksheet SP2L presents a summary of the results. Using the roadway segment length and the AADT, the worksheet
presents the crash rate in miles per year (Column 4) and in million vehicle miles (Column 6).
Worksheet SP2L. Summary Results for Urban and Suburban Roadway Segments
(I) (2) (3) (4)
Frequency, Npmlotod"' Crash Rate
(crashes/year) (crashes/milyear)
(Total) from Roadway Segment Length, L
Crash Severity Level Worksheet SP2K (mi) (2)/(3)
Total 3.411 0.75 4.5
Fatal and injury (FI) 1.082 0.75 1.4
Property damage only (PDQ) 2.329 0.75 3.1
The Site/Facility
A three-leg stop-controlled intersection located on an urban arterial.
The Question
What is the predicted crash frequency of the unsignalized intersection for a particular year?
The Facts
• 1 left-tum lane on one major road approach

• No right-tum lanes on any approach
• AADT of major road is 14,000 veh/day
• AADT of minor road is 4,000 veh/day
Assumptions
Collision type distributions used are the default values from Tables 12-11 and 12-13 and Equations 12-30 and 12-31.
Results
Using the predictive method steps as outlined below, the predicted average crash frequency for the unsignalized
intersection in Sample Problem 3 is determined to be 1.6 crashes per year (rounded to one decimal place).
Steps
Step 1 through 8
Step 9-For the selected site, determine and apply the appropriate safety performance fnnction (SPF) for the
site's facility type and traffic control featnres.
For a three-leg stop-controlled intersection, SPF values for multiple-vehicle, single-vehicle, vehicle-pedestrian, and
vehicle-bicycle collisions are determined. The calculations for vehicle-pedestrian and vehicle-bicycle collisions are
shown in Step 10 since the CMF values are needed for these two models.
Multiple-Vehicle Crashes
The SPF for multiple-vehicle collisions for a single three-leg stop-controlled intersection is calculated from Equation
12-21 and Table 12-10 as follows:
N"•' = exp(a + b x In(AADT=) + c x In(AADT.,,))

Nbim,(<owl) = exp(-13.63 + 1.11 X /n(l4,000) + 0.41 x /n(4,000))
Nbim'iFO = exp(-14.01 + 1.16 X /n(l4,000) + 0.30 x /n(4,000))

Nbim"PDO) = exp(-15.38 + 1.20 x /n(14,000) + 0.51 x /n(4,000))

These initial values for fatal-and-injury (FI) and property-damage-only (PDO) crashes are then adjusted using
Nbimv(FI) = Nbimv(tota1) [ N'

N~imv(FI)
N'
J
bimv(FI) + bimv(PDO)
= 1.892 X ( 0.639 )
0.639 + 1.358
Nbimv(PDO) = Nbimv(toml) - Nbimv(F[)
1.892 - 0.605
1.287 crashes/year

The SPF for single-vehicle crashes for a single three-leg stop-controlled intersection is calculated from Equation 12-24
and Table 12-12 as follows:
= exp(a + b x In(AADTma;) + c x In(AADTmm. ))
Nbisv(row1) = exp(-6.81 + 0.16 x /n(J4,000) + 0.51 x /n(4,000))

Nbi,'iPDO) = exp(-8.36 + 0.25 x /n(l4,000) + 0.55 x /n(4,000))
Since there are no models for fatal-and-injury crashes at a three-leg stop-controlled intersections ' N b1sv(F[)
. is
calculated using Equation 12-27 (in place of Equation 12-25), and the initial value for N br.s~(PDO)
. calculated above is
then adjusted using Equation 12-26 to assure that fatal-and-injury and property-damage-only crashes sum to the
value for total crashes as follows:
N.b1ov{FI) =Nbls~(toml) x<
J bisv
For a three-leg stop-controlled intersection, the default proportion of fatal-and-injury crashes, f61~= 0.31 (see Section
12.6.2, Single-Vehicle Crashes)
0.349 X 0.3]
0 .I 08 crashes/year
= 0.349-0.108
Intersection Left-Turn Lanes (CMF1)

From Table 12-24, for a three-leg stop-controlled intersection with one left-tum lane on the major road, CMF 11 = 0.67.
Intersection Left-Turn Signal Phasing (CMF2)

Forunsignalized intersections, CMF 21 = 1.00.
Intersection Right-Turn Lanes (CMF 3)

Since no right-turn lanes are present, CMF 31 is 1.00 (i.e., the base condition for CMF31 is the absent of right-tum
Right-Turn-on-Red (CMF4)
For unsignalized intersections, CMF41 = 1.00.
Lighting (CMF,)
Since there is no lighting at this intersection, CMF51 is 1.00 (i.e., the base condition for CMF51 is the absence of inter-
section lighting).
Red-Light Cameras (CMF6 )

For unsignalized intersections, CMF 61 is always 1.00.

The predicted average crash frequency of an intersection (excluding vehicle-pedestrian and vehicle-bicycle colli-
sions) for SPF base conditions, N 61 , must be calculated in order to determine vehicle-pedestrian and vehicle-bicycle
crashes. N 61 is determined from Equation 12-6 as follows:
N,, = N,pfl" X (CMFII X CMF21 X ••• X CMF,)
From Equation 12-7, N SP. ,nr can be calculated as follows:

1
= 1.892 + 0.349
N, ~ 2.241 X (0.67)
The SPF for vehicle-pedestrian collisions for a three-leg stop-controlled intersection is calculated from Equation 12-30
as follows:
From Table 12-16, for a three-leg stop-controlled intersection the pedestrian crash adjustment factor, fpodl ~ 0.211.
Npedi ~ 1.501 X 0.021
From Table 12-17, for a three-leg stop-controlled intersection, the bicycle crash adjustment factor fb;k,, ~ 0.016.
Nbikei ~ 1.501 X 0.016

It is assumed in Sample Problem 3 that a calibration factor, C,, of 1.00 has been determined for local conditions.
See Part C, Appendix A. I for further discussion on calibration of the predicted models.

The predicted average crash frequency is calculated using Equation 12-5 based on results obtained in Steps 9
Npredicted Jnr c, X (N,, + Npodl + N,lk.)

1.00 X (1.501 + 0.032 + 0.024)
1.557 crashes/year
WORKSHEETS
average crash frequency for an intersection. To apply the predictive method steps to multiple intersections, a series of
12 worksheets are provided for determining the predicted average crash frequency at intersections. The 12 work-
sheets include:
• Worksheet SP3A (Corresponds to Worksheet 2A)--General Inforination and Input Data for Urban and Suburban
Arterial Intersections
• Worksheet SP3B (Corresponds to Worksheet 2B)-Crash Modification Factors for Urban and Suburban Arterial
Intersections
• Worksheet SP3C (Corresponds to Worksheet 2C)-Multiple-Vehicle Collisions by Severity Level for Urban and
Suburban Arterial Intersections
• Worksheet SP3D (Corresponds to Worksheet 2D)-Multiple-Vehicle Collisions by Collision Type for Urban and
Suburban Arterial Intersections ·
lT
• Worksheet SP3E (Corresponds to Worksheet 2E)-Single-Vehicle Crashes by Severity Level for Urban and Subur-
ban Arterial Intersections
• Worksheet SP3F (Corresponds to Worksheet 2F)-Single-Vehicle Crashes by Collision Type for Urban and Subur-
ban Arterial Intersections
• Worksheet SP3G (Corresponds to Worksheet 2G)-Vehicle-Pedestrian Collisions for Urban and Suburban Arterial
• Worksheet SP3H (Corresponds to Worksheet 2H)-Crash Modification Factors for Vehicle-Pedestrian Collisions
for Urban and Suburban Arterial Signalized Intersections
• Worksheet SP31 (Corresponds to Worksheet 21)-Vehicle-Pedestrian Collisions for Urban and Suburban Arterial
• Worksheet SP3J (Corresponds to Worksheet 2..Q-Vehicle-Bicycle Collisions for Urban and Suburban Arterial
Intersections
• Worksheet SP3K (Corresponds to Worksheet 2K)-Crash Severity Distribution for Urban and Suburban Arterial
Intersections
• Worksheet SP3L (Corresponds to Worksheet 2L)-Summary Results for Urban and Suburban Arterial Intersections
Details of these sample problem worksheets are provided below, except for Worksheets SP3H and SP31 which are only
used for signalized intersections, Blank versions of the corresponding worksheets are provided in Appendix 12A
Worksheet SP3A-Generallnformation and Input Data for Urban and Suburban Arterial Intersections
Worksheet SP3A. General Information and Input Data for Urban and Suburban Arterial Intersections
General Information 1 Location Information
Analyst Roadway
Date Performed 1 Jurisdiction

Analysis Year
Intersection type (3ST, 3SG. 4ST, 4SG) 3ST
(veh/day) 14,000
4,000
(present/not present) not not present
1.00 1.00
. -·····················--
Data for unsignalized intersections only:
Number of major-road approaches with left-turn lanes (0, 1, 2) 0
Number of major-road approaches with right-tum lanes (0, 1, 2) 0 0
Data for signalized intersections only:
Number of approaches with left-turn lanes (0, 1, 2, 3, 4) 0 N/A
Number of approaches with right-turn lanes (0, 1, 2, 3, 4) 0 NIA
Number of approaches with left-rum signal phasing NIA
Number of approaches with right-turn-on-red prohibited 0 NIA
------------·- ····------------+·-·-------- "
Type ofleft-turn signal phasing NIA
Intersection red light cameras (present/not present) not present NIA
Swn of all pedestrian crossing volwnes (PedVol) N/A
.
--~~-~"i'~~. ~~~-=~-~~~~=-~ =:~~~-:~-~--~. ~=-~="~~-~--~~i~~~-~ . . . . . .-.. . . ,. . .-.. _,~
_,_,_, . . . .. NIA
Nwnber of bus stops within 300m (1,000 ft) of the intersection

_____ _____
- - -·----···---""""'_'_ --- .... -.--- - --·---· -·----- ....-..... - .. - - - -·-·---- ...- ............... -- - -
,_, """""""'" - ------~ 0 N/A
Schools within 300m (1,000 ft) of the intersection (present/not present) notpresent N/A
Number of alcohol sales establishments witllln 300m (1 ,000 ft) of the intersection 0 N/A
Worksheet SP3B-Crash Modification Factors for Urban and Suburban Arterial Intersections
Worksheet SP3B. Crash Modification Factors for Urban and Suburban Arterial Intersections
(1) (2) (3) (4) (5) (6) (7)
·~'
CMFfor CMFfor CMFfor
CMFfor ' Left-Turn Signal Right-Turn Right-Turn- CMFfor CMF for Red- I
Left-Turn Lanes I Phasing Lanes on-Red Lighting Light Cameras 1 Combined CMF
CMF,
from Table 12-24
I CMF2,
'fr~'~'"i~~i'~"'l ·;·~25-"'"'1""
- CMF, 1
......................
_,_, ,_,
from Table
- ......~-~.~.:!'.....,... _
from Equation 1- ·r;;3~~;;~~ -1 &~~c~~~~~ j_(;;;(;)~~;~:·(;;;-(;;;-
12-26 12-35 12-36 12-37
i
1
0.67 i
1.00 !.00 1.00 1.00 1.00 o.67 · ·. ·
Worksheet SP3C-Multiple-Vehicle Collisions by Severity Level for Urban and Suburban Arterial Intersections
The SPF for multiple-vehicle collisions at the intersection in Sample Problem 3 is calculated using Equation 12-22 and
entered into Column 4 of Worksheet SP3C. The coefficients for the SPF and the overdispersion parameter associated
with the SPF are entered into Columns 2 and 3; however, the overdispersion parameter is not needed for Sample Prob-
lem 3 (as the EB Method is not utilized). Column 5 of the worksheet presents the proportions for crash severity levels
calculated from the results in Column 4. These proportions are used to adjust the initial SPF values (from Column 4)
to assure that fatal-and-injury (FI) and property-damage-only (PDO) crashes sum to the total crashes as illustrated in
Column 6. Column 7 represents the combined CMF (from Column 7 in Worksheet SP3B), and Column 8 represents
the calibration factor. Column 9 calculates the predicted average crash frequency of multiple-vehicle crashes using the
values in Column 6, the combined CMF in Column 7, and the calibration factor in Column 8.
Worksheet SP3C. Multiple-Vehicle Collisions by Severity Level for Urban and Suburban Arterial Intersections
(1) (2) (3) (4)
SPF Coefficients Overdispersion Parameter, k I Initial N blnw

from Table 12-10
Crash Severity Level a b c from Table 12-10 from Equation 12-22
Total -13.36 1.11 0.41 0.80 1.892
1.16 0.69 0.639
1.20 0.77 1.358
Worksheet SP3C. Continued

(1) (5) (6) (7) (8) (9)
Adjusted Nblmv Combined CMFs Predicted N blm•

Proportion of (7)from Calibration
Crash Severity Level Total Crashes (4),.,.,*(5) Worksheet SP3B Factor, C, (6)*(7)*(8)
Total 1.000 1.892 0.67 1.00 1.268
(4 )F/((4)n+( 4l,Dol
0.320
(S),owi-(S)Fl
0.680
Worksheet SP3D-Multiple-Vehicle Collisions by Collision Type for Urban and

Using the default proportions, the predicted average crash frequency for multiple-vehicle crashes by collision type is
presented in Columns 3 (Fatal and Injury, FI), 5 (Property Damage Only, PDO), and 6 (Total).
These proportions may be used to separate the predicted average crash frequency for multiple-vehicle crashes
(from Column 9, Worksheet SP3C) into components by crash severity and collision type.
Worksheet SP3D. Multiple-Vehicle Collisions by Collision Type for Urban and Suburban Arterial Intersections
(1) (2) (3) (4) (5) (6)
Predicted
Proportion of Predicted Nblm•<FIJ Proportion of I Nblm•(PDO) Predicted Nbjmo(«><ol)
Collision Type IFll (crashes/year) ; Collision Type (PDOI 1
(crashes/year) (crashes/year)
(9)Poo from
(9)F1 from Worksheet (9)PDO from
Collision Type from Table 12-11 Worksheet SP3C from Table 12-11 SP3C Worksheet SP3C
Total 1.000 0.405 1.000 0.862 1.268
1
(2)'(3), (~)'(5),= (3)+(5)
Rear-end collision 0.421 0.171 0.440 0.379 0.550

Head-on collision 0.045 0.018 0.023 0.020 0.038
---------- . i~ -----
!
Angle collision 0.343 0.139 0.262 0.226 0.365
Sideswipe 0.126 0.051 0.040 0.034 0.085
- --- '-""" " ,,,_,,_ - I
Other multiple-vehicle collision i 0.065 0.026 I 0.235 0.203 0.229
Worksheet SP3E-Single-Vehicle Crashes by Severity Level for Urban and Suburban

The SPF for single-vehicle crashes at the intersection in Sample Problem 3 is calculated using Equation 12-25 for
total and property-damage-only (PDO) crashes and entered into Column 4 ofWorksheet SP3E. The coefficients for
the SPF and the overdispersion parameter associated with the SPF are entered into Columns 2 and 3; however, the
overdispersion parameter is not needed for Sample Problem 3 (as the EB Method is not utilized). Since there are no
models for fatal-and-injury crashes at a three-leg stop-controlled intersections, N,o><Fn is calculated using Equation 12-27
(in place of Equation 12-25), and the value is entered into Column 4 and 6 since no further adjustment is required.
Column 5 of the worksheet presents the proportions for crash severity levels calculated from the results in Column 4.
These proportions are used to adjust the initial SPF values (from Column 4) to assure that fatal-and-injury (FI) and
property-damage-only (PDO) crashes sum to the total crashes as illustrated in Column 6. Column 7 represents the
combined CMF (from Column 7 in Worksheet SP3B), and Column 8 represents the calibration factor. Column 9 cal-
culates the predicted average crash frequency of single-vehicle crashes using the values in Column 6, the combined
1.
Worksheet SP3E. Single-Vehicle Collisions by Severity Level for Urban and Suburban Arterial Intersections
(!) (2) (3) (4)
SPF Coefficients Overdispersion Parameter, k

from Table 12~ 12
from Equation 12-25; (Fl)
Crash Severity Level a b c from Table 12-12 from Equation 12-25 or 12-27
Total -6.8! I 0.!6 0.51 I.l4 0.349
Fatal and mjury (FI) I, N/A N/A N/A NIA 0.108
. -·-···-···-· .. ---·- .. -----··- --. ---·-- . ---f
Property damage only (PDO) 1
-8.36 I -Ci:25-T 0.55 1.29 0.244
Worksheet SP 3£. Continued

(I) (5) (6) (7) (8) (9)
Adjusted Nw... Combined CMFs Predicted Nb 1..,.

Proportion of (7) from Calibration
Crash Severity Level Total Crashes (4),...,*(5) Worksheet SP3B Factor, C, (6)*(7)*(8)
Total !.000 0.349 0.67 1.00 0.234
(4M((4),_+(4),DD)
Fatal and injury (FI) 0.!08 0.67 0.072
- -----··=r·--------
I !.00 I
NIA
1 - - - ! . 0 0____ --=~
(S),otoi-(S)F/
1_ _ _ _
Property damage only (PDO) I ······-····· ---·-·-···· 0.242 0.67

0.693
Worksheet SP3F-Single-Vehicle Crashes by Collision Type for Urban and Suburban Arterial Intersections
Worksheet SP3F presents the default proportions for collision type (from Table 12-13) by crash severity level as follows:
presented in Columns 3 (Fatal and Injury, FI), 5 (Property Damage Only, PDO), and 6 (total).
(from Column 9, Worksheet SP3E) into components by crash severity and collision type.
Worksheet SP3F. Single-Vehicle Collisions by Collision Type for Urban and Suburban Arterial Intersections
I
(I) I (2) (3) (4) I
(5) i (6)
Proportion of Predicted
Proportion of Predicted NbJ.,.<FI) I Collision Type Predicted Nbi••<PDOJ Nbi>V(IOtll)
Collision Type (FI) (crashes/year) (PDO) (crashes/year) (crashes/year)
I (9)PDO from
(9)F1 from (9)PDO from Worksheet
Collision Type Table 12~13 Worksheet SP3E Table 12-13 Worksheet SP3E SPJE
Total 1.000 0.072 1.000 ! 0.162 0.234
I (2)'(3)" (4)'(5),00 (3)+(5)
Collision with parked vehicle 0.000 0.003 0.000 .. _,_,,_,,_

0.001 _ ... ... _,_,_,_,_,_,_,_,_
,, , - ............... ...0.000
, _,_,_,_, _,_,, _
Collision with animal 0.003 0.000 O.Q18 0.003 0.003
Collision with fixed object 0.762 0.055 0.834 0.135 0.190
Collision with other object 0.090 0.006 0.092 O.Q15 0.021

--~~~--------
Other single-vehicle collision 0.039 0.003

_.. .....
, ..... -" , ___ -
--- __ 0.023
,_, - ...... ........ 1- .. -
0.004
----- '
0.007
Single-vehicle noncollision 0.105 0.008 0.030 0.005 0.013
Worksheet SP3G-Vehicle-Pedestrian Collisions for Urban and Suburban Arterial

The predicted average crash frequency of multiple-vehicle predicted crashes and single-vehicle predicted crashes
from Worksheets SP3C and SP3E are entered into Columns 2 and 3 respectively. These values are summed in
Column 4. Column 5 contains the pedestrian crash adjustment factor (see Table 12-16). Column 6 presents the
calibration factor. The predicted average crash frequency of vehicle-pedestrian collision (Column 7) is the product
of Columns 4, 5, and 6. Since all vehicle-pedestrian crashes are assumed to involve some level of injury, there are no
Worksheet SP3G. Vehicle-Pedestrian Collisions for Urban and Suburban Arterial Stop-Controlled Intersections
(I) I (2) I (3) I (4) (5) (6) (7)
I :
Predicted Nhim• Predicted Nbl•• Predicted N bJ I fptdl Predicted~
(9) from (9) from from Table Calibration

Crash Severity Level i Worksheet SP3C Worksheet SP3E (2)+(3) I
12-16 Factor, C, (4)*(5)*(6)
Total 1.268 0.234 :
1.502 0.021 I 1.00 0.032
I
Fatal and injury (FI) I - - I - - 1.00 0.032
.I
Worksheet SP3J-Vehicle-Bicycle Collisions for Urban and Suburban Arterial Intersections

from Worksheets SP3C and SP3E are entered into Cohnnns 2 and 3 respectively. These values are summed in Col-
umn 4. Column 5 contains the bicycle crash adjustment factor (see Table 12-17). Column 6 presents the calibration
factor. The predicted average crash frequency of vehicle-bicycle collision (Column 7) is the product of Columns 4, 5,
and 6. Since all vehicle-bicycle crashes are assumed to involve some level of injury, there are no property-damage-
only crashes.
Worksheet SP3J. Vehicle-Bicycle Collisions for Urban and Suburban Arterial Intersections
(1) (2) (3) (4) (5) ! (6) (7)
Predicted N bim• Predicted Nh/_'" Predicted Nhl fblk<l Predicted Ne•dl

(9) from (9) from from Table Calibration
Crash Severity Level Worksheet SP3C Worksheet SP3E (2)+(3) 12-17 Factor, C1 (4)*(5)*(6)
Total 1.268 0.234 1.502 0.016 1.000 0.024
Fatal and injury (FI) !.000 0.024
Worksheet SP3K-Crash Severity Distribution for Urban and Suburban Arterial Intersections
Worksheet SP3K provides a summary of all collision types by severity level. Values from Worksheets SP3D, SP3F, SP3G,
and SP3J are presented and summed to provide the predicted average crash frequency for each severity level as follows:

Worksheet SP3K. Crash Severity Distribution for Urban and Suburban Arterial Intersections
(1) (2) (3) (4)
Property Damage
Fatal and Injury (FI) Only (PDO) Total
I (3) from Worksheet SP3D (6) from Worksheet SP3D
and SP3F; (7) from SP3G (5) from Worksheet and SP3F; (7) from SP3G
Collision Type I
or SP3I and SP3J SP3D and SP3F or SP31 and SP3J
MULTIPLE-VEHICLE COLLISIONS
Rear-end collisions (from Worksheet SP3D) 0.171 0.379 0.550
·····-·····1······································ - ································-I· --
Head-on collisions (from Worksheet SP3D) 0.018 I0.020
---------.---·I·-- .. ----------
0.038
Angle collisions (from Worksheet SP3D) 0.139 0.226 0.365
Sideswipe (from Worksheet SP3D) 0.051 0.034 0.085
Other multiple-vehicle collision (from Worksheet SP3D) 0.026 0.203 0.229
Subtotal 0.405 I
0.862 1.267
SINGLE-VEHICLE COLLISIONS
Collision with parked vehicle (from Worksheet SP3F) 0.000 0.000 0.000
Collision with animal (from Worksheet SP3F) 0.000 0.003 0.003
Collision \\lith fixed object (from Worksheet SP3F) 0.055 0.135 0.190
. . .Collision
. . . . . . . _. _ with
____other
....c...object
:.:.:o:_ :(from : . :. ::.::.:. . :.:..:.:.:.cc._____ - - +- ___ -- ---------
Worksheet SP3F)
. c. :. .:.......... 0.006
-- -·1- - .. - - .0.015
- .. - - . ------ --- - 0.021
Other single~vehide collision (from Worksheet SP3F) 0.003 0.004 0.007
.. -.-.-.
Single-vehicle noncollision (from Worksheet SP3F) 0.008 0.005 0.013

Collision with pedestrian (from Worksheet SP3G or SP31)
________ c __ r ________ 0.032 0.000 0.032
Collision with bicycle (from Worksheet SP3J) 0.024 0.000 0.024
Subtotal 0.128 0.162 0.290
Total 0.533 1.024 1.557
Worksheet SP3L-Summary Results for Urban and Suburban Arterial Intersections

Worksheet SP3L presents a summary of the results.
Worksheet SP3L. Summary Results for Urban and Suburban Arterial Intersections
(I) (2)
Predicted Average Crash Frequency, Np,dlc~«~ lnt (crashes/year)

Crash Severity Level (Total) from Worksheet SP3K
Total 1.557
Fatal and injury (FI) 0.533
Property damage only (PDO) I
1.024
L. .I
The Intersection
A four-leg signalized intersection located on an urban arterial.
The Question
What is the predicted crash frequency of the signalized intersection for a particular year?
The Facts
• I left-tum Jane on each of the two major road approaches
• I right-tum Jane on each ofthe two major road approaches
• Protected/permissive left-tum signal phasing on major road
• AADT of major road is 15,000 veh/day
• AADT of minor road is 9,000 veh/day
• Lighting is present
• No approaches with prohibited right-tum-on-red

• Four-lane divided major road
• Two-lane undivided minor road
• Pedestrian volume is 1,500 peds/day
• The number of bus stops within 1,000 ft of intersection is 2
• A school is present within 1,000 ft of intersection
• The number of alcohol establishments within 1,000 ft of intersection is 6
Assumptions
Collision type distributions used are the default values from Tables 12-11 and 12-13 and Equations 12-28 and 12-31.
The maximum number of lanes crossed by a pedestrian is assumed to be four (crossing two through lanes, one left-
turn lane, and one right-tum lane across one side of the divided major road).
Results
Using the predictive method steps as outlined below, the predicted average crash frequency for the unsignalized
intersection in Sample Problem 4 is determined to be 3.4 crashes per year (rounded to one decimal place).
Steps
Step 1 through 8
through II are conducted. No other steps are necessary because only one roadway segment is analyzed for one year
For a four-leg signalized intersection, SPF values for multiple-vehicle, single-vehicle, vehicle-pedestrian, and
vehicle-bicycle collisions are determined. The calculations for total multiple- and single-vehicle collisions are pre-
sented below. Detailed steps for calculating SPFs for fatal-and-injury (FI) and property-damage-only (PDO) crashes
are presented in Sample Problem 3 (for fatal-and-injury base crashes at a four-leg signalized intersection, Equation
12-25 in place of Equation 12-27 is used). The calculations for vehicle-pedestrian and vehicle-bicycle collisions are
shown in Step 10 since the CMF values are needed for these two models.
Multiple- Vehicle Collisions

The SPF for multiple-vehicle collisions for a single four-leg signalized intersection is calculated from Equation 12-21
and Table 12-10 as follows: ·
Nb,m> = exp(a + b X In(AADTm") + C X In(AADTm1,))

Nblmo;(.ocl) = exp(-10.99 + 1.07 x In(l5,000) + c x In(9,000))

The SPF for single-vehicle crashes for a single four-leg signalized intersection is calculated from Equation 12-24
and Table 12-12 as follows:
Nbl,. = exp(a + b x In(AADTm") + c x In(AADTm 1,))

Nbi"''""'' = exp(-10.21 + 0.68 x In(\5,000) + 0.27 x In(9,000))
Step 10--Multiply the result obtained in Step 9 by the appropriate CMFs to adjust base conditions to site
Each CMF used in the calculation of the predicted average crash frequency of the intersection is calculated below.
CMF 11 through CMF21 are applied to multiple-vehicle collisions and single-vehicle crashes, while CMF 1, through
CMF 3, are applied to vehicle-pedestrian collisions.
Intersection Left-Turn Lanes (CMF)

From Table 12-24, for a four-leg signalized intersection with one left-tum lane on each of two approaches, CMF 11 = 0.81.
Intersection Left-Turn Signal Phasing (CMF,)

From Table 12-25, for a four-leg signalized intersection with protected/permissive left-tum signal phasing for two
approaches, CMF21 = 0.98 (0.99*0.99).

