Journal of Environmental Psychology 51 (2017) 117e131

Contents lists available at ScienceDirect

Journal of Environmental Psychology

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

Minimizing the gender difference in perceived safety: Comparing the

effects of urban back alley interventions
Bin Jiang a, *, Cecilia Nga Sze Mak b, Linda Larsen c, Hua Zhong d
a
Virtual Reality Lab of Urban Environments and Human Health, Division of Landscape Architecture, Faculty of Architecture, The University of Hong Kong,
Hong Kong SAR
b
Division of Landscape Architecture, Faculty of Architecture, The University of Hong Kong, Hong Kong SAR
c
Department of English, University of Illinois at Urbana-Champaign, United States
d
Department of Sociology, Pearl River Delta Social Research Centre, Chinese University of Hong Kong, Hong Kong SAR

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

Article history: Urban alleys are perceived as unsafe, especially by women. We conducted a photograph-questionnaire
Received 24 October 2016 survey to examine gender difference in perceived safety of alley scenes. Photograph simulation tech-
Received in revised form nology was used to create three categories of intervention scenes: Cleaning, Vegetation, and Urban
17 March 2017
Function & Vegetation. For the existing (Baseline) and Cleaning scenes, perceived safety remained low
Accepted 18 March 2017
Available online 21 March 2017
for both genders, though men's perceived safety was signicantly higher than women's. Vegetation
scenes were perceived as moderately safe for both genders, but men's ratings were still signicantly
higher. For Urban Function & Vegetation scenes, perceived safety was high for both genders, and the
Keywords:
High-density city
gender difference largely disappeared. Geometric vegetation yielded higher perceived safety than
Back alley naturalistic vegetation for both genders. These ndings provide clear evidence to support the efforts of
Environmental intervention policy makers, environmental designers, and community associations seeking to create safe and vital
Perceived safety back alley environments for men and women in high-density cities across the world.
Gender difference 2017 Elsevier Ltd. All rights reserved.

1. Introduction Back alleys are associated with multiple environmental and

Urban alleys are neglected but valuable urban spaces in high-
density cities with limited open space. Rapid, intense urban
development that maximizes protable construction areas and
minimizes unprotable public spaces frequently results in hun-
dreds, even thousands of narrow alleys in high-density cities. The
media commonly portrays back alleys as dark, dirty, and dangerous
spacesdhot spots for deviance and crime. The frequent presence of
physical signs of incivility and minor forms of misbehavior
further contribute to the sense that alleys are unsafe (Ferraro, 1996,
p. 175). Fearing victimization, women in particular alter their
routine activities to avoid using back alleys (Ferraro, 1996; Smith &
Torstensson, 1997). Avoidance behaviors lead to further degrada-
tion, as fewer people are present to supervise and maintain back
alleys.
* Corresponding author. Division of Landscape Architecture, Faculty of Architec-
ture, The University of Hong Kong, Hong Kong SAR. Postal Address: 614 Knowles
Building, The University of Hong Kong, Pokfulam Road, Hong Kong.
E-mail addresses: jiangbin@hku.hk (B. Jiang), makngasze@hotmail.com
(C.N.S. Mak), larsen@illinois.edu (L. Larsen), sarazhong@cuhk.edu.hk (H. Zhong).
social problems, but they have received little attention in the elds
of urban planning and landscape architecture. Although some cities
such as Baltimore, Chicago, Hong Kong, and Turin have developed
pilot projects to revitalize back alleys, no empirical studies have
evaluated the effect of proposed back alley interventions on
perceived safety.
Many studies have reported gender disparity in perceived safety
of public spaces such as back alleys. Compared to men, women's
use of public spaces is more often inhibited by worries of personal
safety (Ferraro, 1996; Fujita et al., 2004; Gilchrist, Bannister, Ditton,
& Farral, 2008; Gover, Tomsich, Jennings, & Higgins, 2011;
Jorgensen, Ellis, & Ruddell, 2013; Koskela & Pain, 2000;
Krenichyn, 2004; Madge, 1997; Pain, 1997, 2001; Snedker, 2012;
Sreetheran & van den Bosch, 2014; Steinmetz & Austin, 2014;
Sweet & Ortiz Escalante, 2015; Yavuz & Welch, 2010).
Creating back alley interventions where men and women feel
similarly, if not equally, safe is an important and challenging issue
that has not been addressed in previous literature about urban al-
leys (Seymour, Wolch, Reynolds, & Bradbury, 2010; Wolch et al.,
2010). To ll this gap, we examined the differences in men's and
women's safety perceptions of existing alley scenes and different

