Landscape and Urban Planning 158 (2017) 166–176

Contents lists available at ScienceDirect

Landscape and Urban Planning

journal homepage: www.elsevier.com/locate/landurbplan

Research Paper

Campus score: Measuring university campus qualities

Amir Hajrasouliha
City and Regional Planning Department, Cal Poly State University, San Luis Obispo, CA, United States

h i g h l i g h t s

• Physical campus characteristics can impact student satisfaction and academic performance.
• Campus Score is proposed, representing campus urbanism, greenness, and on-campus living.
• Campus Score has significant associations with freshman retention and graduation rates.

a r t i c l e i n f o a b s t r a c t

Article history: This research proposes an index, called Campus Score, which measures the main physical qualities of uni-
Received 7 September 2015 versity campuses. Campus Score is composed of three latent variables representing Urbanism, Greenness,
Received in revised form 19 October 2016 and On-Campus Living, with 10 indicators. This index has been calculated for 103 research-intensive uni-
Accepted 30 October 2016
versities in the United States of America. Two linear regressions show that Campus Score has significant
Available online 15 November 2016
associations with freshman retention and 6-year graduation rates. It is also interesting to note that, com-
pared to the Academic Ranking of World Universities (Shanghai Ranking), Campus Score has stronger
Keywords:
associations with freshman retention and graduation rates. The one-way ANOVA test and post-hoc anal-
Campus score
Freshman retention
ysis reveal that private universities, on average, have significantly higher Campus Scores than public
Graduation rate universities, Research I universities have significantly higher mean scores than Research II universities,
University ranking and universities in the Northeast census region have significantly higher mean scores than universities in
Campus design other census regions. Exploring the relationships between Campus Score and university objectives, such
Campus planning as student satisfaction, safety, and campus sustainability, can give campus planners fresh insight into the
impacts of campus form.
© 2016 Elsevier B.V. All rights reserved.

1. Introduction and research environment, creating a sustainable environment, and

The United States of America has a rich history of campus plan-
ning and design. Some of the best university campuses in the U.S.
were almost fully developed in the 19th century and early 20th cen-
tury. Before World War II, campus designers would follow certain
formal typologies such as the quadrangle campus (e.g., University
of Washington in Seattle), picturesque campus (e.g., University of
Vermont), or Beaux-Arts campus (e.g., Columbia University). After
World War II, with the vast expansion of university campuses,
the emphasis was more on the design of freestanding buildings
than on campus master plans (Coulson, Roberts, & Taylor, 2010;
Dober, 1996; Turner, 1984). In recent years, most universities have
re-embraced the idea of campus master plans to address their insti-
tutional objectives, such as attracting more students, increasing the
quality of life of current students and faculty, promoting a learning
E-mail address: ahajraso@calpoly.edu
benefiting the surrounding communities (Chapman, 2006; Coulson
et al., 2010, 2014 ; Delbanco, 2014; Disterheft, Caeiro, Azeiteiro, &
Leal Filho, 2014; Dober, 1996; Kenney, Dumont, & Kenney, 2005;
Mitchell & Vest, 2007; Strange & Banning, 2001; Turner, 1984).
Although campus planning and design have received extensive
attention in the profession in recent years, academia has con-
tributed little. However, evidence-based research to quantify the
campus qualities that are at the center of practitioners’ interest
would fill a considerable gap in the field.
Many attempts have been made in the fields of urban planning,
landscape architecture, and architecture to quantify the character-
istics of built environments and to evaluate their impact on a wide
range of environmental, behavioral, and economic variables. For
example, urban sprawl development can be quantified on different
scales, and its impacts on travel behavior, housing affordability, and
public health have been assessed by many researchers (Custinger
& Galster, 2006; Ewing, Pendall, & Chen, 2003; Hajrasouliha &
Hamidi, 2016). Using Space Syntax techniques, the morphology