From Table 12-26, for a four-leg signalized intersection with one right-tum lane on each of two approaches, CMF 31 = 0.92.
Right-Turn-on-Red (CMF)
Since right-tum-on-red (RTOR) is not prohibited on any of the intersection legs, CMF 41 = 1.00 (i.e., the base condi-
tion for CMF41 is permitting a RTOR at all approaches to a signalized intersection).
Lighting (CMF,)
CMF 51 is calculated from Equation 12-36.
CMF51 =I- 0.38 x P,,
From Table 12-27, the proportion of crashes that occur at night, P,, = 0.235.
CMF51 = 1- 0.38 X 0.235
= 0.91
Red-Light Cameras (CMF6)

Since no red light cameras are present at this intersection, CMF" = 1.00 (Le., the base condition for CMF61 is the
absence of red hght cameras).
The combined CMF value applied to multiple- and single-vehicle crashes in Sample Problem 4 is calculated below.
CMFcomb = 0.81 X 0.98 X 0.92 X 0.91
= 0.66
Bus Stop (CMF1J

From Table 12-28, for two bus stops within 1,000 ft of the center of the intersection, CMF 1P = 2.78.
Schools (CMF2J
From Table 12-29, for one school within 1,000 ft of the center of the intersection, CMF,-P = 1.35.
Alcohol Sales Establishments (CMF,J

From Table 12-30, for six alcohol establishments within 1,000 ft of the center of the intersection, CMF 3P = 1.12.

The SPF for vehicle-pedestrian collisions for a four-leg signalized intersection is calculated from Equation 12-28
as follows:
Npedi = ~edbase X CMFlp X CMF2p X CMFJp
Npedbase is calculated from Equation 12-29 using the coefficients from Table 12-14.
Npedbase = exp(a + b xin(AADTj0 taJ) + c xln( AADTmin + d xin(PedVol) +ex nzane.xJ

AADTmaJ
J
9 000
= exp(-9.53 + 0.40x In(24, 000) + 0.26x In( •
15,000
J
+ 0.45 x In(1,500) + 0.04x 4J
The CMF vehicle-pedestrian collision values calculated above are CMF 1p =2. 78, CMF,•P = 1.35, and CMF 3p = 1.12.
Npedl = 0.113 X 2.78 X 1.35 X 1.12

The predicted average crash frequency of an intersection (excluding vehicle-pedestrian and vehicle-bicycle collisions)
for SPF base conditions, N 61 , must be calculated in order to determine vehicle-bicycle crashes. N 61 is determined from
Equation 12-6 as follows:
N 61 = N,pflm X ( CMFJ; X CMF21 X ... X CMF6)
From Equation 12-7, N,pflm can be calculated as follows:

N sp[1nt
. = N b/mv +Nblsv
= 4.027 + 0.297
N,; = 4.324 X (0.66)

CHAPTER 12- PREDICTIVE METHOD FOR URBAN AND SUBURBAN ARTERIALS 12-B9
From Table 12-17, for a four-leg signalized intersection the bicycle crash adjustment factor, f""'' = O.Ql5.
N""'' = 2.854 X 0.015
It is assumed in Sample Problem 4 that a calibration factor, C1, of 1.00 has been determined for local conditions.
See Part C, Appendix A.! for further discussion on calibration of the predicted models.

The predicted average crash frequency is calculated from Equation 12-5 based on the results obtained in Steps 9
Npredictcd int c, X (N,, + Np,d/ + Nbl.,,)

1.00 X (2.854 + 0.475 + 0.043)
WORKSHEETS
average crash frequency for an intersection. To apply the predictive method steps to multiple intersections, a series of
12 worksheets are provided for determining the predicted average crash frequency at intersections. The 12 work-
sheets include:
• Worksheet SP4A (Corresponds to Worksheet 2A)---General Information and Input Data for Urban and Suburban
• ffbrksheet SP4B (Corresponds to ffbrksheet 2B)--Crash Modification Factors for Urban and Suburban Arterial Iotersections
• Worksheet SP4C (Corresponds to Worksheet 2C)-Multiple-Vehicle Collisions by Severity Level for Urban and
• Worksheet SP4D (Corresponds to Worksheet 2D)-Multiple-Vehicle Collisions by Collision Type for Urban and
• Worksheet SP4E (Corresponds to Worksheet 2£)-Single-Vehicle Crashes by Severity Level for Urban and
• Worksheet SP4F (Corresponds to Worksheet 2F)-Sing1e-Vehicle Crashes by Collision Type for Urban and
Suburban Arterial Iotersections
• Worksheet SP4G (Corresponds to Worksheet 2G)-Vehicle-Pedestrian Collisions for Urban and Suburban Arterial
• Worksheet SP4H (Corresponds to Worksheet 2H)-Crash Modification Factors for Vehicle-Pedestrian Collisions
for Urban and Suburban Arterial Signalized Intersections
• Worksheet SP4! (Corresponds to Worksheet 21)-Vehicle-Pedestrian Collisions for Urban and Suburban Arterial
• ffbrksheet SP4J (Corresponds to ffbrksheet 2.1)--Vehicle-Bicycle Collisions for Urban and Suburban Arterial Intersections
• ffbrksheet SP4K (Corresponds to ffbrksheet 2K)-Crash Severity Distribution for Urban and Suburban Arterial Iotersections
• Worksheet SP4L (Corresponds to Worksheet 2L)-Summary Results for Urban and Suburban Arterial Intersections
Details of these sample problem worksheets are provided below, except for Worksheet SP4G which is only used for
stop-controlled intersections. Blank versions of the corresponding worksheets are provided in Appendix 12A.
Worksheet SP4A-Generallnformation and Input Data for Urban and Suburban Arterial Intersections
Worksheet SP4A. General Information and Input Data for Urban and Suburban Anerial Intersections
General Infonnation Location Information
Analyst
Agency or Company
.. "'"" " "-- -·- _, __ , _ - - - - - - - ---- ---- -- """"
Date Performed
Input Data I Base Conditions I Site Conditions

Intersection type (3ST, 3SG, 4ST, 4SG) 4SG
AADT.,1 (vehlday) . 15.000
AADT mm (veh/day) 9,000
Intersection lighting (present/not present) not present present

Calibration factor, LOO 1.00
Number of major-road approaches with left-tum lanes (0, 1. 2) 0 N/A
Number ofmajoHoad approaches with right-tum lanes (0, 1, 2) 0 N/A
Number of approaches with left-tum lanes (0. 1. 2, 3. 4) 0 2

Number of approaches with nght-tum lanes (0. L 2. 3. 4) 0 2
Number of approaches with left-tum signal phasing 2
Number of approaches with right.-..turn

................·. .o. . .". . .-.. '. . e·· · d· · 0p· '· ·o· · ·h· · ·i· · b· · ·i·'· · e·............................................................................................................................
·d o. . . . . . . . . . . . . . . . . . .
0......................................................... ., ..................................
Type of left-tum signal phasing permissive protected/permissive
Intersection red-light cameras (present/not present) not present not present
Sum of all pedestrian crossing volumes (PedVol) 1.500
Maximum n~~e! ?!}~_e_s _c:~s_s:~ :b<y_a_0p..e.:.d.:..e:.':.tri. :.:'.an:.::...:.:.:'"'""················ 4

............................ _ ·- -· _ .......... _ . . . . . . _
Number of bus stops within 300m (1,000 ft) of the intersection 0 2
Schools within 300m (1.000 ft) of the intersection (present/not present) not present present
Number of alcohol sales establishments witllln 300m (1,000 ft) of the intersection 0 6
Worksheet SP4B-Crash Modification Factors for Urban and Suburban Arterial Intersections
Worksheet SP4B. Crash Modification Factors for Urban and Suburban Arterial Intersections
(1) (2) (3) (4) (5) (6) (7)
CMFfor CMFfor CMFfor CMFfor

CMF for Left-Turn Right-Turn Right-Tum- CMFfor Red-Light
Left-Turn Lanes Signal Phasing I Lanes on-Red Lighting Cameras Combined CMF
Worksheet SP4C-Multiple-Vehicle Collisions by Severity Level for Urban and Suburban

The SPF for multiple-vehicle collisions at the intersection in Sample Problem 4 is calculated using Equation 12-22 and
entered into Column 4 of Worksheet SP4C. The coefficients for the SPF and the overdispersion parameter associated
with the SPF are entered into Columns 2 and 3; however, the overdispersion parameter is not needed for Sample
Problem 4 (as the EB Method is not utilized). Column 5 of the worksheet presents the proportions for crash sever-
ity levels calculated from the results in Column 4. These proportions are used to adjust the initial SPF values (from
Column 4) to assure that fatal-and-injury (FI) and property-damage-only (PDO) crashes sum to the total crashes as
illustrated in Column 6. Column 7 represents the combined CMF (from Column 7 in Worksheet SP4B), and Col-
umn 8 represents the calibration factor. Column 9 calculates the predicted average crash frequency of multiple-vehicle
crashes using the values in Column 6, the combined CMF in Column 7, and the calibration factor in Column 8.
Worksheet SP4C. Multiple-Vehicle Collisions by Severity Level for Urban and Suburban Arterial Intersections
(1) (2) (3) (4)
SPF Coefficients Overdispersion Parameter, k Initial Nblmv
from Table 12-10

Total -10.99 1.07 0.23 0.39 4.027
Fatal and injury (FI) 1.18 I 0.22 I

0.33 1.233
Property damage only (PDO) 1.02 0.24 1 0.44 2.647
Worksheet SP4C. Continued

(1) (5) (6) (7) (8) (9)
Adjusted Nbimv Combined CMFs Predicted NbJ,,_

Crash Severity Level Total Crashes (4),.,.,*(5) Worksheet SP4B Factor, C1 (6)'(7)*(8)
Total I 1.000 4.027 0.66 1.00 2.658
I
(4) ,/((4)n+( 4),DO)
0.318
(5),.,,-(5),
0.682
Worksheet SP4D-Multiple-Vehicle Collisions by Collision Type for Urban and Suburban

Using the default proportions, the predicted average crash frequency for multiple-vehicle crashes by collision type is
presented in Columns 3 (Fatal and Injury, FI), 5 (Property Damage Only, PDO), and 6 (Total).
These proportions may be used to separate the predicted average crash frequency for multiple-vehicle crashes
(from Column 9, Worksheet SP4C) into components by crash severity and collision type.
Worksheet SP4D. Multiple-Vehicle Collisions by Collision Type for Urban and Suburban Arterial Intersections
(1) (2) (3) (4) (5) (6)
Proportion of Predicted Nbl.. v(FI) Proportion of Predicted Nblmv (PDOJ Predicted Nb 1,.. (weal)
Collision Type IFf) (crashes/year) Collision Type IPDUI (crashes/year) (crashes/year)
(9}F1 from (9)PDO from (9)PDO from
Collision Type from Table 12-11 Worksheet SP4C from Table 12-11 Worksheet SP4C Worksheet SP4C
Total 1.000 0.845 1.000 1.812 2.658
(2)'(3)" (4 )"(S)PDO (3)+(5)

'
Rear-end collision
Head-on collision
t 0.450
0.049
-'"" . . ...... _,,_,,_---
0.380
_, ... .. - _,_
0.041 ··I _,,_,_
0.483
0.030
0.875
0.054
1.255
0.095
_,, ........ ,_,,_,,_,_
Angle collision 0.347 0.293 0.244 0.442 0.735
Sideswipe 0.099 0.084 0.032 0.058 0.142
"""
Other multlple-vehlcle
0.055 0.046 0.211 0.382 0.428
collision
Worksheet SP4E-Single-Vehicle Crashes by Severity Level for Urban and Suburban Arterial Intersections
The SPF for single-vehicle crashes at the intersection in Sample Problem 4 is calculated using Equation 12-25 for
total and property-damage-only (PDO) crashes and entered into Column 4 of Worksheet SP4E. The coefficients for
the SPF and the overdispersion parameter associated with the SPF are entered into Columns 2, and 3; however, the
worksheet presents the proportions for crash severity levels calculated from the results in Column 4. These propor-
tions are used to adjust the initial SPF values (from Column 4) to assure that fatal-and-injury (FI) and property-
damage-only (PDO) crashes sum to the total crashes as illustrated in Column 6. Column 7 represents the combined
CMF (from Column 7 in Worksheet SP4B), and Column 8 represents the calibration factor. Column 9 calculates the
predicted average crash frequency of single-vehicle crashes using the values in Column 6, the combined CMF in
Worksheet SP4E. Single-Vehicle Collisions by Severily Level for Urban and Suburban Arterial Intersections
(1) (2) (3) (4)
SPF Coefficients Overdispersion Parameter, k Initial Nbl"'

from Table 12-12
from Equation 12~25; (FI)
Crash Severity Level a I b c from Table 12~12 from Equation 12-25 or 12-27
Total -10.21 0.68 1 0.27 0.36 0.297
Worksheet SP4E. Continued

(1) (5) (6) (7) (8) (9)
Adjusted Nbl.•• Combined CMFs Predicted N b~>v

Crash Severity Level Total Crashes (4),..,'(5) Worksheet SP4B Factor, C1 (6)'(7)*(8)
Total 1.000 0.297 0.66 1.000 0.196
Fatal and injury (FI) (4),/(( 4)"+(4),00)

0.085 0.66 1.000 0.056
0.287
I (5),=,-(5)"
I
0.713
Worksheet SP4F-Single-Vehicle Crashes by Collision Type for Urban and Suburban Arterial Intersections
Worksheet SP4F presents the default proportions for collision 1ype (from Table 12-13) by crash severily level as follows:
• Properly-damage-only crashes (Column 4)
Using the default proportions, the predicted average crash frequency for single-vehicle crashes by collision lype is
presented in Columns 3 (Fatal and Injury, FI), 5 (Properly Damage Only, PDO), and 6 (Total).
(from Column 9, Worksheet SP4E) into components by crash severily and collision lype.
Worksheet_SP4E Single-Vehicle Collisions by Collision Type for Urban and Suburban Arterial Intersections
(I) (2) (3) (4) (5) (6)
Proportion of Predicted NbJ.w(Ff) Proportion of Predicted Nhl., (PDQ) IPredicted N61_,., (<oU.IJ
Collision Type IFII (crashes/year) Collision Type (PDQ) 1
(crashes/year) 1 (crashes/year)
(9)n from (9)PDO from (9)Poo from
Collision Type Table 12-13 Worksheet SP4E Table 12-13 ,I Worksheet SP4E Worksheet SP4E
Total 1.000 0.056 0.140 0.196
I 1.000 I
(2)"'(3)Ff (4)'(5)PDO (3)+(5)
I j
0.000
--- --0000 ___ ·r:·-- o.oo1
-000;- ---- ------0000 -
0.000
---~-
1
0.000
--------
0.000
Collision with fixed object 0.744 0.042 0.122 0.164
:~:·:~;~:.:~~:~~;:'ion l____ ~~:~--

Single-vehicle noncollision 0.141
l---oo_:_oo_oo2s__ -+---~o~.-~0:3340 - __ 11___o o_:_~o o_~ s- _____J___ooo .•~Ool :,~- -
Worksheet SP4H-Crash Modification Factors for Vehicle-Pedestrian Collisions for Urban and
Suburban Arterial Signalized Intersections
geometric design and traffic control devices. Section 12.7 presents the tables and equations necessary for deter-
mining the CMF values for vehicle-pedestrian collision. Once the value for each CMF has been determined, all of
the CMFs are multiplied together in Column 4 ofWorksheet SP4H which indicates the combined CMF value for
vehicle-pedestrian collisions.
Worksheet SP4H. Crash Modification Factors for Vehicle-Pedestrian Collisions for Urban and Suburban Arterial
(!) (2) (3) (4)
CMF for Bus Stops CMF for Schools CMF for Alcohol Sales Establishments Combined CMF
". ~~:.!e. . J--··"---·"--"--····-··- (1)'(2)'(3)

from Table 12-28
2.78
·! ". . "". . . . .. from Table 12-30
1.12 4.20
Worksheet SP41-Vehicle-Pedestrian Collisions for Urban and Suburban Arterial Signalized Intersections
The predicted number of vehicle-pedestrian collisions per year for base conditions at a signalized intersection, Nped-
base, is calculated using Equation 12-30 and entered into Column 4 of Worksheet SP41. The coefficients for the SPF
and the overdispersion parameter associated with the SPF are entered into Columns 2 and 3; however, the overdis-
persion parameter is not needed for Sample Problem 4 (as the EB Method is not utilized). Column 5 represents the
combined CMF for vehicle-pedestrian collisions (from Column 4 in Worksheet SP4H). and Column 6 represents the
calibration factor. Column 7 calculates the predicted average crash frequency of vehicle-pedestrian collisions using the
values in Column 4. the combined CMF in Column 5, and the calibration factor in Column 6. Since all vehicle-pedes-
trian crashes are assumed to involve some level of injury, there are no property-damage-only crashes.
Worksheet SP41. Vehicle-Pedestrian Collisions for Urban and Suburban Arterial Signalized Intersections
(1) (2) (3) (4) (5) (6) (7)
Combined Predicted
SPF Coefficients CMF N,"!
Ne:"""·"
from Table 12-14 from (4) from
Crash Severity Overdispersion Equation Worksheet I Calibration I
Level a b c d e Parameter, k 12-30 SP4H Factor, C1 1 (8)*(9)*(10)

Total -9.53 0.40 0.26 0.45 0.04 0.24 0.113 4.20 1.00 I 0.475
Worksheet SP4J-Vehicle-Bicycle Collisions for Urban and Suburban Arterial Intersections

from Worksheets SP4C and SP4E are entered into Columns 2 and 3 respectively. These values are summed in Col-
umn 4. Column 5 contains the bicycle crash adjustment factor (see Table 12-17). Column 6 presents the calibration
factor. The predicted average crash frequency of vehicle-bicycle collision (Column 7) is the product of Columns 4, 5,
and 6. Since all vehicle-bicycle crashes are assumed to involve some level of injury, there are no property-damage-
only crashes.
Worksheet SP4J. Vehicle-Bicycle Collisions for Urban and Suburban Arterial Intersections
(1) i (2) 1 (3) (4) (5) (6) (7)
Predicted Nblmv Predicted Nw.,.,. Predicted NM fb/J.;,I Predicted N dl
(9) f•om (9) f•om from Table Calibration

Crash Severity Level Worksheet SP4C Worksheet SP4E (2)+(3) 12-17 Factor, C1 (4)*(5)*(6)
Total 2.658 0.196 2.854 O.o!5 1.00 0.043

Fatal and injwy (FI) 1.00 0.043
Worksheet SP4K-Crash Severity Distribution for Urban and Suburban Arterial Intersections
Worksheet SP4K provides a summary of all collision types by severity level. Values from Worksheets SP4D, SP4F, SP4I,
and SP4J are presented and summed to provide the predicted average crash frequency for each severity level as follows:

12·96 HIGHWAY SAFETY MANUAL
Worksheet SP4K. Crash Severity Distribution for Urban and Suburban Arterial Intersections
(I) (2) (3) (4)

(3) from Worksheet SP4D (6) from Worksbeet·SP4D
and SP4F; (7) from SP4G or (5) from Worksheet and SP4F; (7) from SP4G
Collision Type SP4I and SP4J SP4D and SP4F or SP4I and SP4J
_R_ear_-en_d_c_oi_Ii_si_oi!Ji_(_fro_m_'M_or_k_sh_e_e_tS_P_4_D_J__,__ _ _ _ _o_.3_8_0_ _ ~ 0.875 !.255
Head-on collisions (from Worksheet SP4D) 0.041 0.054 0.095
Angle collisions (from Worksheet SP4D) 0.293
Subtotal I o.844 L8ll I 2.655
Collision with parked vehicle (from 0.000 0.000 0.000

I I
Worksheet SP4F)
Collision with animal (from Worksheet SP4F)

I
0.000
..... J ..... _, __ ............
,
0.000 0.000
Collision \Vith fixed object 0.042 0.122 0.164

(from Worksheet SP4F) I I
Collision with other object ' 0.004 0.010 0.014
(from Worksheet SP4F) I
I
Other single-vehicle collision 0.002 0.003 0.005

(from Worksheet SP4F) I
Single-vehicle noncollision 0.008 0.005 0.013

(from Worksheet SP4F)
Colliswn with pedestrian 0.475 0.000 0.475

(from Worksheet SP4G or SP41)
collision ~~-bi~~l~ . (fr;~·w~~~~~·~. ·~p~~·-l"'" 0.043 I 0.000 0.043
Subtotal 0.574 0.140 0.714

I
Total 1.418 1.951 3.369
Worksheet SP4L-Summary Results for Urban and Suburban Arterial Intersections

Worksheet SP4L presents a summary of the results.
Worksheet SP4L Summary Results for Urban and Suburban Arterial Intersections
(I) (2)
Predicted Average Crash Frequency, Np.,dlorod 1,,

(crashes/year)
Crash Severicy Level (Total) from Worksheet SP4K
Total 3.369
Fatal and injury (FI) 1.418
Property damage only (PDO) 1.951