http://dx.doi.org/10.1016/j.jenvp.2017.03.012
0272-4944/ 2017 Elsevier Ltd. All rights reserved.
118 B. Jiang et al. / Journal of Environmental Psychology 51 (2017) 117e131
intervened scenes. We ask, to what extent do interventions based
on different criminology theories minimize the gender difference
in perceived safety?
1.1. Gender and perceived safety in urban spaces
Gender has been a consistent predictor of fear of crime in past
research (Maas et al., 2009). Many studies have noted that
perceived safety varies between genders. Compared to men,
women typically have lower perceived safety in public spaces. They
have greater fear of being alone (O'Brien, 2005), of being in con-
cealed spaces (Jorgensen et al., 2013; Steinmetz & Austin, 2014),
and of being in spaces where strangers are loitering (Madge, 1997).
1.1.1. Women's perceived safety in urban public spaces
A widely accepted notion is that although women may be afraid
of many criminal activities, they are most fearful of sexual assault.
Research suggests that compared to women, men's greater physical
strength and stronger sexual drive leads to more sexually indis-
criminate behavior, more casual sexual relationships, and greater
likelihood of committing sexual assault (Felson, 2014). For women,
the fear of sexual assault is the core fear that inuences how they
perceive the safety of public spaces (Ferraro, 1996; Smith &
Torstensson, 1997). This fear is especially strong for young
women. Research found young women are often targeted by po-
tential attackers, even after controlling for opportunities and vul-
nerabilities (Felson, 2014). A study reported that 15-29-year-old
victims of robbery and homicide were frequently sexually assaulted
during the crime (Felson & Cundiff, 2012).
Studies report greater fear of crime for women than men in a
variety of built environments: campuses, urban wooded areas,
public train transit, and public housing complexes (Gover et al.,
2011; Mazey, 1983; Steinmetz & Austin, 2014; Yavuz & Welch,
2010). The gender disparity is especially profound when the envi-
ronment is in a neglected or abused condition (O'Brien, 2005).
Women feel safer in spaces that show clear signs of management.
Women report a higher preference than men for spaces with
visually obvious management and law enforcement (Ho,
Sasidharan, Elmendorf, Willits, Graefe, & Godbey, 2005;
Richardson & Mitchell, 2010). Their fear decreases to a greater
extent when there are other legitimate visitors in the public space
(Jorgensen et al., 2013).
Women are particularly fearful of secluded, visually concealed,
or dark urban spaces. One possible explanation for this fear is that
men prefer to assault female victims in secluded, concealed areas to
avoid societal condemnation and punishment. Many societies
severely condemn men's violence against women and perceive
women as needing protection (Felson, 2014), and perpetrators will
seek secluded spaces to avoid this condemnation.
Avoidance behavior is more prevalent among women than men
when they perceive a risk of victimization (Yavuz & Welch, 2010).
Women tend to spend less time walking in public spaces they
perceive as unsafe, while men's walking behaviors are less inu-
enced by safety concerns (Foster, Hillsdon, & Thorogood, 2004).
These avoidance behaviors further intensify the gender disparity
and may impair women's mental and physical health as they are
less willing to spend time in outdoor public spaces.
One explanation for the gender disparity is that urban spaces are
gendered (Laurie, Claire, Holloway, & Smith, 2000). To some extent,
the built environment is structured according to gendered un-
derstandings of space, particularly in relation to public and
private spheres (Ekinsmyth, 2002; Mazey & Lee, 1983). Very
often, women's spaces are thought of as private and domestic
(homes, residential areas), while men's spaces are thought of as
public (businesses, streets, central areas of cities) (Ekinsmyth,
2002). Women tend to hold a stronger perception than men that
public space is dangerous and private space is safe (Pain, 1997).
1.1.2. Men's perceived safety in urban public spaces
Men's perceived safety in urban public spaces has received far
less attention from researchers than women's perceived safety
(Yavuz & Welch, 2010). Some argue that men may have similar fears
as women, but they tend to underreport their fear in surveys. This is
largely because men are socialized in many cultures to not express
their fear and worries, to be protectors and to be in control in an
environment (Yavuz & Welch, 2010). Indeed, researchers found
women tend to rationally report their fear but men tend to
irrationally discount their fear (Smith & Torstensson, 1997). Re-
searchers also argue women's higher levels of self-reported fear
make sense because women have a stronger tendency than men to
generalize their experiences across time, space, and type of
victimization.
Other literature suggests men may have special safety concerns
that are different from women's. Men may be more sensitive to
losing control in a physical environment or social circumstance.
Thus, men tend to be more concerned about violent assault, direct
confrontation with other men, and lack of knowledge of a place or a
situation (Yavuz & Welch, 2010). Men are more threatened by the
presence of groups of men while women tend to be more threat-
ened by the presence of a single man or being alone (Pain, 1997)
(Crime Concern, 2004). Men showed more trust in electronic sur-
veillance techniques such as CCTV while women show more trust
in the presence of a security guard (Yavuz & Welch, 2010).
1.2. Theories exploring how environments impact perceived safety
Many studies have shown that environmental design can play a
signicant role in promoting perceived safety for both women and
men. In many cases, designers' work is based on intuition and
practical experience, not on criminology theory or literature on
gender and safety. In this study, we used criminology theories to
identify design interventions and evaluate how they impact the
perceived safety of both women and men.
1.2.1. Disorder and the broken windows theory
Compared to traditional criminology theory that mainly focuses
on promoting safety through legal penalties, the broken windows
theory (Wilson & Kelling, 1982) emphasizes promoting safety by
repairing broken windows, a metaphor for disorder or incivility
(Skogan, 2012). The broken windows, or disorder, in a space
decrease the stability of urban neighborhoods, indicate a lack of
social control, deter investment in the neighborhood, and induce
fear of crime (Skogan, 2012).
Disorder can be both physical and social. Physical disorder in-
cludes overt signs of negligence or unchecked decay as well as the
visible consequences of malevolent misconduct (Skogan, 2012, p.
175). Social disorder includes unsettling or potentially threatening
and perhaps unlawful public behaviors (Skogan, 2012, p. 175).
While signs of physical and social disorder can exist independently,
they are often related. For instance, physical disorder (broken
windows) can be a consequence of social disorder (unlawful
behavior), and social disorder is more likely to occur where physical
disorder is present (in abandoned buildings, for instance). There-
fore, the broken windows theory suggests that environmental clues
can reasonably be used to indicate both physical and social disorder
in urban spaces (Skogan, 2012).
Back alleys are notorious for disorder. They deliver a strong
message of social decline and lack of social control. This message
deters legitimate visitors and encourages deviance and crime. So-
cial incivilities (e.g., prostitution, public urination, crime) and
 B. Jiang et al. / Journal of Environmental Psychology 51 (2017) 117e131 119

