Professional Documents
Culture Documents
Thesis Final PDF
Thesis Final PDF
By
April, 2019
The research titled “Exploring the factors affecting road crash severity in Dhaka
city corporation area” submitted by Anik Banerjee Student No: 1415019; Milan
Saha Student No: 1415023 Session: 2017-2018 has been accepted as satisfactory in
partial fulfillment requirement for the degree of Bachelor of Urban and Regional
Planning (BURP) in April, 2019.
Board of Examiners
1. Supervisor
Assistant Professor,
2. Examiner
Assistant Professor,
Professor,
It is hereby declared that this thesis or any part of it has not been submitted elsewhere
for the award of any degree or diploma.
1. Anik Banerjee
2. Milan Saha
At the very beginning all praises and gratitude belongs to the Almighty God, the most
graceful and generous to human beings, whose belongings were always with us and has
given us the opportunity to perform the research successfully. We would like to express
our gratitude to our honorable supervisor Mr. Anindya Kishore Debnath, Assistant
Professor, Department of Urban and Regional Planning, BUET, for his continuous
guidance, supervision, directions and suggestions at all stages which was very helpful to
conduct the research successfully. We express our heartiest gratitude towards him for his
encouragement and inspiration, providing constant support as well as for the opportunity
to undertake this interesting and unique research which has not been done previously in
BURP and MURP thesis of BUET. We are grateful to our examiner Dr. Md. Musleh Uddin
Hasan, Associate Professor, Department of Urban and Regional Planning, BUET, for his
valuable suggestions and feedback regarding our study. Our gratitude also goes to all other
teachers of our department for their valuable directions and guidance.
i
Abstract
The current study examined the impact of demographic, land use, and roadway
characteristics variables and driving characteristics on the severity of road crash in Dhaka
South City Corporation (DSCC) and Dhaka North City Corporation (DNCC) area. Binary
logistic model was used with crash data from 2010 to 2012 maintained by the Accident
Research Institute (ARI) of Bangladesh University of Engineering and Technology
(BUET) to identify the factors associated with the probability of a fatal crash. A total of
1229 road crash data were sampled. Accident severity (the dependent variable) in this study
is a dichotomous variable with two categories, fatal and non-fatal. Year wise Comparisons
between crash severity and explanatory factors were conducted. The results showed that
the involvement of middle aged drivers (individuals older than 30 years of age), absence
of valid fitness certificate, absence of proper traffic control systems and vehicle defect
increased the likelihood of a fatal crash. A higher risk of fatal crash was also seen for
pedestrians who crossed the road than for those who walked along the edges of the road.
Pedestrian collisions with trucks, buses, baby taxis or and car had a higher risk of a fatal
crash than collisions with auto rickshaw and other vehicles. Crashes occurring at locations
with no traffic control and police controlled had a higher risk of a fatality than those
occurring at locations with traffic signals or center line. Besides, the vehicle not having
any valid fitness certificate has a higher risk of involvement with fatal crash. Drivers not
having any driving license also has a higher risk of involvement with fatal crash. Crashes
occurring at locations at night with no street light available had a higher risk of fatality than
crashes occurring at night where street light is available and at dawn time. Besides, the
other explanatory variables like road geometry, surface condition, surface quality, weather
condition and land use of the location where crash occurred were found as the insignificant
factors behind the road crash severity in DSCC and DNCC area. To reduce the severity of
crashes in DSCC and DNCC area, strategies could target specific factors such as valid
fitness certificate of vehicle, traffic control systems, light conditions at night, vehicle
defects and pedestrian action while crossing the road.
ii
Table of contents
Acknowledgement………………………………………………………………………..i
Abstract…………………………………………………………………………………..ii
Table of contents……………………………………………………………………..iii-ix
iii
4.4 Driver and vehicle characteristics ..................................................................... 40
4.4.1 Driving license .............................................................................................. 40
4.4.2 Vehicle defects .............................................................................................. 42
4.4.3 Vehicle type .................................................................................................. 45
4.4.4 Pedestrian action .......................................................................................... 45
4.4.5 Road class ..................................................................................................... 52
4.5 Spatial characteristics ........................................................................................ 54
4.5.1 Land use ....................................................................................................... 54
Chapter 5: Modelling explanatory factors behind road crash severity ..................... 57
5.1 Socio economic characteristics ........................................................................... 57
5.1.1 Influence of driver age on road crash severity............................................ 57
5.2 Vehicle characteristics........................................................................................ 58
5.2.1 Valid fitness certificate ................................................................................ 58
5.2.2 Vehicle defect ............................................................................................... 61
5.2.3 Vehicle type .................................................................................................. 62
5.2.4 Collision type................................................................................................ 63
5.3 Road characteristics ........................................................................................... 64
5.3.1 Junction type ................................................................................................ 64
5.3.2 Traffic control type ...................................................................................... 65
5.4 Identification of association of factors on road crash severity ......................... 66
5.4.1 Selection of variables ................................................................................... 66
5.4.2 Removal of insignificant explanatory variable from the model ................. 71
5.4.3 Reduction of design variables ...................................................................... 71
5.4.4 Precondition test for goodness of fit of model ............................................. 72
5.5 Identification of influential factors behind road crash severity ....................... 73
5.5.1 Road characteristics..................................................................................... 73
5.5.2 Socio economic characteristics .................................................................... 75
5.5.2.1 Influence of driver’s age on road crash severity ...................................... 75
5.5.3 Vehicle characteristics ................................................................................. 76
5.5.3.4 Influence of light condition on road crash severity .................................. 77
5.5.4 Driving characteristics ................................................................................. 83
5.5.5 Spatial characteristics .................................................................................. 85
iv
Chapter 6: Conclusions ............................................................................................... 88
References…………………………………………………………………………. ……90
List of Tables
Table 1: Number of crash according to the type of crash severity and year of
occurrence……………………………………………………………………………...10
Table 6: Percentage of Fatal Crash of 2010, 2011 and 2012 and Time interval…..37
Table 7: Percentage of Grievous Crash of 2010, 2011 and 2012 and Time interval
…………………………………………………………………………………………..38
Table 8: Percentage of Total Crash of 2010, 2011 and 2012 and Time interval
…………………………………………………………………………………………..39
Table No 9.1: Vehicle Defect and Percentage of Fatal Crash in different three years
…………………………………………………………………………………………..44
Table No 9.2: Vehicle Defect and Percentage of Grievous Crash in different three
years…............................................................................................................................45
Table No 9.3: Vehicle Defect and Percentage of Grievous Crash in different three
years…............................................................................................................................45
Table No 10.1: Vehicle Types and Percentage of Grievous Crash in different three
years……………………………………………………………………………………46
Table No 10.2: Vehicle Types and Percentage of Simple Crash in different three years
v
……………………………………………………………………………………….…47
Table No 10.3: Vehicle Types and Percentage of Simple Crash in different three years
……………………………………………………………………………………….…47
Table No 11: Cross tabulation between percentage of Crash severity and Vehicle type
……………………………………………………………………………………….…49
Table No 12.1: Name of Pedestrian Activity and Percentage of Fatal Crash from 2010-
2012…………………………………………………………………………………….51
Table No 12.2: Name of Pedestrian Activity and Percentage of Grievous Crash from
2010-2012…..………………………………………………………………………….52
Table No 12.3: Name of Pedestrian Activity and Percentage of Simple Crash from
2010-
2012…………………………………………………………………………………….52
Table No 12.4: Name of Pedestrian Activity and Percentage of Total Crash from 2010-
2012…………………………………………………………………………………….53
Table No 13: Cross tabulation of percentage between road class and crash severity
………………………………………………………………………………………….53
Table No 14.1: Land Use and Percentage of Fatal Crash from 2010 to 2012……..54
Table No 14.2: Land Use and Percentage of Grievous Crash from 2010 to
2012………………………………………………………………………………….....54
Table No 14.3: Land Use and Percentage of Simple Crash from 2010 to 2012…...55
Table No 15: Cross tabulation between crash severity and valid fitness certificate of
vehicle………………………………………………………………………………….60
Table No 16: Vehicle Types and Percentage of Fatal Crash in different three
years……………………………………………………………………………………62
Table 17: Variables considered in logistic regression modelling…………………..67
Table 18: Summary statistics for the variable used in model specification……….69
vi
Table 19: Statistics for insignificant explanatory variables………………………..71
Table 22: Estimated coefficients, estimated standard errors, and P value for the
main model variables………………………………………………………………....79
Table No 23: Cross tabulation between percentage of crash severity and vehicle
loading………………………………………………………………………………....82
Table No 24: Cross tabulation between percentage of crash severity and driving
license……………………………………………………………………………….…84
List of Figures:
Figure 3.1: Relation in between Junction types and Fatal Crash from 2010 to 2012……22
Figure 3.2: Relation in between Junction types and Grievous Crash from 2010 to
2012………………………………………………………………………………...…….22
Figure 3.3: Relation in between Junction types and Simple Crash from 2010 to
2012…………………………………………………………………………………...….23
Figure 4.1: Relation in between Control Type and Fatal Crash from 2010 to
2012……………………………………………………………………………...……….24
Figure 4.2: Relation in between Control Type and Grievous Crash from 2010 to
2012……………………………………………………………………………...……….25
Figure 4.3: Relation in between Control Type and Simple Crash from 2010 to
2012………………………………………………………………………………...…….25
vii
Figure 5.1: Relation between presence of Median and Fatal Crash from 2010 to
2012…………………………………………………………………………………...….26
Figure 5.2: Relation between presence of Median and Grievous Crash from 2010 to
2012………………………………………………………………………….………..….27
Figure 5.3: Relation between presence of Median and Simple Crash from 2010 to
2012…………………………………………………………………………….………...27
Figure 6.1: Relation between Road Crash and Day Light Availability from 2010 to
2012……………………………………………………………………………………....30
Figure 6.2: Relation between Proportion of Road Crash and Day Light Availability
Figure No 7.1: Relation between Road Geometry and Crash Severity from 2010 to
2012…………………………………………………………………………………...….32
Figure No 10: Relation between Road Crash Severity and Driving License………...….41
Figure No 11: Relation between Road Crash Severity and Vehicle Loading…….…..….42
Figure No 12: Relation between Road Crash Severity and Vehicle Defect……………..43
Figure No 13: Relation between Road Crash Severity and Pedestrian Action…………..49
Figure No 14: Relations between Proportion of Road Crash and Land Use…………….55
Figure No 15: Relationship between Proportion of Road Crash and Different Age
Range…………………………………………………………………………………….57
Figure No 16: Relation between Road Crash Severity and Valid Fitness
Certificate………………………………………………………………………………...58
viii
Figure No 17: Relation between Proportion of Road Crash Severity and Vehicle
Defect…………………………………………………………………………………….61
Figure No 18: Relation between Proportion of Road Crash Severity and types of
Vehicle………………………………………………………………………………..….63
Figure No 19: Relation between Proportion of Road Crash and Collision Type…….….64
Figure No 20: Relation in between Junction types and total crash during 2010 to
2012……………………………………………………………………………………....65
Figure No 21: Relation in between traffic control system and total crash during 2010
to 2012…………………………………………………………………………...…...….66
ix
x
Chapter 1: Introduction
1.1 Background of the study
Road traffic crashes is now a major concern which cause a significant problem in terms of
loss of human life and economic consequences. Uddin, S, & Haque, (2003) found that large
buses and trucks are involved in 61% of all fatal single-vehicle accidents and two or more
vehicles involved in 58.5% of total fatal accident in Bangladesh. Road users are involved
in 90% of all contributing factors behind road crashes in which vehicle driver
characteristics are the major contributing factor (Hoque, Mahmud, & Paul, 2008).
Roadway fault, human error, road unworthy vehicle and failure in system and staring
system are also the contributing factors behind road crash (Hoque, & Hasan, 2007). Besides
the impact of vehicle factor (tyre brust, break failure), behavioral aspect of heavy vehicle
driver, travel pattern and location of garments workers on road accident have been
considered as contributing factors behind road accident in Dhaka (Hoque, & Hasan, 2007;
Hoque, Khondokar, & Hoque, 2007; Hoque, Debnath, & Mahmud, 2006). Kim, Pant, &
Yamashita found that being a male, a tourist, and alcoholic and driving a stolen vehicle are
the major contributing factors behind hit-and-run crashes. Besides roadways features, such
as horizontal alignment, weather, and lighting are also significantly related to hit-and-run
crashes. Stamatiadis, & Deacon, (1995) showed driver age, gender and seat belt as major
contributing factors behind road crashes and the change in the number of road accident
over the year. Another study showed that the sharp decline in crash rates over the past forty
years with the change in road characteristics and vehicle speed (Dumbaugh, 2005).