The Project
A project of interest consists of four sites located on an urban arterial: a three-lane TWLTL segment; a four-lane
divided segment; a three-leg intersection with minor-road stop control; and a four-leg signalized intersection.
(This project is a compilation of roadway segments and intersections from Sample Problems 1 through 4.)
The Question
What is the expected crash frequency of the project for a particular year incorporating both the predicted crash frequen-
cies from Sample Problems 1 through 4 and the observed crash frequencies using the site-specific EB Method?
The Facts
• 2 roadway segments (3T segment, 4D segment)
• 2 intersections (3ST intersection, 4SG intersection)
• 34 observed crashes (3T segment: 7 multiple-vehicle nondriveway, 4 single-vehicle, 2 multiple-vehicle driveway
related; 4D: 6 multiple-vehicle nondriveway, 3 single-vehicle, I multiple-vehicle driveway related; 3ST: 2 multi-
ple-vehicle, 3 single-vehicle; 4SG 6 multiple-vehicle, 0 single-vehicle)
Outline of Solution
predicted crash frequencies for the project using the site-specific EB Method (i.e., observed crashes are assigned to
specific intersections or roadway segments) presented in Section A.2.4 of Part C, Appendix A.
Results
WORKSHEETS
To apply the site-specific EB Method to multiple roadway segments and intersections on an urban or suburban
arterial combined, three worksheets are provided for determining the expected average crash frequency. The three
worksheets include: ·
• Worksheet SP5A (Corresponds to Worksheet 3A)- Predicted Crashes by Collision and Site Type and Observed
Crashes Using the Site-Specific EB Method for Urban and Suburban Arterials.
• Worksheet SP5B (Corresponds to Worksheet 3B)-Predicted Pedestrian and Bicycle Crashes for Urban and
Suburban Arterials.
• Worksheet SP5C (Corresponds to Worksheet 3C)-Site-Specific EB Method Sununary Results for Urban and
Suburban Arterials
Worksheets SPSA-Predicted Crashes by Collision and Site Type and Observed Crashes Using the
Site-Specific EB Method for Urban and Suburban Arterials.
The predicted average crash frequencies by severity level and collision type determined in Sample Problems 1 through
4 are entered into Columns 2 through 4 of Worksheet SP5A. Column 5 presents the observed crash frequencies by site
and collision type, and Column 6 presents the overdispersion parameters. The expected average crash frequency is cal-
culated by applying the site-specific EB Method which considers both the predicted model estimate and observed crash
frequencies for each roadway segment and intersection. Equation A-5 from Part C, Appendix A is used to calculate the
weighted adjustment and entered into Column 7. The expected average crash frequency is calculated using Equation
A-4 and entered into Column 8. Detailed calculation of Columns 7 and 8 are provided below.
Worksheet SPSA. Predicted Crashes by Collision and Site Type and Observed Crashes Using the Site-Specific
EB Method for Urban and Suburban Arterials
(1) (2) (3) (4) (S) (6) (7) (8)
Expected
Average Crash
Predicted Average Crash Frequency Weighted Frequency,
(crashes/year) Adjustment, w N.,.poot<'<l(whldo)
Collision Observed Equation A-5 Equation A-4

Type/Site Crashes, Nobo ..... d I Overdispersion from Part C, from Part C,
Type N NdiOiod(F[)
N redlc«d (PDQ)
,I (crashes/year) Parameter, k Appendix A Appendix A
ROADWAY SEGMENTS
Multiple-Vehicle Nondriveway
Segment 1 4,967 Ll96 3.771 • 7

--- --- . -. -. ··--- --- --···-···-··1- ----·-··-···-···· . - . ---- ----- - ' . --
I 0.66 - -,_I.. 0.234
_,,_,,_,,_,,_,,_,,_,,_
6.524
Segment 2 2.524 1 0.102 1 1.822 6 u2 0.231 5.197
Single-Vehicle
Segment 1 1.182 0.338 0.844
-'"""""""""""""'""'-"'_"_ - _,,_,,
4
........ ,_,,_,,_,_
---I- .
1.37
J ----
0.382
- _,, ......
2.924
Segment 2 0.485 0.085 0.401 3 0.86 0,706 1.224

Multiple-Vehicle Driveway-Related
Segment I
Segment 2
_____o_.7_34
0,!49
______ 0_: _1:-~_ T--~-:-~;--,----~
00 __ +--:::~- 0.553
0.828
1.300
0.295
INTERSECTIONS
Multiple-Vehicle
Intersection 1 1.268 0.405 0.862 2 0.80 0.496 1.637
Intersection 2 2.658 0.845 -·l 1.812 6 0.39
..............!...................
0.491
·····················•················
4.359
Single-Vehicle
Intersection 1 0.234 0.072 ' 0.162 1.14 0.789 0.818
-~- . ;;;~;;--
Intersection 2 0.196 0.056 0.36 0.934 0.183
Combined 14.397 3.920 10.476 34 24.461
(Sum of
Column)
Column 7-WeightedAdjustment
The weighted adjustment, w, to be placed on the predictive model estimate is calculated using Equation A-5 from
w =--7""---"---,
l+kx[ L
all study
Npcediotod'
Y""'
12-99
CHAPTER 12- PREDICTIVE METHOD FOR URBAN AND SUBURBAN ARTERIALS
Segment I
w I = 0.234
1+0.66x(4.967)
Segment2
I
W= 0.231
I+ !.32x (2.524)
Segment I
1
w= = 0.382
1+!.37x(1.182)
Segment2
w I 0.706
1+0.86x(0.485)
Multiple-Vehicle Driveway Related Collisions
Segment I
1
w =0.553
1+1.10x(0.734)
Segment2
I
w 0.828
1+!.39x(O.I49)
Intersection 1
w I 0.496
1+0.80x(1.268)
Intersection 2
1
w 0.491
1+ 0.39 X (2.658)
Intersection I
I
W= 0.789
1+!.149x(0.234)
Intersection 2
I
w= 0.934
1+0.36x(0.196)
Column 8--Expected Average Crash Frequency

The estimate of expected average crash frequency, Nexpecre,, is calculated using""' Equation A -4 from Part C, Appendix
A as follows:
Nexpected = w x Npredicted +(I - w) x Nobserved
Multiple-Vehicle Nondriveway Collisions
Segment IN,,«<,,= 0.234 x 4.967 +(I - 0.234) x 7 = 6.524
Segment2N"'""'=0.231 x 2.524+(1-0.231) x 6=5.197
Segment IN"'""'= 0.382 x 1.182 +(I- 0. 382) x 4 = 2.924
Segment 2 N"'""' = 0.706 x 0.485 + (!- 0.706) x 3 = 1.224
Multiple-Vehicle Driveway Related Collisions
Segment IN""'"'= 0.553 x 0.734 +(I- 0. 553) x 2 = 1.300
Segment 2 N.,,«<" = 0.828 x 0.149 +(I - 0. 828) x I = 0.295
Intersection IN""""= 0.496 x 1.268 + (! - 0. 496) x 2 = 1.637
_. = 0.491 x 2.658 +(I- 0. 491) x 6 = 4.359

Intersection 2 N exp<:et=
Intersection IN"'""'= 0.789 x 0.234 +(I- 0.789) x 3 = 0.818
Intersection 2 N"'""' = 0.934 x 0.196 + (!- 0.934) x 0 = 0.!83

Worksheets SP5B-Predicted Pedestrian and Bicycle Crashes for Urban and Suburban Arterials
Worksheet SP5B provides a summary of the vehicle-pedestrian and vehicle-bicycle crashes determined in Sample
Problems I through 4.
Worksheet SP5B. Predicted Pedestrian and Bicycle Crashes for Urban and Suburban Arterials
(I) (2) (3)
Site Type N~
ROADWAY SEGMENTS
Segment 1 I 0.089 0.048
Segment2 0.212 1 0.041
INTERSECTIONS
Combined (Sum of Column) 0.808 0.156
Worksheets SP5C-Site-Specific EB Method Summary Results for Urban and Suburban Arterials
Worksheet SP5C presents a summary of the results. Column 5 calculates the expected average crash frequency by
severity level for vehicle crashes only by applying the proportion of predicted average crash frequency by severity
level (Column 2) to the expected average crash frequency calculated using the site-specific EB Method. Column 6
calculates the total expected average crash frequency by severity level using the values in Column 3, 4, and 5.
Worksheet SP5C. Site-Specific EB Method Summary Results for Urban and Suburban Arterials
(I) (2) (3) (4) (5) (6)
Crash Severity Level Np...,d~<>,,.. Np,d Nb/U N•• .,..,«<~( .... hlel•l N ••peued
(2).,.,,.,b Worksheet (2)romb Worksheet (3)comb Worksheet

(13)crJmb Worksheet SPSA (3)+(4)+(5)
Total SPSA SP5B SPSB
14.397 i 0.808 0.156 24.461 25.4
(3)comh Worksheet (2),omh Worksheet (3)romh Worksheet

(5),,., '(2)F/(2)total (3)+(4)+{5)
Fatal and injury (FI) SPSA SPSB SP5B
3.920 I
0.808 0.156 6.660 7.6
(4)«>mb Worksheet
Property damage only I (5) •• ,'(2),,J{2J•• , (3)+(4)+(5)
SP5A
(PDO)
10.476 0.000 0.000 17.800 17.8
The Project
A project of interest consists of four sites located on an urban arterial: a three-lane TWLTL segment; a four-lane
divided segment; a three-leg intersection with minor-road stop control; and a four-leg signalized intersection. (This
project is a compilation of roadway segments and intersections from Sample Problems I through 4.)
The Question
average crash frequencies from Sample Problems I through 4 and the observed crash frequencies using the project-
level EB Method?
The Facts
• 2 roadway segments (3T segment, 4D segment)
• 2 intersection (3ST intersection, 4SG intersection)
• 34 observed crashes (but no information is available to attribute specific crashes to specific sites)
Outline of Solution
intersections are not available, but observed crashes are assigned to a facility as a whole) presented in Section A.2.5
Results
WORKSHEETS
To apply the project-level EB Method to multiple roadway segments and intersections on an urban or suburban
arterial combined, three worksheets are provided for determining the expected average crash frequency. The three
worksheets include:
• Worksheet SP6A (Corresponds to Worksheet 4A)-Predicted Crashes by Collision and Site Type and Observed
Crashes Using the Project-Level EB Method for Urban and Suburban Arterials
• Worksheet SP6B (Corresponds to Worksheet 4B)-Predicted Pedestrian and Bicycle Crashes for Urban and
Suburban Arterials
• Worksheet SP6C (Corresponds to Worksheet 4C)-Project-EB Method Summary Results for Urban and Suburban
Arterials
Worksheets SPGA-Predicted Crashes by Collision and Site Type and Observed Crashes Using the
Project-Level EB Method for Urban and Suburban Arterials
The predicted average crash frequencies by severity level and collision type, excluding vehicle-pedestrian and
vehicle-bicycle collisions, determined in Sample Problems I through 4 are entered in Columns 2 through 4 of
Worksheet SP6A. Column 5 presents the total observed crash frequencies combined for all sites, and Column 6
presents the overdispersion parameters. The expected average crash frequency is calculated by applying the project-
level EB Method which considers both the predicted model estimate for each roadway segment and intersection and
the project observed crashes. Column 7 calculates N'"', and Column 8 calculates N. 1• Equations A-1 0 through A-14
from Part C, Appendix A are used to calculate the expected average crash frequency of combined sites. The results
obtained from each equation are presented in Columns 9 through 14. Section A.2.5 in Part C, Appendix A defines all
the variables used in this worksheet. Detailed calculations of Columns 9 through 13 are provided below.
Worksheet SP6A. Predicted Crashes by Collision and Site Type and Observed Crashes Using the Project-Level EB
Method for Urban and Suburban Arterials
(I) (2) (3) (4) (5) (6)
Predicted Crashes
Observed Crashes, Overdispersion I~
Collision Type/Site Type N prodlcted(loiBI) N redlc,..d(FI) Nprodldod(PfJO) Nob••.-1 (crashes/year) Parameter, k
ROADWAY SEGMENTS li
Multiple-Vehicle Nondriveway ii'
Segment I 4.967 !.196 3.771 0.66 !I
Segment 2 2.524 0.702 1.822 1.32
,.
i
!•
Single-Vehicle I'
Segment 1 1.182 0.338 0.844 1.37
I'
1·.
Segment 2 0.485 0.085 0.401 0.86 l1
:!.
Multiple-Vehicle Driveway-Related [:,
!I:
Segment 1 0.734 0.179 0.555 !.10
Segment 2 0.149 0.042 0.107 1.39
INTERSECTIONS
Multiple-Vehicle
Intersection 1 1.268 0.405 0.862 0.80

II
11:1
Intersection 2 2.658 0.845 1.812 0.39
!
Single-Vehicle
i
Intersection l 0.234 0.072 0.162 1.14 I

I
Intersection 2 0.196 0.056 0.140 0.36
II•
I
Combined (Sum of Column) 14.397 3.920 10.476 34 '
I'
Worksheet SP6A continued on next page [.
.. l..!
Worksheet SP6A. Continued

(1) (7) (9) (10) (11) (12) (13)
N ntdlored ...0 w. N" w, N, N,. r..JIC~>mb(..hlclo)

Collision EquationA~S Equation
Type/Site Type (6)*(2)' A-10 EquationA-11 EquationA-12 EquationA-13 EquationA-14
ROADWAY SEGMENTS
Segmen.<..:_ ____
Segment 2
' ___16_~283
8.409"
_ \ _ _ :_81_1_ _[----
\ --1.825 L
_J______________
1
Single-Vehicle
-.------ ---- .----.--- ---·r -- ---·-···-- -- ... -.

Segment 1
Segment 2
I 0.593
0.031
:
_
0.899
0.455 ,
INTERSECTIONS
Multiple-Vehicle
Intersection 2 2.755 1- 1.018
Single-Vehicle
-~;:;:~~:-~- i-- -- ~~~~> ------ -~-;~~ . - -!- -----.--- - - - -

------------'------------r-------------------------- -------------------------------------------------
----1- --- .. -----1--- --- -. -·1- --- ..... --
Combmed 31549 9.716 I 0313 , 27864 0597 22297 25.080
(S f I ' '
Co=) 1 I I
Note: Npredlaod ..o =Predicted number of total crashes assuming that crash frequencies are statistically independent
5 5 5 4 4
NpredictedwO = LkrmjN~; + LkrsjN;sj + LkrdjN;dj + Lkim)Ni~j + LkisJN~j
j=l j=I j=l J=I j=l (A-8)
Nprodlaod .,., = Predicted number of total crashes assuming that crash frequencies are perfectly correlated
5 5 5 4 4
Npredictedwl = L~krmjNrmJ + L~krsjNrsj + LJkrdJNrdj + LJkimjNimj + L,JkrsjNisj

j=l J=l J=l j=l j=l (A-9)
Column9-w0
The weight placed on predicted crash frequency under the asswnption that crashes frequencies for different roadway
elements are statistically independent, w0 , is calculated using Equation A-1 0 from Part C, Appendix A as follows:
I
wo = ----:-,-------
1+ N predicted wO
Npredicted (total)
1
1+ 31.549
14.397
= 0.313
Column 10-N,
N 0, is calculated using Equation A-ll from Part C, Appendix A as follows:
N0 = w0 xNpredietcd(!Otnl) +(1-w)xN
0 obscrved(total)
= 0,313 X 14.397 +(I- 0.313) X 34
= 27,864
Column 11-w,
I
W:t = ---;-;--=------
1+ Npredicted wl
Npredicted (total)
I
I+ 9.716
14.397
= 0.597
Column 12-N,
The expected crash frequency based on the assumption that different roadway elements are perfectly correlated, N" III:
is calculated using Equation A-13 from Part C, Appendix A as follows: II
'I
II
N l = w 1 X N pre<l,cted(tot:ll) +(I- w)
1
X Nobserved(tOml)
= 0.597 X 14.397 +(I - 0.597) X 34
=22.297
Column 13-Nexpectcd/comb
~· ,, is calculated using Equation A-14 from
The expected average crash frequency based of combined sites, N expecteUJcom
N 0 +N1
Nexpected/comb =
2
27.864 + 22.297
2
= 25.080
.... !
,. [ -·
'
Worksheets SP6B-Predicted Pedestrian and Bicycle Crashes for Urban and Suburban Arterials
Worksheet SP6B provides a summary of the vehicle-pedestrian and vehicle-bicycle crashes determined in Sample
Problems I through 4.
Worksheet SP6B. Predicted Pedestrian and Bicycle Crashes for Urban and Suburban Arterials
(I) ! (2) I (3)
Site Type I I Nb,.tc
N ''
ROADWAY SEGMENTS
Segment I 0.089 0.048
- ··---~-------~--------
Segment2 I 0.212 o.o4r
' INTERSECTIONS
~'"~s"c~~~~----
Intersection 2
____ ·-- -I- ----~:03:·---l--· --~·024 0.475 1 0.043
___ --
Combined (Sum of Column) I 0.808 I 0.156
Worksheets SP6C-Project-Level EB Method Summary Results for Urban and Suburban Arterials
Worksheet SP6C presents a summary of the results. Column 5 calculates the expected average crash frequency by
'
severity level for vehicle crashes only by applying the proportion of predicted average crash frequency by severity
I' level (Column 2) to the expected average crash frequency calculated using the project-level EB Method. Column 6
calculates the total expected average crash frequency by severity level using the values in Column 3, 4, and 5.
'
Worksheet SP6C. Project-Level EB Method Summary Results for Urban and Suburban Arterials
i (2) '
(I) I
(3) (4) (5) I (6)
Crash Severity Level Np.-..!lcred I N,., I Nw~ N.xp<rle<lkonrh (vehldo) I Noxpool""
(2)comh Worksheet (2)<omb Worksheet (3)comb Worksheet (13)romh Worksheet SP6A (3)+(4)+(5)
I I I
Total SP6A SP6B SP6B
_,,_,, ................. ,_,_,_ -- _, ........ - _,, ...... ----
!4.397 0.808 0.156 25.080 26.0
. """" - .., ,.,,.,,,,,,.,,,, ''''"'''·- . . ,,,
I····· . ,.,,,,,,,,,,,,,,,,,,,,
(3)comb Worksheet (2)comh Worksheet (3)comh Worksheet (5),,,, '(2),/(2),,, (3)+(4)+(5)

Fatal and injury (Fl) SP6A SP6B SP6B
3.920 0.808 0.156 6.829 7.8
-- -·- - _,.........
,_ _,,_
(4)«>mb Worksheet - - (S),otol "'(2)nx/(2)toool (3)+(4)+(5)
Property damage SP6A
only (PDO) , ..
'I
_ - - - _,_ .....................,_
!0.476 --1·--- - -_,_,_,,_ ............
0.000
,_,,_,,_,_,
I
- _,,_ .,,_,,_,
0.000
-·-" .,,_,,_,,_
·-r - - _,_,_ ......... _,_ --

18,250
"""""'"
_.... _,_ '
1-
-- -- ............. ,_, _
!8.3
-- ""'"
i
:l
12.14. REFERENCES
( I) Bonneson, J. A., K. Zimmerman, and K. Fitzpatrick. Roadway Safety Design Synthesis. Report No. FHWA/
TX-05/0-4703-Pl. Texas Department ofTransportation, Austin, TX, November 2005.
( 2) Clark, J. E., S. Maghsoodloo, and D. B. Brown. Public Good Relative to Right-Turn-on-Red in South Carolina
and Alabama. In Transportation Research Record 926. TRB, National Research Council, 1983.
( 3) Elvik, R. and T. Vaa. The Handbook ofRoad Safety Measures. Elsevier Science, Burlington, MA, 2004.
I
( 4) FHWA.Jnteractive Highway Safety Design Model. Federal Highway Administration, U.S. Department of
Transportation, Washington, DC. Available from http://www.tfhrc.gov/safety/ihsdrnlihsdm.htm.
( 5) FHWA. Planning Glossary. Federal Highway Administration, U.S. Department ofTransportation, Washing-
ton, DC. 2008. Available from http://www.fhwa.dot.gov/planning/glossary/glossary_listing.cfrn?sort=definitio
n&TitleStart=A.
'I.I'
( 6) Harkey, D.L., S. Raghavan, B. Jongdea, EM. Council, K. Eccles, N. Lefler, F. Gross, B. Persaud, C. Lyon, E.
Hauer, and J. Bonneson. National Cooperative Highway Research Program Report 617: Crash Reduction II
Factors for Traffic Engineen·ng and ITS Improvement. NCHRP, Transportation Research Board, Washington, ,,,I
DC,2008.
( 7) Harwood, D. W, K. M. Bauer, I. B. Potts, D. J. Torbic, K. R. Richard, E. R. Kohlman Rabbani, E. Hauer, and L.

E!efteriadou. Safety Effectiveness a/Intersection Left- and Right-Turn Lanes, Report No. FHWA-RD-02-089.
Federal Highway Administration, U.S. Department ofTransportation, Washington, DC, April2002.
( 8) Harwood, D. W, K. M. Bauer, K. R. Richard, D. K. Gilmore, J. L. Graham, I. B. Potts, D. J. Torbic, and E.

Hauer. National Cooperative Highway Research Program Document 129, Phases I and II: Methodology to
Predict the Safety Performance of Urban and Suburban Arterials. (Web Only). NCHRP, Transportation Re-
search Board, Washington, DC, March 2007.
( 9) Harwood, D. W., D. J. Torbic, D. K. Gilmore, C. D. Bokenkroger, J. M. Dunn, C. V. Zegeer, R. Srinivasan, D.

Carter, and C. Raborn. National Cooperative Highway Research Program Document 129, Phase Ill: Methodol-
ogy to Predict the Saftty Peiformance of Urban and Suburban Arterials: Pedestrian Safety Prediction Methodol-
ogy. (Web Only). NCHRP, Transportation Research Board, Washington, DC, March 2008.
( 10) Hauer, E. Left-Turn Protection, Safety, Delay and Guidelines: A Literature Review. Federal Highway Adminis-
tration, U.S. Department ofTransportation, October 2004. Available from http://www.roadsafetyresearch.com.
( II) Lyon, C., A. Haq, B. Persaud, and S. T. Kodama. Development of Safety Performance Functions for Signal-
ized Intersections in a Large Urban Area and Application to Evaluation of Left-Turn Priority Treatment.
Presented at the 84th Annual Meeting of the Transportation Research Board, Washington, DC, January 2005.
( 12) Persaud, B., F. M. Council, C. Lyon, K. Eccles, and M. Griffith. A Multi-Jurisdictional Safety Evaluation of Red-
Light Cameras. 84th Transportation Research Board Annual Meeting, TRB, Washington, DC, 2005. pp. 1-14.
( 13) Srinivasan, R., C. V. Zegeer, F. M. Council, D. L. Harkey, and D. J. Torbic. Updates to the Highway Safety
Manual Part D CMFs. Unpublished memorandum prepared as part of the FHWA Highway Safety Informa-
tion System Project. Highway Safety Research Center, University of North Carolina, Chapel Hill, NC, July
2008.
( !4) Srinivasan, R., F. M. Council, and D. L. Harkey. Calibration Factors for HSM Part C Predictive Models. Un-
published memorandum prepared as part of the FHWA Highway Safety Information System Project. Highway
Safety Research Center, University of North Carolina, Chapel Hill, NC, October 2008.
( 15) Zegeer, C. V., and M. J. Cynecki. Determination of Cost-Effective Roadway Treatments for Utility Pole Ac-
cidents. In Transportation Research Record 970. TRB, National Research Council, Washington, DC, !984 .
.. ~ . '
'-
APPENDIX 12A-WORKSHEETS FOR PREDICTIVE METHOD FOR URBAN AND SUBURBAN ARTERIALS
Worksheet 1A. General Information and Input Data for Urban and Suburban Roadway Segments
General Information ! Location Information I
Analy,_ :':':. . _______________________ --···-····· _______ _:R:_o:_:•:_d::_w_:. :•Y::___ _____ _ ________________ _______________________________ _
1 1
Agency or Company
Analysis Year
Road type (2U, JT, 4U, 4D, 5T)
Length of segment. L (mi)
AADT (veh!day)
Type of on-street parking (none/parallel/angle) none
Proportion of curb length with on-street parking
- - - ~-i~-di,-:-gw_(p_':-:e':(: ".:-1-no_t_p_te_s_en_t_)- - - __ _ ---~ -j-..-. -. .-."-._

.-_. _-_-_-n_o_t:~sent
Auto speed enforcement (present/not present) not present
-=-1=-~-
_,,_,,_ ....... ,_,_ .-.. ·-
Major corrunercial driveways (number)
Minor commercial driveways (number) ______ -·-_-_[--
Major industrial/instrtutional driveways (number)
Minor industrial/~-s~~tu_o_·
o~~~- driveways (~~-b:Q _____ t.
Major residential driveways (number) I ---1-··
Minor residential driveways (number)
Other driveways (number)
Speed Category
Roadside fixed object density (fixed objects/mi) , not present
Offs~t to roadside fi::_d-objects (ft) ----~-+-

