Travel Behaviour and Society 11 (2018) 78–85

Contents lists available at ScienceDirect

Travel Behaviour and Society

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

Gender gap generators for bicycle mode choice in Baltimore college T

campuses
⁎
Farhad Abasahla, Kaveh Bakhsh Kelarestaghib, , Alireza Ermagunc
a
Department of Civil and Environmental Engineering, University of Maryland, 1173 Glenn Martin Hall, College Park, MD 20742-2835, USA
b
Charles Edward Via, Jr. Department of Civil and Environmental Engineering, Smart Urban Mobility Laboratory, National Capital Region, Virginia Tech University, 7054
Haycock Road, Room 425, Falls Church, VA 22043, USA
c
Department of Civil and Environmental Engineering, Northwestern University, Technological Institute, 2145 Sheridan Rd, Evanston, IL 602088, USA

A R T I C L E I N F O A B S T R A C T

Keywords: This study explores the gender equity in bicycle mode choice and obstacles preventing women from bicycling to
Equity promote biking on major college campuses in the Baltimore metropolitan area. Socioeconomic data, travel
Gender preferences, mode accessibility, and individual factors of the surveyed population are used to identify generators
Gap of gender difference. To investigate gender equity, we develop a bivariate statistical analysis along with a two-
College campus
level nested logit model. The results of the bivariate statistical analysis indicate females are about 30% less likely
Sustainability
to bicycle from home to campus and are significantly more sensitive to environmental and infrastructural
Mode choice
conditions. Complementary to the bivariate statistical analysis, the results of the two-level nested logit model
demonstrate that distant trips, longer travel times, not having access to a bicycle, and an unsafe environment
avert females from bicycling. We also find that undergraduate females are less likely to bike to campus than
other group of students. The findings highlight that the integration of bicycle and transit services, advancing
infrastructure to separate bicycle from motorized traffic, improving safety in bicycle facilities, and enhancing
public knowledge about local bicycle routes promote bicycling among females.

1. Introduction infrastructure. As a result, different bicycle facility designs can impact

As an inexpensive, clean, and healthy mode (Handy, 2011), biking
is a way of promoting environmentally sustainable transport. This,
however, requires coping with multi-faceted challenges in mobility,
safety, accessibility, economy, and equity. Transportation cannot be
sustainable unless it assures equitable mobility across gender line
(Hanson, 2010). Hence, as part of efforts to foster bicycling in com-
munities, existing barriers and issues have to be identified and ad-
dressed for both genders. Baker (2009) asserts that women are “in-
dicator species” for bicycle-friendly environments because of their
aversion to risk. This means a higher percentage of female riders in-
dicate safer infrastructure for bicycling.
The gender gap in commuting is partially attributed to the stay-at-
home role some married women take on the household. Nevertheless,
the dimension of this gap is continuously changing through mutual
interchange with needs, policies, and planning (Crane, 2007). Women
are more likely to use the bicycle for shopping and recreational pur-
poses and less for commuting than men (Krizek et al., 2005). They show
a different perception of risk (e.g. unsafe drivers’ behavior) from men
and prefer different bicycle facilities (e.g. bicycle lane) and
different audiences (Krizek et al., 2005).
This study augments the current literature of sustainable transpor-
tation and examines the gender gap in bicycle mode choice for regular
travelers to college campuses located in the Baltimore metropolitan
area, Maryland. Baltimore City with about 622,000 population is the
busiest city in the State of Maryland and the female population is 52.9%
of the total population (City-Data.com, 2016). More than dozen public
and private universities are in the Baltimore Metropolitan area. Uni-
versity students are thought to welcome bicycling as a low-cost mode,
which fulfills their multi-purpose trips. Universities are believed to be
excellent environments for exploring influential factors on bicycling for
several reasons including necessity of promoting new modes due to
financial or environmental concerns, having suitable physical condi-
tions for bicycling, direct aggregate effect by changing the mode choice
in a large number of employees, and positive spill-over effects on
communities through developing positive attitudes in students (Miller
and Handy, 2012). To investigate gender equity, we develop a bivariate
statistical analysis along with a two-level nested logit model. The re-
sults have the potential of contributing to identifying gender gap gen-
erators in college campuses, as well as identifying obstacles that keep

⁎
Corresponding author.
E-mail addresses: farhad@umd.edu (F. Abasahl), kavehbk@vt.edu (K.B. Kelarestaghi), alireza.ermagun@northwestern.edu (A. Ermagun).