Intersection type, minimum angle of intersecting roadway, total number of lanes , number
of through lanes, number of right-turn lanes are considered in the study in urban areas of
Shanghai to identify their relation with road crashes (Xie, Wang, Ozbay, &Yang, 2014).
According to the study in Tucson, Arizona, the factors like population density, number of
employees, intersections density (number of intersection per acre), percentage of miles of
principal arterial, percentage of miles of minor arterials, and percentage of miles of urban
collectors were significantly related to the increase in road crashes (Ladron de Guevara,
Washington, & Oh, 2004). Study in Florida focused on a county-level road safety analysis
where DVMT (Daily vehicle miles traveled in thousands), population (in thousands) and
1
intersection density (number of intersection per road length) were found as the main
responsible factors behind road crash severity (Huang, Abdel-Aty, & Darwiche, 2010).
Various types of land use impact, driver characteristics and behavioral impact on road
accident have been considered as responsible factors behind road crashes (Kim, &
Yamashita, 2002; Kim, Nitz, Richardson, & Li, 1995). Besides, the impacts of landuse,
population, economic activity and employment on road accident have been considered in
the study in Hawaii (Kim, Brunner, & Yamashita, 2006). Driver age, weather condition,
non-use of seat belts, road alignment, vehicle type etc. have also been considered in the
study of Ontario (Wier et al., 2009). As the increase in road crash rate is a major concern,
it is necessary to conduct an exploratory study on the major contributing factors behind
road crashes in the context of Dhaka city. The study mainly focuses on analyzing the road
crashes occurred in Dhaka South and North City Corporation and finding the relation
between road crash severity and the contributing factors behind road crashes.
3
Chapter 2: Literature review
There have been some past studies in the highway safety area in which logistic regression
models had been used to identify various influential factors behind road crash. Jovanis and
Chang (1986) used linear regression in their study to find influential factors behind road
crashes and to predict crash. They found proportional relation with the increase of vehicle
kilometers travelled and crash frequency. Dissanayake and Lu (2002) used binary logistic
regression to predict the crash severity of single-vehicle fixed object crashes involving
young drivers. Influence of alcohol or drugs, ejection in the crash, point of impact, rural
crash locations, existence of curve or grade at the crash location, and speed of the vehicle
are found as the influential factors on increasing the probability of having a more severe
crash. Sarkar, Tay and Hunt (2011) used binary logistic regression to identify association
between the risk of pedestrian mortality and possible contributory factors. They have found
crashes occurring at locations with no traffic control, stop control, pedestrian crossings and
weather as the significant influential factors behind road crashes. Kim et al. (1996)
developed a logistic model to explain the probability of crashes between motorists and
cyclists. Motorist age, cyclist age, cyclist alcohol use, cyclists making turning actions, and
rural locations are found as the significant influential factors behind road crashes between
motorists and cyclists. Kim et al. (1995) used a structural model to relate driver
characteristics and behavior with the type of crash and injury severity. They have identified
the role of driver characteristics and behavior behind road crash. Odd multiplier was used
to estimate the effect of various factors behind road crashes. Stamatiadis and Deacon
(1994) used logistic regression to determine statistical significance of contributory factors
behind road crashes. Driver age, gender are identified as the statistically significant factor
behind road crash. The study also found that middle-aged drivers are safer than younger
drivers and female drivers are safer on average than male drivers and younger female
drivers are safer than younger male drivers. Zhang et al. (2000) used multivariate
unconditional logistic regression to estimate relative risk of contributory factors behind
road crashes. The study found that driver age, sex, failing to yield right-of-way/disobeying
traffic signs, non-use of seat belts, ejection from vehicle, intersection without traffic
controls, roads with higher speed limits, snowy weather, head-on collisions, two-vehicle
4
turning collisions, overtaking, and changing lanes were significantly related to the
increased risk of fatal-injury in crashes. Garcia ferrer, Juan and Poncela (2007) analyzed
the aggregate relationship between traffic crashes and real economic activity in Spain
during the last 30 years (1976-2004). They have found that the number of accidents
depends on the use of cars and other exogenous variables and the level of economic activity
which affects the variation in the stock of cars and degree of utilization. Clifton, Burnier
and Akar (2009) used order probit model to estimate the effects of the personal and location
characteristics on the severity of the injury of road crash. Age, sex, clothing type, signal
disobedience, daylight, weather, road condition, pedestrian location, population and
density are found as the major contributory factors behind road crashes. Law, Noland and
Evan (2009) examined the relationship between economic development and motorcycle
deaths. In the study, motorcycle deaths, percentage of population between 15-24 years age,
gross domestic product per capita and motorcycle helmet law were considered to examine
the relationship where motorcycle deaths was the dependent variable. The study found that
an increase in per capita income is highly associated with a rise in motorcycle deaths, but
decreases once per capita income levels exceed threshold levels. Kim, Pant and Yamashita
(2010) used binomial logistic regression to measure the influence of demographic and land
use variables on road crash. Population, land use, and accessibility measures such as road
length, bus stops, length of bus route, number of intersections, and dead ends were found
as the contributory factors behind crash fatality. The study also found that demographic
variables such as job count and number of people living below the poverty level are
significantly associated with injury crashes and pedestrian and bike crashes. Besides,
business and commercial areas are strongly associated with increased total as well as injury
and fatal crashes. Schneider et al. (2010) used negative binomial regression to examine the
association between roadway intersection characteristics and pedestrian crash risk. The
variable named intersection characteristics included traffic signal, T intersection, traffic
control, cross street lanes, curb radius, missing sidewalks, left turn only lanes and right turn
only lanes. Surrounding land use and transportation system characteristics included total
populations, total employment, school, rail stations and bus stops. Besides, pedestrian
crossing, median income and vehicle proportion also identified as the contributory factors
behind pedestrian crashes. The study found that several intersection characteristics have a
5
significant association with pedestrian crashes. Pulugurthaa, Duddu and Kotagiri (2010)
used chi-square test and negative binomial regression model to identify the association
between land use characteristics and road crash. Demographic/socio-economic
characteristics such as population, the number of house-hold units and employment, traffic
indicators such as trip productions and attractions, and, on-network characteristics such as
center-lane miles by speed limit were observed to be correlated to land use characteristics.
As a result these variables were not considered in the development of travel analysis zone
(TAZ) level crash estimation models. Mixed use development, urban residential, single-
family residential, multi-family residential, business and office district were found to be
strongly associated and statistically significant in estimating TAZ level crashes. Zahabi et
al. (2011) used ordered logit model to determine the effects of the identified variable on
road crash. The impacts of road design, land use, built environment, darkness, vehicle
movement, whether an accident occurred at an intersection, vehicle type, and land use mix
were found to be strongly associated and statistically significant in estimating the severity
of pedestrian injuries from collisions. Siddiqui and Abdel-Aty (2012) used hierarchical
Bayesian model to identify the spatial influence of neighboring zones on crashes that occur
specifically on or near the zonal boundaries. Length of the road, total number of
intersections, hotel unit, total employment and population per square mile have a
significant impact on road crashes according to the study. Long-term parking cost was also
significant in the boundary crash because more expensive parking forces more people to
walk and therefore increases pedestrian activity in the boundary. Ivan et al. (2000) observed
that volume-capacity ratio, segments with no passing zones, the shoulder width, the number
of intersections (or intersection density) and driveways (or driveway density) are
significant variables that can explain single-vehicle crashes, whereas daylight conditions,
the number of intersections and driveways are significant variables that can explain multi-
vehicle crashes. Noland and Oh (2004) found that an increase in the number of lanes and
road widths may lead to an increase in the number of road crashes, where as an increase in
shoulder width may lead to a decrease in the number crashes. Greibe (2003) found that
road environment variables, minor side roads, parking facilities and speed limit are
significant variables which can explain crashes on road links whereas only vehicle traffic
flow was found to be a significant variable for road crashes in junction. Besides traffic flow
6
(motor vehicles, heavy vehicles and vulnerable road users), length of road section, speed
limit, one/two-way traffic, number of lanes, road width, speed reducing measures and
number of minor crossings on the road were found as the major contributory factors behind
road crash. A TAZ level non-parametric safety analysis by Siddiqui et al. (2012) showed
that the total number of intersections per TAZ, light truck productions, total roadway
segment length with 35 mph posted speed limit, total roadway length with 65 mph posted
speed limit, and non-home based work productions are significant variables influencing
severe crashes. Kim et al. (1994) explained the relationship between types of crashes and
injuries occurred in motor vehicle accidents.
A model was built in the study to relate the type of crash (e.g. rollover, head-on, sideswipe,
rear-end, etc.) by using techniques of categorical data analysis and comprehensive data on
crashes in Hawaii during 1990. They also developed an ‘odds multiplier’ that enabled
comparison according to crash type of the odds of particular levels of injury relative to no
injury. The effects of seatbelt use on injury level were also examined, and interactions
among belt use, crash type, and injury level were considered and found as a significant
contributory factors behind road crash. Mercier et al. (1997) used logistic regression to
determine whether either age or gender (or both) was an influencing factor for the severity
of injuries suffered in head-on automobile collisions on rural highways. Besides, individual
variables included age of driver or passenger, position in the vehicle, and form of protection
used also found as influencing factor behind road crash. Pulugurtha and Sambhara (2011)
found that traffic crashes are location specific and are influenced by land uses and network
characteristics within its boundary. Besides on and off network characteristics (road width,
curb width, median, traffic control, pedestrian crossing, and footpath) within 0.25-mile
buffer distance can be used to statistically explain and estimate pedestrian crashes. Lee and
Abdel-Aty (2005) analyzed pedestrian–vehicle crashes at intersections in Florida. They
used an ordered probity model to identify the factors related to the injury severity levels of
pedestrians involved in crashes. Their research found that passenger vehicles are involved
in more crashes than trucks, vans, and buses. Besides they also found that the injury
severity levels of pedestrians involved in crashes is higher for non-passenger vehicles than
passenger vehicles. They also found that pedestrian and driver demographic characteristics
(age, gender, and alcohol), road geometry, traffic, and environmental conditions are the
7
main influential factors behind increasing the frequency and injury severity levels of
pedestrian crashes. Kim et al. (2007) used multinomial logit model to explore the factors
contributing to the injury severity of bicyclists in bicycle–motor vehicle crashes. The
research found that inclement weather, darkness with no streetlights, a.m. peak (06:00 a.m.
to 09:59 a.m.), head-on collision, speeding-involved, vehicle speeds above 48.3 km/h (30
mph), truck involved, intoxicated driver, bicyclist age 55 or over, and intoxicated bicyclist
increases the probability of occurrence of bicycle–motor vehicle crashes. Klop and Khattak
(1999) observed that injury severity increased in fog, after dark on unlighted sections, with
higher speed limit, on road sections with an up/downgrade. These factors were the major
contributing factors behind road crash. Whereas the injury severity decreased with
increasing average annual daily traffic, street lighting, and an interaction of the shoulder
width and speed limit. Epperson (1995) found that socioeconomic factors (education,
employment and income), particularly the percentage of poor households within a
neighborhood, played an important role for increasing the bicycle accident rates. Pless et
al. (1989) used logistic regression to determine the role of social, familial, personal, and
behavioral characteristics behind pedestrian and bicyclists injuries. They found that the
child's personality and behavior are negatively related to pedestrian and bicyclist injuries
whereas family and neighborhood characteristics are strongly related to those injuries.
Family and neighborhood characteristics included mother’s education, number of childrens
in a family, mother employed, and history of accident in a family. Noland and Quddus
(2004) found that alcohol expenditure per capita was significantly correlated with bicyclist
casualties although it was not clear whether this was due to the intoxication of motorists or
bicyclist.
8
Chapter 3: Methodology
3.1 Formulation of Objectives
Research objectives have been formulated to study the location of accident of Dhaka South
City Corporation (DSCC) and Dhaka North City Corporation (DNCC) area. Research
objectives focus on identifying relationship between numbers of crashes and factors that
influence the occurrence of that location.
9
In addition, land use data of DNCC and DSCC area were collected from the shapefile of
Detail area planning (DAP). Then the crash data collected from ARI was overlaid on the
land use data of DCC area using Geographic information software (GIS). After overlapping
the crash data and land use data, the 400 meter buffer for each crash was taken to analyze
the land use impact on road crash. Then the percentage of each types of land use was
calculated within the 400 meter buffer for each crash.