Calibration Factor, Cr i
not present
LO +-----------
Worksheet 1 B. Crash Modification Factors for Urban and Suburban Roadway Segments
(1) (2) (3) (4) (5) (6)
I
CMFfor
CMF for On-Street CMF for Roadside CMFfor Auto Speed
Parking Fixed Objects CMF for Median Width Lighting Enforcement Combined CMF
CMF~r CMF,, CMF1, CMF4, CMFJr CMFcomb

from Equation 12-32 from Equation 12-33 from Table 12-22 from from Section (1 )'(2)'(3)'(4)'(5)
12.7.1
Worksheet 1C Multiple-Vehicle Nondriveway Collisions by Severity Level for Urban and Suburban Roadway Segments
(2) (3) I (4)
(I)
SPF Coefficients Overdispersion Parameter, k Initial Nbmrv
from Table 12-3
Crash Severity Level a b from Table 12-3 from Equation 12-10

Total
--,C-.---------
T-----
Worksheet 1C continued
(I) (5) (6) (7) (8) (9)
Combined Calibration
Adjusted N brno• CMFs Factor
I
(6) from
Crash Severity Level I Proportion of Total Crashes [ Worksheet SPlB I (6)*(7)'(8)
Total
Fatal and injury (FI) - [: __ 5~l,J~~lft<~l,DE) _____
~
1
Property damage only (PDO) I (5\,.,-(s),
Worksheet 1D. Multiple-Vehicle Nondriveway Collisions by Collision Type for Urban and Suburban
Roadway Segments
(I) (2) (3) (4) (5) (6)
Proportion of Predicted Nhrm" (Fl) I Proportion of Predicted Nb""" (PDO) Predicted Nhrmv C<o<Ql)
Collision l)'pe cFn (crashes/year) Collision Type CPDOJ I (crashes/year) I (crashes/year)
(9)F1from (9)PDo from (9)totlll from
Collision Type from Table 12-4 Worksheet 1C I from Table 12-4 Worksheet lC Worksheet lC
Total 1.000 LOOO
I
(2)'(3), (4)'(5)PDO I (3)+(5)
Rear-end collision
Head-on collision
. . . . J. '
I
I
Angle collision
SideS\Vipe. same direction
Sidesv.ripe, opposite
t--+-----1 ------~-·--·-
direction
Other multiple-vehicle
collision 1
Worksheet 1E. Single-Vehicle Collisions by Severity Level for Urban and Suburban Roadway Segments
(I) (2) (3) (4) (5) (6) (7) (8) (9)
I I
SPF Overdispersion Adjusted I Combined Calibration Predicted
Coefficients Parameter, k Initial Nb,.... I
I
CMFs I
Factor N,=
Crash
Severity '--f·-·-~-2T,·;_b_I·_I I E:::.. I
Proportion of
I (6) from
Worksheet
Level a b I from Table 12-5 1 12-13 I Total Crashes (4)•• ,*(5) I SPIB I (6)'(7)*(8)
!
Total
-~~~·"~d""l- -y-- ·-1---- -1--
'
injury
1
(FI) I
1
Property
damage
only
(PDO)
Worksheet 1 F. Single-Vehicle Collisions by Collision Type for Urban and Suburban Roadway Segments
(I) I (2) (3) (4) (5) (6)
Proportion of Predicted Nbr.w(FJ; Proportion of Predicted Nb,.... rPoo; Predicted Nb...,. (rornJ>
Collision Type fF/J (crashes/year) Collision Type rPDoJ (crashes/year) (crashes/year)
(9)F1 from (9)P.oo from I (9),0 ,.1 from Worksheet

Collision Type from Table 12-6 Worksheet lE from Table 12-6 Worksheec IE IE
Total 1.000 1.000
(2)'(3)" (4)'(5),DO (3)+(5)
Collrsron wrth fixed

-obJect
---------
Colhsron With other
--t 1 ---
---------
I
I
I
I
- - - - - - --1---- - - - - - - - - - - - - - - - - - -
I
~-
Qfu~ s~~;l~~;eb~J~-- ------------------ ------------
I --~- --~---- ------------ I --------------- -~----------------
colhsiOn 1
Worksheet 1 G. Multiple-Vehicle Driveway-Related Collisions by Driveway Type for Urban and Suburban
Roadway Segments
(1) (2) (3) (4) (5) (6)

Year,N1 1
Adjustment, t Initial N bnlwv Parameter, k
Number of from Table Equation 12-16

Driveway Type Driveways, n1 from Table 12-7 12·7 n *N *(AADT/15.000)1 from Table 12-7
Major
commercial
-_I
Minor
commercial
Major industriaV
institutional
Minor industrial/
institutional .1 . -··-
Major residential
-
_.:~.~~~~·-i·~~~.~~;~~l- -_.::·.~-·~-~~·.:~-~··-
0ther
-_.,_., _,_,__,_. . ,_,11·-
Total
Worksheet 1 H. Multiple-Vehicle Driveway-Related Collisions by Severity Level for Urban and Suburban
Roadwao' Segments
(1) (2) (3) (4) (5) (6) (7)
Proportion of
Total Crashes Combined
i Initial Nbl"<<....., (f,.)
I
Adjusted Nhrdwv CMFs I Predicted Nhn...,
(6) from
(5),.,Q1 from Worksheet Calibration
Crash Severity Level Worksheet lG from Table 12-7 (Z)Iolol *(3) I 1B Factor, C, (4)*(5)*(6)
Total
_,,_,,,
--- ·- "" - " ...... _,_ -"-""'
..... _,_- "- - -- -·-'"" - -·- - . -- - - . -·- -·- " ""'"""--

----
-
··---- -------- ___ ,,_, __ ,_,_, _ _ ··-·------.... ____,_,_
Property damage only (PDQ) -
Worksheet 1l.Vehicle-Pedestrian Collisions for Urban and Suburban Roadway Segments

(1) (2) (3) (4) (5) (6) (7) (8)
Predicted Predicted
Nbrmv Predicted Nb''" Predicted NbrJwv Nb, f Predicted NprJr
from
Crash I (9) from (9) from (7) from Table Calibration
Severity Level Worksheet I C Worksheet IE Worksheet IH (2)+(3)+(4) 12·8 Factor, C, (5)'(6)'(7)
Total
Fatal and
injury (FI)
Worksheet 1J. Vehicle-Bicycle Collisions for Urban and Suburban Roadway Segments
(I) (2) (3) (4) I (5) I (6) (7) (8)
Predicted
Predicted Nh...,. Predicted N~m~wr N,, fbi""r Predicted Nblk<'r
from
Crash (9) from (9) from (7) from Table Calibration
Severity Level Worksheet I C Worksheet IE Worksheet lH (2}t(3)+(4) 12-9 Factor, C, (5)'(6)'(7)
Total
Fatal and
injury
Worksheet 1 K. Crash Severity Distribution for Urban and Suburban Roadway Segments
(I) (2) (3) (4)
Fatal and Injury (FI) Property Damage Only (PDO) Total
(3) from Worksheet lD and IF; (6) from Worksheet lD and IF;
(7) from Worksheet lH; and (8) (5) from Worksheet lD and IF; (7) from Worksheet lH; and (8)
Collision l)rpe from Worksheet 11 and 1J and (7) from Worksheet lH from Worksheet 11 and lJ
MULTIPLE-VEHICLE
Rear-end collisions I
(from Worksheet ID)
Head-on collisions I
(from Worksheet ID)

I
Angle collisions
(from Worksheet lD)
--~ -- --- ---- ------1--- _______________l ______ ---------------------
Sideswipe. same direction
(from Worksheet lD)
Sideswipe, opposite direction

(from Worksheet lD)
Driveway-related collisions
(from Worksheet IH)
Other multiple-vehicle collision

(from Worksheet lD)
Subtotal
SINGLE-VEHICLE
Collision with animal
(from Worksheet IF)
Collision with fixed object

(from Worksheet IF)
----- -- -·--- --,--------- --··········-····--- -- -----
Collision with other object
(from Worksheet IF)
Other single-vehicle collision

(from Worksheet lF)
Collision with pedestrian

(from Works.:h::•:.•.:.':. :1_!:) _ __ _ _ _ __ _ _ , _ _ _ ____________________ -+- _____________________ _
Collision with bicycle
(from Worksheet 1))
Subtotal
Total
Worksheet 1 L Summary Results for Urban and Suburban Roadway Segments

(I) (2) (3) (4)

Frequency, Np ....dJ ..oo~,.. Crash Rate
(crashes/year) (crasbes/mi/year)
Roadway Segment
Crash Severity Level (total) from Worksheet lK I Length, L (mi) (2)/(3)
.~~-~~-~~-1~~~.-~~.2_. _,_,_,_,_, _,_,_,_,_ . ,_ -_,,_,_, _,_,_ ·-,_,_,_, _, _, _,_,_,_,_,_,_, _, _,_ --,_,_,_...J. . . . . ._,,_,_,_ ·-· _,,_,,. . . . ,_. ,_,_, _,_,_,,. . . . ,_,_,~
Property damage on~y (PDO) I I
Worksheet 2A. General Information and Input Data for Urban and Suburban Arterial Intersections
Gi!neral Information I Location Information
Intersection type (3ST, 3SG, 4ST, 4SG)
AAD~"1 (veh/day)_ -·····

(veh/day)
Intersection lighting (present/not present) not present
Calibration factor, 1.00

Number of major-road approaches with left-turn lanes (0, I, 2) 0
__ Num~_:: of major-ro~~. approach~~.~th right-tum ~~es (0. !_:_.~.~.-_

. J _____ . .·-----~-
I
--1----- . ----.
Number of approaches with left-tum lanes (0. l, 2, 3, 4) 0
Number of approaches with right-tum lanes (0, 1, 2, 3, 4) 0
Number of approaches with left-tum signal phasing
Number of approaches with right-tum-on-red prohibited 0

Type ofleft-turn signal phasing permissive
Intersection red-lighr cameras (present/not present) notpresem
Sum of all pedestrian crossing volumes (PedVol)
Maximum number of lanes crossed by a pedesoian (n1on-.)

Number of bus stops within 300m (1,000 ft) of the intersection 0
Schools within 300m (1,000 ft) of the intersection not present
Number of alcohol sales establishments within 300m (1,000 ft) 0
of the intersection
.•. j
Worksheet 28. Crash Modification Factors for Urban and Suburban Arterial Intersections
(1) (2) (3) (4) I (5) (6) (7)
CMFfor CMFfor CMFfor
CMFfor I Left-Turn Right-Turn CMF for Right- CMFfor Red-Light
Left-Turn Lanes Signal Phasing Lanes Turn-on-Red Lighting Cameras Combined CMF
CMF 11 CMF 2, C.MFJ, CMF4; CMFj, CMF 61 CMFcomb
I
from Table 12-24 I from Table 12-25 from Table from Equation from Equation from Equation (1 )'(2)'(3)'(4)'(5)'( 6)
12-26 12-35 12-36 12-37
Worksheet 2C. Multiple-Vehicle Collisions by Severity Level for Urban and Suburban Arterial Intersections
(1) (2) (3) (4)
SPF Coefficients Overdispersion Parameter, k Initial Nblnr•

from Table 12-10

Total

... J
'·······'······ .
--
--I ·I
Worksheet 2C. Continued
(1) (5) (6) (7) (8) (9)
Adjusted Nhlm> Combined CMFs 1

Predicted Nhlm•
Proportion of Total (7) from Calibration
Crash Severity Level Crashes (4),~,./'(5) Worksheet SP4B Factor, C, I (6)*(7)'(8)
Total

. -.--··---- -·- ....1'-·-- --·······-·------. --·---·-- ----, --------·--
(5)to,.,-(5)FI
Worksheet 20. Multiple-Vehicle Collisions by Collision Type for Urban and Suburban Arterial Intersections
(1) (2) (3) (4) (5) (6)
Predicted
Proportion of i Predicted Nblmv(FI) Proportion of Nblmv(PDO) Predicted Nblmv (<Mal)
Collision Type rFn I
(crashes/year) Collision Type rPDO! (crashes/year) (crashes/year)
(9)n from (9)PDO from

Collision Type from Table 12-11 Worksheet 2C from Table 12-11 Worksheet 2C (9)PDO from Worksheet 2C
Total LOOO 1.000
(2)'(3)" (4)'(5),0 (3)+(5)

Rear-end collision , I
:::~::~~::::_ ·- n·0
__ -_1,_·- , . " ______ . _,_ . _l _________.__ "__ I ___ -·---:-:=::_:1=~:~==--r·-----:-_·
Sideswipe
Other multiple-
vehicle collision
Worksheet 2E. Single-Vehicle Collisions by Severity Level for Urban and Suburban Arterial Intersections
(I) (2) (3) (4)
SPF Coefficients Overd.ispersion Parameter, k
from Table 12-12

from Equation 12-25; (FI)
Crash Severity Level a b I c from Table 12-12 from Equation 12-25 or 12-27
__!o'~~-------------------~'-------~----L _________ ,___________ _ _ __

_)'~ta]_andinj"')'_~F!l ______ _________ -f ______ I ____________ L______________________________ __ __ _ __ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ _ _ ___ -
Property damage only (PDQ) I

1
Worksheet 2E. Continued
(I) (5) (6) (7) (8) (9)
'
Adjusted Nw.-. i
Combined CMFs Predicted Nblw
Crash Severity Level Total Crashes I (4)1011/(5) Worksheet SP4B Factor, C 1 I (6)*(7)*(8)
I
Total I
Property damage only (PDQ) 1- -- _(5),,,~( 5 ),.' - - - - -I
Worksheet 2F. Single-Vehicle Collisions by Collision Type for Urban and Suburban Arterial Intersections
(1) (2) (3) (4) (5) (6)
Proportion Predicted Predicted

Proportion of Predicted Nw.wrFJ; of Collision Nblsv(PDQ)
N
• bls>(<otal)
Collision Type rFJJ (crashes/year) Type (PDOJ (crashes/year) (crashes/year)
(9)n from (9)PDO from (9)PDO from

Table 12-13 Table 12-13
Collision Type Worksheet 2E Worksheet 2E Worksheet 2E
Total LOOO LOOO
(2)'(3), (4)*(5),00 (3)+(5)
Collision with parked vehicle

.
Other smgle-veh1cle colhs10n
Single-vehicle noncollision
1--- -
Worksheet 2G. Vehicle-Pedestrian Collisions for Urban and Suburban Arterial Stop-Controlled Intersections
(I) (2) (3) (4) I (5) I (6) (7)
Predicted N bimv Predicted NMw Predicted N w I Predicted Nv,dl

f '
(9) from (9) from I from Calibration
Crash Severity Level (2)+(3) (4)'(5)*(6)
Worksheet 2C Worksheet 2E Table 12-16 Factor, C 1
Total
Worksheet 2H. Crash Modification Factors for Vehicle-Pedestrian Collisions for Urban and Suburban Arterial
(!) (2) (3) (4)
CMF for Bus Stops CMF for Schools CMF for Alcohol Combined CMF
I
Sales Establishments
CMF 1p I CMF,_ CMF~p
from Table 12-28 from Table 12-29 from Table 12-30 I (1)*(2)*(3)
Worksheet 21. Vehicle-Pedestrian Collisions for Urban and Suburban Arterial Signalized Intersections
(!) (2) (3) (4) (5) (6) (7)
SPF Coefficients N ~a,.,

Combined CMF
from Table 12-14 from Equation (4) from
Crash Overdispersion Calibration (8)'(9)'(10)
Severity Level a b c d e Parameter, k 12-30 Worksheet 2H Factor, C 1
I
Total I
Fatal and
-
I
_I_
injury (Fl) -~-,
Worksheet 2J. Vehicle-Bicycle Collisions for Urban and Suburban Arterial Intersections
(I) (2) (3) (4) (5) (6) (7)
Predicted N blm• Predicted Nbl..v Predicted N w Predicted N Ml
(9) from (9) from from Calibration

(2)+(3) (4)*(5)'(6)
Crash Severity Level Worksheet 2C Worksheet 2E Table 12-17 Factor, Ci
Total
Fatal and injury (Fl)
Worksheet 2K. Crash Severity Distribution for Urban and Suburban Arterial Intersections
(!) (2) (3) (4)
(3) from Worksheet (6) from Worksheet

(5) from Worksheet
2D and 2F; 2D and2F;
2D and 2F
Collision Type I (7) from 2G or 21 and 2J (7) from 2G or 21 and 2J
Rear-end collisions
(from Worksheet 2D)
Head-on collisions
(from Worksheet 2D)
Angle collisions (from Worksheer 2D)
Sideswipe (from Worksheet 2D)

-- -----,-------------··
Other multiple-vehicle collision
(from Worksheet 2D)
Subtotal
Collision with parked vehicle
(from Worksheet 2F)
(from Worksheet 2F)
(from Worksheet 2F)
(from Worksheet 2F)
Other single-vehicle collision
(from Worksheet 2F)
Single-vehicle noncollision
(from Worksheet 2F)
Collision with pedestrian

(from Worksheet 2G or 21)
. -···-···"····-··-···-··-···"····-···-···-···-····-···-···-···-··-··-···-···-···-··-···- -
" - -·- -·- - - -. -··-··-··- -
Collision with brcycle
(from Worksheet 2J)
Subtotal
Total
Worksheet 2L. Summary Results for Urban and Suburban Arterial Intersections
(I) (2)
Predicted Average Crash Frequency, Nprrol<red/nr (crashes/year)

Crash Severity Level (Total) from Worksheet 2K
Total

- '--································--·················· -··················---·················· 1-······················································-
·- l
12-118 HIGHWAY SAFElY MANUAL
Worksheet 3A. Predicted Crashes by Collision and Site Type and Observed Crashes Using the Site-Specific
(I) (2) (3) (4) (5) (6) {7) (8)
Expected
I
I I
' Average Crash
Predicted Average Crash Frequency Weighted Frequency,
(crashes/year) Adjustment, w N exp"'""' (vehicle)
Observed
Crashes, '
Equation A-5 Equation A-4
I
Collision Type/ Noboorwd Overdispersion from Part C, fromPartC,
Site Type N,,.,,,uc,.dltotal) N r<diO<od(Fll Nprodlc!od (PDO)
(crashes/year) I Parameter, k Appendix A Appendix A
ROADWAY SEGMENTS
Segment I
Segment2
Segment 3
- - - ---- -- - - -- --- - - - -- - --- --- --- - - ---1--- --- ------- _ji1,·-···-···-··-·- - - -- --- --
Segment4
Single-Vehicle
:::::~ 1,·--··-·-·-·--·---1 ----······1!--------------

---=-:"'.--'----+---------+----~------'-'--------+--------- -~--~------f--------
Segment 3
Segment4
Segment I
Segment 2
Segment 3
nn--: n n nl n
Segment 4 I
INTERSECTIONS
Multiple-Vehicle
Intersection 1
Intersection 2
Intersection 3
Intersection 4 ·-·--·- ] __
Single-Vehicle
.~;;:~~:~~+ -1--
- .... -...................... "1"''""'""'"
Intersection 3
-
--
-
-----------l-- ---- -----~--- --------
Intersection 4
Combined
. -------·+-- ·+-~----
(Sum of
Column) 1 1
Worksheet 3B. Predicted Pedestrian and Bicycle Crashes for Urban and Suburban Arterials
(I) (2) (3)
Site Type N
ROADWAY SEGMENTS
Segment 1
Segment 2
Segment 3
Segment4
INTERSECTIONS
Intersection 1
Intersection 2
Intersection 3
Intersection 4
Worksheet 3C. Site-Specific EB Method Sununary Results for Urban and Suburban Arterials
(I) (2) (3) (4) (5) (6)
Crash Severity Level Nprodl01od N,m Nb/U N"""""ted (whldo) N••poc<ed
(2)comb Worksheet (3)comb Worksheet (13)comb Worksheet

(2)romb Worksheet 3A (3)+(4)+(S)
Total 3B 3B 3A
(2)comb Worksheet (3)comb Worksheet (3)+(4)+(S)

(3)cDmb Worksheet 3A
Fatal and injury (FI) 3B 3B
-------------+-------------
Property damage only (PDO) (4)romb Worksheet 3A (3)+(4)+(S)
0.000 0.000
12-120 HIGHWAY SAFE1Y MANUAL
Worksheet 4A- Predicted Crashes by Collision and Site Type and Observed Crashes Using the Project-Level
(!) (2) (3) (4) (5) (6) (7)
Predicted Crashes Npredlot<d""'
Observed Crashes,
N ob••,._,_ (crashes/ Overdispersion Equation A-S
Collision Type/Sice Type i N ..,dlclood (totoll I N ....tlct<d (FJ) I N "'dieted IPDOl year) Parameter, k (6)'(2)'
ROADWAY SEGMENTS
Segment 1
Segment 2
Segment 3
Segrnent4
Single-Vehicle
I
Segment 1
- ------··.--.----.-- ___ l _____ -----------
Segment2
Segment 3 1 i
Segment4 --·-----·--- li··············-···l
I
Segment l
Segment 2
...... ···I
Segment 3
Segment 4
11\"TERSECTIONS
Multiple-Vehicle
~--- .. -- r- ---------- --- ----- --··---

Intersection 1 I
Intersection 2
- J -·- - _...............................;. ·· .-. ·"!"-- -
~
Intersection 3
Intersection 4 r -
Single-Vehicle
Intersection 2 I
--- -------------r----··. -------··
Intersection 3
····--
Intersection 4
·-·········--·':··- ------' - - ·-. -· .......-....-. -.. -.. - .............. -·r·-.........................-.. -
1
Combined
(Sum of Column) I
Worksheet 4A. continued
(I) (8) I
(9) (10) (11) (12) (13)
N ellpoOiod!comb
w. N, w, N, (V<hlole)
Collision Type/ Equation A-9 Equation

Site Type (sqrl((6)*(2)) Equation A-10 Equation A-ll EquationA-12 EquationA-13 I A-14
ROADWAY SEGMENTS
Segment 1
Segment 2
Segment 3
Segment 4 I
Single-Vehicle
Segment 1
Segment2 I . ·~--~-L
Segment 3
Segment 4
-- ·----- - -1 - -·1--= -- ! r·
'
--
INTERSECTIONS
Multiple-Vehicle
Intersection 1
Intersection 2
Intersection 3
. -·-··r··---- ---------
~
I
.-·1
l-
'
Intersection 4
Single-Vehicle
Intersection 1
Intersection 2
Intersection 3
-------------································ ·--+--------- ------+-
Intersection 4
Combined
(Sum of Column)
Worksheet 4B. Predicted Pedestrian and Bicycle Crashes for Urban and Suburban Arterials
(I) I (2) (3)
Site Type N
ROADWAY SEGMENTS
Segment 1 I
""'S~;~;·;"_"_,,_,,_,_,,_,,_,,_,_,_,_,_,_,,_,_,_,,_,,_,,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,,_,_,,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,,_,_,
JI - _,_ ·- ,_,_ ·-·- - - - _,_, -·- ·-·-·-·-"-- _,_,,_, _
_,,_,_,,_,_,_,_, __ ,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,,_,_,_,_,_,_,_,_, _,_,_,_,_, __ ,_, __ ,_,_,_,_,_,_,,!..,_,_,_,_,_,_,_, _,_,_,_,_,_,_,_,_,_,_,_,_,_,_,,_,_,,_,,_,_,J. __,_,-- -..-.- -·- - - _,_,_,_, _,_,_,_,_ -
Segment 3
Segment4
INTERSECTIONS
Intersection 1
Intersection 2
Intersection 3
Intersection 4
Worksheet 4C Project-Level EB Melhod Summary Results for Urban and Suburban Arterials
(I) (2) (3) (4) (5) (6)
Total
(3)+(4)+(5)
W._or_k_sh_e_et_4_A+---------~
f-'-(4c<)'""'m"'-" ___ - I (5).,,'(2),DJ(2).,, (3)+(4)+(5)
0.000 I 0.000 I
Appendix A-Specialized Procedures
Common to All Part C Chapters
This Appendix presents two specialized procedures intended for use with the predictive method presented in Chapters
I 0, 11, and 12. These include the procedure for calibrating the predictive models presented in the Part C chapters to
local conditions and the Empirical Bayes (EB) Method for combining observed crash frequencies with the estimate
provided by the predictive models in Part C. Both of these procedures are an integral part of the predictive method in
Chapters 10, 11, and 12, and are presented in this Appendix only to avoid repetition across the chapters.
A.1. CALIBRATION OF THE PART C PREDICTIVE MODELS