https://doi.org/10.1016/j.tbs.2018.01.002
Received 3 April 2017; Received in revised form 7 November 2017; Accepted 12 January 2018
2214-367X/ © 2018 Hong Kong Society for Transportation Studies. Published by Elsevier Ltd. All rights reserved.
F. Abasahl et al.
females away from bicycling and ways to promote sustainable transport
in the Baltimore metropolitan area.
The remainder of the paper is organized as follows. First, we review
the related studies targeting the gender gap determinants, as well as
factors that affect the transportation mode choice of educational set-
tings’ faculty, staff, and students. Second, we show the survey and the
data gathering method along with a descriptive of the sample used for
the analysis. Third, we juxtapose the travel behavior of females with
males using a bivariate statistical analysis. This is then followed by
developing two-level nested logit models to examine variables ex-
plaining the travel mode choice of females and males. We finally pro-
vide an in-depth discussion over the findings and conclude the paper
with summarizing the key findings, representing the study im-
plementations, and recommending future research avenues.
2. Background
A U.S. study (McDonald, 2012) showed boys between 8 and
13 years old bike to school 2–3 times more than girls. This gap is be-
lieved to be related to less independence mobility of girls than boys
among schoolchildren. Trapp et al. (2011) indicated parents’ con-
fidence in their child’s bicycling skills is a significant factor to en-
courage children to bicycle to school. The gender gap is then expected
to diminish when schoolchildren get older and go to college, as they
have higher independency and increased equal opportunities (Ermagun
and Levinson, 2016; Ermagun and Samimi, 2015). In practice, however,
results do not fully correspond to expectations.
Akar and Clifton (2009) conducted research on a target population
of faculty, staff, and students at the University of Maryland. They de-
termined that safety and travel time are major concerns for commuters.
This study suggested that women are generally less likely to use bicycle
for commuting. Incentives and disincentives such as planning of bicycle
lane or bicycle routes, and proper traffic regulations in and around of
campus were proposed to increase the share of biking.
Twaddle et al. (2010) indicated that facility preference is a similar
concern between men and women, while women are often occasional
bicyclists and more concerned about safety than men. This is consistent
with the findings of another study conducted by Wang et al. (2015) at
the Ohio State University, in which they emphasized on higher like-
lihood of commuters being male and a student. They also suggested that
perceived safety, travel cost, and concern for the environment would
impact bicycling.
Handy (2011) demonstrated fear of collision or being attacked
(safety related factors) as primary factors that discourage women from
bicycling. Matsuda et al. (2000) showed that women have different
perceptions of risk and tend to avoid risky practices while bicycling
than men.
Akar et al. (2013) conducted a study in the Ohio State University to
explore factors affecting bicycling decisions for surveyed females in-
cluding faculty, staff, and students. The results are consistent with
previous studies as they revealed different risk perceptions between
men and women in similar environments. The study indicated that
women are more sensitive to being close to bicycle infrastructures. They
suggested that changing policy and improving infrastructure encourage
women to choose bicycling.
Gender differences in bicycling have not been limited to geography.
Various studies found similar characteristics in different countries,
which are associated with the gender gap in bicycling or are likely to
propagate this gap. For example, Garrard et al. (2008) showed that
female bicyclists in Melbourne, Australia prefer to use separate bicycle
routes than shared paths. They proposed that separation between mo-
torized and non-motorized traffic encourages more women to bike.
Another study in China suggested that trip chains for female bicyclists
are longer than men (Zhao et al., 2015).
In the U.S., female bicyclists in urbanized areas only compose 25%
of users while in the Washington, D.C. region this ratio was more
79
Travel Behaviour and Society 11 (2018) 78–85
balanced and reached to 33% (Buehler, 2012). According to data pre-
sented by Capital Bicycle-share (CaBi), the bicycle-sharing system in the
Washington metropolitan area, 52% of casual users in 2010 were fe-
male which is a significant departure from typical female bicyclists in
the whole country (Buehler, 2012). For 2011, 2012, and 2014 the CaBi
female user ratio was reported as 45%, 43%, and 41%, respectively.
Almost 64% of female bicyclists specified their trip purpose as recrea-
tional as opposed to 56% of men. The majority of females were con-
cerned with safety and bicycle facilities (LDA Consulting, 2013; LDA
Consulting, 2015). Citi Bicycle is a bicycle-sharing system in New York
launched in 2013. According to their 2015 report (Kaufman et al.,
2015), 77.7% of users were male. A glimpse into this report shows that
female user’s ratio varies from 14% to 41% depending on location.
Places with higher risk of crashes, heavier traffic, and lack of bicycle
lanes seem to be less desirable for female bicyclists (Kaufman et al.,
2015).
3. Bicycle-to-campus survey
The Bicycle-to-Campus Survey was designed and administered using
in-person interviews and online questionnaire by a group of researchers
at the Morgan State University (MSU). The purpose of the survey was to
identify major issues of bicycling to campus for students, staff, and
faculty, and finding ways to implement policies with the aim of pro-
moting sustainable transport in college campuses. In addition to the
sociodemographic related questions, participants were asked to respond
to some other key questions in the survey. This includes items such as
“access to personal motorized vehicle to commute”, “access to a bicycle
to bicycle-to-campus”, “bicycle skill level”, “frequency of bicycle-to-
campus during a week”, and “whether you encourage others to bicycle-
to-campus”, Participants were also asked to determine if obstacles such
as “Darkness”, “Physical need”, “Condition of bicycle facilities”, “Heavy
traffic”, “Risk of injury and theft”, “Travel time”, and “Air pollution”,
keep them from bicycling between home and campus. In the commu-
nity and campus related section of the questionnaire, the participants
were asked to rank campus and community-related changes to stimu-
late bicycle-to-campus commute. This includes items such as “Bike
sharing programs for campus and/or community”, “Incentive programs
for bikers”, “Increase enforcement and education”, and “Improve bi-
cycle facilities in campus”.
The survey was initially conducted from November 6th, 2014 to
November 26th, 2014 using both in-person interviews and web-based
methods. In-person interviews were held at Johns Hopkins University
(JHU), Towson University, and the University of Maryland – Baltimore
County (UMBC). The on-line survey drew higher number of respondents
and included Morgan State University (MSU), JHU-Homewood Campus
(JHU-H), Towson University, UMBC, JHU-East Baltimore Campus
(JHU-E), and few from other colleges. During the initial 20 days period,
268 responses were collected, of which 255 responses were valid.
However, the web-based survey remained opened until June 6th, 2015
and the number of respondents spiked in the following months due to
the better publicity. Overall, from November 6th, 2014 to June 6th,
2015, we received 780 completed questionnaires. Following the data
processing and elimination of incorrect, or error data, the total number
of valid inputs was reduced to 698, of which 50% are females.
3.1. Sample characteristics
An overview of the data statistics is depicted in Table 1. As shown,
female undergrads are more underrepresented than male undergrads
(37.8% versus 51.3%) whereas female staff has a bolder representation
than their male coworkers (19.2% versus 8.9%). The majority of sur-
veyed population for both genders comes from the University of
Maryland, Baltimore County (41.8% male and 35.8% female). More
female respondents live on campuses (10.6% versus 6.0%) while more
males live 26 miles or farther away from campus (14.1% versus 7.7%).
F. Abasahl et al. Travel Behaviour and Society 11 (2018) 78–85