Table 1: Number of crash according to the type of crash severity and year of
occurrence
10
significant variables and nonsignificant explanatory variables for road crash and chi square
test has been used to identify the goodness of fit of the model.
For the independent variables, age is the only continuous variable whereas the others are
categorical. Since some of the categorical variables have several levels which are coded as
1, 2, 3, and so forth. As a result, dummy variable was needed for representing the
independent variable used in model and match the format of SPSS software used in the
research. Then the dummy variables are coded having k-1 design variables for k level of
the categorical independent variable.
An example of this coding is given in Table 2 for the variable Vehicle defects which has
four levels, so there will be three design variables. When the respondent is ‘No defects’,
the three design variables, D1, D2, and D3, would set to equal zero. When the respondent
is ‘Brakes,’ D1 would be set equal to 1 whereas D2 and D3 would be 0; and so forth for
the other variables.
11
Table 2: Coding of categorical independent variable
Vehicle defects D1 D2 D3
No defects 0 0 0
Brakes 1 0 0
Steering 0 1 0
Multiple 0 0 1
E(Y/x) = β0+β1x,
Where Y denotes the dependent variable, x denotes a value of the independent variable, β0
denotes the value of coefficient and the βi-values denote the model parameters. The
equation denotes the expected value of Y given the value of x. Many distribution functions
have been proposed for use in the analysis of a dichotomous response variable (Hosmer
and Lemeshow, 1989; Agresti, 1984; Feinberg, 1980). The specific form of the logistic
regression model is:
𝑒 𝛽0+𝛽1𝑥
π (x)=
1+𝑒 (𝛽0+𝛽1𝑥)
Here, π(x) = E(Y/x). The transformation of the π(x) logistic function is known as the logit
transformation:
π (x)
g (x)= ln [ ] = β0+β1x.
1−π (x)
The logit function, g(x), can be linear, continuous or may range from minus infinity to plus
infinity, depending on the range of x.
Each accident in the collected data used for the research have been divided into two
categories. One is fatal accident and the other is non-fatal accident.
12
The logistic model used in the research is,
𝑒 𝑔 (𝑥)
P(non-fatal accident)= π(x) = and thus,
1+𝑒 𝑔 (𝑥)
1
p (fatal accident) = 1−P(non-fatal accident) = 1- π(x) = . Here, g (x) refers to the
1+𝑒 𝑔 (𝑥)
function of independent variable. Where, g(x) = β0+ β1x1+ β2x2+ β3x3+……. + βnxn.
Here, D= -2ln (likelihood of the current model - likelihood of the saturated model). The
equation also called the likelihood ratio.
The change in D due to inclusion of the independent variable in the model is measured as
follows:
G= D (for the model without the variable) – D (for the model with the variable)
To identify the significance of an independent variable, the changes in log likelihood ratio
for including the particular independent variable is measured.
13
Table 3: Details of variables defined for data collection
Variable Label
1= not at junction
2= four leg
3=roundabout
4= railway
5= T junction
1= No control
2= police control
3= center line
4= Traffic light
1= Head on
2= Rear end
5= Hit pedestrian
1= Yes
2= No
14
Weather Weather (nominal)
1= Fair
2= Fog
3= Rain
4= Wind
1= Dawn
2= Day
3= Night (lighted)
4= Night (unlighted)
2= Curve only
3= Slope only
1= Dry
2= Wet
1= Sealed
1= Good
15
2= Rough
1= Yes
2= No
1= Yes
2= No
(nominal)
1= Bus
2= Motorcycle
3= Truck
4= Pick up
5= Cycle
6= Private
1= Legal
2= Illegal
1= Brakes
2= Multiple
3= Steering
16
4= None
1= Male
2= Female
1= Fatal
1= None
2= Crossing
17
Chapter 4: Explorative analysis of responsible factors
In order to achieve the first objective in this chapter, different types of exploratory analysis
will be done. Different types of independents factors those are responsible for road crash
will be analyzed in this chapter. To identify the factors severity of road crash different types
of Cross tabulation, Microsoft Excel analysis will be done.
400
350
300
250
200
150
100
50
0
Total Road Grievous
Fatal Crash Simple Crash
Crash Crash
2010 455 296 106 53
2011 400 259 92 49
2012 369 253 79 37
18
The percentages of three types of road crash in 2010 have shown in (Figure 2.1). The
highest percentage belongs to fatal crash and that is 65 percent. The lowest is 12 percent
and that is for simple crash. This calculation represents that the number of fatal crash is
greater than other crash in 2010. The (Figure 2.2) represents the percentage of different
types of road crash in 2011. The percentage of three types of road crash in 2011 is almost
same in previous. There is no change in the percentage of three types of road crashes in
between 2010 and 2011. Again (Figure 2.3) represents the percentage of different types of
road crashes in 2012. Here the percentage of fatal crash has increased in previous two years.
But the percentages of simple and grievous crash have decreased in previous two. The
increasing rate of fatal crush is not so very high. Only 4% of fatal crash has increased in
previous three years, on the other hand the percentages of simple crash and grievous crash
have decreased and those percentages both are only 2%.
12%
Fatal Crash
23% Grievous Crash
Simple Crash
65%
19
Percentage of Different Types of Crash in 2011
12%
Fatal Crash
23% Grievous Crash
Simple Crash
65%
10%
Fatal Crash
21%
Grievous Crash
Simple Crash
69%
20
4.2 Road characteristics
4.2.1 Junction type
(Figure 3.1) represents the relationship in between junction types and fatal crash from 2010
to 2012. Graph represents that the number of fatal crash is the highest at the position of not
at junction. After that T-junction is the second highest position for occurring fatal crash.
Four-leg intersection is another risky point for occurring fatal crash. So, T-junction and
Four-leg junction are very dangerous for occurring fatal crash. Roundabout junction and
staggered junction are free from occurrence of fatal crash. (Figure 3.2) represents the
relation in between junction types and grievous crash from 2010 to 2012. The number of
grievous crash is the highest at the place of not at junction in three years. Within junction
T-junction and Four-leg junction are risky for occurring grievous crash. T-junction is more
dangerous than four-leg intersection for occurring grievous crash. At road intersection or
junction traffic rules and regulations are strictly maintained. The average speed of vehicle
at intersection is comparatively lower than not at junction. For that reasons the number of
crash both fatal and grievous are lower at intersection point. (Figure 3.3) represents that
the relation in between junction types and simple crash from 2010 to 2012. As usual the
number of simple crash is the highest at the position of not at junction. But four-leg
intersection is more responsible than other types of intersection for occurring simple crash.
Within different types of intersection four-leg intersection is the most dangerous place for
occurring simple crash. Second dangerous place for occurring simple crash is T-junction.
Statistics shows that roundabout junction and other types of junctions are almost danger
free intersection for occurring any types of crash. The number of simple crash is decreasing
year by year at four-leg intersection and not at junction, which means people are becoming
conscious about these issues and traffic rules and regulations are also upgrading.
21
Relation in between Juction Types and Fatal Crash
Figure 3.1: Relation in between Junction types and Fatal Crash from 2010 to 2012
Source: Accident Research Institute (ARI)
Number of Grievuous Crash
Figure 3.2: Relation in between Junction types and Grievous Crash from 2010 to 2012
Source: Accident Research Institute (ARI)
22
Number of Grievuous Crash
Relation in between Juction Types and Simple Crash
30
25
20
15
10
5
0
T Roundabo Staggered Not at Railway
Four Leg
Junction ut Junction Junction Junction
2010 6 0 20 0 27 0
2011 7 2 16 1 22 1
2012 7 0 13 0 17 0
Figure 3.3: Relation in between Junction types and Simple Crash from 2010 to 2012
Source: Accident Research Institute (ARI)
23
positive impact over the traffic control system. The number of grievous crash at police
control zone is gradually decreasing. It is very positive. Driver, Pedestrian and Traffic-
Police are gradually increasing consciousness about road crash. (Figure 4.3) represents the
relation in between control type and simple crash from 2010 to 2012. The number of simple
crash of police control and no control zone is almost high than other zone. The number of
simple crash at no control zone is decreasing year by year. It brings a very positive impact.
But number of simple crash at police control intersection does not significantly change.
Overall (Figure: 4.1, 4.2, and 4.3) the major findings are the number of crash simple,
grievous or fatal almost highest where there is no control over the traffic. Crash number is
also at police control intersections. These are general findings overall three types of crash.
200
150
100
2010
50 2011
2012
0
Police Police+ Center Line Traffic Light No Control
Control Traffic Light
Control Type
Figure 4.1: Relation in between Control Type and Fatal Crash from 2010 to 2012
24
Relation in between Control TypeAccident
Source: and Grievous Crash
Research Institute (ARI)
Number Of Grrievous Crash 60
50
40
30
2010
20
2011
10
2012
0
Police Police+ Center Line Traffic Light No Control
Control Traffic Light
Control Type
Figure 4.2: Relation in between Control Type and Grievous Crash from 2010 to 2012
Source: Accident Research Institute (ARI)
25
20
15
2010
10 2011
5 2012
0
Police Police+ Center Line Traffic Light No Control
Control Traffic Light
Control Type
Figure 4.3: Relation in between Control Type and Simple Crash from 2010 to 2012
25
4.2.3 Presence or absence of Median
(Figure 5.1) represents the relation between presence of median and fatal crash from 2010
to 2012. Median separates the face to face vehicle movement. Generally it reduces the face
to face vehicle crash. So it can be assumed that the probability of road crash will be reduced
where the median presents. But this graph represents that presence of median increases the
road crash. Though year wise the number of fatal crash is decreasing both presence of
median and absence of median but the number of fatal crash is always greater where
median presents. Most probably drivers do not obey the traffic rules and regulations and
drive their vehicles at their willing. The highest number of fatal crash because of presence
of median is 237 in the year of 2010. The number of fatal crash is gradually decreasing
because of presence of median. In (Figure 5.2 and Figure 5.3) represent the relation in
between presence of median and grievous crash and simple crash. These graphs give an
assumption that where median presents the number of grievous crash and simple crash is
comparatively greater than where median absence. But year wise both types of road crashes
are decreasing. It is very positive issues that median of road is gradually helpful to reducing
different types of road crash.
200
150
100
50
0
2010 2011 2012
Presence of Median 237 205 164
Absence of Median 59 54 89
Figure 5.1: Relation between presence of Median and Fatal Crash from 2010 to 2012
26
Relationship between presence of Median and Grievous Crash
100
Figure 5.2: Relation between presence of Median and Grievous Crash from 2010 to 2012
45
40
35
30
25
20
15
10
5
0
2010 2011 2012
Presence of Median 45 37 32
Absence of Median 8 12 5
Figure 5.3: Relation between presence of Median and Simple Crash from 2010 to 2012
27
Here, accident severity is classified into two classes one is non-fatal that is the sum of
grievous crash and simple crash and another is fatal crash. (Table 4) represents the cross
tabulation in between crash severity and median. Within crash severity 81% Non-fatal
crashes are responsible for presence of median and 19% Non-fatal crashes are responsible
for absence of median. Within Median 35.7% presence of median is responsible for
occurring Non-fatal crash and 27.6% absence of median is responsible for occurring Non-
fatal crash. Within total crash 33.8% crash is non-fatal and 66.2% crash is fatal. Within
33.8% non-fatal crash, 27.4% non-fatal crash has occurred where median is present and
6.4% non-fatal crash has occurred where median is absent. Within fatal crash 74.5% fatal
has occurred where median presents and 25.5% fatal crash has occurred where median is
absent. Again within median 64.3% median is responsible for occurring fatal crash and
72.4% absence of median is responsible for occurring fatal. Within total crash 66.2% crash
is fatal crash. Within 66.2% fatal crash 49.3% fatal crash has occurred where median is
present and 16.8% fatal has occurred where median is absent. Now, within total crash
76.7% crash has occurred where median is present and 23.3% crash has occurred where
median is absent.
28
Table 4: Cross Tabulation between percentage of Crash Severity and Median
Here, Non-Fatal Crash = Sum of number of grievous crash and number of simple crash.
Median
Presence Absence
of Median of Median Total
29
4.2.4 Light condition of road
(Figure 6.1) represents the relation between road crashes severity and day light availability
from 20010-2012. At day time different types of road crashes are the highest. At day time
more traffic moves in Dhaka City Corporation Area. Office activity and types of movement
have done at day time. For that reasons the number of road crash at day time is larger than
other times of day. The number of fatal crash is larger than other types of crash and the
maximum fatal crashes have occurred at day time. The lowest numbers of fatal crash and
other types of road crash have occurred at the time of night lighted.