The Part C predictive method in Chapters 10, 11, and 12 include predictive models which consist of safety
performance functions (SPFs), crash modification factors (CMFs) and calibration factors and have been developed
for specific roadway segment and intersection types. The SPF functions are the basis of the predictive models and
were developed in HSM-related research from the most complete and consistent available data sets. However, the
general level of crash frequencies may vary substantially from one jurisdiction to another for a variety of reasons
including climate, driver populations, animal populations, crash reporting thresholds, and crash reporting system
procedures. Therefore, for the Part C predictive models to provide results that are meaningful and accurate for
each jurisdiction, it is important that the SPFs be calibrated for application in each jurisdiction. A procedure for
determining the calibration factors for the Part C predictive models is presented below in Section A.1.1.
Some HSM users may prefer to develop SPFs with data from their own jurisdiction for use in the Part C predictive
models rather than calibrating the Part C SPFs. Calibration of the Part C SPFs will provide satisfactory results.
However, SPFs developed directly with data for a specific jurisdiction may provide more reliable estimates for that
jurisdiction than calibration of Part C SPFs. Therefore, jurisdictions that have the capability, and wish to develop
their own models, are encouraged to do so. Guidance on development of jurisdiction-specific SPFs that are suitable
for use in the Part C predictive method is presented in Section A.l.2.
Most of the regression coefficients and distribution values used in the Part C predictive models in Chapters 10, 11, and
12 have been determined through research and, therefore, modification by users is not recommended. However, a few
specific quantities, such as the distribution of crashes by collision type or the proportion of crashes occurring during
nighttime conditions, are known to vary substantially from jurisdiction to jurisdiction. Where appropriate local data are
available, users are encouraged to replace these default values with locally derived values. The values in the predictive
models that may be updated by users to fit local conditions are explicitly identified in Chapters 10, 11, and 12. Unless
explicitly identified, values in the predictive models should not be modified by the user. A procedure for deriving
jurisdiction-specific values to replace these selected parameters is presented below in Section A.l.3.
A.1.1. Calibration of Predictive Models

The purpose of the Part C calibration procedure is to adjust the predictive models which were developed with data
from one jurisdiction for application in another jurisdiction. Calibration provides a method to account for differences
A-1
.I
A-2 HIGHWAY SAFETY MANUAL
between jurisdictions in factors such as climate, driver populations, animal populations, crash reporting thresholds,
and crash reporting system procedures.
The calibration procedure is used to derive the values of the calibration factors for roadway segments and for
intersections that are used in the Part C predictive models. The calibration factor for roadway segments, Cr, is used
in Equations 10-2, 11-2, 11-3, and 12-2. The calibration factor for intersections, C;, is used in Equations 10-3, 11-
4, and 12-5. The calibration factors, C, and C;, are based on the ratio of the total observed crash frequencies for a
selected set of sites to the total expected average crash frequency estimated for the same sites, during the same time
period, using the applicable Part C predictive method. Thus, the nominal value of the calibration factor, when the
observed and predicted crash frequencies happen to be equal, is 1.00. When there are more crashes observed than
are predicted by the Part C predictive method, the computed calibration factor will be greater than 1.00. When there
are fewer crashes observed than are predicted by the Part C predictive method, the computed calibration factor will
be less than 1.00.
It is recommended that new values of the calibration factors be derived at least every two to three years, and some
HSM users may prefer to develop calibration factors on an annual basis. The calibration factor for the most recent
available period is to be used for all assessment of proposed futore projects. If available, calibration factors for the
specific time periods included in the evaluation periods before and after a project or treatment implementation are to
be used in effectiveness evaluations that use the procedures presented in Chapter 9.
If the procedures in Section A.l.3 are used to calibrate any default values in the Part C predictive models to local
conditions, the locally-calibrated values should be used in the calibration process described below.
The calibration procedure involves five steps:

• Step !-Identify facility types for which the applicable Part C predictive model is to be calibrated.
• Step 2-Select sites for calibration of the predictive model for each facility type.
• Step 3-0btain data for each facility type applicable to a specific calibration period.
• Step 4-Apply the applicable Part C predictive model to predict total crash frequency for each site during the
calibration period as a whole.
• Step 5-Compute calibration factors for use in Part C predictive model.
Each ofthese steps is described below.
A.l.l.L Step 1-Identify facility types for which the applicable Part C SPFs are to be calibrated.
Calibration is performed separately for each facility type addressed in each Part C chapter. Table A-1 identifies all of
the facility types included in the Part C chapters for which calibration factors need to be derived. The Part C SPFs
for each of these facility types are to be calibrated before use, but HSM users may choose not to calibrate the SPFs
for particular facility types if they do not plan to apply the Part C SPFs for those facility types.
APPENDIX A-SPECIALIZED PROCEDURES COMMON TO ALL PART C CHAPTERS A-3
Table A-1. SPFs in the Part C Predictive Models that Need Calibration
Calibration Factor to be Derived
Facility, Segment, or Intersection Type Symbol Equation Number(s)
ROADWAY SEGMENTS
Rural Two-Lane, Two-Way Roads
Two-lane undivided segments c, 10-2
Rural Multilane Highways

Undivided segments c, 11-2
Divided segments c, 11-3
Urban and Suburban Arterials
Two-lane undivided segments 12-2
Three-lane segments with center two-way left-twn lane 12-2
Four-lane undivided segments 12-2

Four-lane divided segments c, ··----·--
12-2
Five-lane segments with center two-way left-tum lane 12-2

INTERSECTIONS
Rural Two-Lane, Two-Way Roads·························

Three-leg intersections with minor-road stop control 10-3
10-3
Four-leg signalized intersections c, 10-3
Rural Multilane Highways
Three-leg intersections with minor-road stop control c, 11-4
Four-leg intersections with minor-road stop control c, 11-4
Urban and Suburban Arterials
Three-leg intersections with minor-road stop control 12-5
Three-leg signalized intersections 12-5
_.:.::.:. :. . .:.:.~ . ::e::.:. .::::_:_.::.::.::_:::::::. . -···-··-·- . -· · · · · · - --- ··-·······-- -- -········· ---··--··-- - --- -- --···-··-·'·'·····-·- ---- -- ---- - .... -- --- .. -- --- -----.
Four-leg intersections with minor-road stop control 12-5
A.l.l.2. Step 2-Select sites for calibration of the SPF for each facility type.
For each facility type, the desirable minimum sample size for the calibration data set is 30 to 50 sites, with each
site long enough to adequately represent physical and safety conditions for the facility. Calibration sites should be
selected without regard to the number of crashes on individual sites; in other words, calibration sites should not be
selected to intentionally limit the calibration data set to include only sites with either high or low crash frequencies.
Where practical, this may be accomplished by selecting calibration sites randomly from a larger set of candidate
sites. Following site selection, the entire group of calibration sites should represent a total of at least 100 crashes
per year. These calibration sites will be either roadway segments or intersections, as appropriate to the facility type
being addressed. If the required data discussed in Step 3 are readily available for a larger number of sites, that larger
nuinber of sites should be used for calibration. If a jurisdiction has fewer than 30 sites for a particular facility type,
then it is desirable to use all of those available sites for calibration. For large jurisdictions, such as entire states, with
a variety of topographical and climate conditions, it may be desirable to assemble a separate set of sites and develop
separate calibration factors for each specific terrain type or geographical region. For example, a state with distinct
plains and mountains regions, or with distinct dry and wet regions, might choose to develop separate calibration
factors for those regions. On the other hand, a state that is relatively uniform in terrain and climate might choose to
perfonn a single calibration for the entire state. Where separate calibration factors are developed by terrain type or
region, this needs to be done consistently for all applicable facility types in those regions.
It is desirable that the calibration sites for each facility type be reasonably representative of the range of site
characteristics to which the predictive model will be applied. However, no fonnal stratification by traffic volume or
other site characteristics is needed in selecting the calibration sites, so the sites can be selected in a marmer to make
the data collection needed for Step 3 as efficient as practical. There is no need to develop a new data set if an existing
data set with sites suitable for calibration is already available. If no existing data set is available so that a calibration
data set consisting entirely of new data needs to be developed, or if some new sites need to be chosen to supplement
an existing data set, it is desirable to choose the new calibration sites by random selection from among all sites of the
applicable facility type.
Step 2 only needs to be perfonned the first time that calibration is performed for a given facility type. For calibration
in subsequent years, the same sites may be used again.
A.l.13. Step 3-0btain data for each facility type applicable to a specific calibration period.
Once the calibration sites have been selected, the next step is to assemble the calibration data set if a suitable data set
is not already available. For each site in the calibration data set, the calibration data set should include:
• Total observed crash frequency for a period of one or more years in duration.
• All site characteristics data needed to apply the applicable Part C predictive model.
Observed crashes for all severity levels should be included in calibration. The duration of crash frequency data
should correspond to the period for which the resulting calibration factor, C, or C;, will be applied in the Part C
predictive models. Thus, if an armual calibration factor is being developed, the duration of the calibration period
should include just that one year. If the resulting calibration factor will be employed for two or three years, the
duration of the calibration period should include only those years. Since crash frequency is likely to change over
time, calibration periods longer than three years are not recommended. All calibration periods should have durations
that are multiples of 12 months to avoid seasonal effects. For ease of application, it is recommended that the
calibration periods consist of one, two, or three full calendar years. It is recommended to use the same calibration
period for all sites, but exceptions may be made where necessary.
The observed crash data used for calibration should include all crashes related to each roadway segment or
intersection selected for the calibration data set. Crashes should be assigned to specific roadway segments or
intersections based on the guidelines presented below in SectionA.2.3.
Table A-2 identifies the site characteristics data that are needed to apply the Part C predictive models for each facility
type. The table classifies each data element as either required or desirable for the calibration procedure. Data for
each of the required elements are needed for calibration. If data for some required elements are not readily available,
it may be possible to select sites in Step 2 for which these data are available. For example, in calibrating the
predictive models for roadway segments on rural two-lane, two-way roads, if data on the radii of horizontal curves
are not readily available, the calibration data set could be limited to tangent roadways. Decisions of this type should
be made, as needed, to keep the effort required to assemble the calibration data set within reasonable bounds. For
the data elements identified in Table A-2 as desirable, but not required, it is recommended that actual data be used if
available, but assumptions are suggested in the table for application where data are not available.
Table A-2. Data Needs for Calibration of Part C Predictive Models by Facility Type I
Data Need I'

I
Chapter Data Element Required Desirable Default Assumption

II
ROADWAY SEGMENTS
Segment length X Need actual data
Average annual daily traffic (AADT) X Need actual data
Lengths of horizontal curves and tangents X Need actual data
Radii of horizontal curves X Need actual data
Presence of spiral transition for horizontal curves X Base default on agency design policy
Superelevation variance for horizontal curves X No superelevation variance
Percent X Base default on terrain"
Lane width X Need actual data

10-Rural Two- Shoulder type X Need actual data
Lane, Two-Way
Roads Shoulder width X Need actual data
Presence of lighting X Assume no lighting
Driveway density
_._,_,,_,_,_,_,_,,_, ___ ,_,,_,_,_,_,_ .. _,_, __ ,_,_,_,_,_,_,,_,_,_,_,_, __ X Assume 5 driveways per mile
,_,_,_,,_,_,_,_ .. ,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,, ___ ,_,_,_,_,,_,_,_,,_,,_,_,_,,
Presence of passing lane X Assume not present
Presence of short four-lane section X Assume not present
Presence of center two-way left-tum lane X Need actual data
Presence of centerline rumble strip X Base default on agency design policy
Roadside hazard rating X Assume roadside hazard rating = 3
Use of automated speed enforcement X Base default on current practice
For all rural multilane highways:

Average annual daily traffic (AADT) X Need actual data
Lane'Nidth X Need actual data
Shoulder width X Need actual data

11-Rural
Multilane Presence of lighting X Assume no lighting
Highways
Use of automated speed enforcement X Base default on current practice
For undivided highways only:

Sideslope X Need actual data
For divided hlghways only:
Median width X Need actual data
TableA-2. Continued on next page

Table A-2. Data Needs for Calibration of Part C Predictive Models by Facility Type continued
Data Need
Chapter Data Element Required Desirable Default Assumption
Number of through traffic lanes X Need actual data
Presence of median X Need actual data
Presence of center two~ way left-tum lane X Need actual data
Average annual daily lra:ffic (AADT) X Need actual data
12-Urban Number of driveways by land-use type X Need actual datab

and Suburban Low-speed vs. intermediate or high speed X Need actual data
Arterials
Presence of on-street parking X Need actual data
Type of on-street parking X Need actual data
Roadside fixed object density X database default on fixed-object

offset and density categoriesc
Presence of lighting X Base default on agency practice
Presence of automated speed enforcement X Base default on agency practice

INTERSECTIONS
Number of intersection legs X Need actual data
Type of traffic control X Need actual data
Average annual daily traffic (AADT) for major road X Need actual data
10-Rural Two-
Lane. Tm-Way
Average daily traffic (AADT) for minor road.. ,_,_,_, ________
X __ _____.____ _, ___ __
Need actual data or best estimate
, , , .. ,_,_
-
-
-
-
-
-
·
·
-
·
~
-
-
-
-
-
-
-
-
-
-
Intersection skew angle X Assume no ske~
Roads .... ---· - ......- .. -·-·- .. ,-.. ........
_,_,_,_,,_,,_,_ ..
,_,_,_,_,_,,_,_,_,_,_,_ _,_,,_,,_ .. _,_ ,_,_.._,_,_,_.. ..
,_,_,_,_,_,,_,_,_,_,_,,_,_,._,_,,_,_,_ _,_,,_,_,_,_,_,_,_,_,_,,_,_, _,_,_..,_.._,_,_.. ,_,_,_,_ .. ,_,_
Number of approaches with left-twn lanes X Need actual data
Number of approaches with right-twn lanes X Need actual data
Presence of lighting X Need actual data
For all rural multilane highways:

·~---·-·-·---·---
annual daily traffic (AADT) for major road X Need actual data
11-Rural - --- -~- -·---- - 0 •• - - " _,_ ---'---·-···-··:: __ • --- -- --------- -··- - -·" -- ----- ---- ··---·· -- -- - -- " - " -- -- --- "
Multilane Average annual daily traffic (AADT) for minor road X Need actual data or best estrmate
Highways
Presence of lighting X Need actual datad
Intersection skew angle X Assume no skew
Number of approaches with left-tum lanes X Need acrual data
Number of approaches with right-tum lanes X Need actual data
Table A-2. Continued on next page

Table A-2- Data Needs for Calibration of Part C Predictive Models by Facility Type continued
Data Need
Chapter Data Element Required Desirable DefaultAssumption
For all intersections on arterials:

Average annual daily traffic (AADT) for major road X Need actual data
Average annual daily traffic (AADT) for minor road X Need actual data or best estimate
Number of approaches with left-tum lanes X Need actual data
Number of approaches with right-turn lanes X Need actual data

Presence of lighting X Need actual data
--~--~~------------------------------
For signali::ed intersections only:
12~Urban
Presence of left-tum phasing X Need actual data
and Suburban
Arterials Prefer actual data, but agency
Type ofleft-turn phasing X
practice may be used as a default
-----'-- ~---
Use of right-tum-on-red signal operation X Need actual data
Use of red-light cameras X Need actual data
Pedestrian volume X Estimate with Table 12-21
Maximum number oflanes crossed by pedestrians Estimate from number of lanes and
X
on any approach presence of median on major road
Presence of bus stops within 1,000 ft X Assume not present

Presence of schools within 1,000 ft X Assume not present
Presence of alcohol sales establishments
X Assume not present
within LOOO ft
• Suggested default values for calibration purposes: CMF = 1.00 for level terrain; CMF = 1.06 for rolling terrain; CMF = 1.14 for mountainous terrain
b Use actual data for number of driveways, but simplified land-use categories may be used {e.g., commercial and residential only).
< CMFs may be estimated based on two categories of fixed-object offset {0 )-either 5 or 20ft-and three categories of fixed-object density
1
(D,.,)--0, so, or 100 objects per mile.
d If measurements of intersection skew angles are not available, the calibration should preferably be performed for intersections with no skew.
A.Ll-4. Step 4-Apply the applicable Part C predictive method to predict total crash frequency for each site
during the calibration period as a whole
The site characteristics data assembled in Step 3 should be used to apply the applicable predictive method from
Chapter 10, 11, or 12 to each site in the calibration data set. For this application, the predictive method should be
applied without using the EB Method and, of course, without employing a calibration factor (i.e., a calibration factor
of 1.00 is assumed). Using the predictive models, the expected average crash frequency is obtained for either one,
two, or three years, depending on the duration of the calibration period selected.
A.Ll-5. Step 5-Compnte calibration factors for nse in Part C predictive models
The final step is to compute the calibration factor as:
2: observed crashes
C (or C.) = -c";;'';;;"'o.'---:-:--:----
' ' 2: predicted crashes
all sites
(A-1)
The computation is performed separately for each facility type. The computed calibration factor is rounded to two
decimal places for application in the appropriate Part C predictive model.
Example Calibration Factor Calculation
The SPF for four-leg signalized intersections on rural two-lane, two-way roads from Equation 10-18 1S:
Nspf int ;;;: e[-5.73 + 0.60 )( ln(AADTm,) +0.20 "ln(A4DTmJ,)J
Where:
Nspfint
~ predicted number of total intersection-related crashes per year for base conditions;
AADTm<~J ~ average annual daily entering traffic volumes (vehicles/day) on the major road; and
AADTmin ~ average annual daily entering traffic volumes (vehicles/day) on the minor road.
The base conditiOns are:
• No left-turn lanes on any approach
• No nght-turn lanes on any approach
The CMF values from Chapter 10 are:
• CMF for one approach with a left-turn lane~ 0.82
• CMF for one approach with a right-turn lane~ 0.96
• CMF for two approaches with right-turn lanes~ 0.92
• No lighting present (so l1ghting CMF ~ 1.00 for all cases)
Typical data for erght intersections is shown in an example calculation shown below. Note that for an actual calibration,
the recommended minimum sample size would be 30 to 50 sites that experience at least 100 crashes per year. Thus, the
number of s1tes used here is smaller than recommended, and is intended solely to illustrate the calculations.
For the first intersection ·,n the example the predicted crash frequency for base conditions is:
Nbibase ;;;; e(-5,73 + 0.60 X ln(4000) + 0.20 X ln(2000)) ::; 2.1 52 crashes/year
The intersection has a left-turn lane on the major road, for which CMF,; is 0.67, and a right-turn lane on one approach,
a feature for which CMF, is 0.98. There are three years of data, during which four crashes were observed (shown in
Column 10 of Table Ex-1 ). The predicted average crash frequency from the Chapter 10 for this intersection without
calibration is from Equalion 10-2:
N,, ~ (N,,,,,) x (CMF,;) x (CMF,) x (number of years of data)
~ 2.152 x 0.67 x 0.98 x 3 ~ 4.240 crashes in three years, shown in Column 9.
Similar calculations were done for each intersection in the table shown below. The sum of the observed crash frequenCies
in Column 10 (43) is divided by the sum of the predicted average crash frequencies in Column 9 (45.594) to obtain the
calibration factor, C,. equal to 0.943. It is recommended that calibration factors be rounded to two decimal places, so
cal'rbralion factor equal to 0.94 should be used ·,n the Chapter 10 predictive model for four-leg signalized intersections.
Table Ex-1. Example of Calibration Factor Computation

2 3 4 5 6 7 8 9 10
Intersection Intersection Predicted
Approaches Approaches Years Average Observed
SPF with with of Crash Crash
AADTma! AADTmln Prediction Left-Turn Lanes CMF11 Right-Turn Lane CMF2; Data Frequency Frequency
4000 2000 2.152 0.67 0.98
_,_,,_ .. ..
,, ,_,,_,_,,_,,_,_,
3__,- 4.240 4
_,_,,_,_,,""""""'"""
3000 1500 1.710 0 1.00 2 0.95 2 3.249 5

- - ·······-
5000 3400 2.736 0 1.00 2 0.95 ,__
- _,_,_,_ ,_,_,_,,
3
...........
7.799 10
6500 3000 3.124 0 1.00 2 0.95 3 8.902 5
3600 2300 2.078 0.67 0.98 4.093 2
4600 2.753 4500 0 1.00 2 0.95 3 7.846 8
------ ---------
5700 3300 2.943 0.67 0.98 3 5.796 5
- _,,_,_,_,,_, - .. -
6800 1500 2.794 0.67 0.98 2 3.669 4
Sum 45.594 43
Calibration Factor (C) 0.943
A.1.2. Development of Jurisdiction-Specific Safety Performance Functions for Use in the Part C
Predictive Method
Satisfactory results from the Part C predictive method can be obtained by calibrating the predictive model for each
facility type. as explained in Section A.l.l. However, some users may prefer to develop jurisdiction-specific SPFs
using their agency's own data, and this is likely to enhance the reliability of the Part C predictive method. While
there is no requirement that this be done, HSM users are welcome to use local data to develop their own SPFs, or
if they wish, replace some SPFs with jurisdiction-specific models and retain other SPFs from the Part C chapters.
Within the first two to three years after a jurisdiction-specific SPF is developed, calibration of the jurisdiction-
specific SPF using the procedure presented in Section A.l.l may not be necessary, particularly if other default values
in the Part C models are replaced with locally-derived values, as explained in Section A.l.3.
If jurisdiction-specific SPFs are used in the Part C predictive method, they need to be developed with methods that
are statistically valid and developed in such a manner that they fit into the applicable Part C predictive method.
The following guidelines for development of jurisdiction-specific SPFs that are acceptable for use in HSM Part C
include:
• In preparing the crash data to be used for development of jurisdiction-specific SPFs, crashes are assigned to
roadway segments and intersections following the definitions explained in Section A.2.3 and illustrated in
Figure A-1.
• The jurisdiction-specific SPF should be developed with a statistical technique such as negative binontial regression
that accounts for the overdispersion typically found in crash data and quantifies an overdispersion parameter so
that the model's predictions can be combined with observed crash frequency data using the EB Method.
• The jurisdiction-specific SPF should use the same base conditions as the corresponding SPF in Part C or should be
capable of being converted to those base conditions.
• The jurisdiction-specific SPF should include the effects of the following traffic volumes: average annual
daily traffic volume for roadway segment and major- and minor-road average annual daily traffic volumes for
intersections.
• The jurisdiction-specific SPF for any roadway segment facility type should have a functional form in which
predicted average crash frequency is directly proportional to segment length.
These guidelines are not intended to stifle creativity and innovation in model development. However, a model that
does not account for overdispersed data or that cannot be integrated with the rest of the Part C predictive method will
not be useful.
Two types of data sets may be used for SPF development. First, SPFs may be developed using only data that
represent the base conditions, which are defined for each SPF in Chapters 10, II, and 12. Second, it is also
acceptable to develop models using data for a broader set of conditions than the base conditions. In this approach,
all variables that are part of the applicable base-condition definition, but have non-base-condition values, should be
included in an initial model. Then, the initial model should be made applicable to the base conditions by substituting
values that correspond to those base conditions into the model. Several examples of this process are presented in
Appendix A to Chapter I 0.
A.1.3. Replacement of Selected Default Values in the Part C Predictive Models to local Conditions
The Part C predictive models use many default values that have been derived from crash data in HSM-related research.
For example, the urban intersection predictive model in Chapter 12 uses pedestrian factors that are based on the
proportion of pedestrian crashes compared to total crashes. Replacing these default values with locally derived values
will improve the reliability of the Part C predictive models. TableA-3 identifies the specific tables in Part C that may
be replaced with locally derived values. In addition to these tables, there is one equation-Equation I 0-18-which
uses constant values given in the accompanying text in Chapter I 0. These constant values may be replaced with locally
derived values.
Providing locally-derived values for the data elements identified in Table A-3 is optional. Satisfactory results can be
obtained with the Part C predictive models, as they stand, when the predictive model for each facility type is calibrated
with the procedure given in Section A.l.l. But, more reliable results may be obtained by updating the data elements
listed in Table A-3. It is acceptable to replace some, but not all of these data elements, if data to replace all of them
are not available. Each element that is updated with locally-derived values should provide a small improvement in the
reliability ofthat specific predictive model. To preserve the integrity of the Part C predictive method, the quantitative
values in the predictive models, (other than those listed in Table A-3 and those discussed in Sections A.l.l andA.2.2),
should not be modified. Any replacement values derived with the procedures presented in this section should be
incorporated in the predictive models before the calibration described in Section A. I.! is performed.
Table A-3. Default Crash Distributions Used in Part C Predictive Models Which May Be Calibrated by Users to
Local Conditions
Type of Roadway Element
Table or
Equation Roadway Data Element or Distribution That May Be
Chapter Number Segments Intersections Calibrated to Local Conditions
Table 10-3 X Crash severity by facility type for roadway segments
Table 10-4 X Collision type by facility type for roadway segments
Table 10-5 X Crash severity by facility type for intersections

10---Rural Two-
Table 10-6 X Collision type by facility type for intersections
Lane, Two-Way -··-············· --
Roads Equation 10-18 X Driveway-related crashes as a proportion of total crashes
------'''·'-----
Table 10-12 X Nighttime crashes as a proportion of total crashes by severity level
Nighttime crashes as a proportion of total crashes by severity level