http://dx.doi.org/10.1016/j.landurbplan.2016.10.007
0169-2046/© 2016 Elsevier B.V. All rights reserved.
 A. Hajrasouliha / Landscape and Urban Planning 158 (2017) 166–176
of street networks can be quantified, and their relationship with
both walkability and land use are well-studied research topics
(Hajrasouliha & Yin, 2015; Hillier, 2007, 2009). Certain urban form
variables, such as urban density, land use diversity, and street con-
nectivity, are quantified in the transportation planning field to
assess the impact of built environment qualities on travel behav-
ior (Cervero and Kockelman, 1997; Cervero and Murakami, 2010;
Ewing and Cervero, 2010; Frank et al., 2006). Urban design quali-
ties are also quantified, and their impact on pedestrian activities is
assessed (De Nisco & Warnaby, 2014; Ewing et al., 2013; Ewing
& Handy, 2009; Ewing, Hajrasouliha, Neckerman, Purciel-Hill, &
Green, 2015). Finally, measuring greenness (Gupta, Kumar, Pathan,
& Sharma, 2012; Yao, Liu, Wang, Yin, & Han, 2014) enables a bet-
ter understanding of the role of urban greenness areas in public
health. Together, these cited studies suggest that single-use, low-
density land development, disconnected street networks, and low
neighborhood greenness are characteristics positively associated
with auto dependence and negatively associated with air quality,
walking, transit use, and physical and mental health.
However, the idea of campus is unique. A university campus is
not a city, a neighborhood, or a block. Therefore, describing and ana-
lyzing campus forms should be different. The missions, objectives,
and governance of institutions of higher learning are not compa-
rable to those of neighborhoods or cities. One of the most obvious
distinctions between a campus and a neighborhood is in its pri-
mary purpose of providing a supportive environment for learning.
EDUCAUSE, the Society for College and University Planning (SCUP),
and the PKAL Learning Spaces Collaboratory have aimed to iden-
tify assessment metrics, predictive evidence, and illustrative cases
that relate space to learning. Among the common insights that
have emerged from these efforts are the realizations that experi-
ential learning enhances student engagement and engaging study
behaviors fosters learning (Azevedo, 2015; Gilboy, Heinerichs, &
Pazzaglia, 2015). However, these works focus more on classrooms
and teaching environments (micro-scale design) and less on the
contextual condition of the campus environment (macro-scale
design). Yet, some studies show that certain macro-scale campus
qualities can have impact on students’ quality of life and academic
performance as well. For example, studies have indicated that the
presence of green spaces on campus, along with the perception of
greenness and restorative environment (Kaplan, 1992) of a cam-
pus are associated with student quality of life (Felsten, 2009; Hipp,
Gulwadi, Alves, & Sequeira, 2016). One of the potential impacts of
the physical characteristics of the campus on students’ quality of life
is helping them to cope with college life and to manage challenges
of academic life (Dyson & Renk, 2006). A green campus may create a
pleasant college experience and encourage students to spend time
and socialize on campus, thus reducing their mental fatigue level.
At the same time, an urban-feeling campus may increase students’
perception of social connectedness. Previous studies have shown
both greenness and social connectedness to be associated with stu-
dent retention (Berger & Braxton, 1998; Lounsbury & DeNeui, 1995;
Naretto 1995; Styron 2010).
With the goal of identifying an assessment metric, this research
has generated an index, named Campus Score, by measuring
physical campus qualities that may be associated with students’
retention rate and academic performance. Quantifying physical
campus qualities is an attempt to answer the critical questions,
“Does campus matter?” and “Which qualities of campus matter
most in students’ retention and graduation?” This approach can
also help to determine the relationships between campus qualities
(Campus Score) and other university characteristics such as age,
type, urbanization of setting, region, climate, and academic ranking.
Two studies help to establish the specific conceptual founda-
tion of this article. The first study (Hajrasouliha, 2015), through a
content analysis of 50 campus master plans, conceptualized seven
167
dimensions of campus form. The second study (Hajrasouliha &
Ewing, 2016) operationalized the proposed campus form dimen-
sions and assessed their relationship with the freshman retention
and 6-year graduation rates for 103 research-intensive universities
in the United States. That study determines the significance of the
relationship between the physical campus and student satisfaction
with college life, and ultimately with academic performance. One
of the main differences between Hajrasouliha and Ewing (2016)
research and other evidence-based studies on this subject is the
scale of analysis. Most previous studies are focused on micro-scale
design characteristics of the campus environment and its relation-
ship with learning and engagement. In contrast, following X and Y,
this study focuses on the contextual and macro-scale design char-
acteristics of the physical campus. These relationships have seldom
been explored. Brief summaries of both studies follow:
1.1. The morphology of the “well-designed campus”
In order to construct a theoretical framework for analyzing
campus form, a content analysis of 50 randomly selected uni-
versity campus master plans in the United States was conducted
by Hajrasouliha (2015). The analysis shows significant similarities
between plans in terms of challenges, objectives, and recommen-
dations. To avoid a subjective definition of the “well designed”
campus, the top 100 common recommendations in the selected
master plans were identified. Accordingly, from these recom-
mendations, seven dimensions of campus form were inferred, as
suggested in Table 1.
According to these morphological dimensions, the “well-
designed” campus was conceptualized as a mixed, compact,
well-connected, well-structured, inhabited, green campus in an
urbanized setting (see Fig. 1). These dimensions are measurable;
therefore, it is possible to quantitatively test their relationship to
the desired outcomes.
1.2. The relationship of campus design with student retention and
degree attainment
Hajrasouliha and Ewing (2016) hypothesized that the physi-
cal campus has impact on students’ satisfaction with their college
experience and their overall academic performance. To test this,
they operationalized the seven morphological dimensions of cam-
pus as described in Table 1. Five dimensions were operationalized
quantitatively with one or more variables. However, land use orga-
nization and spatial configuration, were rated qualitatively.
A Structural Equation Model, illustrated in Fig. 2, was used in
that analysis. After controlling for student selectivity, university
resources, student profile and type of institution, data from 103 U.S.
research universities showed strong positive associations between
student retention and graduation rates and three campus qualities:
Campus Living, Greenness, and Urbanism (a composite variable
from dimensions of compactness, connectivity and context). Sig-
nificant associations between the outcome variables and two other
campus qualities, land use organization and spatial configuration,
were not detected. Furthermore, no significant association between
a set of control variables (age of university, campus size, affordabil-
ity of education, city economic status, climate index, safety, and
the degree of urbanization of setting) and student retention and
graduation rates was detected (see Table 2).
2. Methods
Two steps are essential to generating an index that can quantify
campus qualities: adopting a theoretical framework for describing
and analyzing campus form, and identifying campus form quali-
ties that have significant associations with university objectives of
 168
Table 1
Operationalizing the Campus Morphological Dimensions.
Description Scale Data Source
(1) Land Use Land use mix Rating land use organization on campus between 1 and 10, by analyzing campus land-use maps, downloaded from universities’ website. 10 = All Campus land-use map,
Organization uses are mixed on campus, however the major athletic fields, greenhouses, barns and surface parking areas are not located at the campus core. downloaded from
Dimension 5 = land use is neither mixed nor segregated. For example, campus housing is located far from the campus core, but other teaching, research, and universities’ website, The
recreational uses are located in the campus core. 1 = campus has segregated areas away from the campus core for sport, research, residence, and researcher’s rating.a
some academic disciplines.

