Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

International Journal of Sustainable Transportation

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/ujst20

Gender disparities in New York City bike share


usage

Kathleen H. Reilly , Shu Meir Wang & Aldo Crossa

To cite this article: Kathleen H. Reilly , Shu Meir Wang & Aldo Crossa (2020): Gender disparities
in New York City bike share usage, International Journal of Sustainable Transportation, DOI:
10.1080/15568318.2020.1861393

To link to this article: https://doi.org/10.1080/15568318.2020.1861393

Published online: 21 Dec 2020.

Submit your article to this journal

Article views: 49

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=ujst20
INTERNATIONAL JOURNAL OF SUSTAINABLE TRANSPORTATION
https://doi.org/10.1080/15568318.2020.1861393

Gender disparities in New York City bike share usage


Kathleen H. Reilly, Shu Meir Wang, and Aldo Crossa
Bureau of Epidemiology Services, New York City Department of Health and Mental Hygiene, Long Island City, NY, USA

ABSTRACT ARTICLE HISTORY


By understanding gender disparities in bike share usage in New York City, programs can be better Received 23 December 2019
tailored to increase cycling and bike share usage among female members. Data from bike trips Revised 14 September 2020
and a Citi Bike enrollment survey for the period 2013–2018 were used for this analysis. Accepted 28 November 2020
Associations comparing female to male members on selected study variables were examined
KEYWORDS
through the estimation of odds ratios (OR) and 95% confidence intervals (CI) using bivariate logis- Bicycling; bike share;
tic regression models and Wilcoxon signed-rank tests for continuous non-normal data. Spatial gender; physical activity
autocorrelation of bike share station pick ups and drop offs by percentage of female bike share
members was detected using local Moran’s I. This study included 226,237 Citi Bike members; of
these, over one-third (38.1%) self-identified as female. The optional enrollment survey was com-
pleted by 33,945 members; of these, 37.9% self-identified as female. Compared to male gender,
female gender was associated with younger age, higher levels of education completed, being a
student and not employed, and lower household income, as well as social and health reasons for
membership rather than utilitarian reasons. Overall, female members took fewer bike share trips
(median: 46.0 per year vs. 78.5 for males). There was spatial correlation between station usage and
gender, with female members more likely than male members to use Citi Bikes in less dense
neighborhoods. The results from this study highlight the gender disparity in bike share member-
ship and usage in NYC and provide insight into how this gap could be reduced.

Introduction York City Department of Consumer Affairs, 2015). This


expression has also been applied to transit in NYC, where
Riding a bike provides individual benefits such as positive
women often pay more to use private cars and cars-for-hire
health outcomes and an efficient and low-cost means of
rather than use public transit due to caregiver responsibil-
urban transportation (Heinen et al., 2010; Oja et al., 2011;
ities, fear of harassment, and safety concerns (Kaufman
Wilson, 1973) as well as population benefits by reducing
et al., 2018). In much the same way, a “pink tax” could be
pollution and traffic congestion (Saelensminde, 2004). With
thought to apply to cycling since women do not participate
the exception of countries like the Netherlands, Denmark,
at the same rates as men and are, therefore, excluded from
and Germany, the cycling gender disparity persists world-
wide (Pucher & Buehler, 2008). In the US, men are approxi- such advantages as saving money on transportation and
mately four times as likely to bike for transportation than health benefits.
women (Nehme et al., 2016). Population-level surveys of Bike sharing programs have the potential to narrow gen-
adults and public high school students in New York City der disparities in cycling, as the proportion of female bike
(NYC) have found that men of both groups were more share members has been higher than among general cyclists
likely to cycle once a month or more compared with women in several cities (Aldred et al., 2017; Fishman, 2016).
(Crossa et al., 2016). The League of American Bicyclists Reasons for this finding have been speculated, but are not
claims that the “five Cs” (comfort, convenience, confidence, well studied in the current literature (Goodyear, 2013).
consumer products, and community) influence women’s Moreover, little is known about differences in characteristics
decision to bike (Szczepanski, 2013). Evidence indicates that between male and female bike share members, including in
women prefer to ride in protected bike lanes and many NYC. By better understanding gender disparities in bike
avoid cycling due to fear of traffic. Furthermore, women share usage, programs can be tailored to increase cycling
have been considered an “indicator species” for strong bike and bike share usage among women.
infrastructure (Aldred et al., 2017; Baker, 2009; Fowler et al.,
2017; Mitra & Nash, 2019).
Materials and methods
The phrase “pink tax” has been applied to items for
which a female version of a product costs more than the Launched in 2013, New York City’s “Citi Bike” bike share
male version, including toiletries, clothing, and toys (New system is the largest of its kind in the United States. Trip

CONTACT Kathleen H. Reilly kreilly3@health.nyc.gov Bureau of Epidemiology Services, New York City Department of Health and Mental Hygiene, 4209
28th St, CN 07-103, Long Island City, NY 11101.
ß 2020 City of New York
2 K. H. REILLY ET AL.