Table 10-15 X
and by intersection type
Table 11-4 X Crash severity and collision type for undivided segments
Table 11-6 X Crash severity and collision type for divided segments
Table 11-9 X Crash severity and collision type by intersection

11-Rural Nighttime crashes as a proportion of total crashes by severity level
Multilane Table 11-15 X
and by roadway segment type for undivided roadway segments
Highways ---- --·--- ---·-

Table 11-19 X
and by roadway segment type for divided roadway segments
- - ·- ,_,_,,_,,_,,_,,_,,_
Table 11-24 X
Crash severity and collision type for multiple-vehicle nondriveway

Table 12-4 X
collisions by roadway segment type
----------------------- ·--
Crash severity and collision type for single-vehicle crashes by
Table 12-6 X
roadway segment type
...... :.:...............................• -
Table 12-7 X Crash severity for driveway-related collisions by roadway segment type"
--~---------
Table 12-8 X Pedestrian crash adjustment factor by roadway segment type
Table 12-9 X Bicycle crash adjustment factor by roadway segment type
Crash severity and collision type for multiple-vehicle collisions by

12-Urban Table 12-11 X
intersection type
and Suburban - - - ----- -
Arterials Crash severity and collision type for single-vehicle crashes by
Table 12-13 X
intersection type
Pedestrian crash adjustment factor by intersection type for stop-

Table 12-16 X
controlled intersections
Table 12-17 X Bicycle crash adjustment factor by intersection type

Table 12-23 X
and by roadway segment type

Table 12-27 X
• The on!y portion of Table 12-7 that should be modified by the user are the crash severity proportions.
Note: No quantitative values in the Part C predictive models, other than those listed here and those discussed in Sections A.1.1 and A.1.2,
should be modified by HSM users.
Procedures for developing replacement values for each data element identified in Table A-3 are presented below.
Most of the data elements to be replaced are proportions of crash severity levels and/or crash types that are part of a
specific distribution. Each replacement value for a given facility type should be derived from data for a set of sites
that, as a group, includes at least 100 crashes and preferably more. The duration of the study period for a given set
of sites may be as long as necessary to include at least 100 crashes. In the following discussion, the term "sufficient
data" refers to a data set including a sufficient number of sites to meet this criterion for total crashes. In a few cases,
explicitly identified below, the definition of sufficient data will be expressed in terms of a crash category other
than total crashes. In assembling data for developing replacements for default values, crashes are to be assigned to
specific roadway segments or intersections following the definitions explained in Section A.2.3 and illustrated in
Figure A-!.
A.1.3.1. Replacement of Default Values for Rural Two-Lane, Two-Way Roads

Five specific sets of default values for rural two-lane, two-way roads may be updated with locally-derived
replacement values by HSM users. Procedures to develop each of these replacement values are presented below.
Crash Severity by Facility Type

Tables 10-3 and 10-5 present the distribution of crashes by five crash severity levels for roadway segments and
intersections, respectively, on rural two-lane, two-way roads. If sufficient data, including these five severity levels
(fatal, incapacitating injury, nonincapacitating injury, possible injury, and property damage only), are available
for a given facility type, the values in Tables I 0-3 and I 0-5 for that facility type may be updated. If sufficient
data are available only for the three standard crash severity levels (fatal, injury, and property damage only), the
existing values in Tables 10-3 and 10-5 may be used to allocate the injury crashes to specific injury severity levels
(incapacitating injury, nonincapacitating injury, and possible injury).
Collision Type by Facility Type

Table 10-4 presents the distribution of crashes by collision type for seven specific types of single-vehicle crashes
and six specific types of multiple-vehicle crashes for roadway segments, and Table 10-6 presents the distribution of
crashes by collision type for three intersection types on rural two-lane, two-way roads. If sufficient data are available
for a given facility type, the values in Tables 10-4 and I 0-6 for that facility type may be updated.
Driveway-Related Crashes as a Proportion of Total Crashes for Roadway Segments

Equation 10-18 includes a factor, pd..,' which represents the proportion of total crashes represented by driveway-
related crashes. A value for pd..., based on research is presented in the accompanying text. This value may be replaced
with a locally-derived value, if data are available for a set for sites that, as a group, have experienced at least 100
driveway-related crashes.
Nighttime Crashes as a Proportion of Total Crashes for Roadway Segments

Table 10-12 presents the proportions of total nighttime crashes by severity level and the proportion of total
crashes that occur at night for roadway segments on rural two-lane, two-way roads. These values may be replaced
with locally-derived values for a given facility type, if data are available for a set of sites that, as a group, have
experienced at least 100 nighttime crashes.
Nighttime Crashes as a Proportion of Total Crashes for Intersections

Table 10-15 presents the proportion of total crashes that occur at night for intersections on rural two-lane, two-way
roads. These values may be replaced with locally-derived values for a given facility type, if data are available for a
set of sites that, as a group, have experienced at least 100 nighttime crashes.
A.I.3.2. Replacement of Default Values for Rural Multilane Highways

Five specific sets of default values for rural multilane highways may be updated with locally-derived replacement
values by HSM users. Procedures to develop each of these replacement values are presented below.
Crash Severity and Collision Type for Undivided Roadway Segments

Table 11-4 presents the combined distribution of crashes for four crash severity levels and six collision types. If
sufficient data are available for undivided roadway segments, the values in Table 11-4 for this facility type may be
updated. Given that this is a joint distribution of two variables, sufficient data for this application requires a set of sites
of a given type that, as a group, have experienced at least 200 crashes in the time period for which data are available.
Crash Severity and Collision Type for Divided Roadway Segments

Table ll-6 presents the combined distribution of crashes for four crash severity levels and six collision types. If
sufficient data are available for divided roadway segments, the values in Table 11-6 for this facility type may be
updated. Given that this is a joint distribution of two variables, sufficient data for this application requires sites that
have experienced at least 200 crashes in the time period for which data are available.
Crash Severity and Collision Type by Intersection Type

Table 11-9 presents the combined distribution of crashes at intersections for four crash severity levels and six
collision types. If sufficient data are available for a given intersection type, the values in Table 11-9 for that
intersection type may be updated. Given that this is a joint distribution of two variables, sufficient data for this
application requires a set of sites of a given type that, as a group, have experienced at least 200 crashes in the time
period for which data are available.

Tables 11-15 and 11-19 present the proportions of total nighttime crashes by severity level and the proportion of total
crashes that occur at night for undivided and divided roadway segments, respectively, on rural multilane highways.
These values may be replaced with locally-derived values for a given facility type, if data are available for a set of
sites that, as a group, have experienced at least I 00 nighttime crashes.

Table 11-24 presents the proportion of total crashes that occur at night for intersections on rural multilane highways.
These values may be replaced with locally-derived values for a given facility type, if data are available for a set of
sites that, as a group, have experienced at least I 00 nighttime crashes.
A.l.3.3. Replacement of Default Values for Urban and Suburban Arterials

Eleven specific sets of default values for urban and suburban arterial highways may be updated with locally-derived
replacement values by HSM users. Procedures to develop each of these replacement values are presented below.
Crash Severity and Collision Type for Multiple- Vt!hicle Nondriveway Crashes by Roadway Segment Type
Table 12-4 presents the combined distribution of crashes for two crash severity levels and six collision types. If
sufficient data are available for a given facility type, the values in Table 12-4 for that facility type may be updated.
Given that this is a joint distribution of two variables, sufficient data for this application requires a set of sites of a
given type that, as a group, have experienced at least 200 crashes in the time period for which data are available.
Crash Severity and Collision Type for Single- Vehicle Crashes by Roadway Segment Type
Crash Severity for Driveway-Related Collision by Roadway Segment Type

Table 12-7 includes data on the proportions of driveway-related crashes for two crash severity levels (fatal-and-
injury and property-damage-only crashes) by facility type for roadway segments. If sufficient data are available for a
given facility type, these specific severity-related values in Table 12-7 for that facility type may be updated. The rest
ofTable 12-7, other than the last two rows of data which are related to crash severity, should not be modified.
Pedestrian Crash Adjustment Factor by Roadway Segment Type

Table 12-8 presents a pedestrian crash adjustment factor for specific roadway segment facility types and for two
speed categories: low speed (traffic speeds or posted speed limits of30 mph or less) and intermediate or high
speed (traffic speeds or posted speed limits greater than 30 mph). For a given facility type and speed category, the
pedestrian crash adjustment factor is computed as:
(A-2)
Where:
J' pedr ~ pedestrian crash ad;ustment

~
factor;
KP'' observed vehicle-pedestrian crash frequency; and

K,, observed frequency for all crashes not including vehicle-pedestrian and vehicle-bicycle crash.
The pedestrian crash adjustment factor for a given facility type should be determined with a set of sites of that speed
type that, as a group, includes at least 20 vehicle-pedestrian collisions.
Bicycle Crash Adjustment Factor by Roadway Segment Type

Table 12-9 presents a bicycle crash adjustment factor for specific roadway segment facility types and for two speed
categories: low speed (traffic speeds or posted speed limits of 30 mph or less) and intermediate or high speed (traffic
speeds or posted speed limits greater than 30 mph). For a given facility type and speed category, the bicycle crash
adjustment factor is computed as:
K bd<e
.{"
lbiker -K
-
non (A-3)
Vlhere:
J;;,., ~ bicycle crash adjustment factor;

K,;,. observed vehicle-bicycle crash frequency; and
K,,, observed frequency for all crashes not including vehicle-pedestrian and vehicle-bicycle crashes.
The bicycle crash adjustment factor for a given facility type should be determined with a set of sites of that speed
type that, as a group, includes at least 20 vehicle-bicycle collisions.
Crash Severity and Collision Type for Multiple-Vehicle Crashes by Intersection Type
Crash Severity and Collision Type for Single- Vehicle Crashes by Intersection Type
Pedestrian Crash Adjustment Factor by Intersection Type

Table 12-16 presents a pedestrian crash adjustment factor for two specific types of intersections with stop control
on the minor road. For a given facility type and speed category, the pedestrian crash adjustment factor is computed
using Equation A-2. The pedestrian crash adjustment factor for a given facility type is determined with a set of sites
that, as a group, have experienced at least 20 vehicle-pedestrian collisions.
Bicycle Crash Adjustment Factor by Intersection Type

Table 12-17 presents a bicycle crash adjustment factor for four specific intersection facility types. For a given facility
type and speed category, the bicycle crash adjustment factor is computed using Equation A-3. The bicycle crash
adjustment factor for a given facility type is determined with a set of sites that, as a group, have experienced at least
20 vehicle-bicycle collisions.

Table 12-23 presents the proportions of total nighttime crashes by severity level for specific facility types for
roadway segments and the proportion of total crashes that occur at night. These values may be replaced with locally-
derived values for a given facility type, if data are available for a set of sites that, as a group, have experienced at
least I 00 nighttime crashes.

Table 12-27 presents the proportions of total nighttime crashes by severity level for specific facility types for
intersections and the proportion of total crashes that occur at night. These values may be replaced with locally-
derived values for a given facility type, if data are available for a set of sites that, as a group, have experienced at
least 100 nighttime crashes.
A.2. USE OF THE EMPIRICAL BAYES METHOD TO COMBINE PREDICTED AVERAGE CRASH
FREQUENCY AND OBSERVED CRASH FREQUENCY
Application of the EB Method provides a method to combine the estimate using a Part C predictive model and
observed crash frequencies to obtain a more reliable estimate of expected average crash frequency. The EB Method
is a key tool to compensate for the potential bias due to regression-to-the-mean. Crash frequencies vary naturally
from one time period to the next. When a site has a higher than average frequency for a particular time period, the
site is likely to have lower crash frequency in subsequent time periods. Statistical methods can help to assure that
this natural decrease in crash frequency following a high observed value is not mistaken for the effect of a project or
for a true shift in the long-term expected crash frequency.
There are several statistical methods that can be employed to compensate for regression-to-the-mean. The EB
Method is used in the HSM because it is best suited to the context of the HSM. The Part C predictive models include
negative binomial regression models that were developed before the publication of the HSM by researchers who
had no data on the specific sites to which HSM users would later apply those predictive models. The HSM users
are generally engineers and planners, without formal statistical training, who would not generally be capable of
developing custom models for each set of the sites they wish to apply the HSM to and, even if there were, would
have no wish to spend the time and effort needed for model development each time they apply the HSM. The EB
Method provides the most snitable tool for compensating for regression-to-the-mean that works in this context.
Each of the Part C chapters presents a four-step process for applying the EB Method. The EB Method assumes
that the appropriate Part C predictive model (see Section 10.3.1 for rural two-lane, two-way roads, Section 11.3.1
for rural multilane highways, or Section 12.3.1 for urban and suburban arterials) has been applied to determine the
predicted crash frequency for the sites that make up a particular project or facility for a particular past time period of
interest. The steps in applying the EB Method are:
• Determine whether the EB Method is applicable, as explained in Section A.2.1.
• Determine whether observed crash frequency data are available for the project or facility for the time period for
which the predictive model was applied and, if so, obtain those crash frequency data, as explained in Section
A.2.2. Assign each crash instance to individual roadway segments and intersections, as explained in SectionA.2.3.
• Apply the EB Method to estimate the expected crash frequency by combining the predicted and observed crash
frequencies for the time period of interest. The site-specific EB Method, applicable when observed crash frequency
data are available for the individual roadway segments and intersections that make up a project or facility, is
presented in SectionA.2.4. The project-level EB Method, applicable when observed crash frequency data are
available only for the project or facility as a whole, is presented in SectionA.2.5.
• Adjust the estimated value of expected crash frequency to a future time period, if appropriate, as explained in
Section A.2.6.
Consideration of observed crash history data in the Part C predictive method increases the reliability of the estimate
of the expected crash frequencies. When at least two years of observed crash history data are available for the facility
or project being evaluated, and when the facility or project meets certain criteria discussed below, the observed crash
data should be used. When considering observed crash history data, the procedure must consider both the existing
geometric design and traffic control for the facility or project (i.e., the conditions that existed during the before
period while the observed crash history was accumulated) and the proposed geometric design and traffic control for
the project (i.e., the conditions that will exist during the after period, the period for which crash predictions are being
made). In estimating the expected crash frequency for an existing arterial facility in a future time period where no
improvement project is planned, only the traffic volumes should differ between the before and after periods. For an
arterial on which an improvement project is planned, traffic volumes, geometric design features, and traffic control
features may all change between the before and after periods. The EB Method presented below provides a method to
combine predicted and observed crash frequencies.
A.2.1 Determine whether the EB Method is Applicable

The applicability of the EB Method to a particular project or facility depends on the type of analysis being
performed and the type of future project work that is anticipated, If the analysis is being performed to assess the
expected average crash frequency of a specific highway facility, but is not part of the analysis of a planned future
project, then the EB Method should be applied. If a future project is being planned, then the nature of that future
project should be considered in deciding whether to apply the EB Method.
The EB Method should be applied for the analyses involving the following future project types:
• Sites at which the roadway geometries and traffic control are not being changed (e.g., the "do-nothing"
alternative);
• Projects in which the roadway cross section is modified but the basic number of through lanes remains the
same (This would include, for example, projects for which lanes or shoulders were widened or the roadside was
improved, but the roadway remained a rural two-lane highway);
• Projects in which minor changes in alignment are made, such as flattening individual horizontal curves while
leaving most of the alignment intact;
• Projects in which a passing lane or a short four-lane section is added to a rural two-lane, two-way road to increase
passing opportunities; and
• Any combination of the above improvements.
The EB Method is not applicable to the following types of improvements:

• Projects in which a new alignment is developed for a substantial proportion of the project length; and
• Intersections at which the basic number of intersection legs or type of traffic control is changed as part of a project
The reason that the EB Method is not used for these project types is that the observed crash data for a previous time
period is not necessarily indicative of the crash experience that is likely to occur in the future after such a major
geometric improvement Since, for these project types, the observed crash frequency for the existing design is not
relevant to estimation of the future crash frequencies for the site, the EB Method is not needed and should not be
applied. If the EB Method is applied to individual roadway segments and intersections, and some roadway segments
and intersections within the project limits will not be affected by the major geometric improvement, it is acceptable
to apply the EB Method to those unaffected segments and intersections.
If the EB Method is not applicable, do not proceed to the remaining steps. Instead, follow the procedure described in
the Applications section of the applicable Part C chapter.
APPENDIX A-SPECIALIZED PROCEDURES COMMON TO ALL PART C CHAPTERS
A.2.2. Determine whether Observed Crash Frequency Data are Available for the Project or Facility
and, if so, Obtain those Data
If the EB Method is applicable, it should be determined whether observed crash frequency data are available
for the project or facility of interest directly from the jurisdiction's crash record system or indirectly from
another source. At least two years of observed crash frequency data are desirable to apply the EB Method. The
best results in applying the EB Method will be obtained if observed crash frequency data are available for each
individual roadway segment and intersection that makes up the project of interest. The EB Method applicable
to this situation is presented in Section A.2.4. Criteria for assigning crashes to individual roadway segments
and intersections are presented in Section A.2.3. If observed crash frequency data are not available for
individual roadway segments and intersections, the EB Method can still be applied if observed crash frequency
data are available for the project or facility as a whole. The EB Method applicable to this situation is presented
in Section A.2.5.
If appropriate crash frequency data are not available, do not proceed to the remaining steps. Instead, follow the
procedure described in the Applications section of the applicable Part C chapter.
A.2.3. Assign Crashes to Individual Roadway Segments and Intersections for Use in the EB Method
The Part C predictive method has been developed to estimate crash frequencies separately for intersections and
roadways segments. In the site-specific EB Method presented in section A.2.4, observed crashes are combined
with the predictive model estimate of crash frequency to provide a more reliable estimate of the expected average
crash frequency of a particular site. In Step 6 of the predictive method, if the site-specific EB Method is applicable,
observed crashes are assigned to each individual site identified within the facility of interest. Because the predictive
models estimate crashes separately for intersections and roadway segments, which may physically overall in some
cases, observed crashes are differentiated and assigned as either intersection related crashes or roadway segment
related crashes.
Intersection crashes include crashes that occur at an intersection (i.e., within the curb limits) and crashes that occur
on the intersection legs and are intersection-related. All crashes that are not classified as intersection or intersection-
related crashes are considered to be roadway segment crashes. Figure A-1 illustrates the method used to assign
crashes to roadway segments or intersections. As shown:
• All crashes that occur within the curbline limits of an intersection (Region A in the figure) are assigned to that
intersection.
• Crashes that occur outside the curbline limits of an intersection (Region B in the figure) are assigned to either
the roadway segment on which they occur or an intersection, depending on their characteristics. Crashes that are
classified on the crash report as intersection-related or have characteristics consistent with an intersection-related
crash are assigned to the intersection to which they are related; such crashes would include rear-end collisions
related to queues on an intersection approach. Crashes that occur between intersections and are not.related to an
intersection, such as collisions related to turning maneuvers at driveways, are assigned to the roadway segment on
which they occur.
Segment Length
(center of intersection to center of ·mtersection)
A All crashes that occur with1n this region are classified as intersection crashes.
B Crashes in this region may be segment or intersection related~ depending on

the characteristiCS of the crash.
Figure A-1. Definition of Roadway Segments and Intersections
In some jurisdictions, crash reports include a field that allows the reporting officer to designate the crash as
intersection-related. When this field is available on the crash reports, crashes should be assigned to the intersection
or the segment based on the way the officer marked the field on the report. In jurisdictions where there is not a field
on the crash report that allows the officer to designate crashes as intersection-related, the characteristics of the crash
may be considered to make a judgment as to whether the crash should be assigned to the intersection or the segment.
Other fields on the report, such as collision type, number of vehicles involved, contributing circumstances, weather
condition, pavement condition, traffic control malfunction, and sequence of events can provide helpful information
in making this determination.
If the officer's narrative and crash diagram are available to the user, they can also assist in making the determination.
The following crash characteristics may indicate that the crash was related to the intersection:
• Rear-end collision in which both vehicles were going straight approaching an intersection or in which one vehicle
was going straight and struck a stopped vehicle
• Collision in which the report indicates a signal malfunction or improper traffic control at the intersection
The following crash characteristics may indicate that the crash was not related to the intersection and should be
assigned to the segment on which it occurred:
• Collision related to a driveway or involving a turning movement not at an intersection
• Single-vehicle run-off-the-road or fixed object collision in which pavement surface condition was marked as wet
or icy and identified as a contributing factor
These examples are provided as guidance when an "intersection-related" field is not available on the crash report;
they are not strict rules for assigning crashes. Information on the crash report should be considered to help make the
determination, which will rely on judgment. The information needed for classifying crashes is whether each crash is,
or is not, related to an intersection. The consideration of crash type data is presented here only as an example of one
approach to making this determination.
Using these guidelines, the roadway segment predictive models estimate the average frequency of crashes that
would occur on the roadway if no intersection were present. The intersection predictive models estimate the average
frequency of additional crashes that occur because of the presence of an intersection.
A.2.4. Apply the Site-Specific EB Method

Equations A-4 and A-5 are used directly to estimate the expected crash frequency for a specific site by combining
the predictive model estimate with observed crash frequency. The value of Nexpected from Equation A-4 represents the
expected crash frequency for the same time period represented by the predicted and observed crash frequencies.
. N observed. and N expected all represent either total .crashes
N pred1cted' .
or a specific severity
. .
level or. collision type of
interest. The expected average crash frequency cons1denng both the prediCtive model estrmate and observed crash
frequencies for an individual roadway segment or intersection is computed as:
N expected ~ w x N predicted + (1- w) x N observed (A-4)
1
w = -------,=----
1+k X ( L
Np<edicred )
all study
(A-5)
Y'""
Where:
N~"'"d ~ estimate of expected average crashes frequency for the study period;
N,reru.red ~ predictive model estimate of average crash frequency predicted for the study period under the given
conditions;
Nobscrved observed crash frequency at the site over the study period;
w ~ weighted adjustlnent to be placed on the predictive model estimate; and
k = overdispersion parameter of the associated SPF used to estimate Npredicred·
When observed crash data by severity level is not available. the estimate of expected average crash frequency for
fatal-and-injury and property-damage-only crashes is calculated by applying the proportion of predicted average
crash frequency by severity level (Npredicrcd(FJ)/Npredicted(roml) and Nprcdicted(Pna/Npredicreci(tot:J.l)) to the total expected average
crash frequency from Equation A-4.
Equation A-5 shows an inverse relationship between the overdispersion parameter, k, and the weight, w. This implies
that when a model with little overdispersion is available; more reliance will be placed on the predictive model
estimate, Npredicted' and less reliance on the observed crash frequency, NobSI!rvoo· The opposite is also the case; when
a model with substantial overdispersion is available, less reliance will be placed on the predictive model estimate,
Npred!cted' and more reliance on the observed crash frequency, Nobserved"
It is important to note in Equation A-5 that, as Np,....,!cte

__ •. d increases. there is less weight placed on Npred.Jc""'
·-• and more
on N,b"~'d" This might seem counterintuitive at first. However, this implies that for longer sites and for longer study
periods, there are more opportunities for crashes to occur. Thus, the observed crash history is likely to be more
meaningful and the model prediction less important. So. as N,redio<od increases. the EB Method places more weight
on the number of crashes that actually occur, Nob•=·'· When few crashes are predicted, the observed crash frequency,
is not likely to be meaningful, in statistical terms, so greater reliance is placed on the predicted crash
Nobserved'
frequency, Npredicred"
The values of the overdispersion parameters, k, for the safety performance functions used in the predictive models
are presented with each SPF in Sections 10.6, 11.6. and 12.6.
Since application of the EB Method requires use of an overdispersion parameter, it cannot be applied to portions of
the prediction method where no overdispersion parameter is available. For example, vehicle-pedestrian and vehicle-
bicycle collisions are estimated in portions of Chapter 12 from adjustlnent factors rather than from models and
should, therefore, be excluded from the computations with the EB Method. Chapter 12 uses multiple models with
different overdispersion parameters in safety predictions for any specific roadway segment or intersection. Where
observed crash data are aggregated so that the corresponding value of predicted crash frequency is determined as the
sum of the results from multiple predictive models with differing overdispersion parameters, the project-level EB
Method presented in Section A.2.5 should be applied rather than the site-specific method presented here.
Chapters 10, II, and 12 each present worksheets that can be used to apply the site-specific EB Method as presented
in this section.
Section A.2.6 explains how to update N ~''"'' to a future time period, such as the time period when a proposed future
project will be implemented. This procedure is only applicable if the conditions of the proposed project will not be
substantially different from the roadway conditions during which the observed crash data was collected.
A.2.5. Apply the Project-Level EB Method