250
200
150
100
50
0
Griev Simp Griev Simp Griev Simp
Fatal
Fatal Fatal
ous le ous le ous le
2010 2011 2012
Dawn 38 2 5 21 5 3 30 3 1
Day 181 86 48 199 78 45 175 66 35
Night (UnLight) 53 17 0 21 9 1 32 9 1
Night (Light) 24 1 0 18 0 0 16 1 0
Figure 6.1: Relation between Road Crash and Day Light Availability from 2010 to 2012
Source: Accident Research Institute (ARI)
(Figure 6.2) represents the relation between proportion of road crash and day light
availability from 2010 to 2012. Maximum road crash almost three-fourth of total crash has
occurred at day time. The proportion of road crash at day time is 0.746. That means within
total road crash 74.6% road crash has occurred at day time. The lowest portion of road
crash is belonging to the night lighted time and that value is 0.049. That means within total
road crash only 4.9% road crash has occurred at night lighted time. So at day time the
portion of road crash is highest than other times of day.
30
Comparison between Proportion of Road Crash and Day Light
Availability
0.800
0.700
0.600
Proportion
0.500
0.400
0.300
0.200
0.100
0.000
Night
Dawn Day Night (Light)
(UnLight)
Proportion 0.088 0.746 0.117 0.049
Figure 6.2: Relation between Proportion of Road Crash and Day Light Availability from
2010 to 2012 Source: Accident Research Institute (ARI)
31
Number of Different Types of Road
Relation in between Road Geometry and Road Crash Severity
350
300
250
200
Crash 150
100
50
0
Grie Simp Grie Simp Grie Simp
Fatal Fatal Fatal
vous le vous le vous le
2010 2011 2012
Curve Only 3 2 0 4 3 0 1 2 0
Curve and Slope 1 0 0 0 0 0 3 0 0
Straight and Flat 291 103 53 255 89 49 249 77 36
Slope Only 1 1 0 0 0 0 0 0 1
Figure No 7.1: Relation between Road Geometry and Crash Severity from 2010 to 2012
Source: Accident Research Institute (ARI)
According to (Figure No: 7.2) the highest portion of road crash is 0.982 and it has occurred
because road geometry is straight and flat. Because of straight and flat geometry, almost
98.2% road crash has occurred. Straight and flat road is responsible for increasing the
vehicle velocity.
0.600
0.400
0.200
0.000
Curve and Straight and
Curve Only Slope Only
Slope Flat
Proportion 0.012 0.003 0.982 0.002
Figure No 7.2: Relation between Proportion of Road Crash and Road Geometry
Source: Accident Research Institute (ARI)
32
In (Table No: 05) cross tabulation in between accident severity and road geometry has
represented. Within total crash 33.8% crash is non-fatal and 66.2% crash is fatal. Within
total non-fatal crash almost 97.8% non-fatal crash has occurred on straight and flat road.
Again within road geometry 66.7% slope road faces to non-fatal crash and 46.7% curve
road faces to non-fatal crash. On another hand 33.7% straight road faces to non-fatal crash.
So for occurring non-fatal crash slope road is more responsible. Again within fatal crash,
98.4% fatal crash has occurred on straight and flat road. Within road geometry 33.3% slope
road faces to fatal crash, 53.3% curve road faces to fatal crash and 66.3% straight road
faces to fatal crash. Within total non-fatal crash 33.1% non-fatal crash has occurred on
straight and flat road. Within total fatal crash 65.1% fatal crash has occurred on straight
and flat road. Within total crash 98.2% crash has occurred on straight and flat road.
33
Table No 05: Cross tabulation between percentage of Crash severity and Road
geometry
Road geometry
34
4.3 Temporal factors
4.3.1 Hour of the day
Here this table (Table No: 06) represents the percentages of fatal crash in different three
years on the basis of time interval. Time interval is two hours that means on the basis of
two hours difference the whole day is divided into twelve portions. The percentage of fatal
crash of each portion has calculated and whole values are represented on the table. On the
basis of three years crash data, for fatal crash the most risky intervals are 10.01 am to 12.00
pm and another one is 4.01 pm to 6.00 pm. because these are the peak time. At these time
traffic volume is very high. The highest percentage of fatal crash in 2012 is 17.615% and
it is the highest value all over the three years data and the time interval is 10.01 am to 12.00
pm. This time is most risky because in this time the percentage of fatal crash is very high.
City activity and traffic flow will have increased in this time.
Again table (Table No: 07) represents the percentage of grievous crash in different three
years on the basis of time interval. The most grievous crash prone time interval is 10.01
am to 12.00 pm. According to the percentage of grievous crash data in 2010, 2011 and
2012 the highest percentages are 24.30%, 26.09% and 25.32% at the time interval of 10.01
am to 12.00 pm. According this table the second highest grievous crash interval time is
12.01 pm to 2.00 pm. This time the percentages of grievous crash in 2010, 2011 and 2012
are 16.82%, 16.305 and 17.72%.
According to (Table No: 06) the highest percentages of total crash in 2010, 2011 and 2012
are 14.85%, 16.25% and 17.62% and all these crashes had occurred at the time interval of
10.01 A.M to 12.00 P.M. According to the information of percentage of total crash another
time interval of 4.01 P.M to 6.00 P.M is the second highest crash prone time. Because this
time interval is peak time and high volume of traffic moves of this time.
35
Table No 06: Percentage of Fatal Crash of 2010, 2011 and 2012 and Time interval
36
Table No 07: Percentage of Grievous Crash of 2010, 2011 and 2012 and Time interval.
37
Table No 08: Percentage of Total Crash of 2010, 2011 and 2012 and Time interval
38
Comparison between Proportion of Road Crash and Time
0.18
0.16
Proportion 0.14
0.12
0.1
0.08
0.06
0.04
0.02
0
10.0 10.0
8.01 12.0 8.01 12.0
6.01 1 2.01 4.01 6.01 1 2.01 4.01
AM- 1 PM- 1
AM- AM- PM- PM- PM- PM- AM- AM-
10.0 PM- 10.0 AM-
8.00 12.0 4.00 6.00 8.00 12.0 4.00 6.00
0 2.00 0 2.00
AM 0 PM PM PM 0 AM AM
AM PM PM AM
PM AM
Proportion 0.07 0.09 0.16 0.13 0.13 0.15 0.12 0.05 0.03 0.03 0 0.03
4.3.2 Month
Relations between proportion of road crash and months have shown on (Figure No: 16).
There has no significance change on road crash of different months. But the highest
proportion of road crash is 0.10 and the months are January, March and August. In these
three months the highest portion of road crash are existing. The lowest proportion of road
crash is 0.06 and the months are May, June and September.
39
Relations between Proportion of Road Crash and
Months
0.12
Proportion 0.1
0.08
0.06
0.04
0.02
0
Sept Nov Dec
Janu Febr Mar Apri Aug Octo
May June July emb emb emb
ary uary ch l ust ber
er er er
Proportion 0.11 0.08 0.11 0.08 0.07 0.07 0.08 0.11 0.06 0.07 0.1 0.07
40
Relation in between Road Crush Severity and Driving License
Number of Different Types of Road Crash
250
200 Having
Drivin
150 g
Licens
100 e
50
0
Simple
Simple
Simple
Grievous
Grievous
Grievous
Fatal
Fatal
Fatal
2010 2011 2012
Year
Figure No 10: Relation between Road Crash Severity and Driving License
Source: Accident Research Institute (ARI)
Important findings have come out from (Figure No: 11). This figure represents the relation
between road crash severity and vehicle loading. In general concept when a vehicle bears
the load beyond its capacity, the probability of road crash will increase. In this graph same
phenomena has observed. The number of fatal crash, grievous crash and simple crash are
almost greater where vehicle loading beyond its capacity. So, to minimize the road crash
vehicle loading must be controlled and the vehicle loading must be existed within capacity
otherwise the number of road crash cannot be controlled.
41
Relation in between Road Crush Severity and Vehicle
Loading
300
Figure No 11: Relation between Road Crash Severity and Vehicle Loading
Source: Accident Research Institute (ARI)
4.4.2 Vehicle defects
According to (Figure No: 12) vehicle defects are classified into three classes and those are
brakes, multiple and another is steering. Within three defects of road crash defects of break
is the most visible phenomena and its number is the highest. So this statistics gives that
because of brake failure different types of road crashes have occurred.
42
Relation in between Road Crush Severity and Vehicle Defect
Figure No 12: Relation between Road Crash Severity and Vehicle Defect
Source: Accident Research Institute (ARI)
Here (Table No: 9.1; 9.2; 9.3 and 9.4) represents the vehicle defect and percentage of fatal,
grievous, simple and total crash. According to the (Table No: 9.1) the highest percentage
of fatal crash in different three years are 40.54%, 49.03% and 57.71%. Brake failure of
vehicle is responsible factor for occurring highest percentage of fatal crash in every year.
Similar results have found on other tables.
Table No 9.1: Vehicle Defect and Percentage of Fatal Crash in different three years.
43
Table No 9.2: Vehicle Defect and Percentage of Grievous Crash in different three
years.
Table No 9.3: Vehicle Defect and Percentage of Simple Crash in different three years.
44
4.4.3 Vehicle type
According to (Table No: 10.1) the percentage of grievous crash for bus is the highest. The
second highest percentage of grievous crash is for car. So, the most grievous crash prone
vehicle is bus and second vehicle is car. Again according to (Table No: 10.2) the highest
simple crash prone vehicle is bus and the percentage of simple crash is for bus in different
three years is 49.05%, 48.98% and 45.94%. That means within simple crash almost half
times of simple crash occur because of bus. Car and Truck are also simple crash prone
vehicles. According to (Table No: 10.3) the highest crash prone vehicle is bus. The
percentage of total crash is for bus in different three years is 58.02%, 63.75% and 60.16%.
Truck and car is the second and third crash prone vehicles
Table No 10.1: Vehicle Types and Percentage of Grievous Crash in different three
years
45
Table No 10.2: Vehicle Types and Percentage of Simple Crash in different three years.
Table No 10.3: Vehicle Types and Percentage of Total Crash in different three years.
46
In (Table No: 11) cross tabulation in between crash severity and types of vehicle has shown.
Within non-fatal crash 67.1% non-fatal crash occur where bus is involved. Within vehicles
77.8% baby-taxi faces to non-fatal crash and 32.7% bus faces to non-fatal crash. To occur
a non-fatal crash, baby-taxi is the most crash prone vehicle. Within fatal crash 70.7% fatal
crash occur where bus is involved. Within vehicles 67.3% bus faces to fatal crash and
70.7% track faces to fatal crash. 58.7% car faces to fatal crash. Within total crash 69.5%
crash occur where bus is involved. So the major findings after completing cross-tabulation
in between crash severity and different types of vehicles, for occurring non-fatal crash the
percentage of baby-taxi is the highest and that is 77.8%. It means within vehicles 77.8%
baby-taxi faces to non-fatal crash. On another hand within vehicles 70.7% truck faces to
fatal crash. So truck is risk for fatal crash. Again within fatal and non-fatal crash in both
cases the percentage of crash is the highest for bus. So in this perception, bus faces more
fatal and non-fatal crash than other types of vehicles.
47
Table No 11: Cross tabulation between percentage of Crash severity and Vehicle type
Vehicle type
Baby- Motor
Taxi Bus Truck Car Pickup Cycle Total
48
4.4.4 Pedestrian action
In (Figure No: 13) the relation between road crash severity and pedestrian action has
shown. While a vehicle hits on pedestrian, that’s time pedestrian engages some kinds of
activity. These activities are classified into four categories. But one phenomena has
observed the most and that is while a pedestrian crosses the road, that’s time vehicle hits
the pedestrian. This number is almost high. Thus causes fatal, grievous or simple crash
occurs. Another one is while vehicle hits the pedestrian, pedestrian has no activity. It is the
second highest in overall activities.
Figure No 13: Relation between Road Crash Severity and Pedestrian Action
Source: Accident Research Institute (ARI)
According to (Table No: 12.1) the highest percentage of fatal crash has occurred while
pedestrians cross the road. According to this table the highest percentage of fatal crash is
78.04% in 2010 and this has happened while pedestrians cross the road. Not only in 2010
but also 2011 and 2012 these same phenomena has happened. So the pedestrians are not
serious about traffic rules and regulations. Pedestrians and driver of vehicles do not
maintain zebra crossing. Pedestrians do not maintain the traffic rules while crossing the
49
road. Almost 70.75% fatal crash has occurred during crossing the road of pedestrian in
2012.