Figure 1. NYC bike share and social–ecological model for impacts on human behavior.

utilization data are collected electronically when members active” or “somewhat active” were considered physically
unlock and return bikes to docking stations. Citi Bike col- active, whereas those who reported that they were “not very
lects limited demographic information on its members (gen- active” or “not active at all” were considered not physically
der [male, female, other, or refuse to answer], birth year, active. Survey respondents were also asked, “Would you say
and billing ZIP code). Additional data came from an in general your health is excellent, very good, good, fair, or
optional online enrollment survey of people signing up for poor?” Those who reported that their health was “excellent,”
an annual Citi Bike subscription for the first time. The sur- “very good,” or “good” were considered to be in good
vey, designed by the New York City Department of Health health, whereas those who reported that their health was
and Mental Hygiene (NYC DOHMH), includes questions on “fair” or “poor” were considered to not be in good health.
race/ethnicity, highest level of education completed, employ- Trip utilization data were aggregated to total trip counts for
ment status, household income, private bicycle ownership, each member and created into rate variables for trips per
private car ownership, self-reported physical activity status, year. Trips were excluded if they started and ended in the
self-reported health status, and reasons for membership. same station and had trip durations less than two minutes
Unique membership ID numbers were used to link survey since these could reflect issues unlocking bikes rather than
data to membership and trip utilization data. This study actual trips. Peak season was defined as May through
considered a socio-ecological framework for how aspects of October and rush hour was considered 7–10 a.m. and 4–7
bike share at different levels influence behavior (Figure 1). p.m. during weekdays.
Data were restricted to those who reported either male or
female gender and further limited to Citi Bike members
Statistical analysis
with a billing ZIP code in NYC who had a subscription last-
ing at least 365 days in the period from May 20, 2013 (the The frequencies and percentages for each level of categorical
date of Citi Bike launch) to December 31, 2018. Yearly sub- variables were calculated, stratified by male and female
scriptions with breaks greater than 10 days were right cen- members. Differences in survey response were examined by
sored in order to ensure that only continuous usage data gender, residence in a Zip Code Tabulation Area (ZCTA)
were examined. Analyses of trip data included all members with a Citi Bike station (Chi-square tests), and age
meeting the criteria above; some analyses were further (Wilcoxon signed-rank test), since these are the variables
restricted to members who completed the enrollment survey. that Citi Bike collects from its members. Associations com-
Due to a technical error following migration to a new survey paring female to male members on selected study variables
platform, approximately 20% of surveys could not be linked were examined through Chi-square tests for categorical vari-
to membership IDs. ables and Wilcoxon signed-rank tests for continuous non-
normal data. For categorical variables with more than two
levels, Chi-square post hoc multiple comparison tests were
Measures
conducted with Bonferroni adjustment to determine the dif-
Dichotomous variables were created for self-reported phys- ferences between levels that were significant. Multiplicative
ical activity and overall health. Survey respondents were interaction between year of Citi Bike service and gender was
asked, “In general, how physically active are you? Would examined using Poisson regression to determine if trip
you say very active, somewhat active, not very active, or not counts varied by gender over time. A multivariable model
active at all?” Those who reported that they were “very was created to determine if the significance of the
INTERNATIONAL JOURNAL OF SUSTAINABLE TRANSPORTATION 3

interaction between year and gender persisted after control- median: 32 years of age [IQR: 27, 41], respectively; Wilcoxon
ling for potential confounders among those who completed p < 0.0001). Overall the, age and trip counts varied between
the enrollment survey. Variables that were significantly asso- those who reported binary gender and those who reported
ciated with trip counts while controlling for year were con- other (Wilcoxon p < 0.0001 for both), where those who
sidered for inclusion in the multivariable model. The final reported “other” gender were younger (median 32 (IQR: 25,
multivariable model was selected through backward selection 41)) vs. those reporting binary gender (median 32 (IQR: 27,
of the full model including the gender and year interaction 42). Those reporting “other” also made fewer trips (median:
term using Bayesian Information Criterion. The percentage 74.5 (IQR: 14, 215) vs. median: 99.0 (IQR: 30, 278).
of female members for bike share station pick ups and drop There were 12,865 female and 21,080 male members who
offs was mapped with Voronoi polygons around each bike completed the enrollment survey; none of survey respond-
share station. Spatial autocorrelation for the percentage of ents reported “other” gender or refused to answer. Table 1
female members for bike share station pick ups and drop shows the differences in characteristics between female and
offs was tested using Moran’s I and p values were estimated male bike share members at the time of survey completion.
using Monte Carlo tests with 999 simulations. Additionally, The proportion of female members that were Latina was sig-
the relationship between traffic volume (annual million nificantly lower than the proportion of men that were
vehicle miles traveled per km2 per year [2016] from the Latino (8.5% vs. 9.8%). A larger proportion of female mem-
New York Metropolitan Transportation Council Best bers were in the <30 years and >45 years age groups com-
Practice Model) and percentage of station usage by female pared with male members; this difference was significant for
bike share members at the Community District level the comparison of all age group categories with the excep-
(approximations of the U.S. Census Public Use Microdata tion that there was no difference in the proportion of female
Area [PUMA] geographies) was examined. The spatial cor- members between the 30–35 years of age and >45 years of
relation structure was determined by fitting a variogram of age categories. Greater proportions of female members had
the null model with percentage of station usage by female completed higher levels of education compared with male
bike share members as the outcome. A likelihood ratio (LR) members. The comparison of less than college and some
test was used to determine improvement of the non-linear college was not significantly different between genders; how-
mixed effects model incorporating spatial correlation over ever, significant differences were found between all other
the null model. The association between traffic volume and levels. Proportions of female cyclists who were students
percentage of station usage by female bike share members (6.0% vs. 4.9%) or not employed (4.3% vs. 2.8%) were
was measured using Pearson’s correlation coefficient. higher compared with those who were employed (89.7% vs.
Analyses were conducted using SAS Enterprise Guide 7.1 92.3%); there was no difference in the proportion of female
(Cary, NC, USA) and the sp, rgdal, rgeos, nmle, dismo, members who were students compared with not employed.
dplyr, ggplot2, and spdep packages in R 3.5.2 (Vienna, Female members had greater proportions with lower house-
Austria). Statistical significance was determined if the two- hold income, however, differences in gender did not vary
sided p value was less than 0.05. significantly within the lowest household income brackets
(<$25,000 vs. $25,000–$49,000 vs. $50,000–$99,999).
Female members were less likely to own a working
Ethics
bicycle (46.2% vs. 50.4%) or a car (35.7% vs. 41.2%) and
This study was reviewed by the New York City Department were more likely to report being physically active (88.3% vs.
of Health and Mental Hygiene’s Institutional Review Board 88.0%), in good health (95.4% vs. 93.9%), and to request
and was deemed to be a public health program evaluation safety tips from Citi Bike (17.1% vs. 10.1%). There were also
that is not research. differences by gender for their primary reason for subscrib-
ing to Citi Bike, with a higher proportion of female mem-
bers more likely to cite friends and family as motivation to
Results
join (3.3% vs. 1.5%); enjoy riding a bike (12.6% vs. 8.9%);
Citi Bike data for the study period contained 533,874 sub- personal health and fitness (11.2% vs. 8.5%); to reduce pol-
scriptions; of these, 486,481 (91.1%) did not have a break of lution and help the environment (3.7% vs. 3.2%); and as an
more than 10 days in membership. There were 253,313 alternative to public transportation (15.0% vs. 13.3%).
(52.1%) unique members with 252,473 (99.7%) having a Female members were less likely to report their primary rea-
subscription for at least 365 days; of these, 227,284 (90.0%) son for membership as a faster way to get around town
had a billing ZIP code in a NYC ZCTA. Only 41,471 mem- (19.2% vs. 23.2%) and to save money on transportation
bers (18.2%) answered the survey and 33,955 (81.9%) could (6.1% vs. 6.8%).
be linked to other membership data. There was no differ- During the study time period, there were 140,059 male
ence in survey response by gender (female members: 14.9% members, 86,178 female members, and 1100 (<1%) mem-
vs. male members: 15.0%, p ¼ 0.45), but survey respondents bers who reported “other” gender. Table 2 shows the differ-
were more likely to live in a ZCTA with a Citi Bike station ences in trip utilization between female and male bike share
(15.6% vs. 11.8% for those who did not live in a ZCTA with members. Overall, female bike share members used Citi Bike
a Citi Bike station, p < 0.0001) and were somewhat older less than male bike share members, in terms of duration of
than non-respondents (median: 34 years [IQR: 28, 45] vs. membership (median 1.0 years [IQR: 1.0, 2.0] vs. median 1.3
4 K. H. REILLY ET AL.