(2) Compactness Mass Density Computing the total area of building footprints divided by campus areab . Note that “campus area” is not considered as the total land owned by OpenStreetMap, Google Earth
Dimension the university. The area of the undeveloped lands, the agricultural lands, and the scattered low-density service facilities on the periphery of images
campus are not calculated. This calculation process for the multi-patch campuses is the same as uni-patch campuses.
Proximity Conducting average nearest neighborhood distance tool in ArcGIS. The input data are building footprints. OpenStreetMap, Google Earth
images
c
Pervious open spaces Computing the percentage of pervious open spaces in a quarter mile buffer around campus buildings. NLCD2011
Surface parking Computing the total area of surface parking divided by the campus area. Rooftop parking and parking structures are not included. OpenStreetMap, Google Earth
images

(3) Connectivity Campus connectivity 1) Downloading census street lines at the county level 2) Refining the maps according to Google Earth Images 3) Export maps as dxf files from Census Tiger 2010, street

A. Hajrasouliha / Landscape and Urban Planning 158 (2017) 166–176

Dimension ArcGIS and open it in Depthmap (Space Syntaxd Software) 4) Angular integration analysis with radius of 3, weighted by segment length; 5) lines
Averaging integration values of campus street segments
Campus centrality Dividing the average integration value of campus street segments with radius 3 by the average integration value of county street segment with Census Tiger 2010, street
same radius.e lines

(4) Configuration Campus spatial structure Rating the strength of campus spatial structure from 1 to 10. 10 = the entire campus has organized around most of these principles: Buildings are Google Map and Google Earth
Dimension defining open spaces. Campus spaces are connected through main corridors, courtyards, or quads. Campus has a main central space such as a images, The researcher’s
plaza or a lawn, long view corridors with a land mark at the focal point, enclosed open spaces, and the entire master plan is relatively symmetric ratingf .
and geometric. 5 = The campus has neither organized nor disorganized layout. For example, the historical part of campus has a defined spatial
structure, but the rest of the campus are free-standing buildings in open, landscaped ground. 1 = the campus has a disorganized layout.

(5) Campus Living On-campus living Computing the percentage of students living on-campus US News and World Report
Dimension

(6) Greenness Dimension Tree canopy Computing the average percentage of tree canopy in a quarter mile buffer around campus buildings.g NLCD2011
Pervious open spaces Described under compactness dimensionh
Surface parking Described under compactness dimensioni

(7) Context Dimension Activity density Computing the density of population and employment of all census tracts neighboring the campusj Longitudinal Employment
Household Dynamic 2010-
Census 2010
Land use entropy Computing land use entropy of all census tracts neighboring the campus. Land use entropy was computed with the formula: LED 2010
Entropy = −[residential share × ln (residential share) + retail share × ln (retail share) + office share × ln (office share)]/LN (3)
Intersection density Computing intersection density of all census tracts neighboring the campus, computed as the number of intersections within all census tracts Census Tiger 2010, street
neighboring the campus divided by the area of census tracts lines and census tracts
a
To test the reliability of this scale, two persons rated 40 campuses. The interrater reliability test shows that this measure is reliable. A two-way mixed effects model was used, where people effects are random and measure
effects are fixed. The intraclass correlation coefficient of interrater reliability is 0.865 which is above the acceptable threshold of 0.7. Please note that the two dimensions that have “researcher’s rating” as their source, are the
only two dimensions that do not have significant associations with either of the outcome variables. Therefore, these two dimensions are not used in the calculation of Campus Score.
b
Mapping campus figure-ground in ArcGIS is used to calculate campus mass density. If the GIS file does not exist, the base-map of OpenStreetMap – an openly licensed map of the world – can be used to map main physical
features, such as building footprints, and campus core boundary. These maps can be refined according to the Google Earth images to increase the accuracy of the GIS maps.
c
Note that the quarter mile buffer is generated only around campus buildings and not the sport fields.
d
Space Syntax is a set of theories and techniques for measuring the spatial configuration of street networks. The basic element in Space Syntax is the street segment between intersections, which can be derived from road
center line data. With the help of Depthmap software, developed by Space Syntax., the street segments can be translated into a graph, in which segments and their connections turn into nodes and links. The most commonly used
Space Syntax measure is known as integration, which measures “how close each segment is to all others under each definition of distance” (Hillier, 2009).
e
This measure shows the relative connectivity of campus streets to the county street network.

 
f
The interrater reliability test shows that this measure is reliable, with an intraclass correlation coefficient of 0.885.
g
TNC = total number of 30 × 30 meter cells in the buffer; TCi = the percentage of tree canopy in a given cell i; “Tree Canoy Percentage” = ni=0 TCi /TNC .
h
The percentage of pervious open space is positively related to greenness and negatively related to compactness. As an exogenous variable in SEM, this dimension can be loaded into two different latent variables with generating
separate coefficient estimates.
i
The percentage of surface parking areas is negatively related to both greenness and compactness.
j
Note that activity density, land use entropy, and intersection density are calculated only for census tracts neighboring the campus with population density more than 100 per square mile. Census tract that contains the core
of campus and census tracts with population density less than 100 per square mile are not included.
 A. Hajrasouliha / Landscape and Urban Planning 158 (2017) 166–176
Fig. 1. Campus form dimensions.
high retention and graduation rates. As a first step toward gen-
erating Campus Score, the theory of the “well-designed campus”
(Hajrasouliha, 2015) has been adopted. For the second step, the
three qualities of urbanism, greenness, and campus living have
been selected to generate the Campus Score, based on the find-
ings of Hajrasouliha and Ewing (2016). In addition, this study used
the same dataset as Hajrasouliha and Ewing (2016).
The research population consisted of 206 universities in the
United States with high or very high levels of research activ-
ity according to the 2010 Carnegie Classification. One hundred
and three campuses were randomly selected, stratified by census
regions, Northeast, South, Midwest, and West, and type: Research I
(very high research activity) and Research II (high research activity).
Universities with more than one campus that are formally very dif-
ferent were not selected. The University of Michigan in Ann Arbor
was the only case with this quality in the sample, and was there-
169
fore replaced by another university. Fig. 3 shows the location of the
selected campuses in the United States.
In order to calculate Campus Score, a composite score was devel-
oped from the three latent variables of Urbanism, Greenness, and
Campus Living (see Fig. 2). These three latent variables were first
standardized with a mean of 0 and a standard deviation of 1. The
composite score is generated with the following formula:
CampusScore = 0.177 × Urban + 0.215 × Green + 0.251 × Living
The multipliers are the standardized regression weights on
freshman retention rate, obtained from modeling freshman reten-
tion rate with maximum likelihood estimation (see Hajrasouliha &
Ewing, 2016). For ease of interpretation, the overall score and latent
variables were converted to have a mean of 100 with variance of
50.
170 A. Hajrasouliha / Landscape and Urban Planning 158 (2017) 166–176

Fig. 2. Modeling student retention and graduation rates.