According to (Table No: 12.2) the percentage of grievous crash is the highest while
pedestrians cross the road. The percentage of grievous crash while pedestrians cross the
road is 70.17% in 2012. That means almost 70% grievous crash occurs while pedestrians
cross the road. So while pedestrians cross the road they must be followed traffic signals
and must be used zebra crossing while crossing the road.
According to (Table No: 12.3) the highest percentage of simple crash occurs while
pedestrians do not take any activity. The highest percentage of simple crash in 2012 is
91.89%. So it can be said that because of taking no actions of pedestrians, simple crash has
occurred.
According to (Table No: 12.4) the highest percentages of crash in 2010, 2011 and 2012 are
67.03%, 63% and 60.98%.This figure gives a relation in between pedestrian activity and
road crash.
Table No 12.1: Name of Pedestrian Activity and Percentage of Fatal Crash from 2010-
2012
50
Table No 12.2: Name of Pedestrian Activity and Percentage of Grievous Crash from
2010-2012
Table No 12.3: Name of Pedestrian Activity and Percentage of Simple Crash from
2010-2012
51
Table No 12.4: Name of Pedestrian Activity and Percentage of Total Crash from 2010-
2012
52
Table No 13: Cross tabulation of percentage between road class and crash severity
Road class
53
4.5 Spatial characteristics
4.5.1 Land use
According to (Table No: 14.1) the relation between land use and percentage of fatal crash
in total years has shown. The highest percentage of fatal crash has occurred in the area of
transport land and that percentage is 24.11%. The second highest fatal has occurred in
residential area and the percentage is 21.71%. Again, the relation between land use and
percentage of grievous crash has shown on (Table No: 14.2). As usual, the highest
percentage of grievous crash has occurred in the area of transport land and the second
highest percentage of grievous crash has occurred in residential area. Again, In (Table No
14.2), the relation between land use and percentage of simple crash has shown. Same
phenomena have observed in this case. The highest percentage of simple crash has occurred
in the area of transport land and the percentage is 24.84%. So after observed the crash
severity in different types of land, it can be said that the highest crash prone area is transport
land and second highest crash prone area is residential land.
Table No 14.1: Land Use and Percentage of Fatal Crash from 2010 to 2012
Name of Land Use Percentage of Fatal Crash from 2010 to
2012
Commercial 5.81
Residential 21.71
Transport 24.11
Mixed Use 16.77
Education 14.43
Water body 17.18
Source: Accident Research Institute (ARI) and Detail Area Plan (DAP), 2010
Table No 14.2: Land Use and Percentage of Grievous Crash from 2010 to 2012
Name of Land Use Percentage of Grievous Crash from
2010 to 2012
Commercial 6
Residential 22.68
Transport 24.84
Mixed Use 15.75
Education 13.78
Water body 16.96
Source: Accident Research Institute (ARI) and Detail Area Plan (DAP), 2010.
54
Table No 14.3: Land Use and Percentage of Simple Crash from 2010 to 2012
Name of Land Use Percentage of Simple Crash from 2010
to 2012
Commercial 5.69
Residential 21.71
Transport 23.67
Mixed Use 16.37
Education 14.06
Water body 18.51
Source: Accident Research Institute (ARI) and Detail Area Plan (DAP), 2010.
In (Figure: 14) the relation between proportion of road crash and land use has shown. The
proportion of road crash 0.356 and that is associated with transport land. The second
highest ratio of road crash is 0.345 and it is associated with residential land. That means
besides residential land the number of occurrence of road crash is comparatively larger
than commercial, education are mixed used land. The lowest of road crash has observed
beside water body.
0.200
0.150
0.100
0.050
0.000
Comme Residen Transpo Mixed Educati Waterb
rcial tial rt use on ody
Proportion of Road Crash 0.039 0.345 0.356 0.109 0.095 0.057
Figure No 14: Relations between Proportion of Road Crash and Land Use
Source: Accident Research Institute (ARI) and Detail Area Plan (DAP), 2010
55
In this chapter some influential factors have come those are highly associated with road
crash. Road crashes have occurred within junction and out of junction but the highest
number of road crash has occurred at out of junction. That means not at junction is the most
crash prone area. No control of vehicle is another factor that influences on road crash.
Presence of median is another influential factor to occur road crash. Most of road crashes
have happened at the time of day. Road geometry is classified into four classes but the
number of occurrence of road crash is the highest in straight and flat road. Again, those
drivers have no driving license they more responsible for occurring road crash. Maximum
road crashes happens because of brake failure of vehicles. Carrying load beyond capacity
of vehicle is one of the most influential factors to occur a road crash and bus is associated
with more road crash than other types of vehicles. Maximum pedestrians fall into road
crash while they cross the road and within different types of road class the maximum
numbers of road crashes are associated with city road rather than other types of road. Last
of all within different uses of land, transport land is more associated with crash related
incidents. But the relation between crash severity and month is not very significant. There
has no significance outcome in this relation.
56
Chapter 5: Modelling explanatory factors behind road crash
severity
In order to explore the relationship between road crash severity with spatial, demographic
and driving characteristics, statistical model has been used and explorative analysis of
different variables has been done in the chapter. The section identifies the independent
variables which have significant impact on road crash severity through using statistical
model and explorative analysis of variables.
0.4
0.3
0.2
0.1
0
<= 20 21-30 31-40 41-50 >50
Proportion 0.04003268 0.187091503 0.566176471 0.202614379 0.004084967
Figure No 15: Relationship between Proportion of Road Crash and Different Age Range
Source: Accident Research Institute (ARI)
57
5.2 Vehicle characteristics
5.2.1 Valid fitness certificate
According to (Figure No: 16) those vehicles have valid fitness certificate it faces to road
crash more than other vehicles those have not valid faintness certificate. This calculation
gives information and that is within Dhaka City Corporation area the fitness vehicles face
to more road crash.
Figure No 16: Relation between Road Crash Severity and Valid Fitness Certificate
Source: Accident Research Institute (ARI)
58
Cross tabulation in between valid fitness certificate and crash severity has been showing
on (Table No: 15). Within total non-fatal crash 73.3% non-fatal crash has happened where
the vehicles do not have valid fitness certificate and 26.7% non-fatal crash has happened
where the vehicles have valid fitness certificate. On the other hand, 39.1% vehicles not
having valid fitness certificate are associated with non-fatal crash and 24.8% vehicles those
have valid fitness certificate are associated with non-fatal crash. Within total non-fatal
crash 24.8% non-fatal crash of total crash occur where vehicle fitness certificate is absent
and 9% non-fatal crash occur where there have valid fitness certificates. Now, within fatal
crash, 58.5% fatal crash has occurred where vehicles do not have valid fitness certificate
and 41.5% fatal crash has occurred where vehicles have valid fitness certificate. Again
60.9% vehicles having valid fitness certificate are associated with fatal crash and 75.2%
vehicles not having valid fitness certificate are associated with fatal crash. Within total fatal
crash 38.7% fatal crash have occurred for vehicles not having valid fitness certificate and
27.4% fatal crash have occurred for unfit vehicle. Now within total crash both fatal and
non-fatal, 63.5% crash occur those vehicles are unfit and 36.5% crash occurring those
vehicles are fit.
59
Table No 15: Cross tabulation between crash severity and valid fitness certificate of
vehicle
Do not
Have Valid Have Valid
Fitness Fitness
Certificate Certificate Total
Crash severity Non fatal % within Crash severity 73.3% 26.7% 100.0%
60
5.2.2 Vehicle defect
According to (Figure No: 17), the proportion of road crash for brake failure of vehicle is
0.507. That means because of brake failure of vehicle almost 50.7% vehicles face to road
crash. The ratio of proportion of multiple defects is 0.302. That means because of multiple
defect almost 30.2% vehicles face to road crash. The proportion of steering defect is 0.19.
That means because of steering defect almost 19% vehicles face to road crash. Here,
Example:
0.3
0.2
0.1
0
Brakes Multiple Steering
Proportion 0.507352941 0.302287582 0.190359477
Figure No 17: Relation between Proportion of Road Crash Severity and Vehicle Defect
61
5.2.3 Vehicle type
According to (Table No: 16) bus faces more fatal crash than other types of vehicles. This
percentage is gradually increasing year by year. Here in 2010 the percentage of fatal crash
for bus is 55.74% and in 2012 this value is 61.66% .The result indicates in between years,
the percentage of fatal crash for bus has increased by 5.92%.
Table No 16: Vehicle Types and Percentage of Fatal Crash in different three years.
According to (Figure No: 18) most of the portion of total crash has belongs to bus. The
proportion of total for bus is 0.60.5. That means within total road crash in Dhaka City
Corporation area about 60.5% road crashes occur where bus is involved. Bus plays a vital
factor to occur road crash in study area. To reduce road crash the license of bus driver and
valid fitness certificate of each bus must be checked in a proper way. This great portion
where bus is responsible has a great significant. This massive portion where bus involved
gives information and that is bus drivers are not conscious about road and they drive their
vehicle at their willing and avoid traffic rules and regulations. Proper steps should be taken
to reduce road crash.
62
Relationship between Proportion of Road Crash and Types of Vehicle
0.7
0.6
Proportion 0.5
0.4
0.3
0.2
0.1
0
Motor-
Bus Truck Car Pickup Baby-Taxi
Cycle
Proportion 0.6053921570.0285947710.1715686270.1519607840.0228758170.019607843
Figure No 18: Relation between Proportion of Road Crash Severity and types of Vehicle
Source: Accident Research Institute (ARI)
63
Relationship between Proportion of Road Crash and Collision
Type
0.7
0.6
0.5
Proportion
0.4
0.3
0.2
0.1
0
Hit Hit
Hit Hit
Head object Object Rear Right Side
Parked Pedestr Other
on on off End Angle Swipe
vehicle ian
Road Road
Proportion 0.068630.017970.023690.021240.656860.18546 0.0098 0.007350.00899
Figure No 19: Relation between Proportion of Road Crash and Collision Type
Source: Accident Research Institute (ARI)
64
Relation between Proportion of Road Crash and Junction Type
Figure No 20: Relation in between Junction types and total crash during 2010 to 2012
Source: Accident Research Institute (ARI)
65
Relation between proportion of road crash and trffic control type
0.7
0.5
0.4
0.3
0.2
0.1
0
No control Police control Traffic light Police controlled Center line
and traffic light
Traffic control system Proportion of
road crash
Figure No 21: Relation in between traffic control system and total crash during 2010
to 2012
Source: Accident Research Institute (ARI)
66
Table 17: Variables considered in logistic regression modelling
Variables Interpretation
Binary response =1 Binary response =0
Crash severity Fatal crash Non-fatal crash
No control (dum) Do not have any traffic Have traffic control
control
Police control (dum) Have police controlled Do not have police
traffic system controlled traffic system
Not at junction (dum) Crash occurred outside Crashes occurred at
junction junction
Roundabout (dum) Crash occurred at Crash occurred outside
roundabout junction roundabout junction
Hit_object_off_road (dum) Crashes occurred for Crashes occurred for not
hitting object off road hitting object of road
Hit_pedestrian (dum) Crashes occurred for Crashes occurred for not
hitting pedestrian hitting pedestrian
Dawn (dum) Crashes occurred at dawn Crashes occurred at other
time
Night_unlighted (dum) Crash occurred at night Crash occurred at night
with no street light where street light is
avilable
Straight_and_flat (dum) Road geometry is straight Road geometry is not
and flat straight and flat
Curve_only (dum) Road geometry is curved Road geometry is not
curved
Bus (dum) Bus is responsible for road Other vehicles are
crash responsible for crash.
Heavy_truck (dum) Heavy truck is responsible Other vehicles are
for road crash responsible for crash.
Car (dum) Car is responsible for road Other vehicles are
crash responsible for crash.
Brakes (dum) Crashes occurred for brake Crashes occurred for other
failure of vehicle defects in vehicle
67
Multiple (dum) Crashes occurred for Crashes occurred for other
multiple defect in vehicle defects in vehicle
Fitness certificate (dum) Do not have any fitness Have fitness certificate of
certificate of vehicle vehicle
Driving license (dum) Do not have any driving Have driving license of
license of drivers vehicle
City_road (dum) Crashes occurred at city Crashes occurred at other
road road
Feeder road (dum) Crashes occurred at feeder Crashes occurred at other
road road
Driver age Age of the driver
The summary of the variables used in crash severity analysis has been shown in the table
18. The standard deviation of police controlled traffic system, roundabout junction, hit
object of road (dum), light condition, curve only (dum) road geometry, vehicle type,
vehicle defect and feeder road are greater than their mean which indicates that data for
those variables are very much dispersed. In case of all other dummy variables used in
model, the value of standard deviation are smaller than their mean which indicates that the
data for those variables are very much clustered. The mean age of drivers is 36 years with
a standard deviation of 6.78 years which indicates that middle aged drivers have more
involvement with road crash than any other driver age group in DCC area.