Table 1. Characteristics of Citi Bike members in NYC and associations with female gender (2013–2018).
Female Male
n (%) n (%)
Variable n ¼ 12,865 n ¼ 21,080 p Value
Race/Ethnicity (totals) n ¼ 8304 n ¼ 12,021 0.004
White 6078 (73.2%) 8749 (72.8%)
Asian and Pacific Islander 1089 (13.1%) 1489 (12.4%)
Hispanic or Latino 705 (8.5%) 1182 (9.8%)
Black 277 (3.3%) 353 (2.9%)
Other 154 (1.9%) 248 (2.1%)
Age (years) (totals) n ¼ 12,811 n ¼ 21,008 <0.0001
<30 4245 (33.1%) 6100 (29.0%)
30–35 2854 (22.3%) 4829 (23.0%)
36–45 2604 (20.3%) 5084 (24.2%)
>45 3108 (24.3%) 4995 (23.8%)
Highest level of education completed n ¼ 8738 n ¼ 12,701 <0.0001
Less than college 180 (2.1%) 429 (3.4%)
Some college 562 (6.4%) 1134 (8.9%)
College 3733 (42.7%) 5582 (43.9%)
Graduate school 4263 (48.8%) 5556 (43.7%)
Employment status n ¼ 8544 n ¼ 12534 <0.0001
Employed 7664 (89.7%) 11,573 (92.3%)
Student 514 (6.0%) 610 (4.9%)
Not employed 366 (4.3%) 351 (2.8%)
Household income n ¼ 6962 n ¼ 10,343 <0.0001
<$25,000 356 (5.1%) 383 (3.7%)
$25,000–$49,999 546 (7.8%) 662 (6.4%)
$50,000–$99,999 1939 (27.9%) 2148 (20.8%)
$1,00,000–$199,999 2388 (34.3%) 3576 (34.6%)
>$200,000 1733 (24.9%) 3573 (34.5%)
Owns a working bike n ¼ 8784 n ¼ 12,840 <0.0001
4059 (46.2%) 6473 (50.4%)
Did not ride a bike in NYC in 12 months prior to membership n ¼ 9415 n ¼ 13,993 0.01
2593 (27.5%) 3649 (26.1%)
Owns a car n ¼ 8855 n ¼ 12,964 <0.0001
3162 (35.7%) 5342 (41.2%)
Self-reported physically active n ¼ 9571 n ¼ 14,147 0.05
8503 (88.3%) 12,448 (88.0%)
Self-reported good health n ¼ 9596 n ¼ 14,174 <0.0001
9152 (95.4%) 13,308 (93.9%)
Requested safety tips from Citi Bike n ¼ 12,865 n ¼ 21,080 <0.0001
2206 (17.1%) 2130 (10.1%)
Primary reason for membership n ¼ 12,817 n ¼ 20,965
Faster way to get around town 2456 (19.2%) 4857 (23.2%) <0.0001
Alternative to public transportation 1919 (15.0%) 2797 (13.3%) <0.0001
Enjoy riding a bike 1609 (12.6%) 1835 (8.9%) <0.0001
Personal health and fitness 1438 (11.2%) 1772 (8.5%) <0.0001
Save money on transportation 781 (6.1%) 1419 (6.8%) 0.01
Reduce pollution help environment 479 (3.7%) 663 (3.2%) 0.005
Friends and family joining 420 (3.3%) 305 (1.5%) <0.0001
Support the program 291 (2.3%) 513 (2.4%) 0.30
Better than riding own bike 368 (2.9%) 543 (2.6%) 0.12
Do not drive a car 104 (0.8%) 213 (1.0%) 0.06
Lives in a NYC ZCTA with a Citi Bike station n ¼ 12,865 n ¼ 21,080 <0.0001
11,168 (86.8%) 17,744 (84.2%)