Table 2
Control variables and their data source.

Variable Proxy Variable Data Source

University Resources Percentage of classes with fewer than 20 students US News and World Report
the average faculty pay American Association of University Professors (AAUP)
Student Selectivity SAT Score College Navigator
Age of university Year founded US News and World Report
Campus Size Campus Size OpenStreetMap refined by Google Earth images
Type of Institution Public, Private for-Profit, Private not-for-profit Carnegie Classification 2010
Research II (high research activities), Research I (very high Carnegie Classification 2010
research activities)
Student Profile Enrollment Profile Classification 2010 0 = (Not classified) Carnegie Classification 2010
1 = ExU2: Exclusively undergraduate two-year 2 = ExU4:
Exclusively undergraduate four-year 3 = VHU: Very high
undergraduate 4 = HU: High undergraduate 5 = MU:
Majority undergraduate 6 = MGP: Majority
graduate/professional 7 = ExGP: Exclusively graduate or
professional 0,1,2, and 7 are not in the samples
Percentage of undergraduate enrollment: the total number US News and World Report
of undergraduates divided by the total number of students
Affordability of Education Average total indebtedness of 2013 graduating class US News and World Report
City Economic Status The median household income of city 2013 Census Bureau
Climate Index the total cooling days of city in 2014 NOAA’s National Climatic Data Center
the total heating days of city in 2014 NOAA’s National Climatic Data Center
Safety Crime rate of City in 2013 FBI Uniform Crime Reports
The degree of urbanization of setting 4 = City 3 = Suburban 2 = Town 1 = Rural Carnegie Classification 2010

Next, the predictive power of the Campus Score was tested 3. Results
for student retention and graduation rates. Two linear regression
analyses were conducted. First, the freshman retention rate was The final rankings of all 103 campuses with their scores are
modeled with Campus Score and the same control variables as in presented in Table 3.
the Hajrasouliha and Ewing (2016) study: the average SAT score,
the percentage of classes with fewer than 20 students, university
3.1. The relationship of campus score with student retention and
type (Research I or II), and the total number of undergraduates.
graduation rates
Second, the 6-year graduation rate was modeled with the fresh-
man retention rate and Campus Score. None of the other control
Scatter plots (see Fig. 4) suggest strong quadratic relationships
variables was significant in this model.
between both Campus Score and freshman retention rate and Cam-
 A. Hajrasouliha / Landscape and Urban Planning 158 (2017) 166–176 171

Table 3
Ranking 103 research universities based on their Campus Score.

Rank University Name Urban Score Green Score Living Score Campus Score (Composite score)

1 Columbia University 248.84 93.52 216.79 238.25

2 Princeton University 100.17 187.36 223.12 226.66
3 Lehigh University 97.86 219.43 161.89 204.03
4 Duke University 63.93 193.08 191.45 193.20
5 Emory University 69.06 223.07 159.78 191.68
6 Boston University 200.79 89.84 176.67 188.92
7 Stanford University 106.34 126.42 210.45 187.96
8 Yale University 153.21 86.96 202.01 182.08
9 College of William & Mary 36.62 217.00 172.45 181.83
10 Clarkson University 15.08 209.99 193.56 181.79
11 Case Western Reserve University 112.79 91.77 206.23 169.19
12 Brandeis University 82.29 140.63 185.12 168.70
13 University of Connecticut 35.84 195.96 170.33 168.61
14 Fordham University 204.34 91.97 138.66 167.37
15 New York University 310.10 29.44 117.55 167.32
16 Washington University in St. Louis 132.01 94.16 183.00 164.32
17 University of Notre Dame 93.06 119.44 187.23 163.30
18 Syracuse University 122.25 98.13 176.67 158.04
19 Rice University 130.59 89.93 168.22 151.91
20 Cornell University 52.27 186.92 134.44 148.11
21 U of Massachusetts Amherst 42.95 172.06 153.44 147.92
22 U of New Hampshire, Main Campus 41.27 187.75 140.77 147.66
23 Binghamton University 48.26 177.14 142.89 146.34
24 Tufts University 159.42 75.27 151.33 146.08
25 University of Dayton 87.30 111.31 170.33 145.45
26 University of Virginia, Main Campus 72.44 178.72 104.88 133.81
27 University of Vermont 81.49 135.88 123.88 126.58
28 University of California, Los Angeles 137.01 108.46 100.66 121.74
29 University at Albany, SUNY 55.85 124.54 142.89 120.96
30 University of Rhode Island 38.80 174.15 111.21 120.19
31 Miami University 68.41 139.63 117.55 118.68
32 Illinois Institute of Technology 134.05 51.03 142.89 115.95
33 Georgia Institute of Technology 151.54 51.06 128.11 114.41
34 U of California, Santa Barbara 75.58 147.91 98.54 114.30
35 Carnegie Mellon University 149.96 85.60 98.54 113.71
36 University of Denver 129.23 84.81 113.32 113.38
37 University of Maryland, College Park 75.96 130.28 111.21 112.92
38 U of Illinois at Urbana Champaign 102.53 91.64 123.88 111.83
39 University of Maine 33.93 162.83 102.77 106.39
40 North Carolina State University 55.92 170.70 75.32 103.07
41 Pennsylvania State University 77.73 126.18 96.43 102.02
42 University of Florida 69.70 166.82 66.87 101.75
43 University of California, San Jose 82.61 98.25 113.32 99.72
44 George Mason University 53.84 161.37 75.32 97.02
45 University of California, Irvine 90.93 90.51 109.10 96.53
46 Indiana University Bloomington 87.43 129.61 77.43 96.12
47 Ball State University 77.55 105.58 104.88 96.05
48 Portland State University 157.20 47.12 98.54 95.89
49 U of North Carolina at Greensboro 120.11 91.80 85.88 95.54
50 University of Wisconsin, Madison 104.34 119.54 71.10 94.18
51 Missouri University of Science 76.43 95.94 109.10 92.96
52 Drexel University 194.81 38.92 73.21 92.16
53 Oklahoma State U, Stillwater 84.26 80.41 113.32 90.69
54 Bowling Green State University 69.46 90.64 111.21 88.27
55 Temple University 216.54 33.31 56.31 88.09
56 University of Washington, Seattle 118.37 99.23 68.98 88.03
57 University of California, Davis 90.52 113.24 71.10 84.49
58 San Diego State University 176.84 61.56 52.09 82.96
59 University of Tennessee 96.20 75.07 96.43 82.35
60 U of Cincinnati, Main Campus 144.93 69.69 62.65 79.77
61 Texas A&M University 70.97 119.06 71.10 78.87
62 University of Iowa 72.43 114.62 71.10 77.09
63 Ohio State University, Main Campus 128.05 65.56 73.21 76.65
64 Southern Illinois University 28.34 134.80 83.76 76.37
65 University of Kansas 65.95 114.80 71.10 74.27
66 University of Oregon 116.40 83.78 60.54 73.27
67 University of Alabama 71.37 105.51 73.21 72.97
68 Virginia Commonwealth University 157.57 34.70 71.10 71.71
69 University of Colorado Boulder 84.91 83.43 77.43 69.69
70 University of Memphis 63.79 134.88 45.76 68.09
71 Rutgers–Newark 193.58 37.36 37.31 67.80
72 Kansas State University 71.47 96.94 71.10 66.97
73 University of California, Riverside 73.78 79.83 83.76 66.75
74 University of Wyoming 78.16 94.71 66.87 66.07
75 Idaho State University 70.28 63.41 98.54 65.63
76 University of Minnesota, Twin Cities 113.66 61.54 66.87 63.91
172 A. Hajrasouliha / Landscape and Urban Planning 158 (2017) 166–176