68
Table 18: Summary statistics for the variable used in model specification
Traffic control
Junction type
Pedestrian action .
Light conditions
Road geometry
69
Driver age 1229 18.00 57.00 36.0382 6.78048
Vehicle type
Vehicle defect
Availability of valid
fitness certificate
Availability of
driving license
Road class
70
5.4.2 Removal of insignificant explanatory variable from the model
The backward selection process has been done to identify variables which have significant
impact on road crash. All the variables with no interactions were tested on the basis of their
significance (p) value respectively. The goal was to eliminate, at the beginning, those
variables that were not significant and then continue with testing interaction effects with
only significant variables. Removal of weather condition, road condition, and surface
condition from the model did not produce much change in the deviance, and those variables
were not significant at the 0.05 level as shown in Table 19.
For example, the design variables for junction type were reduced from five (six levels) to
two (three levels) after it was shown in the correlation test that the impact of T junction,
staggered junction and four leg intersection on road crash were not statistically significant
at the 5% confidence interval. Number of design variables are shown in table 20.
71
Table 20: Number of design variables after reduction
The r-square value for the model is 0.312 which indicates that 31.2 percent variation in the
dependent variable can be explained by the predictor variables.
The classification accuracy rate of the model for fatal crash 91 percent. In case of overall
classification rate of model, 78.2 percent of the sample were correctly predicted to fall into
their respective groups.
Table 21: Statistical test for identifying the goodness of fit of the model
72
5.5 Identification of influential factors behind road crash severity
5.5.1 Road characteristics
5.5.1.1 Influence of junction type on road crash severity
The influence of junction type on road crash severity have been explored in the study. The
categorical independent variable has three levels. So the number of design variable for the
junction type is two. Crashes occurred outside the junction and crashes occurred in round
about junction are considered as design variables and crashes occurred at four leg
intersection is considered as omitted variable in the model.
According to table 22, road crash occurred outside the junction considered as the design
variable is found to be statistically significant with respect to the four leg intersection which
is considered as the omitted variable. The p value for road crash occurred outside the
junction is 0.000 which is less than 0.05. That indicates the variable have significant
influence on occurring fatal accident at 5 % confidence interval. The value of the relative
odds (odds ratio) of the independent variable (crashes occurred outside the junction) is
2.415 which indicates, for every one unit increase in the crashes occurred outside the
junction, the odds of occurring fatal crash increases by a factor of 2.415 subject to the non-
fatal crash. There is positive coefficient value (B) for crash occurred outside the junction
which is 0.882 indicates that with the increase of location outside the junction, the
likelihood of occurring fatal crash also increases.
On the other hand, crashes occurred in round about junction is found to be statistically
insignificant with respect to the four leg intersection which is considered as the omitted
variable. Because the The p value for road crash occurred at roundabout junction is 0.450
which is more than 0.05. That indicates the variable does not have any significant influence
on road crash severity at 5 % confidence interval.
73
According to table 22, fatal crash occurred in the place having no traffic control and having
police controlled traffic are found to be statistically significant with respect to the place
having only traffic light which is considered as the omitted variable. The respective p value
for those two variables, a place having no traffic control and having police controlled traffic
are 0.025 and 0.031 which is less than 0.05. That indicates the variables have significant
influence on occurring fatal accident at 5 % confidence interval. The value of the Exp (B)
known as relative odds (odds ratio) for the place having no traffic control is 2.467 which
indicates, for every one unit increase in the place having no traffic control, the probability
of occurring fatal crash increases by a factor of 2.467 subject to the non-fatal crash. There
is positive coefficient (B) value for place having no traffic control which is 0.903 indicates
that, with the increase of place having no traffic control, the likelihood of occurring fatal
crash also increases.
On the other hand, the value of the Exp (B) known as relative odds (odds ratio) for the
place having police controlled traffic is 2.303 which indicates, for every one unit increase
in the place having police controlled traffic, the probability of occurring fatal crash
increases by a factor of 2.303 subject to the non-fatal crash. There is positive coefficient
(B) value for place having police controlled traffic which is 0.834 indicates that, with the
increase of place having police controlled traffic, the likelihood of occurring fatal crash
also increases.
74
increase in the collision type ‘hit object off road’, the probability of occurring fatal crash
increases by a factor of 0.158 subject to the non-fatal crash.
On the other hand, the value of the Exp(B) known as relative odds (odds ratio) for the
collision type, ‘hit pedestrian’ is 4.002 which indicates, for every one unit increase in the
collision type, ‘hit pedestrian’, the probability of occurring fatal crash increases by a factor
of 4.002 subject to the non-fatal crash. There is positive coefficient (B) value for the
collision type, ‘hit pedestrian’ which is 1.387 indicates that, with the increase of the
collision type, ‘hit pedestrian’, the likelihood of occurring fatal crash also increases.
75
5.5.3 Vehicle characteristics
5.5.3.1 Influence of vehicle defects on road crash severity
The categorical independent variable, vehicle defects has three levels which was used as
dummy variable in the model. So the number of design variable for vehicle having defects
is two. Crashes occurred for brake fail and having multiple defect in vehicle are considered
as design variables and crashes occurred for not having any defect in vehicle is considered
as omitted variable in the model.
According to table 22, crashes occurred for brake fail of the vehicle is found to be
statistically significant with respect to the omitted variable ‘not having any defect in
vehicle’. The p value for brake failure of vehicle is 0.000 which is less than 0.05. The p
value indicates that the variable have significant influence on occurring fatal accident at 5
% confidence interval. The value of the relative odds (odds ratio) of the independent
variable (brake failure of the vehicle) is 17.711 which indicates, for every one unit increase
in brake failure of the vehicle, the probability of occurring fatal crash increases by a factor
of 17.711 subject to the non-fatal crash. There is positive coefficient value (B) for brake
failure of the vehicle which is 2.874 indicates that with the increase of brake failure of the
vehicle, the likelihood of occurring fatal crash also increases.
On the other hand, crashes occurred for multiple defect in vehicle is found to be statistically
insignificant with respect to the omitted variable ‘not having any defect in vehicle’.
Because the p value for multiple defect in vehicle is 0.880 which is more than 0.05. That
indicates the variable does not have any significant influence on road crash severity at 5 %
confidence interval.
In the model, crashes occurred by bus is found to be statistically significant with respect to
the omitted variable crashes occurred by rickshaw shown in table 22. The p value for road
76
crash occurred by bus is 0.036 which is less than 0.05. That indicates the variable have
significant influence on occurring fatal accident at 5 % confidence interval. The value of
the relative odds (odds ratio) of the independent variable (crashes occurred by bus) is 0.635
which indicates that crashes occurred by car increases the probability of occurring fatal
crash by a factor of 0.635 subject to the non-fatal crash.
On the other hand, crashes occurred by bus and heavy truck is found to be statistically
insignificant with respect to the crashes occurred by rickshaw which is considered as the
omitted variable. Because the p value for road crash occurred by bus and heavy truck are
0.375 and 0.657 which are more than 0.05. That indicates these variables do not have any
significant influence on road crash severity at 5 % confidence interval.
5.5.3.3 Influence of not having valid fitness certificate of vehicle on road crash severity
Vehicle fitness certificate, categorical independent variable, has two levels. So the number
of design variable for vehicle fitness certificate is one. Not having any vehicle fitness
certificate is considered as design variables and having vehicle fitness certificate is
considered as omitted variable in the model.
In the model, vehicles not having any fitness certificate is found to be statistically
significant with respect to the omitted variable ‘drivers having vehicle fitness certificate’.
The p value for the variable is 0.000 which is less than 0.05 shown in table 22. That
indicates the variable have significant influence on occurring fatal crash at 5 % confidence
interval. The value of the relative odds (odds ratio) of the variable (drivers not having any
vehicle fitness certificate) is 2.337 which indicates that drivers not having vehicle fitness
certificate increases the probability of occurring fatal crash by a factor of 2.337 subject to
the non-fatal crash . Besides, there is positive coefficient (B) value for the variable, which
is 0.849 indicates that with the increase of vehicles not having valid fitness certificate, the
likelihood of occurring fatal crash also increases.
77
According to table 22, crashes occurred at day and crash occurred at night with no street
light are found to be statistically significant with respect to the omitted variable day time.
The respective p value for these two variables, crash occurred at dawn and crash occurred
at night with no street light are 0.005 and 0.016 which are less than 0.05. That indicates the
variables have significant influence on occurring fatal accident at 5 % confidence interval.
The value of the Exp (B) known as relative odds (odds ratio) for the variable ‘crash
occurred at day’ is 2.317 which indicates that crashes occurred at day increases the
probability of occurring fatal crash by a factor of 0.158 subject to the non-fatal crash. There
is positive coefficient (B) value for crashes occurred at dawn, which is 0.840 indicates that
with the increase of crash occurred at day, the likelihood of occurring fatal crash also
increases.
On the other hand, the value of the Exp (B) known as relative odds (odds ratio) for the
crash occurred at night with no street light is 5.878 which indicates that crashes occurred
at night with no street light increases the probability of occurring fatal crash by a factor of
5.878 subject to the non-fatal crash. There is positive coefficient (B) value for crashes
occurred at night with no street light, which is 1.771 indicates that with the increase of
crash occurred at night with no street light, the likelihood of occurring fatal crash also
increases.
78
Table 22: Estimated coefficients, estimated standard errors, and P value for the
main model variables
Variable Estimated Standard Degree p- Odds ratio 95 %
of value
name coefficient error Exp(B) C.I for
freedom
(B) (S.E.) (df) Exp(B)
Upper Lower
Junction
type .881 .210 1 .000 2.412 1.599 3.640
Not at .755 1 .404 1.878 .428 8.245
.630
junction*
Round about
Traffic
control
No control* .885 .403 1 .028 2.423 1.100 5.338
Police .819 .386 1 .034 2.268 1.064 4.834
controlled*
Collision
type
Hit 1.407 0.156 1 .000 4.082 3.007 5.541
pedestrian*
Hit object off -1.933 0.613 1 .002 0.145 .044 0.481
road*
Light
condition
Day* .840 0.299 1 .005 2.317 1.290 4.160
Night 1.771 0.736 1 .016 5.878 1.389 24.87
(unlighted)* 6
Road
geometry
Curve only -1.281 1.070 1 .231 .278 .034 2.261
79
Straight and -1.168 .878 1 .184 .311 .056 1.739
flat
Vehicle type
Heavy truck .115 .260 1 .657 1.122 .675 1.867
Car -.159 .179 1 .375 .853 .601 1.212
Bus* -.454 .216 1 .036 .635 .416 .971
Vehicle
defects
Brakes* 2.874 .784 1 .000 17.711 3.808 82.38
Multiple -.167 1.113 1 .880 .846 .096 0
7.488
Fitness
certificate of
vehicle
Do not have
fitness .849 0.152 1 .000 2.337 1.735 3.147
certificate*
Driving
license
Do not have .191 .164 1 .244 1.210 .878 1.669
driving
license
Road class
City road -.163 .152 1 .284 .850 .631 1.144
National road 1.674 1.286 1 .193 5.334 .429 66.36
6
Age of .026 .011 1 .018 1.026 1.004 1.048
driver*
* Statistically significant at 5% confidence interval
80
5.5.3.5 Influence of vehicle loading on road crash severity
The variable, vehicle loading’ has not been included in the statistical model. Because the
variable was found to have insignificant relation with the dependent variable ‘crash
severity’ in the correlation test.
Here table 23 represents the cross tabulation in between crash severity and vehicle loading.
Within total non-fatal crash 97.6% non-fatal crash occur where vehicles carry load beyond
in capacity and only 2.4% non-fatal crash occur where vehicles carry load within capacity.
So, vehicle loading is an important factor for occurring non-fatal crash. Within total crash
33.0% non-fatal crash occur where vehicles carry load beyond its capacity and only 0.8%
non-fatal crash occur where vehicles carry load within its capacity. Again within total fatal
crash 95.6% fatal crash occur where vehicles carry load beyond its capacity. Within vehicle
loading 65.7% vehicles carry load beyond its capacity and face to fatal crash. Again 78.3%
vehicles carry load within its capacity and face to fatal crash. Within total crash 63.2% fatal
crashes occur where vehicles carry the load beyond its capacity. Within total crash almost
96.3% road crashes occur where vehicles carry load beyond its capacity.