physical incivilities (e.g., littering and abandoned buildings) go
hand in hand. Social incivilities often occur in places that are hidden
and disorderly, places people tend to avoid. Physical disorder can
provide clues about the likelihood of social incivilities in a space as
well as people's willingness to visit (Keizer, Lindenberg, & Steg,
2008; Seymour et al., 2010; Wilson & Kelling, 1982). Studies
based on this theory have tended to evaluate physical trails of ci-
vility or incivility rather than behaviors themselves.
The broken windows theory suggests that disorderly environ-
mental features encourage incivility or crime and can be conta-
gious, spreading like an epidemic (Welsh, Braga, & Bruinsma, 2015).
Disorder encourages avoidance, and avoidance leads to further
disorder and abandonment of nearby spaces. Design interventions
rooted in the broken windows theory mainly address the visual
impression or physical appearance of a place rather than
creating a new social dynamic for that place (Welsh et al., 2015). In
daily urban management practice, environmental clean-up or
adding green plants would belong to this type of intervention. In a
narrow back alley, these interventions would make the place more
attractive and orderly. However, their impact on perceived safety is
questionable because they do not improve the uncertain function of
a back alley or alter people's daily use. In our study, we base our
Cleaning and Vegetation interventions on the broken windows
theory.
alleys in ways that will help both men and women feel safe.
1.4. Research questions
To ll in the knowledge gap, we examine three main questions:
First, to what extent are men's and women's attitudes and behav-
iors different concerning disorderly alley conditions? Second, to
what extent are men's and women's safety perceptions different for
a variety of environmental interventions based on broken windows
theory or routine activity theory? Third, to what extent are men's
and women's safety perceptions different for alley scenes with the
very similar content but different forms?
2. Methods
In this study, we evaluated and compared gender differences in
perceived safety of back alley scenes and interventions by con-
ducting a photograph-questionnaire survey. We examined 16 types
of scenes including an existing alley scene and 15 simulated scenes.
We also conducted a survey to obtain respondents' demographic
information and attitudes regarding existing back alleys (See Fig. 1
as an example).
2.1. Background of site district