Table 3 (Continued)

Rank University Name Urban Score Green Score Living Score Campus Score (Composite score)

77 Auburn University, Main Campus 57.96 111.08 62.65 63.22

78 Colorado State University 79.57 82.53 71.10 62.73
79 Brigham Young University 107.16 74.08 58.43 62.45
80 Northern Arizona University 73.85 67.77 85.88 61.52
81 Cleveland State University 197.32 22.00 37.31 61.07
82 Louisiana State University 53.22 101.11 71.10 61.03
83 University of North Dakota 74.56 77.17 75.32 60.24
84 University of Illinois at Chicago 158.17 25.50 54.20 56.14
85 University of Wisconsin, Milwaukee 122.07 53.82 52.09 54.02
86 University of Alaska Fairbanks 43.55 90.79 75.32 53.72
87 Oregon State University 71.76 83.72 58.43 51.76
88 University of Louisville 100.04 40.56 75.32 51.68
89 Utah State University 76.44 90.88 45.76 49.69
90 University of Akron, Main Campus 135.60 36.27 47.87 47.81
91 Arizona State University, Tempe 103.29 37.79 66.87 46.22
92 University of Utah 72.93 83.16 45.76 43.88
93 University of North Texas 63.61 74.08 58.43 42.81
94 University of Missouri, Kansas 120.36 51.25 35.20 41.04
95 University of Missouri, St. Louis 67.64 91.95 37.31 40.91
96 Indiana University, Purdue U 114.46 48.00 39.42 39.30
97 University of Houston 110.98 29.67 56.31 38.49
98 New Mexico State U, Main Campus 53.13 68.82 56.31 33.85
99 Wayne State University 130.37 9.73 39.42 25.51
100 University of Texas at San Antonio 32.94 101.29 28.87 24.98
101 University of Texas at Arlington 78.95 41.56 41.53 21.11
102 University of Colorado Denver 118.89 22.05 28.87 20.33
103 University of Nevada, Las Vegas 121.34 17.87 28.87 19.15

pus Score and 6-year graduation rate. Campus Score has a skewed
distribution; therefore, natural log transformation was conducted
on Campus Score before exploring its predictive power.
The first linear regression model had an adjusted R2 of 0.802,
and all coefficient estimates were significant at 0.05 or beyond.
The unstandardized coefficient of Campus Score was 6.893, and the
standardized coefficient was 0.414. This means that a 1% increase
in Campus Score is related to an increased freshman retention rate
of 0.0689. The second model had an adjusted R2 of 0.917, and all
coefficient estimates were significant at 0.001. The unstandardized
coefficient of the freshman retention rate was 1.483, and standard-
ized coefficient was 0.745. The unstandardized coefficient of the
natural log-transformed Campus Score was 8.989, and standard-
ized coefficient was 0.271. This means that a 1% increase in Campus
Score is related to an increased 6-year graduation rate of 0.0899.
No multicollinearity or outlier impact has been detected in these
models.
3.2. The relationship of campus score with the external factors:
age, urbanization of setting, climate, region, and academic ranking
Fifteen external factors such as university resources, affordabil-
ity, safety, and campus size are considered in modeling Campus
Score (see Table 2). Only four present statistically significant asso-

Fig. 3. Distribution of the selected campuses.

 A. Hajrasouliha / Landscape and Urban Planning 158 (2017) 166–176 173

Fig. 4. Left: Scatter plot. X: Campus Score, Y: Freshman Retention Rate. R2 = 0.530; Scatter plot. X: Campus Score, Y: 6-Year Graduation Rate. R2 = 0.663.

Table 4
The relationship of Campus Score and the age of campus and the urbanization of setting. Gray cell indicates that the mean of a given group is significantly higher than other
groups.