Table 1 participants prefer non-active transportation modes (67%), and (7) al-
Statistics Overview. most 70% of the participants found darkness as an obstacle to their
bicycle choice while only 34% found air pollution as an obstacle.
Male Female
Fig. 1 reveals some aspects of the users of each mode choice. Fig. 1a
N % N % shows that most bicycle users are between 25 and 36 years old. In
Fig. 1b, stating race as White indicates a stronger tendency to bike more
Academia Status Undergraduate student 179 51.3 132 37.8
than other races, while Black race indicates using walk more and less
Graduate student 106 30.4 108 30.9
Staff 31 8.9 67 19.2 automobile than the others. Fig. 1c shows that the majority of bicyclists
Faculty 33 9.5 42 12.0 turn out to feel confident in their mode choice and in the Fig. 1d we can
Campus MSU 89 25.5 73 20.9
observe the dependency of mode choice to the distance.
JHU-H 50 14.3 54 15.5
JUH-E 17 4.9 24 6.9 4. Bivariate statistical analysis
UMBC 146 41.8 125 35.8
Loyola University 16 4.6 42 12.0
Bivariate analysis determines what factors or variables have statis-
Maryland
Towson University 24 6.9 17 4.9 tically significant differences between female and male respondents.
Other 7 2.0 14 4.0 Pearson’s chi-squared test is used for this purpose and variables with
Distance from Campus On-campus 21 6.0 37 10.6 5% level of significance in the difference between genders are shown in
D ≤ 5 miles 159 45.6 142 40.7 bold in Table 3. Among the socioeconomic variables, marriage and
5 < D ≤ 15 miles 80 22.9 102 29.2 having a graduate degree have a significant difference between men
16 < D ≤ 25 miles 40 11.5 41 11.7 and women. As far as built environment related variables are con-
26 < D ≤ 50 miles 38 10.9 23 6.6
D > 50 miles 11 3.2 4 1.1
cerned, variables associated with participants facility preference to ride
on (bicycle lane and separated facilities) and also the condition of bi-
Own or have access to No 221 63.3 204 58.5
cycle facilities indicate significant difference among genders. Among
bicycle Yes 128 36.7 145 41.5
the barriers and obstacles, several variables were found with a statis-
Bicycle to Campus per Never 222 63.6 244 69.9
tically significant difference between female and male respondents.
week Rarely 42 12.0 47 13.5
1–2 days 20 5.7 23 6.6 This includes: risk of injury, risk of theft, weather condition, travel
3–4 days 34 9.7 21 6.0 time, darkness, air pollution, need for trip change, need to accompany
5 days and more 31 8.9 14 4.0 someone and/or cargo, and physical need.
Mode Choice Walk 45 12.9 44 12.6 A higher percentage of surveyed females have graduate degrees
Bicycle 84 24.1 56 16.0 than men. Married female percentage is also higher. These can be re-
Transit 41 11.7 46 13.2 lated to presence of higher percentage of female faculty or staff in
Automobile 179 51.3 203 58.2
surveyed population.
On average, male respondents live farther from campus and claim
About 41.5% of surveyed females compared to 36.7% males own or better bicycling skills than females. Moreover, a higher percentages of
have access to bicycle for personal use. About 70% of surveyed females males’ who bike to campus are involved in bicycle incidents, or prefer
never used bicycle for campus trips as opposed to 63.3% of male re- bicycle lanes.
spondents. In terms of mode choice, some respondents indicated mul- Higher percentages of women who use motorized mode for campus
tiple modes. trips, prefer bicycle separation, or know someone who was involved in
Four modes of travel, including walking, biking, taking public a bicycle incident. Bicycling obstacles with significant differences be-
transit, and using private car are considered the main modes of travel to tween two genders are marked by higher percentage of surveyed fe-
campus. Few assumptions are made to distinguish active from non-ac- males than males.
tive transportation modes. Active modes include bicycle and walk On the other side, there are variables that show no statistically
modes, which include all respondents who bicycle to campus at least significant difference between two groups. Age, income, walk mode,
once a week and those who solely walk to campus. Almost 24% of male access to personal bicycle, access to personal vehicle, travel time and
participants use bicycle as opposed to 16% of females while the per- convenience (access to shower and changing room) are among those
centages of female and male walkers are almost the same (about 13%). variables. Between 6 and 7 percent of both groups did not know how to
Non-active mode includes respondents who chose public transit or bicycle and there is no significant difference between genders for this
private car for their trips to campus. The difference between female and variable.
male bicyclist is about 8% and they use automobile (almost 7%) or
transit (almost 1%). 5. Mode choice analysis: a two-level nested logit model
Table 2 summarizes variables used for the analysis. Variable are
classified as socioeconomics (e.g. age, income, position), spatial (e.g. Four modes of travel are considered for developing a mode choice
participant distance from campus), travel preferences (e.g. motorized, model: Walk, Bicycle, Transit, and Automobile. The first two are active
walk, bicycle), individual factors (e.g. bike access), individual attitude modes of travel, while the last two are categorized as motorized or non-
(e.g. public opinion about bicycling), and physical or environmental active modes of travel. To examine factors affecting the mode choice
obstacles (e.g. darkness, air pollution). The stat (average and standard decision among females and males in campus trips, we develop three
deviation) for each variable is presented in Table 2. For instance, the distinct two-level nested logit models. This provides an understanding
statistics represent that: (1) on average participants’ age is about over the gender gap generators and addresses some of the differences
25–36 years old, (2) on average the participants’ income is in the range between men and women in bicycle mode choice. To elucidate these
of $25 k–$50 K, (3) on average participants indicated their bicycling differences, three models are developed: pooled model, female model,
skill level as “Interested but concern” or higher, (4) participants’ per- and male model.
ception over the risk measures (risk of theft and injury) are almost the Among the possible tree structures, we selected the tree structure
same (61% and 67%, respectively), (5) Although only 40% of the portrayed in Fig. 2 using the model selection criteria that include: (1)
participants were involved in a bicycle incident, almost 60% of them goodness-of-fit measures, (2) inclusive value (IV) parameters, and (3)
knows someone that had an incident before, (6) majority of the significance and rationality of the estimates (Hensher et al., 2005). The
selected structure has two limbs in its upper level that cleave into active