HSM users may not always have location specific information for observed crash data for the individual
roadway segments and intersections that make up a facility or project of interest. Alternative procedures are
available where observed crash frequency data are aggregated across several sites (e.g., for an entire facility
or project). This requires a more complex EB Method for two reasons. First, the overdispersion parameter, k,
in the denominator of Equation A-5 is not uniquely defined, because estimate of crash frequency from two or
more predictive models with different overdispersion parameters are combined. Second, it cannot be assumed,
as is normally done, that the expected average crash frequency for different site types are statistically correlated
with one another. Rather, an estimate of expected average crash frequency should be computed based on the
assumption that the various roadway segments and intersections are statistically independent (r = 0) and on the
alternative assumption that they are perfectly correlated (r = I). The expected average crash frequency is then
estimated as the average of the estimates for r = 0 and r = I.
The following equations implement this approach, summing the first three terms, which represent the three roadway-
segment-related crash types, over the five types of roadway segments considered in the (2U, 3T, 4U, 4D, 5T) and the
last two terms, which represent the two intersection-related crash types, over the four types of intersections (3ST,
3SG, 4ST, 4SG):
5 5 5 4 4
Npredicted (total) =L: N}=I

predicted nn; +L: N j=l
predicted rsj +L: N
]=1
predicted rdj +L: N
}""l
predicted imj +L: N
}'=1
predicted is}
(A-6)
5 5 5 4 4
Nobscrvcd (total) =L: }=I

Nobservedrmj +L: N j=I
observed rsj +L: N
j=l
observed rdj +L: N
j=l
observed vnj +L:
j=l
Nobservedisj
(A-7)
5 5 5
Npredicted wO =L: }=I

krmj N?m; +L: krsj N?."} +L: k,dj N;dj +
j=l j=l
(A-8)
4 4
L: kimJ Ni~J +L: kisJ N;;;

j=l j=l
Nprcdicted wl =L:
5
)""I
Jkrmj Nrmj
5
+L: Jk~j +L:

)=1
Nisj
5
j=I
Jk,dj N,dj + (A-9)
4 4
L: Jkw.j)=1
Nimj +L: Jk,,j N~J
j=l
1
w, = ----:-:,------
1+ Nprediored wO (A-10)
Npredicted {total)
(A-ll)
1
WI = ---:-::-----
1+ Npredicted wl (A-12)
Npredicted (total)
N, = WI Npredicted (total)+ (1-w I) Nobserved (total) (A-13)
(A-14)
Where:
Nprcdicted (total)
~ predicted number of total crashes for the facility or project of interest during the same period for
which crashes were observed;
~ Predicted number of multiple-vehicle nondriveway collisions for roadway segments of type j,j ~ 1... , 5,
Npredicted rmj
during the same period for which crashes were observed;
Nprc<hctcd rsj
~ Predicted number of single-vehicle collisions for roadway segments of type j, during the same period
for which crashes were observed;
Npredicted rdj
~ Predicted number of multiple-vehicle driveway-related collisions for roadway segments of type j,
during the same period for which crashes were observed;
Npredicled imj
~ Predicted number of multiple-vehicle collisions for intersections of type j,j ~ !..., 4, during the same
period for which crashes were observed;
Npredicled isj
~ Predicted number of single-vehicle collisions for intersections of type j, during the same period for
which crashes were observed;
Nobserved (total)
Observed number of total crashes for the facility or project of interest;
Nobserved rmj
Observed number of multiple-vehicle nondriveway collisions for roadway segments of type j;
Nobserved rsj
Observed number of single-vehicle collisions for roadway segments of type j;
Observed number of driveway-related collisions for roadway segments of type j;
Nobserved imj
Observed number of multiple-vehicle collisions for intersections of type j;
Observed number of single-vehicle collisions for intersections of type j;
Npredicted W 0 Predicted number of total crashes during the same period for which crashes were observed under the
assumption that crash frequencies for different roadway elements are statistically independent (p ~ 0);
Overdispersion parameter for multiple-vehicle nondriveway collisions for roadway segments of type j;
Overdispersion parameter for single-vehicle collisions for roadway segments of type j;
~ Overdispersion parameter for driveway-related collisions for roadway segments of type j;

Overdispersion parameter for multiple-vehicle collisions for intersections of type j;
kisj Overdispersion parameter for single-vehicle collisions for intersections of type j;
Npredictc.d wl = Predicted number of total crashes under the assumption that crash frequencies for different roadway
elements are perfectly correlated (p = I);
w, = weight placed on predicted crash frequency under the assumption that crash frequencies for different
roadway elements are statistically independent (r = 0);
weight placed on predicted crash frequency under the assumption that crash frequencies for different
roadway elements are perfectly correlated (r = I);
N, expected crash frequency based on the assumption that different roadway elements are statistically
independent (r = 0);
expected crash frequency based on the assumption that different roadway elements are perfectly
correlated (r = I); and
N er.pecledicomb expected average crash frequency of combined sites including two or more roadway segments or
intersections.
All of the crash terms for roadway segments and intersections presented in Equations A-6 through A-9 are used for
analysis of urban and suburban arterials (Chapter 12). The predictive models for rural two-lane, two-way roads and
multilane highways (Chapters 10 and II) are based on the site type and not on the collision type. Therefore, only one of
the predicted crash terms for roadway segments (N .•.• , rm;., Npre,.JCl=
pr<.'<~l~oc
_, rs;, N _•.•.• ro.1
pro.~!~=
_,,),one of the predicted crash terms for
intersections (Npredictc.ct imP Npredie~ed '-'1), one of the observed crash terms for roadway segments (Nobserved rmj' Nobscrvcd. l").F
No~ed ro,)• and one of the observed crash terms for intersections (Nobser-~ed imp Nobscrvcd L·) is used. For rural t\:vo-lane, two-
way roads and multilane highways, it is recommended that the multiple-vehicle collision terms (with subscripts rmj and
im;) be used to represent total crashes; the remaining unneeded terms can be set to zero.
Chapters 10, 11, and 12 each present worksheets that can be used to apply the project-level EB Method as presented
in this section.
The value of N expcctcUicom

., , from Equation A-14 represents the expected average crash frequency for the same time
period represented by the predicted and observed crash frequencies. The estimate of expected average crash
frequency of combined sites for fatal-and-injury and property-damage-only crashes is calculated by multiplying the
proportion of predicted average crash frequency by severity level (NprediC1ed(FiNpredicted(t.otalJ and NpredJctcd(PDo/Npredicted(toi;JJ))
to the total expected average crash frequency of combined sites from Equation A-14. Section A.2.6 explains
how to update N expccrewcom
., , to a future time period, such as the time period when a proposed future project will be
implemented.
A.2.6. Adjust the Estimated Value of Expected Average Crash frequency to a Future Time Period,
If Appropriate
The value of the expected average crash frequency (Napected) from Equation A -4 or Nexpected/comb from Equation A-14
represents the expected average crash frequency for a given roadway segment or intersection (or project, for
Nexpeeteutcombl
., 'during the before period. To obtain an estimate of expected average crash frequency in a future period
(the after period), the estimate is corrected for (I) any difference in the duration of the before and after periods;
(2) any growth or decline in AADTs between the before and after periods; and (3) any changes in geometric design
or traffic control features between the before and after periods that affect the values of the CMFs for the roadway
segment or intersection. The expected average crash frequency for a roadway segment or intersection in the after
period can be estimated as:
(A-15)
'Where:
= expected average crash frequency during the future time period for which crashes are being forecast for
the segment or intersection in question (i.e., the after period);
= expected average crash frequency for the past time period for which observed crash history data were
available (i.e., the before period);
number of crashes forecast by the SPF using the future AADT data, the specified nominal values for
geometric parameters, and-in the case of a roadway segment-the actual length of the segment;
= number of crashes forecast by the SPF using the past AADT data, the specified nominal values for
geometric parameters, and-in the case of a roadway segment-the actual length of the segment;
= value of the nth CMF for the geometric conditions planned for the future (i.e., proposed) design; and
= value of the nth CMF for the geometric conditions for the past (i.e., existing) design.
Because of the form of the SPFs for roadway segments, if the length of the roadway segments are not changed,
the ratio N,jN,, is the same as the ratio of the traffic volumes, AADTJAADT,. However, for intersections, the ratio
N,JN,, is evaluated explicitly with the SPFs because the intersection SPFs incorporate separate major- and minor-
roadAADT terms with differing coefficients. In applying EquationA-15, the values of N,,, N,f' CMF,,, and CMF,1
should be based on the average AADTs during the entire before or after period, respectively.
In projects that involve roadway realignment, if only a small portion of the roadway is realigned, the ratio N/N,,
should be determined so that its value reflects the change in roadway length. In projects that involve extensive
roadway realignment, the EB Method may not be applicable (see discussion in Section A.2.1 ).
EquationA-15 is applied to total average crash frequency. The expected future average crash frequencies by severity
level should also be determined by multiplying the expected average crash frequency from the before period for each
severity level by the ratio~!«,.
In the case of minor changes in roadway alignment (i.e., flattening a horizontal curve), the length of an analysis
segment may change from the past to the future time period, and this would be reflected in the values of N,, and N,F
Equation A-15 can also be applied in cases for which only facility- or project-level data are available for observed
crash frequencies. In this situation, Nexpectedlcomb should be used instead of Nexpected in the equation.
Glossary
This chapter defines the terms used in the manual.
85th-percentile speed-the speed at or below which 85 percent of the motorists drive a given road. The speed is
indicative of the speed that most motorists consider to be reasonably safe under normal conditions.
AADT-annual average daily traffic. (See traffic, average annual daily.)
acceleration lane-a paved auxiliary lane, including tapered areas, allowing vehicles to accelerate when entering the
through-traffic lane of the roadway.
acceptable gap--the distance to nearest vehicle in oncoming or cross traffic that a driver will accept to initiate a
turning or crossing maneuver 50 percent of the time it is presented, typically measured in seconds.
access management-the systematic control of the location, spacing, design, and operation of driveways, median
openings, interchanges, and street connections to a roadway, as well as roadway design applications that affect
access, such as median treatments and auxiliary lanes and the appropriate separation of traffic signals.
accessible facilities-facilities where persons with disabilities have the same degree of convenience, connection,
and safety afforded to the public in general. It includes, among others, access to sidewalks and streets, including
crosswalks, curb ramps, street furnishings, parking, and other components of public rights-of-way.
accommodation (visua(!-the ability to change focus from instruments inside the vehicle to objects outside the vehicle.
all-way stop-controlled-an intersection with stop signs at all approaches.
approach-a lane or set of lanes at an intersection that accommodates all left-tum, through, and right-tum move-
ments from a given direction.
auxiliary lane-a lane marked for use, but not assigned for use by through traffic.
base model-a regression model for predicting the expected average crash frequency in each HSM prediction proce-
dure given a set of site characteristics. The base model, like all regression models, predicts the value of a dependent
variable as a function of a set of independent variables. The expected average crash frequency is adjusted for changes
to set site characteristics with the use of a CMF.
Bayesian statistics-statistical method of analysis which bases statistical inference on a number of philosophical
underpinnings that differ in principle from frequentist or classical statistical thought. First, this method incorporates
G-1
G-2 HIGHWAY SAFETY MANUAL
knowledge from history or other sites. In other words, prior knowledge is formally incorporated to obtain the "best"
estimation. Second, the method considers the likelihood of certain types of events as part of the analysis process.
Third, it uses Bayes' theorem to translate probabilistic statements into degrees of belief (e.g., the belief that we are
more certain about something than others) instead of the classical confidence interval interpretation.
before-after study--the evaluation of implemented safety treatments, accomplished by comparing frequency or

severity of crashes before and after implementation. There are several different types of before-after studies. These
studies often develop CMFs for a particular treatment or group of treatments. Also known as BA studies.
bicycle facility--a road, path, or way specifically designated for bicycle travel, whether exclusively or with other
vehicles or pedestrians.
breakaway support-a design feature which allows a device such as a sign, luminaire, or traffic signal support to
yield or separate upon impact.
bus lane-a highway or street lane designed for bus use during specific periods.
calibration factor-a factor to adjust crash frequency estimates produced from a safety prediction procedure to
approximate local conditions. The factor is computed by comparing existing crash data at the state, regional, or local
level to estimates obtained from predictive models.
channelization-the separation of conflicting traffic movements into definite travel paths. Often part of access man-
agement strategies.
clear zone-the total roadside border area, starting at the edge of the traveled way, available for use by errant vehicles.
climbing lane-a passing lane added on an upgrade to allow traffic to pass heavy vehicles whose speeds are reduced.
closing speed-movement of objects based on their distance as observed from the driver.
coding--organization of information into larger units such as color and shape (e.g., warning signs are yellow, regula-
tory signs are white).
collision-see crash.
collision diagram-a schematic representation of the crashes that have occurred at a site within a given time period.
comparison group-a group of sites, used in before-and-after studies, which are untreated but are similar in nature
to the treated sites. The comparison group is used to control for changes in crash frequency not influenced by the
treatment.
comparison ratio-the ratio of expected number of "after" to the expected number of "before" target crashes on the
comparison group.
condition diagram-a plan view drawing of relevant site characteristics.
conflict-to-crash ratio--number of conflicts divided by the number of crashes observed during a given period.
conspicuity-relates to the ability of a given object or condition to attract the attention of the road user.
context sensitive design (CSD)-a collaborative, interdisciplinary approach that involves all stakeholders to develop
a transportation facility that fits its physical setting and preserves scenic, aesthetic, historic, and environmental
resources, while maintaining safety and mobility.
GLOSSARY G-3
continuous variable-a variable that is measured either on the interval or ratio scale. A continuous variable can
theoretically take on an infinite number of values within an interval. Examples of continuous variables include
measurements in distance, time, and mass. A special case of a continuous variable is a data set consisting of counts
(e.g., crashes), which consist of non-negative integer values.
contrast sensitivity--the ability to distinguish between low-contrast features. Ability to detect slight differences in
luminance (level of light) between an object and its background (e.g., worn lane lines, concrete curbs).
control group-a set of sites randomly selected to not receive safety improvements.
control task-a major subtask of the driving task model consisting of keeping the vehicle at a desired speed and
heading within the lane. Drivers exercise control through the steering wheel, accelerator or brake.
corner clearance-minimum distance required between intersections and driveways along arterials and collector
streets.
cost-effectiveness-a type of economic criteria for assessing a potential implementation of a countermeasure or

design to reduce crashes. This term is generally expressed in terms of the dollars spent per reduction of crash
frequency or crash severity.
cost-effectiveness index-ratio of the present value cost to the total estimated crash reduction.
count data-data that are non-negative integers.
countermeasure-a roadway-based strategy intended to reduce the crash frequency or severity, or both at a site.
countermeasure, proven--countermeasures that are considered proven for given site characteristics because scientifi-
cally rigorous evaluations have been conducted to validate the effectiveness of the proposed countermeasure for the
given site characteristics.
countermeasure, tried and experimental-countermeasures for which a scientifically rigorous evaluation

has not been conducted or because an evaluation has not been performed to assess the effectiveness of such
countermeasures.
crash-a set of events not under human control that results in injury or property damage due to the collision of at
least one motorized vehicle and may involve collision with another motorized vehicle, a bicyclist, a pedestrian or an
object.
crash cushion (impact attenuator)-device that prevents an errant vehicle from impacting fixed objects by gradu-
ally decelerating the vehicle to a safe stop or by redirecting the vehicle away from the obstacle in a manner which
reduces the likelihood of injury.
crash estimation-any methodology used to forecast or predict the crash frequency of an existing roadway for
existing conditions during a past period or future period; an existing roadway for alternative conditions during a past
or future period; a new roadway for given conditions for a future period.
crash evaluation-determining the effectiveness of a particular treatment or a treatment program after its
implementation. The evaluation is based on comparing results obtained from crash estimation.
crash frequency-number of crashes occurring at a particular site, facility, or network in a one year period and is
measure in number of crashes per year.
crash mapping-the visualization of crash locations and trends with computer software such as Geographic Infor-
mation System (GIS).
crash modification factor (CMF)-an index of how much crash experience is expected to change following a
modification in design or traffic control. CMF is the ratio between the number of crashes per unit of time expected
after a modification or measure is implemented and the number of crashes per unit of time estimated if the cbange
does not take place.
crash prediction algorithm-procedure used to predict average crash frequency, consisting of three elements. It has
two analytical components: baseline models and crash modification factors, as well as a third component: crash
histories.
crash rate-the number of crashes per unit of exposure. For an intersection, this is typically the number of crashes
divided by the total entering AADT; for road segments, this is typically the number of crashes per million vehicle-
miles traveled on the segment.
crash rate method-a method that normalizes the frequency of crashes against exposure (i.e., traffic volume for the
study period for intersections, and traffic volume for the study period and segment length for roadway segments).
Also known as accident rate method.
crash reduction factor (CRF)-the percentage crash reduction that might be expected after implementing a
modification in design or traffic control. The CRF is equivalent to (I - CMF).
crash severity-the level of injury or property damage due to a crash, commonly divided into categories based on
the K.ABCO scale.
critical rate method (CRM)-a method in which the observed crash rate at each site is compared to a calculated criti-
cal crash rate that is unique to each site.
cross-sectional studies-studies comparing the crash frequency or severity of one group of entities having some
common feature (e.g., stop-controlled intersections) to the crash frequency or severity of a different group of entities
not having that feature (e.g., yield-controlled intersections), in order to assess difference in crash experience between
the two features (e.g., stop versus yield sign).
cycle-a complete sequence of signal indications (phases).
cycle length-the total time for a traffic signal to complete one cycle.
dark adaptation (visual)-the ability to adjust light sensitivity on entering and exiting lighted or dark areas.
deceleration lane-a paved auxiliary lane, including tapered areas, allowing vehicles leaving the through-traffic lane
of the roadway to decelerate.
decision sight distance (DSD)-the distance required for a driver to detect an unexpected or otherwise difficult-to-
perceive information source, recognize the object, select an appropriate speed and path, and initiate and complete the
maneuver efficiently and without a crash outcome.
delay-the additional travel time experienced by a driver, passenger, or pedestrian in comparison to free flow conditions.
delineation-methods of defining the roadway operating area for drivers.

GLOSSARY G-5
dependent variable-in a function given as Y= j{X1, ••• , X), it is customary to refer to X1, ••• , Xn as independent or
explanatory variables, and Y as the dependent or response variable. In each crash frequency prediction procedure, the
dependent variable estimated in the base model is the annual crash frequency for a roadway segment or intersection.
descriptive analysis-methods such as frequency, crash rate, and equivalent property damage only (EPDO), which
summarize in different forms the history of crash occurrence, type or severity, or both, at a site. These methods do
not include any statistical analysis or inference.
design consistency-(!) the degree to which highway systems are designed and constructed to avoid critical driving
maneuvers that may increase crash risk; (2) the ability of the highway geometry to conform to driver expectancy;
(3) the coordination of successive geometric elements in a manner to produce harmonious driver performance
without surprising events.
design speed-a selected speed used to determine the various geometric design features of the roadway. The as-
sumed design speed should be a logical one with respect to the topography, anticipated operating speed, the adjacent
land use, and the functional classification of highway. The design speed is not necessarily equal to the posted speed
or operational speed of the facility.
diagnosis-the identification of factors that may contribute to a crash.
diamond interchange-an interchange that results in two or more closely spaced surface intersections, so that one
connection is made to each freeway entry and exit, with one connection per quadrant.
discount rate-an interest rate that is chosen to reflect the time value of money.
dispersion parameter-see overdispersion parameter.
distribution (data analysis and modeling related)-the set of frequencies or probabilities assigned to various out-
comes of a particular event or trail. Densities (derived from continuous data) and distributions (derived from discrete
data) are often used interchangeably.
driver expectancy-the likelihood that a driver will respond to common situations in predictable ways that the driver
has found successful in the past. Expectancy affects how drivers perceive and handle information and affects the
speed and nature of their responses.
driver workload---.surrogate measure of the number of simultaneous tasks a driver performs while navigating a roadway.
driveway density-the number of driveways per mile on both sides ofthe roadway combined.
driving task model-the simultaneous and smooth integration of a number of sub-tasks required for a successful
driving experience.
dynamic programming-a mathematical technique used to make a sequence of interrelated decisions to produce an
optimal condition.
economically valid project-a project in which benefits are greater than the cost.
Empirical Bayes (EB) methodology-method used to combine observed crash frequency data for a given site with
predicted crash frequency data from many similar sites to estimate its expected crash frequency.
entrance ramp-a ramp that allows traffic to enter a freeway.

I,
equivalent property damage only (EPDO) method-assigns weighting factors to crashes by severity (fatal, injury,
property damage only) to develop a combined frequency and severity score per site. The weighting factors are
calculated relative to Property Damage Only (PDO) crash costs. Crash costs include direct costs such as ambulance
service, police and fire services, property damage, insurance and other costs directly related to the crashes. Crash
costs also include indirect costs, i.e., the value society would place on pain and suffering or loss oflife associated
with the crash.
exit ramp--a ramp that allows traffic to depart a freeway.
expected average crash frequency-the estimate of long-term expected average crash frequency of a site, facility, or
network under a given set of geometric conditions and traffic volumes (AADT) in a given period of years. In the Em-
piracal Bayes (EB) methodology, this frequency is calculated from observed crash frequency at the site and predicted
crash frequency at the site based on crash frequency estimates at other similar sites.
expected average crash frequency, change in-the difference between the expected average crash frequency in the
absence of treatment and with the treatment in place.
expected crashes-an estimate of long-range average number of crashes per year for a particular type of roadway or
intersection.
expected excess crash method-method in which sites are ranked according to the difference between the adjusted
observed crash frequency and the expected crash frequency for the reference population (e.g., two-lane rural seg-
ment, multilane undivided roadway, or urban stop-controlled intersection).
experimental studies-studies where sites are randomly assigned to a treatment or control group and the differences
in crash experience can then be attributed to a treatment or control group.
explanatory variable (predictor)-a variable which is used to explain (predict) the change in the value of another
variable. An explanatory variable is often defined as an independent variable; the variable which it affects is called
the dependent variable.
facility-a length of highway that may consist of connected sections, segments, and intersections.
first harmfol event-the first injury or damage-producing event that characterizes the crash.
freeway-a multilane, divided highway with a minimum of two lanes for the exclusive use of traffic in each direction
and full control of access without traffic interruption.
I
lr frequency method-a method that produces a ranking of sites according to total crashes or crashes by type or
severity, or both.
I"
I
frequentist statistics-statistical philosophy that results in hypothesis tests that provide an estimate of the probabil-
ity of observing the sample data conditional on a true null hypothesis. This philosophy asserts that probabilities are
obtained through long-run repeated observations of events.
gap--the time, in seconds, for the front bumper of the second of two successive vehicles to reach the starting point
of the front bumper of the first vehicle. Also referred to as headway.
gap acceptance-the process by which a vehicle enters or crosses a vehicular stream by accepting an available gap
to maneuver.
geometric condition-the spatial characteristics of a facility, including grade, horizontal curvature, the number and
width oflanes, and lane use.
GLOSSARY G-7
goodness-of-fit (GOF) statistics-the goodness of fit of a statistical model describes how well it fits a set of
observations. Measures of goodness of fit typically summarize the discrepancy between observed values and the
values expected under the model in question. There are numerous GOF measures, including the coefficient of
determination R2, the F test, and the chi-square test for frequency data, among others. Unlike F-ratio and likelihood-
ratio tests, GOP measures are not statistical tests.
gore area-the area located immediately between the edge of the ramp pavement and the edge of the roadway
pavement at a merge or diverge area.
guidance task-a major subtask of the driving task model consisting of interacting with other vehicles (following,
passing, merging, etc.) through maintaining a safe following distance and through following markings, traffic control
signs, and signals.
Haddon Matrix-a framework used for identifying possible contributing factors for crashes in which contributing
factors (i.e., driver, vehicle, and roadway/environment) are cross-referenced against possible crash conditions before,
during, and after a crash to identify possible reasons for the events.
headway--see gap.
Heinrich Triangle-concept founded on the precedence relationship that "no injury crashes" precedes "minor injury
crashes." This concept is supported by two basic ideas: (I) events oflesser severity are more numerous than more
severe events, and events closer to the base of the triangle precede events nearer the top; and (2) events near the base
of the triangle occur more frequently than events near the triangle's top, and their rate of occurrence can be more
reliably estimated.
high-occupancy vehicle (HOV)-a vehicle with a defined minimum number of occupants (may consist of vehicles
with more than one occupant).
high proportion of crashes method-the screening of sites based on the probability that their long-term expected
proportion of crashes is greater than the threshold proportion of crashes.
Highway Safety Improvement Program (HSIP)-SAFETEA-LU re-established the Highway Safety Improvement
Program (HSIP) as a core program in conjunction with a Strategic Highway Safety Plan (SHSP). The purpose of the
HSIP is to reduce the number offatal and serious/life-changing crashes through state-level engineering measures.
holistic approach-a multidisciplinary approach to the reduction of crashes and injury severity.
homogeneous roadway segment-a portion of a roadway with similar average daily traffic volumes (vehlday),
geometric design, and traffic control features.
human factors-the application of knowledge from human sciences, such as human psychology, physiology, and
kinesiology, in the design of systems, tasks, and environments for effective and safe use.
incremental benefit-cost ratio-the incremental benefit-cost ratio is an extension of the benefit-cost ratio method.
Projects with a benefit-cost ratio greater than one are arranged in increasing order based on their estimated cost.
independent variables-a variable which is used to explain (predict) the change in the value of another variable.
Indiana Lane Merge System (ILMS)-advanced dynamic traffic control system designed to encourage drivers to
switch lanes well in advance of the work zone lane drop and entry taper.
indirect measures of safety-see surrogate measures.