[IQR: 1.0, 2.9], Wilcoxon p < 0.0001), number of trips per month, and day of the week are shown in Figure 2(a, b, c,
year (median: 46.0 [IQR: 15.0, 117.0] vs. median: 78.5 [IQR: and d), respectively. The proportions of trips by hour of day
27.8, 187.8], Wilcoxon p < 0.0001), percentage of rush hour and month followed similar patterns for female and male
trips (median: 28.8% [IQR: 16.7% vs. 44.6%] vs. median: members. However, male members had greater proportions
30.3% [IQR: 18.9%, 46.2%], Wilcoxon p < 0.0001), and per- of trips in the late evening and early morning and greater
centage of peak season trips (median: 77.3% [IQR: 62.9%, proportions of trips in the winter. Female members had a
91.7%] vs. median: 72.4% [IQR: 60.0%, 86.1%], Wilcoxon greater proportion of trips on weekends compared with
p < 0.0001). However, the percentage of weekend and holi- male members.
day trips was higher among female members (median: Among those who completed the enrollment survey, the
30.0% [IQR: 17.3%, 47.4%] vs. median: 25.9% [IQR: 14.6%, number of trips increased for each year of Citi Bike service
39.3%], Wilcoxon p < 0.0001). Female bike share members (b ¼ 0.05, S.E. ¼ 0.0002, p < 0.0001); the coefficient compar-
also had longer trips (median: 12.4 minutes [IQR: 9.3, 17.2] ing female to male members was negative (b ¼ 0.45,
vs. median: 10.0 [IQR: 7.5, 13.8], both Wilcoxon, S.E. ¼ 0.0002, p < 0.0001), however the coefficient for the
p < 0.0001). The proportion of trips stratified by gender for interaction term for female gender and year of Citi Bike ser-
weekday hour of day, weekend and holiday hour of day, vice was positive (b ¼ 0.01, S.E. ¼ 0.0004, p < 0.0001).
INTERNATIONAL JOURNAL OF SUSTAINABLE TRANSPORTATION 5

Table 2. Characteristics of Citi Bike trips in NYC by gender (2013–2018).


Female Male
Median (IQR) Median (IQR)
Variable n ¼ 86,178 n ¼ 140,059 Wilcoxon p Value
Duration of membership (years) 1.0 (1.0, 2.0) 1.3 (1.0, 2.9) <0.0001
Number of trips per year 46.0 (15.0, 117.0) 78.5 (27.8, 187.8) <0.0001
Percentage of trips during rush hour 28.8% (16.7%, 44.6%) 30.3% (18.9%, 46.2%) <0.0001
Percentage of trips on weekends and holidaysa 30.0% (17.3%, 47.4%) 25.9% (14.6%, 39.3%) <0.0001
Percentage of trips during peak seasonb 77.3% (62.9%, 91.7%) 72.4% (60.0%, 86.1%) <0.0001
Median duration of ride (minutes) 12.4 (9.3, 17.2) 10.0 (7.5, 13.8) <0.0001
a
Holidays include New Year’s Eve, New Year’s Day, Martin Luther King Jr. Day, Memorial Day, Independence Day, Labor Day, Columbus Day,
Veterans Day, Thanksgiving, the day after Thanksgiving, Christmas Eve, and Christmas Day.
b
Peak season: May–October.

Figure 2. (a) Proportion of trips on weekdays by hour of the day and gender; (b) proportion of trips on weekends and holidays by hour of the day and gender; (c)
proportion of trips by month and gender; (d) proportion of trips by day of the week and gender.

Table 3 shows the results of the multivariable analysis. After users was similar between pick-up and drop off stations,
controlling for other study variables, the number of trips indicating that origin and destination stations did not vary
increased for each year of service (b ¼ 0.05, S.E. ¼ 0.0003, by gender. The percentage of pick ups and drop offs by
p < 0.0001); the coefficient comparing females to males female members for all stations was below 50%; however,
remained negative (b ¼ 0.51, S.E. ¼ 0.002, p < 0.0001), the neighborhoods with clusters of higher usage by female
however the coefficient for the interaction term for female members included parts of Brooklyn and the Upper West
gender and year of Citi Bike service was positive (b ¼ 0.01, Side, Manhattan. Female usage was lowest in the Financial
S.E. ¼ 0.0006, p < 0.0001). District and Midtown in Manhattan and parts of Queens.
Figure 3(a and b) shows the percentage of female mem- The percentages of pick ups and drops offs by female mem-
bers for bike share station pick ups and drop offs, respect- bers were geographically correlated (Moran’s I 0.62
ively. The spatial distribution of the percentage of female [p ¼ 0.001] and 0.58 [p ¼ 0.001], respectively). In examining
6 K. H. REILLY ET AL.