80
F. Abasahl et al. Travel Behaviour and Society 11 (2018) 78–85

Table 2
Variable descriptions.

Variable Descriptive Avg. Std. Dev.

AGE 1: 18 to 24, 2: 25 to 36, 3: 37 to 48, 4: Above 49 1.94 0.99

AGE_1834 1: Age of [18–34], 0: Otherwise 0.76 0.42
AGE_1824 1: Age of [18–24], 0: Otherwise 0.40 0.49
INCOME 1: Under 15 k, 2:15–25 k, 3: 25–50 k, 4: 50–75 k; 5: 75–100 k, 6: Above 100 k 3.69 1.72
HIGHINC 1: Income of $50,000 and above 0.53 0.49
UNDERGR 1: Undergraduate student, 0: Otherwise 0.44 0.49
GRAD_DEG 1: Have Master or/and PhD degree, 0: Otherwise 0.37 0.48
BIKE_SKILL 1: No way, no how, 2: Interested but concern, 3: Enthused & confident, 4: Strong & fearless 2.68 0.82
STRNG&FRLES 1: Bicycle skill level: Strong & Fearless, 0: Otherwise 0.16 0.37
ENTH&CONF 1: Bicycle skill level: Enthused & Confident, 0: Otherwise 0.40 0.49
D2C 1: On-campus, 2: < 5, 3: 5–15, 4: 16–25, 5: 26–50, 6: Above 50 2.76 1.17
ONCAMP 1: Living on-campus, 0: Otherwise 0.08 0.27
D2C_5 1: House distance to campus is less than 5 miles, 0: Otherwise 0.43 0.49
D2C_15 1: House distance to campus is [5–15] miles, 0: Otherwise 0.26 0.43
MARRIED 1: Married, 0: Otherwise 0.29 0.46
INC_I 1: I have experienced bicycling incident, 0: Otherwise 0.40 0.49
INC_O 1: I know someone had bicycling incident, 0: Otherwise 0.60 0.49
VEH_AC 1: Access to personal vehicle, 0: Otherwise 0.75 0.42
TRANS_AC 1: Access to transit 0: Otherwise 0.70 0.45
BIKE_AC 1: Access to bicycle for personal use, 0: Otherwise 0.60 0.48
ENCRGB 1: Encourage others to bicycle, 0: Otherwise 0.56 0.49
BIKETREX 1: Bicycle on-board transit experience, 0: Otherwise 0.11 0.32
ROADFAM 1: Familiarity w/local bicycle routes, 0: Otherwise 0.63 0.48
BFREQ_N 1: Bicycle-to-campus frequency: Never, 0: Otherwise 0.66 0.47
MOTORIZED 1: Mode choice: Motorized vehicle, 0: Otherwise 0.67 0.47
WALK 1: Mode choice: walk, 0: Otherwise 0.13 0.33
BICYCLE 1: Mode choice: Bicycle, 0: Otherwise 0.20 0.40
NO_PREF 1: No preferred bicycle facility, 0: Otherwise 0.15 0.36
PREF_LN 1: Preferred bicycle lane, 0: Otherwise 0.71 0.46
PREF_SEP 1: Preferred bicycle separation, 0: Otherwise 0.14 0.35
PUBLIC_OP 1: feel more comfortable if others bike too, 0: Otherwise 0.17 0.37
OB_RI 1: Risk of injury keeps me from bicycling, 0: Otherwise 0.67 0.47
OB_TR 1: Traffic is too fast and heavy keeps me from bicycling, 0: Otherwise 0.78 0.41
OB_DR 1: Incautious motorists keeps me from bicycling, 0: Otherwise 0.80 0.39
OB_WE 1: Bad weather keeps me from bicycling, 0: Otherwise 0.88 0.33
OB_AIR 1: Exposure to air pollution keeps me from bicycling, 0: Otherwise 0.34 0.47
OB_C_FAC 1: Condition of bicycle facility keeps me from bicycling, 0: Otherwise 0.73 0.44
OB_C_R 1: Condition of roads keeps me from bicycling, 0: Otherwise 0.77 0.42
OB_AMEN 1: Not having access to showers & changing rooms keeps me from bicycling, 0: Otherwise 0.52 0.50
OB_COMP 1: Need to accompany someone and/or cargo keeps me from bicycling, 0: Otherwise 0.55 0.50
OB_PHY 1: Physical needs (energy and comfort) keeps me from bicycling, 0: Otherwise 0.46 0.50
OB_SHAPE 1: I am not in shape keeps me from bicycling, 0: Otherwise 0.29 0.45
OB_BIKE 1: I don’t own or have access to a bicycle, 0: Otherwise 0.33 0.47
OB_KNW 1: Not knowing how to ride a bicycle keeps me from bicycling, 0: Otherwise 0.064 0.25
OB_RTEFT 1: Risk of bicycle theft keeps me from bicycling, 0: Otherwise 0.61 0.48
OB_TRIPCH 1: Need to go to other destinations keeps me from bicycling, 0: Otherwise 0.60 0.48
OB_DARK 1: Darkness keeps me from bicycling, 0: Otherwise 0.67 0.46
OB_TT 1: Travel time keeps me from bicycling, 0: Otherwise 0.58 0.49