1.2.2. Routine activity theory The sample alleys we used in our photo-questionnaire are in the
The broken windows theory evaluates perceived safety by Yau Tsim Mong District, which is a part of the Kowloon peninsula in
looking at environmental clues of disorder and emphasizing visual Hong Kong (Fig. 2). This district was one of the earliest and most
appearance. The routine activity theory, on the other hand, evalu- heavily developed harbor and commercial areas with a large
ates safety not just through the appearance of the place, but by
examining the human activity that exists in a place. It connects the
functions of an urban space with human activity. Cohen and Felson
(1979) dene routine activities as recurrent and prevalent activ-
ities which provide for basic population and individual needs. By
adjusting urban functions offered in a place, environmental de-
signers can facilitate legitimate routine activities and deter unciv-
ilized behavior. Routine activities create a strong guardianship,
preventing incivility and crime and improving perceived safety
(Meghan, Marcus, & Brandon, 2013).
The routine activity theory promotes safety and deters crime by
altering the function of the urban space. This generates a stronger,
more benign social dynamic and more layers of supervision. Unlike
interventions grounded in the broken windows theory, routine
activity interventions directly alter the land use in a space. Dark,
disorderly back alleys, which function primarily as passageways,
can be transformed into inviting destinations for daily recreational,
entertainment, or educational activities. It is plausible that this type
of intervention is more effective at promoting perceived safety than
interventions that merely change the appearance of the place. In
our study, our Urban Function and Vegetation design interventions
are mainly based on the routine activity theory.