Table 3. Coefficient and standard errors (SE) for the relationship between trip
counts and the interaction between gender and year for those who completed
the NYC Citi Bike enrollment survey (2013–2018) after controlling for other
study variables in a multivariable Poisson regression model.
Variable Coefficient (SE)
Female gender 0.51 (0.002)a
Year 0.05 (0.0003)a
Female gendera Year 0.01 (0.0006)a
Race/ethnicity
White 1.0
Asian and Pacific Islander 0.16 (0.001)a
Hispanic or Latino 0.15 (0.002)a
Black 0.15 (0.003)a
Other 0.06 (0.003)a
Age (years)
<30 0.31 (0.001)a
30–35 0.32 (0.001)a
36–45 0.19 (0.001)a
>45 1.0
Highest level of education completed
Less than college 0.15 (0.003)a
Some college 0.12 (0.002)a
College 1.0
Graduate school 0.02 (0.001)a
Employment status
Employed 1.0
Student 0.06 (0.003)a
Not employed 0.18 (0.003)a
Household income
<$25,000 0.69 (0.002)a
$25,000–$49,999 0.46 (0.002)a
$50,000–$99,999 0.39 (0.001)a
$100,000–$199,999 0.22 (0.001)a
>$200,000 1.0
Owns a working bike 0.02 (0.001)a
Did not ride a bike in NYC in 12 months prior to membership 0.03 (0.001)a
Owns a car 0.16 (0.001)a
Self-reported physically active 0.35 (0.002)a
Self-reported good health 0.16 (0.003)a
Requested safety tips from Citi Bike 0.03 (0.001)a
Primary reason for membership
Faster way to get around town 0.07 (0.002)a
Alternative to public transportation 0.02 (0.002)a
Enjoy riding a bike 0.01 (0.002)a
Personal health and fitness 0.14 (0.002)a
Save money on transportation 0.32 (0.002)
Reduce pollution help environment N/A
Friends and family joining 0.19 (0.004)a
Support the program 0.04 (0.003)a
Better than riding own bike N/A
Do not drive a car 0.11 (0.004)a
Lives in a NYC ZCTA with a Citi Bike station 0.52 (0.002)a
a
p < 0.0001.

the relationship between traffic volume and percentage of sta-


tion usage by female bike share members at the Community
District level, the Gaussian spatial correlation structure best fit
the data. A log LR test determined that the non-linear mixed
effects model accounting for spatial correlation did not
improve the fit of the data over the model that did not
account for spatial correlation (LR: 7.5  1011, df ¼ 1,
p > 0.99), so spatial correlation was not considered in this
analysis. The correlation was Pearson r ¼ 0.35, indicating a
modest inverse relationship between traffic volume and use of
bike share stations by female bike share members.

Figure 3. (a) Percentage of bike share station pick ups by female members. (b)
Discussion Percentage of bike share station drop offs by female members.

This study highlights the disparities in bike share usage


between female and male members in NYC and elucidates
INTERNATIONAL JOURNAL OF SUSTAINABLE TRANSPORTATION 7