and non-active limbs. The active nest encompasses walk and bike
modes, while transit and automobile modes are located in the lower
level of the non-active limb.
The probability of choosing each travel mode based on the two-level
NL model is presented in Eq. (1) (Coldren and Koppelman, 2005). The
utility of travel mode i is the sum of systematic utility (Vi ) and an error
term (εi ) . The IV parameters are used for estimation of procedure, and
these parameters should meet certain conditions of nested logit to be
consistent with global utility maximization theory. In the two-level NL
model, the IV parameters must be positive and less than one.
mode i.Pn is the participant share allocated to limb n. Pi | n is the parti-
cipant share allocated to transportation mode i given limb n. μ is the
inverse logsum parameter also known as coefficient of the IV.
The final models are depicted in Table 4. We include only the
variables that are significant at the 90% interval. The IV parameters of
the active and non-active limbs for all three models are statistically
positive and less than one, according to Wald test. The values of IV
parameter for the case of active and non-active limbs are, respectively,
0.53 and 0.52 (for the pool model), 0.68 and 0.38 (for the male model),
0.82 and 0.82 (for the female model).
As for the pooled model, the negative coefficients of ONCAMP and

Pi = Pn × Pi | n =
exp ( τ ) × exp(μV )
1
μ n i
D2C_5 indicate that it is less likely to use automobile or transit if people
live on campus or within its 5-mile radius. Access to personal vehicle or
exp( τ ) ∑ exp(μV )
1
∑n′∈N μ n′ i′∈ n
i′
transit has positive effects on choosing them for campus trips. The re-
sults also show younger traveler (age between 18 and 34) are less likely
and to make their trips with automobile. Odds are that STRNG&FRLES and
ENTH&CONF bicycle riders are less likely to choose non-active modes
⎛ ⎞ of travel. There is also a strong probability that those who select au-
τn = ln⎜∑ exp(μVi′)⎟
⎝ i ⎠ (1) tomobile or transit for their campus trip, have never bicycled to
campus. Higher income individuals are not willing to use transit
In which, Pi is the participants share allocated to the transportation

81
F. Abasahl et al.
Fig. 1. Age, race, bicycle skill and travel distance of four mode respondents.
services (Ermagun et al., 2015; Ermagun and Levinson 2017).
Having access to personal bicycle has positive impacts on choosing
this mode of travel. However, travel time substantially has negative
effects on bicycling and walking. Respondents with bicycle on-board
transit experience and those who are familiar with local bicycle routes
are more likely to make their campus trips with bicycle.
As for the male model, respondents who live on campus or within its
5 miles share the views of the whole group toward automobile and
transit. Having a personal access to automobile is a decisive factor for
those who choose this mode and they probably never use bicycle for
their campus trips. Access to transit is very important for its users and
those who live 5–15 miles around the campus probably use this service,
while they may never use bicycle for their trips. It seems transit is less
attractive for male undergraduates. Bicycle access is an important
factor for bicyclists. Bicyclists encourage others to utilize bicycle in
their trips and presumably have experience with bicycle on-board
transit.
As for the female model, respondents who live on campus or its
vicinity are less probable to use automobile or transit. In a bigger
neighborhood, up to 15 miles of campus, females are less probable to
drive personal vehicles for campus trips. Female respondents who use
automobile show prominent access to this mode and they have a ne-
gative view over their capability and confidence of riding bicycle.
Females aged between 18 and 34 are less likely to use automobile for
their commute. The results show that darkness is one of the reasons that
female's use automobile or transit. Many women who bicycle encourage
others to bike, benefit from transit on-board services, and are familiar
with local bicycle routes. It also appears that undergraduate female
students are not likely to ride bicycles in campus trips. Travel time and
risk of bicycle theft are among reasons that prevent women from
choosing bicycle.
82
Travel Behaviour and Society 11 (2018) 78–85
6. Discussion
Analyzing the sample, we have found that women choose bicycle
30% less than men for their campus trips and they seem to fill this gap
with motorized modes of travel. Our study shows that this gap is not
rooted in less accessibility, ownership issues, and knowledge how to
bike but other factors. The rate of people who do not know how to
bicycle is about the same (6–7%) for both gender groups. A smaller
percent of surveyed females have personally experienced bicycle in-
cidents (about 20%) than males but higher percent of them (about 15%)
know someone who already experienced some sort of bicycle incident.
These can affirm that women are more prone to avoid risky bicycling
while are more alert to incidents with other bicyclists. On average,
females live closer to campus than men, but both groups perceive dis-
tance as an obstacle to choosing bicycle mode. The preferred bicycle
facility between surveyed women and men is significantly different;
men prefer bicycle lanes but women are pro separate bicycle routes.
This is consistent with previous studies (Krizek et al., 2005; Twaddle
et al., 2010; Wang et al., 2015). Significantly, a higher percentage of
women surveyed believe risks associated with collision, theft, injury,
bad weather, air pollution, darkness, and physical condition of the road
are main factors which keep them away from bicycling.
To provide an in-depth discussion over gender gap in campus trips,
we calculate the ratio of coefficients for similar variables in the male
and female models. Fig. 3 depicts the results. As shown, travel time for
walking mode is more discouraging for females than males in walking,
while for bicycle mode, it is more encouraging for females. Surprisingly,
female bicyclists, who are willing to encourage others to bicycle-to-
campus and have experience taking bicycle on transit, are less likely to
use bike for their commute, compared to their male counterpart. Fe-
males’ access to transit have slightly more contribution toward
F. Abasahl et al. Travel Behaviour and Society 11 (2018) 78–85