influence area (freeway)----;m area that incurs operational impacts of merging (diverging) vehicles in Lanes I and 2
of the freeway and the acceleration (deceleration) lane for I ,500 ft from the merge (diverge) point downstream.
influence area (intersection)-functional area on each approach to an intersection consisting of three elements:
(I) perception-reaction distance, (2) maneuver distance, and (3) queue storage distance.
integer programming-a mathematical optimization technique involving a linear programming approach in which
some or all of the decision variables are restricted to integer values.
interchange-intersections that consist of structures that provide for the cross-flow of traffic at different levels with-
out interruption, thus reducing delay, particularly when volumes are high.
interchange ramp terminal-a junction with a surface street to serve vehicles entering or exiting a freeway.
intersection-general area where two or more roadways or highways meet, including the roadway, and roadside
facilities for pedestrian and bicycle movements within the area.
intersection functional area-area extending upstream and downstream from the physical intersection area including
any auxiliary lanes and their associated channelization.
intersection related crash-a crash that occurs at the intersection itself or a crash that occurs on an intersection
approach within 250ft (as defined in the HSM) of the intersection and is related to the presence of the intersection.
intersection sight distance-the distance needed at an intersection for drivers to perceive the presence of potentially
conflicting vehicles in sufficient time to stop or adjust their speed to avoid colliding in the intersection.
KABCO-an injury scale developed by the National Safety Council to measure the observed injury severity for any
person involved as determined by law enforcement at the scene of the crash. The acronym is derived from (Fatal
injury (K), Incapacitating Injury (A), Non-Incapacitating Injury (B), Possible Injury (C), and No Injury (0).) The
scale can also be applied to crashes: for example, a K crash would be a crash in which the most severe injury was a
fatality, and so forth.
lateral clearance-latera! distance from edge of traveled way to a roadside object or feature.
level ofservice ofsafety (LOSS) method-the ranking of sites according to their observed and expected crash
frequency for the entire population, where the degree of deviation is then labeled into four classes oflevel of service.
median-the portion of a divided highway separating the traveled ways from traffic in opposite directions.
median refoge island-an island in the center of a road that physically separates the directional flow of traffic and
that provides pedestrians with a place of refuge and reduces the crossing distance of a crosswalk.
meta analysis-a statistical technique that combines the independent estimates of crash reduction effectiveness from
separate studies into one estimate by weighing each individual estimate according to its variance.
method of moments-method in which a site's observed crash frequency is adjusted based on the variance in the
crash data and average crash counts for the site's reference population.
minor street-the lower volume street controlled by stop signs at a two-way or four-way stop-controlled intersection;
also referred to as a side street. The lower volume street at a signalized intersection.
Model Minimum Inventory ofRoadway Elements (MMIRE)-set of guidelines outlining the roadway information
that should be included in a roadway database to be used for safety analysis.
GLOSSARY G-9
Model Minimum Unifonn Crash Criteria (MMUCC)-set of guidelines outlining the minimum elements in crash,
roadway, vehicle, and person data that should ideally be in an integrated crash database.
most hannful event-event that results in the most severe injury or greatest property damage for a crash event.
motor vehicle crash-any incident in which bodily injury or damage to property is sustained as a result of the move-
ment of a motor vehicle, or of its load while the motor vehicle is in motion. Also referred to as a motor vehicle crash.
multilane highway-a highway with at least two lanes for the exclusive use of traffic in each direction, with no control,
partial control, or full control of access, but that may have periodic interruptions to flow at signalized intersections.
multivariate statistical modeling-statistical procedure used for cross-sectional analysis which attempts to account
for variables that affect crash frequency or severity, based on the premise that differences in the characteristics of
features result in different crash outcomes.
navigation task-activities involved in planning and executing a trip from origin to destination.
net benefit-a type of economic criteria for assessing the benefits of a project. For a project in a safety program, it is
assessed by determining the difference between the potential crash frequency or severity reductions (benefits) from
the costs to develop and construct the project. Maintenance and operations costs may also be associated with a net
benefit calculation.
net present value (NPV) or net present worth (NPff)-this method is used to express the difference between
discounted costs and discounted benefits of an individual improvement project in a single amount. The term
"discounted" indicates that the monetary costs and benefits are converted to a present-value using a discount rate.
network screening-nernrork screening is a process for reviewing a transportation network to identify and rank sites
from most likely to least likely to benefit from a safety improvement.
non-monetary factors-items that do not have an equivalent monetary value or that would be particularly difficult to
quantify (i.e., public demand, livability impacts, redevelopment potential, etc.).
observational studies--{)ften used to evaluate safety performance. There are two forms of observational studies:
before-after studies and cross-sectional studies.
offset-lateral distance from edge of traveled way to a roadside object or feature. Also known as lateral clearance.
operating speed-the 85th percentile of the distribution of observed speeds operating during free-flow conditions.
overdispersion parameter-an estimated parameter from a statistical model that when the results of modeling are
used to estimate crash frequencies, indicates how widely the crash counts are distributed around the estimated mean.
This term is used interchangeably with dispersion parameter.
p-value-the level of significance used to reject or accept the null hypothesis (whether or not a result is statistically valid).
passing lane-a lane added to improve passing opportunities in one or both directions of travel on a two-lane highway.
peak searching algorithm-a method to identify the segments that are most likely to benefit from a safety
improvement within a homogeneous section.
pedestrian-a person traveling on foot or in a wheelchair.
pedestrian crosswalk-pedestrian roadway crossing facility that represents a legal crosswalk at a particular location.
pedestrian refuge--an at-grade opening within a median island that allows pedestrians to wait for an acceptable gap
in traffic.
pedestrian traffic control-traffic control devices installed particularly for pedestrian movement control at
intersections; it may include illuminated push buttons, pedestrian detectors, countdown signals, signage, pedestrian
channelization devices, and pedestrian signal intervals.
perception-reaction time (PR1)-time required to detect a target, process the information, decide on a response, and
initiate a response (it does not include the actual response element to the information). Also known as perception-
response time.
perception-response time-see perception-reaction time.
performance threshold-a numerical value that is used to establish a threshold of expected number of crashes (i.e.,
safety performance) for sites under consideration.
peripheral vision-the ability of people to see objects beyond the cone of clearest vision.
permitted plus protected phase-compound left-turn protection that displays the permitted phase before the
protected phase.
perspective, engineering-the engineering perspective considers crash data, site characteristics, and field conditions
in the context of identifying potential engineering solutions that would address the potential safety concern. It may
include consideration of human factors.
perspective, human factors-the human factors perspective considers the contributions of the human to the contributing
factors of the crash in order to propose solutions that might break the chain of events leading to the crash.
phase-the part of the signal cycle allocated to any combination of traffic movements receiving the right-of-way
simultaneously during one or more intervals.
positive guidance-when information is provided to the driver in a clear manner and with sufficient conspicuity to
allow the driver to detect an object in a roadway environment that may be visually cluttered, recognize the object and
its potential impacts to the driver and vehicle, select an appropriate speed and path, and initiate and complete the
required maneuver successfully.
potential for safety improvement (PS.l)--estimates how much the long-term crash frequency could be reduced at a
particular site.
predicted average crash frequency-the estimate of long-term average crash frequency which is forecast to occur
at a site using a predictive model found in Part C of the HSM. The predictive models in the HSM involve the use of
regression models, known as Safety Performance Functions, in combination with Crash Modification Factors and
calibration factors to adjust the model to site-specific and local conditions.
predictive method-the methodology in Part C of the manual used to estimate the 'expected average crash
frequency' of a site, facility, or roadway under given geometric conditions, traffic volumes, and period of time.
primacy-placement of information on signs according to its importance to the driver. In situations where
information competes for drivers' attention, unneeded and low-priority information is removed. Errors can occur
when drivers shred important information because of high workload (process less important information and miss
more important information).
GLOSSARY G-11
programming, dynamic-a mathematical technique used to make a sequence of interrelated decisions to produce
an optimal condition. Dynamic programming problems have a defined beginning and end. While there are multiple
paths and options between the beginning and end, only one optimal set of decisions will move the problem from the
beginning to the desired end.
programming, integer-an instance of linear programming when at least one decision variable is restricted to an
integer value.
programming, linear-a method used to allocate limited resources (funds) to competing activities (safety
improvement projects) in an optimal manner.
project development process-typical stages of a project from planning to post-construction operations and
maintenance activities.
project planning-part of the project development process in which project alternatives are developed and analyzed
to enhance a specific performance measure or a set of performance measures, such as, capacity, multimodal
amenities, transit service, and safety.
quantitative predictive analysis-methodology used to calculate an expected number of crashes based on the
geometric and operational characteristics at the site for one or more of the following: existing conditions, future
conditions, or roadway design alternatives.
queue-a line of vehicles, bicycles, or persons waiting to be served by the system in which the flow rate from the
front of the queue determines the average speed within the queue.
randomized controlled trial-experiment deliberately designed to answer a research question. Roadways or facilities
are randomly assigned to a treatment or control group.
ranking methods, individual-the evaluation of individual sites to determine the most cost-effective countermeasure
or combination of countermeasures for the site.
ranking methods, systematic-the evaluation of multiple safety improvement projects to determine the combination
of projects that will provide the greatest crash frequency or severity reduction benefit across a highway network
given budget constraints.
rate-see crash rate.
rate, critical--compares the observed crash rate at each site with a calculated critical crash rate unique to each site.
reaction time (R1)-the time from the onset of a stimulus to the beginning of a driver's (or pedestrian's) response to
the stimulus by a simple movement of a limb or other body part.
redundancy-providing information in more than one way, such as indicating a no passing zone with signs and
pavement markings.
regression analysis-a collective name for statistical methods used to determine the interdependence of variables for
the purpose of predicting expected average outcomes. These methods consist of values of a dependent variable and
one or more independent variables (explanatory variables).
regression-to-the-mean (RTM)-the tendency for the occurrence of crashes at a particular site to fluctuate up or
down, over the long term, and to converge to a long-term average. This tendency introduces regression-to-the-mean
bias into crash estimation and analysis, making treatments at sites with extremely high crash frequency appear to be
more effective than they truly are.
relative severity index (RSI)-a measure of jurisdiction-specific societal crash costs.
relative severity index (RSI) method-an average crash cost calculated based on the crash types at each site and then
compared to an average crash cost for sites with similar characteristics to identify those sites that have a higher than
average crash cost. The crash costs can include direct crash costs accounting for economic costs of the crashes only;
or account for both direct and indirect costs.
roadside-the area between the outside shoulder edge and the right-of-way limits. The area between roadways of a
divided highway may also be considered roadside.
roadside barrier-a longitudinal device used to shield drivers from natural or man-made objects located along either
side of a traveled way. It may also be used to protect bystanders, pedestrians. and cyclists from vehicular traffic under
special conditions.
roadside hazard rating--considers the clear zone in conjunction with the roadside slope, roadside surface roughness,
recoverability of the roadside, and other elements beyond the clear zone such as barriers or trees. As the RHR
increases from l to 7, the crash risk for frequency and/or severity increases.
road-use culture-each individual road user's choices and the attitudes of society as a whole towards transportation
safety.
roadway-the portion of a highway, including shoulders, for vehicular use.
roadway cross-section elements-roadway travel lanes, medians, shoulders, and sideslopes.
roadway environment-a system in which the driver, the vehicle, and the roadway interact with each other.
roadway, intermediate or high-speed-facility with traffic speeds or posted speed limits greater than 45 mph.
roadway, low-speed-facility with traffic speeds or posted speed limits of 30 mph or less.
roadway safety management process-a quantitative, systematic process for studying roadway crashes and charac-
teristics of the roadway system and those who use the system, which includes identifying potential improvements,
implementation, and the evaluation of the improvements.
roadway segment-a portion of a road that has a consistent roadway cross-section and is defined by two endpoints.
roundabout-an unsignalized intersection with a circulatory roadway around a central island with all entering
vehicles yielding to the circulating traffic.
rumble strips--<levices designed to give strong auditory and tactile feedback to errant vehicles leaving the travel way.
running speed-the distance a vehicle travels divided by running time, in miles per hour.
rural areas-places outside the boundaries of urban growth boundary where the population is less than 5,000
inhabitants.
Safe, Accountable, Flexible, Efficient Transportation Equity Act: A Legacy for Users (SAFETEA-LU)-a federal
legislature enacted in 2005. This legislature elevated the Highway Safety Improvement Program (HSIP) to a core
FHWA program and created requirement for each state to develop a State Highway Safety Plan (SHSP).
safety-the number of crashes, by severity, expected to occur on the entity per unit of time. An entity may be a
signalized intersection, a road segment, a driver, a fleet of trucks, etc.
• I
I'
I
GLOSSARY G-13
safety management process-process for monitoring, improving, and maintaining safety on existing roadway
networks.
safety performance fimction (SPF)-an equation used to estimate or predict the expected average crash frequency
per year at a location as a function of traffic volume and in some cases roadway or intersection characteristics (e.g.,
number of lanes, traffic control, or type of median).
segment-a portion of a facility on which a crash analysis is performed. A segment is defined by two endpoints.
selective attention-the ability, on an ongoing moment-to-moment basis while driving, to identify and allocate
attention to the most relevant information, especially within a visually complex scene and in the presence of a
number of distracters.
service life-number of years in which the countermeasure is expected to have a noticeable and quantifiable effect
on the crash occurrence at the site.
severity index-a severity index (SI) is a number from zero to ten used to categorize crashes by the probability of
their resulting in property damage, personal injury, or a fatality, or any combination of these possible outcomes. The
resultant number can then be translated into a crash cost and the relative effectiveness of alternate treatments can be
estimated.
shoulder-a portion ofthe roadway contiguous with the traveled way for accommodation of pedestrians, bicycles,
stopped vehicles, emergency use, as well as lateral support of the subbase, base, and surface courses.
sight distance-the length of roadway ahead that is visible to the driver.
sight triangle-in plan view, the area defined by the point of intersection of two roadways, and by the driver's line of
sight from the point of approach along one leg of the intersection to the farthest unobstructed location on another leg
of the intersection.
site-project location consisting of, but not limited to, intersections, ramps, interchanges, at-grade rail crossings,
roadway segments, etc.
sites with potential for improvement-intersections and corridors with potential for safety improvements and
identified as having possibility of responding to crash countermeasure installation.
skew angle, intersection-the deviation from an intersection angle of 90 degrees. Carries a positive or negative sign
that indicates whether the minor road intersects the major road at an acute or obtuse angle, respectively.
slalom effect-dynamic illusion of direction and shape used to influence traffic behavior.
sliding-window approach-analysis method that can be applied when screening roadway segments. It consists
of conceptually sliding a window of a specified length (e.g., 0.3 mile) along the road segment in increments of a
specified size (e.g., 0.1 mile). The method chosen to screen the segment is applied to each position of the window,
and the results of the analysis are recorded for each window. The window that shows the most potential for safety
improvement is used to represent the total performance of the segment.
slope-the relative steepness of the terrain expressed as a ratio or percentage. Slopes may be categorized as positive
(backslopes) or negative (foreslopes) and as parallel or cross slopes in relation to the direction of traffic.
speed adaptation-phenomenon experienced by drivers leaving a freeway after a long period of driving, and having
difficulty conforming to the speed limit on a different road or highway.
.I
speed choice-speed chosen by a driver that is perceived to limit the risk and outcome of a crash.
spreading-where all the information required by the driver cannot be placed on one sign or on a number of signs
at one location, spread the signage out along the road so that information is given in small amounts to reduce the
information load on the driver.
stopping sight distance (SSD)-the sight distance required to permit drivers to see a stationary object soon enough
to stop for it under a defined set of worst-case conditions, without the performance of any avoidance maneuver or
change in travel path; the calculation of SSD depends upon speed, gradient, road surface and tire conditions, and
assumptions about the perception-reaction time of the driver.
Strategic Highway Safety Plan (SHSP)-a comprehensive plan to substantially reduce vehicle-related fatalities and
injuries on the nation's highways (AASHTO). All departments of transportation are required by law to develop,
implement, and evaluate a Strategic Highway Safety Plan for their state, in coordination with partner groups as
stipulated in federal regulations.
suburban environment-an area with a mixture of densities for housing and employment, where high-density non-
residential development is intended to serve the local community.
superelevation-the banking of a roadway in a curve to counteract lateral acceleration.
surrogate measure-an indirect safety measurement that provides the opportunity to assess safety performance when
crash frequencies are not available because the roadway or facility is not yet in service or has only been in service
for a short time, or when crash frequencies are low or have not been collected, or when a roadway or facility has
significant unique features
system planning-the first stage of the project development process, in which network priorities are identified and
assessed.
systematic prioritization-the process used to produce an optimal project mix that will maximize crash frequency
and severity reduction benefits while minimizing costs, or fitting a mixed budget or set of policies.
systematic reviews-process of assimilating knowledge from documented information.
taper area-an area characterized by a reduction or increase in pavement width, typically located between mainline
and ramp or areas with lane reductions.
total entering volume-sum of total major and minor street volumes approaching an intersection.
total million entering vehicles (TMEY)-measurement for total intersection traffic volume calculated from total
entering vehicles (TEV) for each intersection approach.
traffic, annual average daily-the counted (or estimated) total traffic volume in one year divided by 365 days/year.
traffic barrier-a device used to prevent a vehicle from striking a more severe obstacle or feature located on the
roadside or in the median or to prevent crossover median crashes. As defined herein, there are four classes of traffic
barriers, namely, roadside barriers, median barriers, bridge railings, and crash cushions.
traffic calming-measures that are intended to prevent or restrict traffic movements, reduce speeds, or attract drivers'
attention, typically used on lower speed roadways.
traffic conflict-an event involving two or more road users, in which the action of one user causes the other user to
make an evasive maneuver to avoid a collision.
GLOSSARY G-15
Transportation Safety Planning (TSP)-the comprehensive, systemwide, multimodal, proactive process that better
integrates safety into surface transportation decision making.
traveled way-lanes, excluding the shoulders.
urban environment-an area typified by high densities of development or concentrations of population, drawing
people from several areas within a region.
usefo.lfield of view (UFOV)-a subset of the total field of view where stimuli can not only be detected, but can be
recognized and understood sufficiently to permit a timely driver response. As such, this term represents an aspect of
visual information processing rather than a measure of visual sensitivity.
visual acuity-the ability to see details at a distance.
visual demand-aggregate input from traffic, the road, and other sources the driver must process to operate a motor
vehicle. While drivers can compensate for increased visual demand to some degree, human factors experts generally
agree that increasing visual demand towards overload will increase crash risk.
volume-the number of persons or vehicles passing a point on a lane, roadway, or other traffic-way during some time
interval, often one hour, expressed in vehicles, bicycles, or persons per hour.
volume, annual average daily traffic-the average number of vehicles passing a point on a roadway in a day from
both directions, for all days of the year, during a specified calendar year, expressed in vehicles per day.
.I

Highway Safety Manual - 1st Edition Volume 2 2010

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Highway Safety Manual - 1st Edition Volume 2 2010

Uploaded by

Copyright:

Available Formats

HIGHWAY

REGION II: Dan Flowers, Arkansas, One-Year Term

REGION III: Nancy l Richardson, One-Year Term

REGION IV: Paula Hammond, Washington, One-Year Term

ALABAMA, William Adams, MINNESOTA, Mukhtar Thakur VERMONT, Kevin Marshia

MICHIGAN, Kathy S. Farnum, UTAH, Robert E. Hull, David Beach,

DISTRICT OF COLUMBIA, NORTH CAROLINA, Terry Hopkins, NHTSA, Bill Watada

FLORIDA, Marianne A. Trussell NORTH DAKOTA, Christopher Holzer,

MARYLAND, Ron Lipps, TENNESSEE, Kendell Poole,

MASSACHUSETTS, TEXAS, Margaret (Meg) A. Moore,

ALABAMA, Stacey N. Glass MISSOURl, Eileen Rackers U.S. DOT MEMBER

ARJZONA, Mike Manthey, NEBRASKA, Daniel J. Waddle

MICIDGAN, Mark W. Bott WEST VIRGINIA, Cindy Cramer

MINNESOTA, Susan M. Groth WISCONSIN, Thomas N. Notbohm

MISSISSIPPI, Robert "Wes" Dean WYOMING, Joel Meena

NORTH CAROLINA, J. Kevin Lacy WASHINGTON, Theodore Trepanier

• NCHRP 17-18(04): Development of an HSM-DraftTable of Contents for the HSM

• NCHRP 17-27: Prepare Parts I and II of the HSM

• NCHRP 17-34: Prepare Parts IV and V of the Highway Safety Manual

• NCHRP 17-38: Highway Safety Manual Implementation and Training Materials

TRB Staff Representative

Chapter 2-Human Factors

Part B-Roadway Safety Management Process

Chapter 6-Select Countermeasures

Chapter 7-Economic Appraisal

Chapter 8-Prioritize Projects

Chapter 9-Safety Effectiveness Evaluation

Chapter 11-Predictive Method for Rural Multilane Highways

Chapter 12-Predictive Method for Urban and Suburban Arterials

Chapter 16-Special Facilities and Geometric Situations

Chapter 17-Road Networks

PART (-INTRODUCTION AND APPLICATIONS GUIDANCE .••••••••..................................... C-1

C.2. Relationship to Parts A, B, and D of the HSM . ..C-2

C.3. Part C and the Project Development Process .. ...... C-2

C.4. Overview of the HSM Predictive Method .. ... C-3

C.5. The HSM Predictive Method .. . ....... C-5

C.6. Predictive Method Concepts .. . .... C-12

C. B. Limitations of the HSM Predictive Method .. ................. C-19

C.9. Guide to Applying Part c. ................ C-20

C.10. Summary .... .... C-20

10.5. Roadway Segments and Intersections .... ····· 10-11

10.7. Crash Modification Factors ... .... 10-22

10.8. Calibration of the SPFs to Local Conditions ...... . . ... 10-33

10.9. Limitations of Predrctive Method in Chapter 10 .. ............. 10-34

10.1 0. Application of Chapter 10 Predictive Method .. ······· 10-34

10.12. Sample Problems .. ............... 10-35

10.13. References .. ... 10-66

CHAPTER 11-PREDICTIVE METHOD FOR RURAL MULTILANE HIGHWAYS .........•.•.••.•. 11-1

11.4. Predictive Method for Rural Multilane Highways .. .. ............. 11-4

11.7. Crash Modification Factors ................................................................ ... 11-24

11.8. Calibration to Local Conditions ......................... . .. 11-35

11.13. References .... .. .. 11-61

12.7. Crash Modification Factors ..................... . .. 12-39

12 .14. References .......... . .. .. 12-106

APPENDIX A-SPECIALIZED PROCEDURES COMMON TO ALL PART C CHAPTERS ..••••••. A-1

CHAPTER 11-PREDICTIVE METHOD FOR RURAL MULTILANE HIGHWAYS .................. 11-1

Figure 12-2. Definition of Roadway Segments and Intersections .. ... 12-15

APPENDIX A-SPECIALIZED PROCEDURES COMMON TO ALL PART C CHAPTERS---------A-1

CHAPTER 11-PREDICTIVE METHOD FOR RURAL MULTILANE HIGHWAYS .................. 11-1

APPENDIX A-SPECIALIZED PROCEDURES COMMON TO ALL PART C CHAPTERS ......... A-1

CHAPTER 11-PREDICTIVE METHOD FOR RURAL MULTILANE HIGHWAYS ........•......... 11-1

Worksheet 3B. Site-Specific EB Method Summary Results........ .. .. 11-68

Project-Level EB Method Summary Results ...... .. ..... 11-69

PURPOSE OF THE HSM