characteristics associated with them. Among NYC bike share Socioeconomic factors were also associated with gender
users, female gender was associated with demographic and among bike share members. Previous studies show that
socioeconomic variables, as well as social and health reasons women had greater representation in bike share programs
for membership rather than utilitarian reasons. with increasing levels of education completed. For example,
Almost twice the proportion of females requested safety a qualitative study in London found that cycling among
tips from Citi Bike compared with male members. Research women represented a “bourgeois sensibility” in parallel with
has found that safety is a major factor in determining other middle class lifestyle attributes (Steinbach et al., 2011).
whether female members will ride a bicycle and female However, the current study found that female members had
members have a greater preference for routes that are sepa- lower incomes compared with men, which could reflect the
rated from motor vehicle traffic (Aldred et al., 2017; Emond underlying gender gap in salaries between men and women
et al., 2009; Heesch et al., 2012). The spatial correlation of (Bureau of Labor Statistics, 2019). It follows that female
station usage by gender demonstrates a significant difference members were more likely to be students or unemployed
in travel patterns between male and female members. As than male members, whereas the proportion of men who
demonstrated in the current study’s correlation analysis, the were employed full-time was higher, which may also reflect
areas in NYC where bike share station usage was lower underlying population demographics, as there are more men
among female compared with male members (such as in the labor force compared to women (U.S. Census
Midtown Manhattan) correspond with high traffic neighbor- Bureau). Similarly, female members were less likely to own a
hoods, which women may not perceive as safe. The NYC bike or car, which may also reflect differences in gender in
Department of Transportation (DOT) considers improving ownership of these forms of transportation. The authors
bicycle infrastructure in Midtown Manhattan a “gender were not able to find any data on gender differences in car
imperative” (Kuntzman, 2019). NYC has actively expanded ownership in NYC, however, according to the results of the
bike infrastructure throughout the city with an added 308 2016 NYC Community Health Survey, 52.7% of men in
miles of conventional (Class II and Class III) bike lanes and NYC drove a car in the past 30 days compared with 30.1%
74 miles of protected (Class I) bike lanes from 2006 to 2017 of women (New York City Department of Health and
Mental Hygiene). Self-perceived health and fitness may also
(Getman et al., 2017). Despite these improvements to the
influence women’s decision to become bike share members
cycling infrastructure, lack of traffic enforcement can dimin-
more than for men; women who do not perceive themselves
ish the effectiveness of these lanes (Basch et al., 2019;
to be physically active or in good physical health may not
Conway et al., 2013). Improving cycling infrastructure with
sign up for a bike share program if they perceive that they
protected bike paths may reduce disparities in usage by gen-
are not fit enough to ride a bike (Steinbach et al., 2011).
der and promote bicycling in NYC to levels similar to the
Female members were also more likely to live in a ZCTA
Netherlands, Denmark, and Germany (Pucher &
with a Citi Bike station, which may reflect that convenience
Buehler, 2008).
is a more influential motivator for membership
The lower proportion of female to male Hispanic or
among women.
Latino members might represent a cultural difference in
Although the top reasons for getting a Citi Bike member-
cycling by gender among particular racial/ethnic groups, ship were the same for female and male members, women
such as a greater preference to bike socially with family and were more likely to get a membership because of social and
friends (Lusk et al., 2017). Bike share programs may inhibit personal improvement reasons such as friends and family
social riding if there are not enough bikes or docking spaces joining, because they enjoy riding a bike, for health and fit-
available from desired stations, if friends or family do not ness, and for environmental concerns. Men were more likely
also have memberships, or if bikes are not compatible with to cite their primary reason for membership as more utili-
transporting young children; therefore, if Latinas are more tarian, such as a faster way to get around the city and to
likely to ride bikes for social reasons, this could be why bike save money on transportation. Heesch et al. found similar
share usage is lower among this group (People for Bikes, gender differences in cycling with women, citing more per-
2017). Additional research is needed to better understand sonal, social, and environmental reasons; however, unlike
why bike share usage is lower among Latinas in NYC. the current study, women were more likely to cite cycling as
Similar to the findings from a study of gender differences an inexpensive transportation option as a primary motiv-
in cycling in Queensland, Australia (Heesch et al., 2012), the ator (2012).
current study found a greater proportion of younger mem- Male members had a longer duration of membership,
bers among females. This age difference may be reflective of greater number of trips per year, and greater percentage of
a generational shift; an analysis of U.S. national data from rush hour trips. These disparities in usage have been found
2015 found there was no gender gap in cycling for millenni- in other countries, including the UK, Australia, and China
als (Dill, 2017). Another potential explanation for this gener- (Guo et al., 2017; Heesch et al., 2012; Ogilvie & Goodman,
ational gap in bike share usage may be based on changes in 2012). However, female bike share members had greater per-
life events, such as having children, which may explain why centages of weekend and holiday trips and peak season trips,
the proportion of female members in the current study was which parallels the gendered usage of bike share usage in
lower in the 36–45 age group compared with those older London (Beecham & Wood, 2014). In addition, female