81
Table No 23: Cross tabulation between percentage of crash severity and vehicle
loading
Vehicle loading
Beyond in Within
Capacity Capacity Total
Crash severity Non fatal % within Crash severity 97.6% 2.4% 100.0%
82
5.5.4 Driving characteristics
5.5.4.1 Influence of not having driving license on road crash severity
Driving license, categorical independent variable, has two levels. So the number of design
variable for driving license is one. Driver not having any driving license, is considered as
design variables and drivers having driving license is considered as omitted variable in the
model.
In the study, according to table 24, Within total non-fatal crash, 47.1% non-fatal crash
occur those drivers have license on driving and 52.9% non-fatal crash occur those drivers
have no license on driving. Again within driving license 46.2% drivers those have driving
license are responsible for non-fatal crash and 27.3% drivers those have no driving license
are responsible for non-fatal crash. Within total non-fatal crash 15.9% non-fatal crash occur
where driving license is available and 17.9% non-fatal crash occur where drivers have no
driving license. Now within fatal crash 28% fatal crash occur where drivers have driving
license and 72% fatal crash occur where divers have no driving license. Within driving
license 53.8% drivers those have driving license face to fatal crash and 72.7% drivers those
have no driving license face to fatal crash. Within total fatal crash 18.6% fatal crash occur
where drivers have license on driving and 47.6% fatal crash occur where drivers have no
license on driving. Within total crash 34.5% road crash occur where drivers have driving
license and 65.5% road crash occur where drivers have no driving license.
83
Table No 24: Cross tabulation between percentage of crash severity and driving
license
Driving License
Do not
Having Have
Driving Driving
License License Total
Crash severity Non fatal % within Crash severity 47.1% 52.9% 100.0%
84
5.5.5 Spatial characteristics
5.5.5.1 Influence of road class on road crash severity
Road class, categorical independent variable, has three levels. So the number of design
variable for Vehicle fitness certificate is two. National road and city road are considered as
design variables and feeder road is considered as omitted variable in the model.
According to figure 22 most of the road crash has occurred at national road. The proportion
of road crash in national road is found 0.75 in the study. That means almost three-fourth of
total road crash has occurred at national road. At city road, the ratio of road crash is 0.236.
The highest proportion of road crash is 0.75 and that has occurred at national road. High
vehicles move on national road and the vehicle volume on national road is comparatively
high. For that reasons the proportion of road crash in national road is comparatively high
than other roads.
85
Relationship between Proportion of Road Crash and Road Class
0.8
0.7
0.6
Proportion
0.5
0.4
0.3
0.2
0.1
0
City Road Feeder Road National Road Regional Road
Proportion 0.236928105 0.004084967 0.75 0.008986928
Figure No 22: Relationship between Proportion of Road Crash and Road Class
Kim, Pant and Yamashita (2010) found ‘land use’ as an influential factor behind the
increasing of pedestrian and motor vehicle crash. Besides, Zahabi et. Al (2011) also found
that the land use mix has significant impact and affected the severity of pedestrian injuries
from road crash.
That is why land use of the location of crash has been included in the study to analyze
whether the land use is an influential factors behind road crash severity or not. But land
use was found to have little impact on road crash severity in DSCC and DNCC area because
of the difference in the context with other countries.
86
In the chapter, driver’s age, not having proper vehicle fitness certificate, location outside
junction, no traffic control system and brake failure of vehicles are found as the most
influential factors behind road crash severity through using binary logistic regression
model and other explorative analysis. On the other hand, vehicle loading, not having
driving license, road geometry, weather condition, surface condition etc. are found having
little influence on road crash severity.
87
Chapter 6: Conclusions
Driver characteristics are the major contributing factors behind road crashes. Road users
are also involved in ninety (90%) percent of all contributing factors behind road crashes in
Dhaka. Besides the impact of vehicle factor (tyre burst, break failure), behavioral aspect of
heavy vehicle driver, travel pattern of road users on road accident have been considered as
contributing factors behind road accident in Dhaka. Seventy seven (77) percent of traffic
accident fatalities were found to be pedestrians and (50) fifty percent of these fatalities
involved buses in Dhaka. Large buses and trucks are also identified in involving in 61% of
all fatal single-vehicle accidents in Bangladesh.
In the study, the probability of occurring fatal crash have been investigated with respect to
the socio-economic factors, spatial and driving characteristics. Binary logistic model has
been used to identify the major influential factors behind road crashes in DSCC and DNCC
area. Absence of valid fitness certificate, driver’s age, light condition and vehicle defects
are found as the most influential factors behind the severity of road crash. Bus are involved
with 60.5 % of all road crash where car and truck are involved with 15 % and 17% of all
road crash in Dhaka city. Besides, collision type of vehicle was also found as significant
factors behind the road crash severity where Sixty five percent (65%) of all fatal crash has
also been occurred for hitting pedestrian while crossing the road. In case of the variable
‘junction type’, sixty five percent (65%) of all fatal crash has occurred at the location
outside the junction. Besides, almost sixty two percent (62%) of all crash has occurred at
the location where no traffic control system was available and thirty five percent (35%) of
all crash has occurred at place having police controlled traffic system. In case of the
variable, ‘vehicle defect’ which has been found as the significant factor behind road crash,
almost 50.7% vehicles face to road crash for brake failure of the vehicle.
On the other hand, the spatial factors like land use at the location of road crash had been
found as the significant factors behind road crash in the other country. That is why, the
factor was included in the study to find its impact on road crash severity in Dhaka. But in
the study, the impact of spatial factor (land use) are found as an insignificant factors behind
the road crash severity in DNCC and DSCC area. Different types of land use like,
residential, educational, mixed use and transport had little impact on road crash severity.
88
The decision making agency like Dhaka Transport coordination Authority (DTCA), Roads
and Highway department (RHD) do not take proper action to reduce the impact of these
influential factor on road crash in Dhaka. There is no reflection in the policy of higher
authority of the Government to reduce the impact of these factors. Proper measures should
be taken by incorporating of these influential factors while making policy decision
regarding reduction of road crash.
89
References
Jovanis, P.P., Chang, H.(1986). Modeling the relationship of accidentsto miles traveled.
In: Transportation Research Record 1068,TRB, National Research Council, Washington,
DC, pp. 42–51.
Dissanayake, S., & Lu, J. (2002). Analysis of severity of young driver crashes: sequential
binary logistic regression modeling. Transportation Research Record, (1784).
Sarkar, S., Tay, R., & Hunt, J. D. (2011). Logistic regression model of risk of fatality in
vehicle–pedestrian crashes on national highways in Bangladesh. Transportation research
record, 2264(1), 128-137.
Kim, K., Lawrence, N., Richardson, J., Li, L.(1996). Modeling faultamong bicyclists and
drivers involved in collisions in Hawaii1986–1991. In: Transportation Research Record
1538, TRB, NationalResearch Council, Washington, DC, pp. 75–80.
Kim, K., Lawrence, N., Richardson, J., Li, L.(1995). Personal andbehavioral predictors of
automobile crash and injury severity.Accident Analysis and Prevention 27 (4), 469–481.
Stamatiadis, N., & Deacon, J. A. (1995). Trends in highway safety: effects of an aging
population on accident propensity. Accident Analysis & Prevention, 27(4), 443-459.
Zhang, J., Lindsay, J., Clarke, K., Robbins, G., & Mao, Y. (2000). Factors affecting the
severity of motor vehicle traffic crashes involving elderly drivers in Ontario. Accident
Analysis & Prevention, 32(1), 117-125.
García‐ferrer, A., De Juan, A., &Poncela, P. (2007). The relationship between road traffic
accidents and real economic activity in Spain: common cycles and health issues. Health
Economics, 16(6), 603-626.
Clifton, K.J., Burnier, C.V., Akar, G. (2009). Severity of Injury Resulting from Pedestrian-
Vehicle Crashes: What Can We Learn from Examining the Built Environment?
Transportation Research Part D: Transport and Environment, Vol. 14, No. 6, 2009, pp.
425-436.
90
Law, T. H., Noland, R. B., & Evans, A. W. (2009). Factors associated with the relationship
between motorcycle deaths and economic growth. Accident Analysis & Prevention, 41(2),
234-240.
Kim, K., Pant, P., & Yamashita, Y. (2010). Accidents and accessibility: measuring
influences of demographic and land use variables in Honolulu, Hawaii.
Schneider, R. J., Diogenes, M. C., Arnold, L. S., Attaset, V., Griswold, J., & Ragland, D.
R. (2010). Association between roadway intersection characteristics and pedestrian crash
risk in Alameda County, California. Transportation Research Record, 2198(1), 41-51.
Pulugurtha, S. S., Duddu, V. R., &Kotagiri, Y. (2013). Traffic analysis zone level crash
estimation models based on land use characteristics. Accident Analysis & Prevention, 50,
678-687.
Zahabi, S. A. H., Strauss, J., Manaugh, K., & Miranda-Moreno, L. F. (2011). Estimating
Potential Effect of Speed Limits, Built Environment, and Other Factors on Severity of
Pedestrian and Cyclist Injuries in Crashes (No. 11-1247).
Siddiqui, C., & Abdel-Aty, M. (2012). Nature of modeling boundary pedestrian crashes at
zones. Transportation Research Record, 2299(1), 31-40.
Ivan, J. N., Wang, C., & Bernardo, N. R. (2000). Explaining two-lane highway crash rates
using land use and hourly exposure. Accident Analysis & Prevention, 32(6), 787-795.
Noland, R. B., & Oh, L. (2004). The effect of infrastructure and demographic change on
traffic-related fatalities and crashes: a case study of Illinois county-level data. Accident
Analysis & Prevention, 36(4), 525-532.
Greibe, P. (2003). Accident prediction models for urban roads. Accident Analysis &
Prevention, 35(2), 273-285.
Siddiqui, C., Abdel-Aty, M., & Huang, H. (2012). Aggregate nonparametric safety analysis
of traffic zones. Accident Analysis & Prevention, 45, 317-325.
91
Kim, K., Lawrence, N., Richardson, J., Li, L. (1994). Analyzing therelationship between
crash types and injury severity in motorvehicle collisions in Hawaii. In: Transportation
Research Record1467, TRB, National Research Council, Washington, DC, pp.9–13.
Mercier, C.R., Shelley, M.C., Rimkus, J., Mercier, J.M.(1997). Ageand gender as
predictors of injury severity in head-on highwayvehicular collisions. In: Transportation
Research Record 1581,TRB, National Research Council, Washington, DC.
Pulugurtha, S.S., Sambhara, V.R., 2011. Pedestrian crash estimation models for signalized
intersections. Accident Analysis.Prev.43 (1),439–446.
doi:http://dx.doi.org/10.1016/j.aap.2010.09.014.
Kim, J. K., Kim, S., Ulfarsson, G. F., &Porrello, L. A. (2007). Bicyclist injury severities
in bicycle–motor vehicle accidents. Accident Analysis & Prevention, 39(2), 238-251.
Klop, J.R., Khattak, A.J., (1999). Factors influencing bicycle crash severity ontwo-lane,
undivided roadways in North Carolina. Transport. Res. Rec. 1674,78–85.
Pless, I.B., Verreault, R., Tenina, S., (1989). A case–control study of pedestrianand
bicyclist injuries in childhood. Am. J. Public Health 79 (98), 995–998.
Noland, R.B., Quddus, M.A., (2004). An analyses of pedestrian and bicycle casualties
using regional panel data. Transport. Res. Rec. 1897, 28–33.
Hoque M. M.,Mahmud S. M. S., Paul S., (2008), "The Cost of Road Traffic Accidents in
Bangladesh", Publication in 10th Pacific Regional Science Conference Organization
(PRSCO) Summer Institute 2008, hosted by Bangladesh Regional Science Association
(BRSA),15-17 May 2008, Dhaka, Bangladesh, P.N. -88, Pg-88 (Abstract), Paper 6b3 (CD
proceedings).
92
Uddin, S. R., & Hoque, M. S. (2003, October). STUDY OF HEAVY
VEHICLES’DRIVER BEHAVIOR IN ROAD ACCIDENTS OF BANGLADESH. In
AUSTRALASIAN TRANSPORT RESEARCH FORUM (ATRF), 26TH, 2003,
WELLINGTON, NEW ZEALAND.