Frequency Mean of Urban Score Mean of Green Score Mean of Living Score Mean of Campus Score

Age of campus
Pre Civil War 40 103.78 119.36 122.77 126.87
1800s, Post Civil War 35 94.10 91.4089 90.64 86.65
1900s, Pre WWII 14 103.72 83.6265 80.74 80.39
Post WWII 14 100.20 82.5194 77.58 76.18
Total 103 100 100 100 100

Setting
City Large 34 138.92 66.78 86.99 91.04
City Midsize 16 102.62 89.24 95.37 92.33
City Small 25 79.65 113.25 92.97 93.59
Suburb Large 13 82.70 137.85 142.23 139.94
Suburb Midsize and Small 8 52.85 145.68 107.51 108.57
Town and Rural 7 63.57 116.06 111.81 99.92
Total 103 100 100 100 100

ciations with student retention and graduation rates (see Fig. 2).
However, ANOVA test and Post-hoc analysis show that pre-Civil
War campuses have a statistically higher mean Campus Score com-
pared to newer campuses (see Table 4). Comparing the means of
Urban, Green, and Living Scores reveals that the difference between
historic campuses (pre Civil War) and newer campuses is mainly
due to their high Living Score. Commuter campuses became more
prevalent in the twentieth century, quite possibly explaining the
lower Living Score of the newer campuses.
Although campuses in large suburbs generally have a higher
Campus Score, there is no statistically significant difference in Cam-
pus Score for different urbanization settings. However, as expected,
campuses in large cities have a significantly higher Urban Score,
and suburban campuses have significantly higher Green and Living
Scores.
Climate is an interesting external factor. When investigating its
impact on Campus Score the first question that comes to mind is
whether it makes sense to compare the greenness of a campus in
the northeast region with, for example, a desert campus in Arizona Fig. 5. Means of Campus Score for each census region and university type.
or Nevada. Is a different degree of greenness required for optimum
satisfaction in various climates? Perhaps. Nevertheless, although
many studies have found associations between mental and physi- Interestingly, a closer look at these four campuses shows that their
cal health and greenness, there is no study to test these associations Green Scores are low, but they are not necessarily the l“east green
across different climates. However, from this sample of 103 cam- campuses. Except for one case, their overall Campus Score ranking
puses only four are in a desert or semi-arid climate (Table 5). The is lower than their Green Score ranking.
sample size of desert campuses is thus too small to test any mod- The distribution of campuses with higher scores is not geo-
ification in the definition of greenness or weighting by climate. graphically consistent. The one-way ANOVA test of means and
Post-hoc analysis reveal large and significant differences between
174 A. Hajrasouliha / Landscape and Urban Planning 158 (2017) 166–176

Fig. 6. a: Scatter Plot of Campus Score Ranking and Shanghai Ranking, R2 = 0.125; b: Scatter Plot of Campus Score Ranking and Freshman Retention Rate, R2 = 0.539; c: Scatter
Plot of Shanghai Ranking and Freshman Retention Rate, R2 = 0.464; d: Scatter Plot of Campus Score Ranking and Graduation Rate, R2 = 0.661; e: Scatter Plot of Shanghai
Ranking and Graduation Rate, R2 = 0.427.

Table 5
Campuses in desert and semi-arid climate, their Green Score, and Campus Score.

Campuses in desert and semi-arid climate Rank of Green Score Green Score Rank of Campus Score Campus Score

New Mexico State University 77 68.82 98 33.85

Northern Arizona University 78 67.77 80 61.52
Arizona State University, Tempe 92 37.79 91 46.22
University of Nevada, Las Vegas 102 17.87 103 19.15