than 45 (Bonham & Wilson, 2012). members had longer ride durations. This difference in ride
8 K. H. REILLY ET AL.

duration might reflect women preferring leisurely rides or standards of dress and hygiene, experiencing street harass-
riding at a perceived safer speed (Wheeler et al., 2010). ment, and having caregiver and other household responsibil-
Childcare responsibilities during rush hour might be a rea- ities, may also contribute to gender disparities in bike share
son that female members are less likely to ride during usage (Garrard et al., 2006; Steinbach et al., 2011; Stop
these times. Street Harassment, 2014).
The total number of Citi Bike trips increased over time The results from this study highlight gender disparities in
for both male and female members, and although female bike share membership in NYC and provide insight into
members consistently made fewer trips than male members how these gaps could be reduced. In November 2019, the
each year, the interaction between female gender and year NYC Department of Transportation and Lyft (operator of
was positive indicating that women are increasingly taking Citi Bike) formed an Equity Advisory Board for Citi Bike
more trips each year compared with men. As the Citi Bike consisting of 20 members from community organizations
coverage area continues to expand, it is expected that mem- and city government. This board explores Citi Bike’s equity
bership will increase and that the proportion of trips made strategy as the program expands throughout the city (“NYC
by female members will also increase over time. However, a DOT and Lyft Announce New Citi Bike Equity Advisory
study in the UK found that overall increases in cycling did Board to Ensure Inclusive and Equitable Bike Share
not correlate with a decrease in the gender gap, indicating a Expansion to Additional NYC Neighborhoods,” 2019).
need for a targeted approach to increase cycling among Future research should examine if other factors such as care-
women (Aldred et al., 2016). giver responsibilities, household errands, standards of dress
This is the first study to examine gender differences in and hygiene, and street harassment differentially impact bike
bike share members in North America’s largest bike share share usage by gender, and if so, how these factors could be
system, however, this study was subject to several limita- better accommodated through bike share programs. Safety
tions. The optional member survey at enrollment had a low concerns related to infrastructure may also play a part in
response rate (<20%) and those who responded to the sur- determining where women decide to bike in NYC.
vey might have differed in study characteristics compared Considering infrastructure improvements in areas with low
with those who did not complete the survey. Although Citi rates of station usage by female members may also be
Bike enrollment information contains an option to report a instrumental in closing the cycling gender gap in NYC.
gender outside the binary, all survey respondents reported
male or female gender, so we were unable to conduct
detailed analyses about utilization for members who self- Acknowledgements
identified as “other” gender. Better collection of sex and The authors acknowledge Sungwoo Lim, Hannah Gould, Charon
gender identity data would help to elucidate utilization Gwynn, James Hadler, and Hannah Helmy for reviewing previous
among all members. Furthermore, Citi Bike members do drafts of this article.
not necessarily represent all cyclists or the general popula-
tion (Crossa et al., 2019); the gender differences found in
Declaration of interest statement
the current study may not be generalizable to other popula-
tions. The analyses conducted were descriptive and, aside No potential conflict of interest was reported by the authors. The
from the analysis examining factors associated with trip authors did not receive any specific funding for this work.
counts, did not adjust for potential confounders.
Furthermore, correlated variables do not imply a causal rela- References
tionship. Survey questions were based on self-report and,
therefore, may be subject to social desirability bias, which Aldred, R., Elliott, B., Woodcock, J., & Goodman, A. (2017). Cycling
provision separated from motor traffic: A systematic review
might vary by gender (Hebert et al., 1997). Billing ZIP code
exploring whether stated preferences vary by gender and age.
was used as a proxy for residential ZIP code, but this may Transport Reviews, 37(1), 29–55. https://doi.org/10.1080/01441647.
not have been accurate for all members. In addition, ZIP 2016.1200156
code was only available for the time of enrollment and did Aldred, R., Woodcock, J., & Goodman, A. (2016). Does more cycling
not account for changes in residence. Geographic analyses mean more diversity in cycling? Transport Reviews, 36(1), 28–44.
were limited to station locations; if route data had been https://doi.org/10.1080/01441647.2015.1014451
available, these data would have provided a more nuanced Baker, L. (2009). How to get more bicyclists on the road. Scientific
American, 301(4), 28–29. https://doi.org/10.1038/scientificameri-
insight into road usage differences between men and can1009-28
women. Also, the analysis examining the relationship Basch, C. H., Ethan, D., & Basch, C. E. (2019). Bike lane obstructions
between traffic volume and percentage of station usage by in Manhattan, New York City: Implications for bicyclist safety.
female bike share members were based on 21 Community Journal of Community Health, 44(2), 396–399. https://doi.org/10.
Districts; the availability of traffic data at a smaller geo- 1007/s10900-018-00596-4
graphical level would have produced a more precise result. Beecham, R., & Wood, J. (2014). Exploring gendered cycling behav-
iours within a large-scale behavioural data-set. Transportation
The current study was a secondary analysis of data collected
Planning and Technology, 37(1), 83–97. https://doi.org/10.1080/
for a general public health survey and not specifically to 03081060.2013.844903
examine gender disparities. Other gender-related factors that Bonham, J., & Wilson, A. (2012). Bicycling and the life course: The
were not captured in this study, including maintaining start-stop-start experiences of women cycling. International Journal
INTERNATIONAL JOURNAL OF SUSTAINABLE TRANSPORTATION 9