Table 3 time is still considered to negatively influence bicycling across both

Bivariate Analysis. genders (Akar and Clifton, 2009; Fishman et al., 2014). While having
access to personal automobile, transit, or bicycle positively affects their
Type Variable Male Female p-Value
choice, those that bicycle have a positive role to encourage others to
Socioeconomics AGE 1.96 2.11 0.291 bicycle. On the positive side, experiencing bicycle on-board transit and
INCOME 3.58 3.81 0.101 familiarity with local bicycle routes at trip origin and destination has a
GRAD_DEG 31.81% 42.41% 0.004
positive role on bicycling to campus. Our study shows that bicyclists
MARRIED 26.36% 33.52% 0.039
reflect their positive views toward this mode choice by encouraging
Spatial factor D2C 2.85 2.66 0.010 others to bicycle.
Travel preferences MOTORIZED 63.04% 71.35% 0.019 Risks of theft, which has a considerable impact on the utility of
WALK 12.89% 12.61% 0.910 bicycling (Ji et al, 2017), prevent a significant number of females from
BICYCLE 24.07% 16.05% 0.008
riding. This, to some degree, could be related to the negative impact of
NO_PREF 14.04% 15.76% 0.524
PREF_ LN 74.50% 66.76% 0.025 undergraduate students on bicycling for surveyed females. This variable
PREF_ SEP 11.46% 17.48% 0.024 can be interpreted along with women’s sensitivity to risks, so they may
Individual factors BIKE_AC 63.32% 58.45% 0.187 feel more vulnerable to ride a bicycle. We suggest further studies on this
VEH_AC 73.93% 77.36% 0.290 aspect.
BIKE_SKILL 2.89 2.47 ≤0.001 In terms of other barriers for female bicyclists, risks associated with
INC_I 45.27% 35.24% 0.007 bicycle theft significantly impact the behaviour of female respondents.
INC_O 55.01% 65.33% 0.005
These findings are consistent with previous studies (Kelarestaghi et al.,
Individual attitude, and Physical OB_BIKE 30.09% 35.82% 0.107 2018; Handy, 2011; Krizek et al., 2005; Akar and Clifton, 2009;
or Environmental obstacles OB_KNW 6.59% 6.30% 0.878
Twaddle et al., 2010; Matsuda et al., 2000; Akar et al., 2013).
for bicycling PUBLIC_OP 15.47% 18.05% 0.362
OB_TT 55.01% 61.03% 0.107 The experience of bicycle on-board transit has positive impacts on
OB_RI 57.59% 76.79% ≤0.001 bicycling for female as well as male respondents. In addition, being
OB_RTEFT 54.73% 67.91% ≤0.001 familiar with bicycle routes has a positive influence on bicycle mode
OB_TR 70.20% 86.53% ≤0.001
choice. These may indicate the importance of developing public
OB_DR 74.50% 87.11% ≤0.001
OB_W 82.81% 92.26% ≤0.001
transportation systems and improving chain supply management at
OB_AIR 28.65% 40.11% 0.001 transit hubs. Increasing bicycle on-board capacity in transit services,
OB_DARK 57.31% 77.08% ≤0.001 safe bicycle racks, lockers at transit hubs, and boosting bicycle routes
OB_C_FAC 66.48% 80.23% ≤0.001 may be effective to encourage more women to bicycle.
OB_C_R 67.62% 85.67% ≤0.001
We found that being a female undergraduate negatively impacts
OB_AMEN 52.15% 51.00% 0.762
OB_COMP 51.00% 58.45% 0.048 bicycling for women. Education may change attitude a and eventually
OB_TRIPCH 53.87% 67.05% ≤0.001 result in mode shift.
OB_PHY 37.54% 53.58% ≤0.001
OB_SHAPE 20.92% 36.68% ≤0.001
7. Conclusion

commute by transit compare to males. In terms of the housing effect,
more females than males who live on-campus or within a 5-mile dis-
tance use automobile or transit for their transport. Access to the per-
sonal vehicle contributes to the higher automobile utility for the male
participants (60% higher). However, females with bike skills of “strong
and fearless” tend to commute by automobile less than males.
Our gender pooled model reaffirms that having access to personal
bicycle makes it more likely to choose bicycle for campus trips. While
close distances lessen the popularity of automobile and transit, travel
College campuses are considered as an interesting community for
transportation mode choice analysis for several reasons including: ir-
regular scheduling, population diversity, financial or environmental
concerns, and being a setting that is apt for physical conditions for
bicycling. Despite necessities for investigating the transportation mode
choice behavior of individuals on college campuses and to stipulate
suitable plans and policies for them, rigorous research is lacking. We
attempted to delve into the understanding of transportation mode
choice determinants in college campuses and to provide new

Fig. 2. Nested logit structure.

83
F. Abasahl et al. Travel Behaviour and Society 11 (2018) 78–85

Table 4
Gender pooled and specific nested models.