the means of Northeast campuses and campuses in the other three
census regions. The mean Campus Score for Northeast universities
is 149.03, while only 85.65 for the Midwest and 91.16 for the West.
The lowest mean Campus Score is 75.91 for Southern universities.
The high mean score for the Northeast may driven by the built envi-
ronment that is in general more urbanized, the universities that
are the most historic and well-established in the country and their
long tradition of on-campus housing. Significant difference was also
found using a one-way ANOVA between the mean Campus Score for
Research I universities (112.32) and Research II Universities (86.41)
and between Campus Score for private universities (161.68) for
public universities (80.23). Fig. 5 shows the means plot of Cam-
pus Scores stratified by type and region. The highest mean belongs
to Northeast Research I universities with 163.02, and the lowest
mean belongs to South Research II with 62.79.
The relationship of Campus Score with a well-known university
quality index, Academic Ranking of World Universities, also known
as the Shanghai Ranking (as a proxy for academic and research
 A. Hajrasouliha / Landscape and Urban Planning 158 (2017) 166–176
Fig. 7. standardized values of morphological dimensions of University of Utah’s Campus based on the total samples (left) and just Research I universities in the West region
(right).
performance) was also investigated. The Shanghai Ranking is con-
ducted by researchers at the Center for World-Class Universities
of Shanghai Jiao Tong University (CWCU). This index considers
Nobel prizes, Fields medalists, highly cited researchers, and papers
published in Nature or Science. Out of the 103 universities in
this sample, 73 were ranked among the top 500 universities in
the world. Bivariate correlation was conducted between Shanghai
Ranking and the ranking based on Campus Score. This resulted in
a correlation coefficient of 0.35, significant at the 0.01 level. Yet,
as Fig. 6 reveals, rankings based on Campus Score have a stronger
association with freshman retention rate and graduation rate than
the Shanghai Ranking. The relationship of other commercial rank-
ings such as that of U.S. News and World Report with retention and
graduation rates is not relevant to this study, since retention and
graduation rates are the main indicators of their rankings.
4. Discussion: the application of campus score for campus
planners
Exploring the relationships between Campus Score and desired
outcomes, such as student satisfaction, safety, and campus sustain-
ability, can give campus planners fresh insight into the possible
impacts of campus form. In addition, quantifying dimensions of
campus form can inform campus planners about the norms of cam-
pus design for different university types in different census regions.
For example, the percentage of surface parking areas or the per-
centage of pervious surfaces of one campus can be compared to the
mean value of these variables for similar institutions. However, first
and foremost, Campus Score can inform campus planners about
how their campus compares to peers. For example, the University
of Utah’s ranking is 93 out of 103 campuses. By looking at its three
scores on Urban, Green, and Living, it becomes clear why the Uni-
versity of Utah has such a low rank: its Urban Score is 73, its Green
Score is 83, and its Living Score is 46. As a reminder, note that the
mean value of scores for all samples is 100. Although all three scores
are below average, the Living Score is considerably low. In addition,
note that the three scores are computed from 10 campus-form vari-
ables (see Table 1). Therefore, for more detailed analysis, one can
compare the value of these 10 variables for the University of Utah to
the mean of all samples. Fig. 7 shows standardized values of mor-
175
phological dimensions of the University of Utah’s campus based
on the total sample (left) and Research I universities in the West
region (right). Positive values are above the mean and negative val-
ues are below the mean. Fig. 7 shows that the biggest strength of
the University of Utah’s campus is ‘high percentage of land covered
by tree canopy,’ and its biggest weaknesses are qualities such as
“high percentage of land covered by surface parking areas,” “low
campus mass density,” and “low percentage of students living on
campus.” Recognizing the strengths and weaknesses of a campus
can also help physical planners to identify the main opportunities
and threats. Such quantitative metrics provide a helpful starting
point for campus design assessments, but they are less reliable as
measures of what is needed in the future. Qualitative data gathered
through individual and group interviews would constitute a com-
plementary analysis to establish a clear picture of user needs and
identify critical campus projects for optimal efficiency.
5. Conclusions
The biggest limitation of this research is data availability.
Although data on institutional characteristics are diverse and rela-
tively accessible, very little information on the built environment
characteristics of university campuses is available. Information
such as building height and architectural quality of campus build-
ings is not publicly available and therefore not included in this
research. Moreover, the administrative policies of universities can
have a critical impact on the achievement of different institutional
objectives. Proxies such as “type of institution” or “enrollment
profile” cannot fully represent the complexity of administrative
policies in addressing institutional missions.
Universities in the Northeast census region have significantly
higher mean scores than universities in other census regions, but
this may not be because of the eminence of their institutions; it
may be related to the type of urbanism in that region and, most
importantly, the age of the campuses. Most campuses in that region
were almost fully developed before the mid-20th century when the
mode of campus planning was to follow certain established formal
typologies rather than be dominated by auto-oriented develop-
ment patterns and star architects. However, not all universities in
the northeast region have high Campus Scores. For example, the
176 A. Hajrasouliha / Landscape and Urban Planning 158 (2017) 166–176
Rutgers campus ranking is only 71, and seven of the top 20 cam-
puses are not in the northeast: Emory University (rank 5), Stanford
University (rank 7), College of William & Mary (rank 9), Case West-
ern Reserve University (rank 11), Washington University in St. Louis
(rank 16), University of Notre Dame (rank 17), and Rice University
(rank 19). In addition, it is an interesting finding that Campus Score
ranking, compared to Shanghai ranking, had stronger associations
with freshman retention and graduation rates. This indicates that
compared with its overall reputation, an institution’s physical cam-
pus qualities may have a stronger impact on student satisfaction
and academic performance.
Private universities have significantly higher mean scores than
public schools, and Research I universities have significantly higher
mean scores than Research II universities. This finding may raise
certain questions. Primarily, whether campus form has any true
influence on freshman retention and 6-year graduation rates, or is
it simply that better institutions have better campuses, and physi-
cal campus has no true influence on students’ overall satisfaction or
graduation rate. While Campus Score had significant associations
with freshman retention and graduation rates after controlling for
a set of external factors (as described in Table 2), the causal rela-
tionship is uncertain.
There might be a simple explanation for these relationships.
Although it is hard to imagine that one student decides not to con-
tinue his or her education solely due to the campus qualities, it is
much more likely that a green, urban-feeling, and livable campus
provides a “restorative environment” (Kaplan, 1992) for the stu-