of Sustainable Transportation, 6(4), 195–213. https://doi.org/10.1080/ Lusk, A. C., Anastasio, A., Shaffer, N., Wu, J., & Li, Y. (2017). Biking
15568318.2011.585219 practices and preferences in a lower income, primarily minority
Bureau of Labor Statistics. (2019, December 16). Women’s earnings in neighborhood: Learning what residents want. Preventive Medicine
New York – 2018. Retrieved from https://www.bls.gov/regions/new- Reports, 7, 232–238. https://doi.org/10.1016/j.pmedr.2017.01.006
york-new-jersey/news-release/womensearnings_newyork.htm#chart2 Mitra, R., & Nash, S. (2019). Can the built environment explain gender
Conway, A., Cheng, J., Peters, D., & Lownes, N. (2013). Characteristics gap in cycling? An exploration of university students’ travel behavior
of multimodal conflicts in urban on-street bicycle lanes. in Toronto, Canada. International Journal of Sustainable
Transportation Research Record: Journal of the Transportation Transportation, 13(2), 138–147. https://doi.org/10.1080/15568318.
Research Board, 2387(1), 93–101. https://doi.org/10.3141/2387-11 2018.1449919
Crossa, A., Noyes, P., Parton, H., Lim, S. (2016, October). Cycling in Nehme, E. K., Perez, A., Ranjit, N., Amick, B. C., & Kohl, H. W.
New York City, 2007 to 2014. New York City Department of Health (2016). Sociodemographic factors, population density, and bicycling
and Mental Hygiene: Epi Data Brief (78). Retrieved from https:// for transportation in the United States. Journal of Physical Activity
www1.nyc.gov/assets/doh/downloads/pdf/epi/databrief78.pdf and Health, 13(1), 36–43. https://doi.org/10.1123/jpah.2014-0469
Crossa, A., Reilly, K. H., Wang, S. M., Liu, S. Y., Krassner, A., & Lim, New York City Department of Consumer Affairs. (2015, December).
S. (2019). A comparison of first-time Citi Bike members to New From cradle to cane: The cost of being a female consumer. A study of
York City cyclists, 2013–2018. Reimagining Health in Cities: gender pricing in New York City. The New York Department of
Challenges and New Directions, Philadelphia, PA, USA. Consumer Affairs. https://www1.nyc.gov/assets/dca/downloads/pdf/
Dill, J. (2017). Millenials and Bicycling: Do they go together like fixies partners/Study-of-Gender-Pricing-in-NYC.pdf
and messenger bags?. Paper presented at the 96th Annual Meeting New York City Department of Health and Mental Hygiene. (2016).
of the Transportation Research Board, Washington, DC, USA. Community health survey.
January 8-12, 2017). NYC DOT and Lyft Announce New Citi Bike Equity Advisory Board
Emond, C. R., Tang, W., & Handy, S. L. (2009). Explaining gender dif- to Ensure Inclusive and Equitable Bike Share Expansion to
ference in bicycling behavior. Transportation Research Record: Additional NYC Neighborhoods. (2019). https://www1.nyc.gov/html/
Journal of the Transportation Research Board, 2125(1), 16–25. dot/html/pr2019/pr19-071.shtml
https://doi.org/10.3141/2125-03 Ogilvie, F., & Goodman, A. (2012). Inequalities in usage of a public
Fishman, E. (2016). Bikeshare: A review of recent literature. Transport bicycle sharing scheme: Socio-demographic predictors of uptake and
Reviews, 36(1), 92–113. https://doi.org/10.1080/01441647.2015. usage of the London (UK) cycle hire scheme. Preventive Medicine,
1033036
55(1), 40–45. https://doi.org/10.1016/j.ypmed.2012.05.002
Fowler, S. L., Berrigan, D., & Pollack, K. M. (2017). Perceived barriers Oja, P., Titze, S., Bauman, A., de Geus, B., Krenn, P., Reger-Nash, B.,
to bicycling in an urban US environment. Journal of Transport &
& Kohlberger, T. (2011). Health benefits of cycling: A systematic
Health, 6, 474–480.
review. Scandinavian Journal of Medicine & Science in Sports, 21(4),
Garrard, J., Crawford, S., & Hakman, N. (2006). Revolutions for women:
496–509. https://doi.org/10.1111/j.1600-0838.2011.01299.x
Increasing women’s participation in cycling for recreation and trans-
People for Bikes. (2017, July 14). Latinos and family biking: Study after
port: Summary of key findings. School of Health and Social
study finds a link. https://peopleforbikes.org/blog/latinos-and-family-
Development, Deakin University.
biking-study-after-study-finds-a-link/
Getman, A., Gordon-Koven, L., Hostetter, S., & Viola, R. (2017, July).
Pucher, J., & Buehler, R. (2008). Making cycling irresistible: Lessons
Safer cycling: Bicycle ridership and safety in New York City. New
from the Netherlands, Denmark and Germany. Transport Reviews,
York City Department of Transportation. http://www.nyc.gov/html/
28(4), 495–528. https://doi.org/10.1080/01441640701806612
dot/downloads/pdf/bike-safety-study-fullreport2017.pdf
Saelensminde, K. (2004). Cost–benefit analyses of walking and cycling
Goodyear, S. (2013, August 13). Bike-share is key to closing the cycling
gender gap. https://www.citylab.com/solutions/2013/08/bike-share- track networks taking into account insecurity, health effects and
may-be-one-key-closing-cycling-gender-gap/6580/ external costs of motorized traffic. Transportation Research Part A:
Guo, Y., Zhou, J., Wu, Y., & Li, Z. (2017). Identifying the factors Policy and Practice, 38(8), 593–606. https://doi.org/10.1016/j.tra.
affecting bike-sharing usage and degree of satisfaction in Ningbo, 2004.04.003
China. PLoS One, 12(9), e0185100. https://doi.org/10.1371/journal. Steinbach, R., Green, J., Datta, J., & Edwards, P. (2011). Cycling and
pone.0185100 the city: A case study of how gendered, ethnic and class identities
Hebert, J. R., Ma, Y., Clemow, L., Ockene, I. S., Saperia, G., Stanek, can shape healthy transport choices. Social Science & Medicine
E. J., Merriam, P. A., & Ockene, J. K. (1997). Gender differences in (1982), 72(7), 1123–1130. https://doi.org/10.1016/j.socscimed.2011.
social desirability and social approval bias in dietary self-report. 01.033
American Journal of Epidemiology, 146(12), 1046–1055. https://doi. Stop Street Harassment. (2014, Spring). Unsafe and harassed in public
org/10.1093/oxfordjournals.aje.a009233 spaces: A national street harassment report. http://www.stopstreethar-
Heesch, K. C., Sahlqvist, S., & Garrard, J. (2012). Gender differences in assment.org/wp-content/uploads/2012/08/National-Street-Harassmen
recreational and transport cycling: A cross-sectional mixed-methods t-Report-November-29-20151.pdf
comparison of cycling patterns, motivators, and constraints. The Szczepanski, C. (2013, August 6). Women on a roll: Benchmarking
International Journal of Behavioral Nutrition and Physical Activity, women’s bicycling in the United States—And five keys to get more
9(1), 106. https://doi.org/10.1186/1479-5868-9-106 women on wheels. https://www.bikeleague.org/sites/default/files/
Heinen, E., Van Wee, B., & Maat, K. (2010). Commuting by bicycle: WomenBikeReport(web)_0.pdf
An overview of the literature. Transport Reviews, 30(1), 59–96. U.S. Census Bureau. American Community Survey, 2017 American
https://doi.org/10.1080/01441640903187001 Community Survey 5-Year Estimates. Table S2301. https://factfinder.cen-
Kaufman, S. M., Polack, C. F., & Campbell, G. A. (2018, November). sus.gov/bkmk/table/1.0/en/ACS/17_5YR/S2301/1600000US3651000
The pink tax on transportation: Women’s challenges in mobility. Wheeler, N., Conrad, R., & Figliozzi, M. A. (2010). A statistical analysis
https://wagner.nyu.edu/files/faculty/publications/Pink%20Tax%20Rep of bicycle rider performance: The impact of gender on riders’ per-
ort%2011_13_18.pdf formance at signalized intersections. Paper Presented at the
Kuntzman, G. (2019, March 26). Community board panel OK’s Proceedings 89th Annual Meeting Transportation Research Board,
more protected bike lanes in midtown. https://nyc.streetsblog.org/ January.
2019/03/26/community-board-panel-oks-more-protected-bike-lanes- Wilson, S. (1973). Bicycle technology. Scientific American, 228(3),
in-midtown/ 81–91. https://doi.org/10.1038/scientificamerican0373-81

You might also like