1.3. A critical knowledge gap

A number of studies examining the gender difference in

perceived safety have reported that women experience lower
perceived safety than men (Richardson & Mitchell, 2010; Steinmetz
& Austin, 2014; Yavuz & Welch, 2010). However, there is a critical
knowledge gap: These studies do not address differences in men's
and women's safety perceptions of back alleys, a common but often
neglected urban space. We have no evidence identifying the impact
on men's and women's perceived safety of a variety of environ-
mental design interventions. Lack of this knowledge prevents ur-
ban environment managers, urban planners, and landscape Fig. 1. A disorderly alley in Hong Kong. Urban back alleys are common but neglected
architects from making science-based decisions to revitalize back spaces in high-density cities.
120 B. Jiang et al. / Journal of Environmental Psychology 51 (2017) 117e131
Fig. 2. The Yau Tsim Mong District (up) and the site area (bottom) in the central area of Hong Kong city. The highly dense buildings creates hundreds of urban back alleys in this
district (Photo source: Google Earth Pro, 2016).
amount of socio-economic, cultural, and built environment di-
versity. The 2011 Hong Kong City population census shows 307,878
citizens resided in this 6.55 square kilometer district and 54% were
women (165,070), which suggests it is an extremely high-density
urban district. Most residents are Chinese (88%) but with a di-
versity of minorities, including Filipino (2%), Nepalese (2%), Indo-
nesian (2%), Indian (1%), White (3%), and others (3%). Most
minorities except White are low-income workers (such as home
helpers) or unemployed. A variety of illegal businesses thrive in the
district, including gambling, prostitution, fake goods dealing, and
drug dealing. The high density of buildings and complexity of land
use is representative of the central Hong Kong city area. Hundreds
of back alleys were created by this type of heavy development.
This district has one of the highest crime rates in Hong Kong
(Hong Kong Police Force, 2014). In our search of online ofcial
media reports published between January 2014 and September
2015, we found reports of 60 crimes in this district that were
associated with back alleys including burglary (16 cases), robbery
(8 cases), rape (6 cases), arson (5 cases), cruelty to animals (5 cases),
and serious assault (4 cases).
2.2. Site selection
Three experts in landscape architecture and urban design
reviewed back alleys through Google Earth Street and chose 58
candidate back alleys. Then one investigator visited the back alleys
and took a panoramic photograph for each alley. The investigator
stood by the sidewalk of the main street and faced the entry of an
alley and then took an eye-level panoramic photograph. The dis-
tance between the investigator and entry was approximately 2 m.
The angle of view shed was approximately 120
to include not only
the alley space but also the 10e15 m-wide main street facade at
both sides of the alley entry.
Next, photographs of all potential sites were evaluated by three
experts. A series of criteria was used to remove scenes containing
unusual objects or information, including unique looking buildings,
unusual or special vegetation, the presence of birds or animals,
unusual lighting conditions, unusual architectural decorations, and
humans with an unusual or attractive appearance. Any scene
deemed unacceptable by at least one expert was removed, and 35
photographs were removed from the sample pool. Microsoft Excel
was used to randomly select ve scenes from the remaining pho-
tographs for the survey (Fig. 3).
2.3. Three categories of design interventions
Three categories of design interventions were developed based
on the broken windows theory and routine activity theory. The
Cleaning and Vegetation intervention categories are grounded in
the broken windows theory because they mainly alter the
 B. Jiang et al. / Journal of Environmental Psychology 51 (2017) 117e131 121

Fig. 3. Five scenes of urban alleys were selected as research sites by three experts through a restrictive process.

appearance of the place. The Urban Function & Vegetation inter-
vention category is mainly grounded in the routine activity theory
because these interventions alter the function of the space, in
addition to altering the appearance. Including the existing disor-
derly alley scenes (the Baseline category), four categories of scenes
were examined in this study.
Our rst category, the Baseline category, shows scenes of
existing alleys with dirt, litter, grafti, and posts. The Cleaning
category shows scenes of the alleys after litter, dirt, grafti, and
posts were removed from the existing alley scenes. For the third
category, the Vegetation scenes, we assess the impact of adding
various doses and styles of green vegetation to the cleaned alley
scenes. The Vegetation category includes four types of in-
terventions (Fig. 4: 2e5): naturalistic green wall, geometric
green wall, naturalistic green wall & ground, and geometric
green wall.
The fourth category, Urban Function & Vegetation, includes in-
terventions that invite new functions for legitimate users. We
included small business functions, community facilities and ser-
vices, and municipal public space. The rst two are often bottom up
interventions developed by land owners, commercial tenants, or
community associations. Small businesses such as restaurants,
122 B. Jiang et al. / Journal of Environmental Psychology 51 (2017) 117e131

Fig. 4. The baseline image and 15 simulated images for one site. 0. baseline (existing), 1. cleaning, 2. naturalistic green wall, 3. geometric green wall, 4. naturalistic green wall &
ground, 5. geometric green wall & ground, 6. green art gallery, 7. green shop, 8. green cycling, 9. green gym, 10. naturalistic green platform, 11. geometric green platform, 12.
, 13. geometric green cafe
naturalistic green cafe , 14. naturalistic green park, and 15. geometric green park. (For interpretation of the references to colour in this gure legend, the
reader is referred to the web version of this article.)