Mode Variables Pool Model Male Model Female Model

Coefficient t-test Coefficient t-test Coefficient t-test

Automobile ONCAMP −10.869*** −3.20 −12.634** −2.39 −6.711** −2.48

D2C_5 −5.212*** −2.85 −7.413** −2.5 −2.730* −1.93
D2C_15 −2.665* −1.81 – – −3.092** −2.13
VEH_AC 3.494*** 7.95 4.633*** 5.8 2.916*** 4.84
AGE_1834 −0.836** −2.24 – – −1.151*** −2.96
STRNG&FRLES −2.070** −2.14 −1.034* −1.72 −2.223* −1.93
ENTH&CONF −1.035* −1.59 – – −1.222* −1.82
BFREQ_N 3.319*** 3.08 4.299** 2.04 – –
DARK 0.972* 1.62 – – 1.212* 1.81

Transit Constant −1.849** −2.18 −1.863* −1.58 −2.945** −2.55

ONCAMP −11.767*** −3.30 −11.975** −2.18 −7.759** −2.44
D2C_5 −4.971*** −2.68 −6.308** −2.13 −2.283* −1.62
D2C_15 −2.665* −1.81 1.327* 1.69 – –
TRANS_AC 3.395*** 4.98 3.743*** 3.8 3.381*** 3.22
STRNG&FRLES −2.148** −2.11 – – −3.279** −2.18
ENTH&CONF −1.035* −1.59 – – −1.222* −1.82
BFREQ_N 3.374*** 2.94 4.834** 2.1 – –
HIGHINC −0.558* −1.76 −0.870* −1.67 – –
UNDERGR −0.680** −1.97 −1.103** −2 – –
DARK 0.972* 1.62 – – 1.212* 1.81

Bicycle Constant −6.052*** −3.23 −4.564*** −3.23 −4.428*** −2.58

BIKE_AC 4.356*** 5.33 4.348*** 4.93 – –
ENCRGB 1.839*** 3.81 2.277*** 3.55 1.342** 2.27
OB_TT −2.562*** −2.73 −2.516*** −2.7 −1.399* −1.85
AGE_1824 −0.837* −1.58 −1.749*** −3.35 – –
BIKETREX 2.071*** 3.21 2.673*** 2.96 1.808** 2.46
UNDERGR −1.027** −2.00 – – −1.925*** −3.12
ROADFAM 0.737* 1.73 – – 2.972*** 3.79
OB_RTEFT −0.631* −1.68 – – −0.991** −1.98

Walk Constant −1.554* −0.98 −0.599* −0.6 −2.917* −1.87

OB_TT −2.998*** −2.93 −2.191** −2.17 −3.198*** −3.13
OB_TRIPCH −0.765* −1.81 – – – –
BFREQ_N – – – – 1.138* 1.78

Inclusive value parameters

Active 0.53*** 3.30 0.68*** 2.99 0.82*** 3.11
Non-active 0.52*** 3.70 0.38*** 2.53 0.82*** 3.09
McFadden Pseudo Adjusted R2 0.57 0.59 0.55
Sample size 698 349 349

*** ** *
Note: , , means significance at 1 percent, 5 percent, 10 percent level.

dimensions to an old issue. The focus is gender gap in choosing bicycle
as preferred mode in campuses and it is not aimed to study soon-to-be
launched bicycle share program in Baltimore city. However, that pro-
gram may remove some of obstacles introduced in this study for
Baltimore residents, and not just campus population. These barriers
include access or ownership issues and concerns associated with bicycle
theft. The key findings of this study are summarized as follows:
• Subjects with access to the bicycle and those willing to encourage
others to bike are contributing to the bicycle mode choice. This
impact is higher for male participants for both variables. In addition,
experience of bicycle on-board transit is associated with bicycle
transportation mode choice among both men and women. This
factor has less contribution toward female participants than their
male counterpart.
• Two obstacles are found to negatively impact walking transporta-
tion mode choice. These are: (1) the need to go to other destinations
and (2) travel time. For the travel time obstacle, the impact is higher
among the female participants. Similarly, travel time obstacle is
found to contribute to lower proclivity for bicycle mode choice
among both female and male participants. However, in contrast
with walking, travel time affects male participants in a greater scale.
• Two-level NL model result indicates that the impacts of several
variables are common between male and female participants when
it comes to the active transportation mode choice. As far as bicycling
mode choice is concerned, (1) travel time obstacles, (2) encouraging
others to bike, and (3) experience of bicycle on-board transit have
similar impacts (negative or positive) but with a different magni-
tude. As far as walking mode choice is concerned, the travel time
obstacle shakes both male and female intentions toward non-active
transportation mode choice.
• As far as non-active transportation mode choice is concerned, five
factors play an important role among male and female participants:
(1) Access to the transit and personal vehicle strongly sway both
male and female participants to choose non-active modes over the
active ones; (2) Living on campus or within 5 miles of campus are
averting factors regarding the non-active modes of transportation;
(3) These factors’ impacts are common between genders, however
they have higher weight on male participants than their female
counterparts; (4) It is less likely for participants with strong and
fearless bicycle skills to choose automobile transportation mode
over active modes; (5) Higher bicycling skill stimulates female
participation toward bicycling mode choice with higher extent than
male participants.
Moreover, some suggestions seem to be effective to reduce the
gender gap. The idea of minimizing bicycle mixture with vehicular
traffic considering financial, environmental, planning, and physical