Hoque, M.S. and Hasan, M.R., (2007), Involvement of Vehicle Factors in Road
Accidents, Journal of Civil Engineering (JCE), The Institute of Engineers Bangladesh
(IEB), Vol. 35 (1), pp 29-45.
Kim, K., Pant, P., & Yamashita, E. Y. (2008). Hit-and-run crashes: Use of rough set
analysis with logistic regression to capture critical attributes and determinants.
Transportation research record, 2083(1), 114-121.
Stamatiadis, N., & Deacon, J. A. (1995). Trends in highway safety: effects of an aging
population on accident propensity. Accident Analysis & Prevention, 27(4), 443-459.
Dumbaugh, E. 2005a. Safe streets, livable streets. Journal of the American Planning
Association 71 (3): 283-98.
Xie, K., Wang, X., Ozbay, K., & Yang, H. (2014). Crash frequency modeling for
signalized intersections in a high-density urban road network. Analytic methods in
accident research, 2, 39-
Ladron de Guevara, F., Washington, S., & Oh, J. (2004). Forecasting crashes at the
planning level: simultaneous negative binomial crash model applied in Tucson, Arizona.
Transportation Research Record: Journal of the Transportation Research Board,
(1897), 191-199.
Huang, H., Abdel-Aty, M., & Darwiche, A. (2010). County-level crash risk analysis in
Florida: Bayesian spatial modeling. Transportation Research Record: Journal of the
Transportation Research Board, (2148), 27-37.
93
Islam, M. R., & Sharmeen, N. (2011). Road Accidents: Contemporary Scenario and Policy
Issues in Bangladesh. Joumal ol Bangladesh lnstirute of Planners Vol. 4. December 20 |
I, pp. 45-5-5.
Hoque, M. S., & Hasan, M. R. (2007). Involvement of vehicle factors in road accidents.
Journal of Civil Engineering (IEB), 35(1), 17-27.
Hoque, M. S., Khondaker, B., & Hoque, M. M. (2007). Behavioral habits and attitudes of
heavy vehicle drivers towards road safety. Journal of Civil Engineering (IEB), 1, 29-45.
Hoque, M. S., Debnath, A. K., & Mahmud, S. M. (2006, August). Impact of garment
industries on road safety in metropolitan Dhaka. In Proceedings of international
conference on traffic safety in developing countries.
Kim, K., Brunner, I., & Yamashita, E. (2006). Influence of land use, population,
employment, and economic activity on accidents. Transportation Research Record:
Journal of the Transportation Research Board, (1953), 56-64.
Kim, K., & Yamashita, E. (2002). Motor vehicle crashes and land use: empirical analysis
from Hawaii. Transportation Research Record: Journal of the Transportation Research
Board, (1784), 73-79.
Kim, K., Nitz, L., Richardson, J., & Li, L. (1995). Personal and behavioral predictors of
automobile crash and injury severity. Accident Analysis & Prevention, 27(4), 469-481.
Wier, M., Weintraub, J., Humphreys, E. H., Seto, E., & Bhatia, R. (2009). An area-level
model of vehicle-pedestrian injury collisions with implications for land use and
transportation planning. Accident Analysis & Prevention, 41(1), 137-145.
Ahsan, H. M., Raihan, M. A., Rahman, M. S., & Arefin, N. H. (2011). Reporting and
recording of road traffic accidents in Bangladesh. Proceedings of 4th Annual Paper Meet
and 1st Civil Engineering Congress, Dhaka, Bangladesh.
94
Feinberg, S., (1980). The Analysis of Cross-Classified CategoricalData. MIT Press,
Cambridge, MA.
Hosmer, D.W., Lemeshow, S., (1989). Applied Logistic Regression.Wiley, New York.
95
APPENDIX
Correlations
Accident severity Pearson 1 .187** -.012 -.177** .329** .107* .146** -.030 .020 .035 .058* - .050 .106** -.020 .190** .325** .050 .145* .109*
*
Correlati .066* * *
on
Sig. (2- .000 .680 .000 .000 .000 .000 .292 .480 .215 .042 .020 .077 .000 .495 .000 .000 .079 .000 .000
tailed)
N 1229 1229 1229 1229 1229 1229 1229 1229 1229 1229 1229 1229 1229 1229 1229 1229 1229 1229 1229 1229
Not at junction Pearson .187** 1 -.125** .048 .108** -.024 .164** -.060* .084** .036 .005 .054 - -.073* .011 -.055 .084** .026 - -.031
*
Correlati .146 .068*
*
on
Sig. (2- .000 .000 .094 .000 .410 .000 .035 .003 .210 .874 .059 .000 .010 .693 .052 .003 .361 .017 .280
tailed)
N 1229 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1229
Roundabout Pearson -.012 -.125** 1 -.014 -.068* .132* -.020 -.010 .013 -.004 -.022 -.013 -.018 -.016 -.005 -.029 -.064* -.006 -.012 -.031
*
Correlati
on
Sig. (2- .680 .000 .616 .017 .000 .473 .725 .655 .875 .444 .660 .532 .577 .856 .305 .026 .824 .673 .274
tailed)
N 1229 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1229
Hit_object_off_road Pearson -.177** .048 -.014 1 -.219** .025 -.036 .078** -.054 -.008 - - -.031 -.028 -.009 -.140** -.199** -.011 - .077*
Correlati .070* .066* .087* *
*
on
Sig. (2- .000 .094 .616 .000 .372 .210 .006 .059 .784 .014 .020 .275 .329 .752 .000 .000 .698 .002 .007
tailed)
N 1229 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1229
Hit_pedestrian Pearson .329** .108** -.068* -.219** 1 -.024 .163** -.013 .059* -.034 .177* .053 .079* - .011 .014 .930** .051 .301* .094*
* *
Correlati .083** * *
on
Sig. (2- .000 .000 .017 .000 .399 .000 .639 .038 .237 .000 .062 .005 .003 .696 .617 .000 .076 .000 .001
tailed)
N 1229 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1229
Dawn Pearson .107** -.024 .132** .025 -.024 1 -.070* -.008 .002 -.015 - - .423* -.038 -.018 .087** -.005 -.022 .058* .106*
Correlati .141* .074* * *
* *
on
Sig. (2- .000 .410 .000 .372 .399 .014 .772 .939 .591 .000 .010 .000 .185 .535 .002 .855 .447 .042 .000
tailed)
N 1229 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1229
Night_unlighted Pearson .146** .164** -.020 -.036 .163** - 1 -.025 .032 -.011 .182* - -.045 -.040 -.013 .022 .155** -.016 -.046 .144*
Correlati .070* *
.095* *
*
on
Sig. (2- .000 .000 .473 .210 .000 .014 .378 .263 .695 .000 .001 .118 .162 .651 .442 .000 .579 .109 .000
tailed)
N 1229 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1229
Curve_only Pearson -.030 -.060* -.010 .078** -.013 -.008 -.025 1 -.788** -.005 -.004 -.047 -.022 -.020 -.006 -.059* -.009 -.008 .039 .018
Correlati
on
Sig. (2- .292 .035 .725 .006 .639 .772 .378 .000 .847 .880 .103 .443 .493 .824 .038 .760 .785 .170 .529
tailed)
N 1229 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1229
Straight_and_flat Pearson .020 .084** .013 -.054 .059* .002 .032 -.788** 1 .007 .048 .043 .028 .025 .008 .083** .028 .010 .009 -.002
Correlati
on
Sig. (2- .480 .003 .655 .059 .038 .939 .263 .000 .807 .090 .135 .330 .384 .778 .004 .323 .729 .753 .945
tailed)
N 1229 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1229
Artic_truck Pearson .035 .036 -.004 -.008 -.034 -.015 -.011 -.005 .007 1 -.042 -.021 -.010 .088** -.003 .001 -.031 -.003 -.003 -
Correlati .078*
*
on
Sig. (2- .215 .210 .875 .784 .237 .591 .695 .847 .807 .141 .468 .733 .002 .921 .963 .273 .903 .912 .006
tailed)
N 1229 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1229
Bus Pearson .058* .005 -.022 -.070* .177** - .182** -.004 .048 -.042 1 - - -.015 .038 .075** .171** -.012 .209* .124*
Correlati .141* .356* .115* * *
* * *
on
Sig. (2- .042 .874 .444 .014 .000 .000 .000 .880 .090 .141 .000 .000 .610 .180 .009 .000 .670 .000 .000
tailed)
N 1229 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1229
Car Pearson -.066* .054 -.013 -.066* .053 - -.095** -.047 .043 -.021 - 1 - - .016 -.107** .046 .069* - -
Correlati .074* .356* .083* .074** .061* .061*
* * *
on
Sig. (2- .020 .059 .660 .020 .062 .010 .001 .103 .135 .468 .000 .004 .010 .573 .000 .108 .016 .031 .032
tailed)
N 1229 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1229
Illegal Pearson .050 -.146** -.018 -.031 .079** .423* -.045 -.022 .028 -.010 - - 1 .091** .064* .026 .059* -.014 .154* .165*
*
Correlati .115* .083* * *
* *
on
Sig. (2- .077 .000 .532 .275 .005 .000 .118 .443 .330 .733 .000 .004 .001 .025 .357 .037 .629 .000 .000
tailed)
N 1229 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1229
Brakes Pearson .106** -.073* -.016 -.028 -.083** -.038 -.040 -.020 .025 .088** -.015 - .091* 1 -.010 .088** -.085** -.012 .015 .055
Correlati .074* *
*
on
Sig. (2- .000 .010 .577 .329 .003 .185 .162 .493 .384 .002 .610 .010 .001 .725 .002 .003 .666 .595 .053
tailed)
N 1229 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1229
Multiple Pearson -.020 .011 -.005 -.009 .011 -.018 -.013 -.006 .008 -.003 .038 .016 .064* -.010 1 -.078** .013 -.004 .046 -.055
Correlati
on
Sig. (2- .495 .693 .856 .752 .696 .535 .651 .824 .778 .921 .180 .573 .025 .725 .006 .640 .889 .108 .053
tailed)
N 1229 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1229
No_dummy Pearson .190** -.055 -.029 -.140** .014 .087* .022 -.059* .083** .001 .075* - .026 .088** -.078** 1 .052 -.047 .270* -
* * * *
Correlati .107 .063*
*
on
Sig. (2- .000 .052 .305 .000 .617 .002 .442 .038 .004 .963 .009 .000 .357 .002 .006 .068 .098 .000 .028
tailed)
N 1229 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1229
Crossing_the_road Pearson .325** .084** -.064* -.199** .930** -.005 .155** -.009 .028 -.031 .171* .046 .059* - .013 .052 1 -.093** .300* .101*
*
Correlati .085** * *
on
Sig. (2- .000 .003 .026 .000 .000 .855 .000 .760 .323 .273 .000 .108 .037 .003 .640 .068 .001 .000 .000
tailed)
N 1229 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1229
Walking_along_the_r Pearson .050 .026 -.006 -.011 .051 -.022 -.016 -.008 .010 -.003 -.012 .069* -.014 -.012 -.004 -.047 -.093** 1 -.004 -.038
oad Correlati
on
Sig. (2- .079 .361 .824 .698 .076 .447 .579 .785 .729 .903 .670 .016 .629 .666 .889 .098 .001 .876 .180
tailed)
N 1229 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1229
No Pearson .145** -.068* -.012 -.087** .301** .058* -.046 .039 .009 -.003 .209* - .154* .015 .046 .270** .300** -.004 1 .047
*
Correlati .061* *
on
Sig. (2- .000 .017 .673 .002 .000 .042 .109 .170 .753 .912 .000 .031 .000 .595 .108 .000 .000 .876 .101
tailed)
N 1229 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1231 1229
Driver age 1 Pearson .109** -.031 -.031 .077** .094** .106* .144** .018 -.002 -.078** .124* - .165* .055 -.055 -.063* .101** -.038 .047 1
* *
Correlati .061* *
on
Sig. (2- .000 .280 .274 .007 .001 .000 .000 .529 .945 .006 .000 .032 .000 .053 .053 .028 .000 .180 .101
tailed)
N 1229 1229 1229 1229 1229 1229 1229 1229 1229 1229 1229 1229 1229 1229 1229 1229 1229 1229 1229 1229
Missing Cases 2 .2
a. If weight is in effect, see classification table for the total number of cases.
Non fatal 0
Fatal 1
Classification Tablea,b
Predicted
Score df Sig.
No 25.912 1 .000
Chi-square df Sig.
Model Summary
Predicted