dents where they can reduce their overall level of mental fatigue,
and consequently perform better academically.
The increased pressure on students for “direct attention” (Stuss
& Benson, 1986) leaves them susceptible to fatigue. A well-designed
campus can be a fascinating place to be and allows students to rest
and enjoy their college life. Taking a short walk in a park-like setting
of campus (Green dimension), or visiting coffee shops and galleries
adjacent to campus, or simply watching a vital street (Urban dimen-
sion) or taking a short nap in an accessible dorm (Living dimension)
may have a positive impact on students’ mental health. The per-
ceived restorative qualities of campuses can be measured with
self-report rating scales. Therefore, as a subject of future research
it would be possible to investigate whether or not the restora-
tive qualities of campuses can explain the observed relationship
between Campus Score and retention/graduation rates.
Literature suggests that the physical campus should be designed
and managed as a restorative environment for students to reduce
their mental fatigue and consequently improve their quality of life
and academic performance. Accordingly, understanding the poten-
tial payoffs for student satisfaction and academic performance is of
significant importance when an institution plans its physical cam-
pus. In other words, better understanding of these relationships
would be of value for campus planners in an evidence-based prac-
tice. Campus Score can act as a tool to pave the path for investigation
of particular campus design approaches and to identify the effects
of investments in specific campus forms/strategies.
References
Azevedo, R. (2015). Defining and measuring engagement and learning in science:
Conceptual, theoretical, methodological, and analytical issues. Educational
Psychologist, 50(1), 84–94.
Berger, J. B., & Braxton, J. M. (1998). Revising Tinto’s interactionalist theory of
student departure through theory elaboration: Examining the role of
organizational attributes in the persistence process. Research in Higher
Education, 39(2), 103–119.
Cervero, R., & Kockelman, K. (1997). Travel demand and the 3Ds: Density, diversity,
and design. Transportation Research Part D: Transport and Environment, 2(3),
199–219.
Cervero, R., & Murakami, J. (2010). Effects of built environments on vehicle miles
traveled: Evidence from 370 US urbanized areas. Environment and Planning A,
42(2), 400–418.
Chapman, M. P. (2006). American places: In search of the twenty-first century
campus. Westport, CT: Praeger.
Coulson, J., Roberts, P., & Taylor, I. (2010). University planning and architecture: The
search for perfection. Abingdon, UK: Routledge.
Coulson, J., Roberts, P., & Taylor, I. (2014). University Trends: Contemporary Campus
Design. Abingdon, UK: Routledge.
Custinger, J., & Galster, G. (2006). There is No sprawl syndrome: A new typology of
metropolitan land use patterns. Urban Geography, 27(3), 228–252.
De Nisco, A., & Warnaby, G. (2014). Urban design and tenant variety influences on
consumers’ emotions and approach behavior. Journal of Business Research,
67(2), 211–217.
Delbanco, A. (2014). College: What it was, is, and should be. Princeton, NJ: Princeton
University Press.
Disterheft, A., Caeiro, S., Azeiteiro, U. M., & Leal Filho, W. (2014). Sustainable
universities – a study of critical success factors for participatory approaches.
Journal of Cleaner Production.
Dober, R. P. (1996). Campus planning. New York: Reinhold Publishing Corporation.
Dyson, R., & Renk, K. (2006). Freshmen adaptation to university life: Depressive
symptoms, stress, and coping. Journal of Clinical Psychology, 62(10), 1231–1244.
Ewing, R., & Cervero, R. (2010). Travel and the built environment: A meta-analysis.
Journal of the American Planning Association, 76(3), 265–294.
Ewing, R., Connors, M.B., Goates, J.P., Hajrasouliha, A., Neckerman, K., Nelson, A.C.,
Greene, W., 2013. Validating urban design measures. In: Transportation
research board 92nd annual meeting (No. 13-1662).
Ewing, R., & Handy, S. (2009). Measuring the unmeasurable: Urban design qualities
related to walkability. Journal of Urban Design, 14(1), 65–84.
Ewing, R., Pendall, R., & Chen, D. (2003). Measuring sprawl and its impact: The
character & consequences of metropolitan expansion, report for Smart Growth
America.. Retrieved June 15, 2006, from http://www.smartgrowthamerica.org
Ewing, R., Hajrasouliha, A., Neckerman, K., Purciel-Hill, M., & Green, W. (2015).
Streetscape features related to pedestrian activity. Journal of Planning
Education and Research, 36(1), 5–15.
Felsten, G. (2009). Where to take a study break on the college campus: An
attention restoration theory perspective? Journal of Environmental Psychology,
29(1), 160–167.
Frank, L. D., Sallis, J. F., Conway, T. L., Chapman, J. E., Saelens, B. E., & Bachman, W.
(2006). Many pathways from land use to health: Associations between
neighborhood walkability and active transportation, body mass index, and air
quality. Journal of the American Planning Association, 72(1), 75–87.
Gilboy, M. B., Heinerichs, S., & Pazzaglia, G. (2015). Enhancing student engagement
using the flipped classroom. Journal of Nutrition Education and Behavior, 47(1),
109–114.
Gupta, K., Kumar, P., Pathan, S. K., & Sharma, K. P. (2012). Urban Neighborhood
Green Index—A measure of green spaces in urban areas. Landscape and Urban
Planning, 105(3), 325–335.
Hajrasouliha, A., & Yin, L. (2015). The impact of street network connectivity on
pedestrian volume. Urban Studies, 52(13), 2483–2497.
Hajrasouliha, A. (2015). The morphology of the “well-designed campus”: campus
design for a sustainable and livable learning environment. Doctoral dissertation,
THE UNIVERSITY OF UTAH.
Hajrasouliha, A., & Ewing, R. (2016). Campus does matter: The relationship of
student retention and degree attainment to campus design. Planning for Higher
Education.
Hajrasouliha, A. H., & Hamidi, S. (2016). The typology of the American metropolis:
monocentricity, polycentricity, or generalized dispersion? Urban Geography,
1–25.
Hillier, B. (2007). Space is the machine: A configurational theory of architecture.
Cambridge: Cambridge University Press.
Hillier, B. (2009), Spatial sustainability in cities: Organic patterns and sustainable
forms. in Koch, D., Marcus, L., & Steen, J. (eds.), Proceedings of the 7th
international space syntax symposium.
Hipp, J. A., Gulwadi, G. B., Alves, S., & Sequeira, S. (2016). The relationship between
perceived greenness and perceived restorativeness of university campuses and
student-reported quality of life. Environment and Behavior, 48(10), 1292–1308.
Kaplan, S. (1992). The restorative environment: Nature and human experience. In
Role of horticulture in human well-being and social development: A national
symposium (pp. 134–142).
Kenney, D. R., Dumont, R., & Kenney, G. S. (2005). Mission and place: Strengthening
learning and community through campus design. Westport, CT: Praeger
Publishers.
Lounsbury, J. W., & DeNeui, D. (1995). Psychological sense of community on
campus. College Student Journal, 29(3), 270–277.
Mitchell, W. J., & Vest, C. M. (2007). Imagining MIT: Designing a campus for the
twenty-first century. Cambridge, MA: MIT Press.
Naretto, J. A. (1995). Adult student retention: The influence of internal and
external communities. NASPA Journal, 32(2), 90–97.
Strange, C. C., & Banning, J. H. (2001). Education by design: Creating campus learning
environments that work. San Francisco, CA: Jossey-Bass.
Stuss, D. T., & Benson, D. F. (1986). The frontal lobes. New York: Raven Pr.
Styron, R., Jr. (2010). Student satisfaction and persistence: Factors vital to student
retention. Research in Higher Education Journal, 6, 1.
Turner, P. V. (1984). Campus: An american planning tradition. Cambridge, MA: MIT
Press.
Yao, L., Liu, J., Wang, R., Yin, K., & Han, B. (2014). Effective green equivalent—A
measure of public green spaces for cities. Ecological Indicators, 47, 123–127.