84
F. Abasahl et al.
Fig. 3. Ratio of common coefficients for male and female.
constraints is supported. Another suggestion is to improve transit ser-
vices and hubs to accommodate the higher number of bicyclists.
Elevating public knowledge about local bicycle routes and facilities is
also proposed. Creating safe environments for bicyclists is a powerful
tool to encourage more women to bicycle.
One should note that this study was conducted on college campuses
where sample groups cannot fully represent people living in urban
areas, but there might be some similar patterns with larger commu-
nities which needs to be investigated in further studies. Some limita-
tions in this study result from gender composition of the surveyed
group. For example, male undergrads composed about 50% of re-
spondents as opposed to 38% female undergrads. Similarly, female staff
formed about 19% of respondents compared with 9% of male staff.
These differences may have influenced the age variable in the female
specific model. Further studies are needed to identify gender related
issues with higher focus on each campus with its spatial characteristics
and at a larger scale in communities. Future impact studies of in-
dividuals’ interdependencies, household data, space, time and costs on
gender gap are suggested.
Acknowledgement
The authors would like to thank Dr. Safieh Laaly, Maryam
Sheykholmolouki, and Seyedehsan Dadvar, research team at Morgan
State University for their efforts to conduct the survey and supports to
use data for this study. In addition, we would like to thank two anon-
ymous reviewers for their valuable comments.
References
Akar, G., Clifton, K., 2009. Influence of individual perceptions and bicycle infrastructure on
decision to bike. Transp. Res. Rec. J. Transp. Res. Board 2140, 165–172.
Akar, G., Fischer, N., Namgung, M., 2013. Bicycling choice and gender case study: the Ohio
State University. Int. J. Sustainable Transp. 7 (5), 347–365.
Baker, L., 2009. How to get more bicyclists on the road to boost Urban bicycling, figure out
what women want. Sci. Am. 301, 28–29.
Buehler, R., 2012. Capital Bikeshare Sudy; A Closer Look at Causal Users and Operations.
Virginia Tech. http://www.bikearlington.com/tasks/sites/bike/assets/File/vt-bike-share-
study-final3.pdf.
City-Data.com. 2016. Accessed March 25. http://www.city-data.com/city/Baltimore-Maryland.
html.
Coldren, G.M., Koppelman, F.S., 2005. Modeling the competition among air travel itinerary
85
Travel Behaviour and Society 11 (2018) 78–85
shares: GEV model development. Transp. Res. A: Policy Pract. 39 (4), 345–365.
Crane, R., 2007. Is there a quiet revolution in women’s travel? Revisiting the gender gap in
commuting. J. Am. Plan. Assoc. 73 (3), 298–316.
Ermagun, A., Samimi, A., 2015. Promoting active transportation modes in school trips. Transp.
Policy 37, 203–211.
Ermagun, A., Levinson, D., 2016. Intra-household bargaining for school trip accompaniment of
children: A group decision approach. Transp. Res. Part A: Policy Pract. 94, 222–234.
Ermagun, A., Levinson, D., 2017. Public transit, active travel, and the journey to school: a cross-
nested logit analysis. Transp. metrica A: Trans. Sci. 13 (1), 24–37.
Ermagun, A., Rashidi, T.H., Lari, Z.A., 2015. Mode choice for school trips: long-term planning
and impact of modal specification on policy assessments. Transport. Res. Record: J.
Transport. Res. Board 2513, 97–105.
Fishman, E., Washington, S., Haworth, N., Mazzei, A., 2014. Barriers to bikesharing: an analysis
from Melbourne and Brisbane. J. Transp. Geogr. 41, 325–337.
Garrard, J., Rose, G., Lo, S.K., 2008. Promoting transportation cycling for women: the role of
bicycle infrastructure. Prev. Med. 46 (1), 55–59.
Handy, S.L., 2011. The Davis bicycle studies: why do i bicycle but my neighbor doesn’t?
ACCESS Mag. 1 (39).
Hanson, S., 2010. Gender and mobility: new approaches for informing sustainability. Gender,
Place Culture 17 (1), 5–23.
Hensher, D.A., Rose, J.M., Greene, W.H., 2005. Applied Choice Analysis: A Primer. Cambridge
University Press.
Ji, Y., Fan, Y., Ermagun, A., Cao, X., Wang, W., Das, K., 2017. Public bicycle as a feeder mode to
rail transit in China: The role of gender, age, income, trip purpose, and bicycle theft ex-
perience. Int. J. Sustain. Transp. 11 (4), 308–317.
Kaufman, S.M., Gordon-Koven, L., Levenson, N., Moss, M.L., 2015. Citi Bike: The First Two
Years. The Rudin Center for Transportation Policy and Management. http://wagner.nyu.
edu/rudincenter/wp-content/uploads/2015/06/Citi_Bike_First_Two_Years_RudinCenter.
pdf.
Kelarestaghi, K.B., Ermagun, A., Heaslip, K., 2018. Cycling Usage and Frequency Determinants
in College Campuses. Transp. Res. Board 18–06565.
Krizek, K.J., Johnson, P.J., Tilahun, N., 2005. Gender differences in bicycling behavior and
facility preferences. Res. Women’s Issues Transp. 2, 31–40.
LDA Consulting, 2013. Capital Bikeshare Member Survey Report 2013, Washington DC.
LDA Consulting 2014 Capital Bikeshare Member Survey Report 2015, Washington DC.
Matsuda, F., Ikegami, T., Kishida, K., Kimotsuki, K., 2000. Effects of gender, age and experience
on bicycle riding behavior. In: Proceedings of the Human Factors and Ergonomics Society
Annual Meeting. SAGE Publications, pp. 417.
McDonald, N.C., 2012. Is there a gender gap in school travel? an examination of US children
and adolescents. J. Transp. Geogr. 20 (1), 80–86.
Miller, J., Handy, S., 2012. Factors that influence university employees to commute by bicycle.
Transp. Res. Rec. J. Transp. Res. Board 2314, 112–119.
Trapp, G.S., Giles-Corti, B., Christian, H.E., Bulsara, M., Timperio, A.F., McCormack, G.R.,
Villaneuva, K.P., 2011. On your bike! a cross-sectional study of the individual, social and
environmental correlates of cycling to school. Int. J. Behav. Nutr. Phys. Act. 8 (1), 123.
Twaddle, H., Hall, F., Bracic, B., 2010. Latent bicycle commuting demand and effects of gender
on commuter cycling and accident rates. Transp. Res. Rec. J. Transp. Res. Board 2190,
28–36.
Wang, C.H., Akar, G., Guldmann, J.M., 2015. Do your neighbors affect your bicycling choice? a
spatial probit model for bicycling to the Ohio State University. J. Transp. Geogr. 42,
122–130.
Zhao, J., Wang, J., Deng, W., 2015. Exploring bikesharing travel time and trip chain by gender
and day of the week. Transp. Res. Part C: Emerging Technol. 58, 251–264.