news stalls, coffee shops, bookstores, grocery stores, and snack bars
might spatially and functionally t in narrow alleys. Community
facilities or services might include outdoor exercise space, bike
parking, or community art galleries.
Municipal public space interventions are often large-scale,
involving a fundamental change of building structure or street
layout. These large-scale interventions demand nancial support,
leadership, negotiation among stake holders, careful assessment,
and approval. In this study, we consider two smaller-scale munic-
ipal interventions that are more feasible: transforming the entry
area into a small pocket park, and creating an overhead public
platform.
In all, we developed 10 interventions in the Urban Function &
Landscape category (Fig. 4: 5e15). For the shop, cycling, exercise
facility, and art gallery interventions, we added green vegetation
, and public platform
without a distinctive style. For the park, cafe
interventions, we added green vegetation in either a natural or a
geometric style.
2.4. Photo simulations
In this study, we used Adobe Photoshop 6.0 to produce a
panoramic photograph for each of the ve existing alleys. Using
these photographs, we produced visual simulations of the
proposed interventions. In order to avoid confounding factors,
we followed four rules: rst, we chose common green plant
species without distinctive cultural meaning or aesthetic value;
second, we chose human gures with a normal appearance;
 B. Jiang et al. / Journal of Environmental Psychology 51 (2017) 117e131 123

third, we used common construction materials for pavement
and building simulation, nothing rough, delicate, or unusual. In
all, we created a photomontage set of four categories with 15
types of alley revitalization scenes for each of the ve alley
sites. In total, we produced 80 panoramic photos for the photo-
questionnaire survey, including ve existing scenes and 75
simulated scenes.
Our choice of these methods are supported by previous studies.
Photo simulation has been widely used in environmental psy-
chology research as a reliable surrogate of real environments
(Junker & Buchecker, 2008; Kuo, Bacaicoa, & Sullivan, 1998;
Sullivan & Lovell, 2006; Todorova, Asakawa, & Aikoh, 2004).
Panoramic photographs are frequently used in research because
they better represent humans' viewshed than photographs with
common aspect ratios (4:3 or 3:2) (Jiang, Larsen, Deal, & Sullivan,
2015; Li et al., 2015; Raanaas, Patil, & Hartig, 2012).
2.5. Photo-questionnaire and background survey
We used snowball sampling and convenient sampling methods
to recruit participants online or on-site. Online participants were all
current Hong Kong residents accessed through public email and
social networking apps. On-site participants were citizens walking
on the main streets adjacent to the ve sites who were randomly
invited to participate. Each participant was randomly assigned to
view 30 out of 80 photos. They reported their perceived safety for
each photo using a ve-point Likert scale ranging from very unsafe
(1), unsafe (2), uncertain (3), safe (4), to very safe (5). The
uncertain (3) rating suggests a neutral perception, and we use it
as a cutoff value between negative and positive perceived safety.
Respondents were also required to complete a background infor-
mation survey with demographic questions and questions about
their familiarity with and psychological or behavioral responses to
back alleys (see Appendix Table A).
2.6. Survey respondents
We recruited 218 adults for the questionnaire survey93 on-site
and 125 online participants. Table 1 shows the demographic char-
acteristics of the sample. 124 (57%) respondents were women, and
94 (43%) were men. 81 (86.2%) male and 111 (89.5%) female re-
spondents were local Chinese.
2.7. Data analysis
We analyzed the data in two main sections. First, we analyzed
the data from the background survey to answer our rst question
through comparison of histograms and one-way ANOVA. Second,
we analyzed data from the photo-questionnaire survey to answer
questions 2, 3, and 4. Results of the photo-questionnaire move from
general to specic. First, we checked whether gender difference
among safety perception scores are signicant using paired t-tests
and ANOVA. Second, we examined the gender difference in safety
perception for the four theoretical categories, using an F-test. Third,
we conducted a summary t-test and a paired t-test to examine the
gender differences within each of the four categories. Fourth, we
examined the gender differences for each of the 16 interventions
and baseline scenes using a summary t-tests. Fifth, we tested the
possibility that there are ve statistical categories, rather than four,
by conducting k-means analysis. Finally, we examined the gender
differences of preference for naturalistic or geometric vegetation
using a summary t-test.

Table 1
Sample demographic characteristics for men and women.

Demographic variables Men (percentage, %) Women (percentage, %)

Sample Size 94 124

Age
18e20 5 (5.3) 1 (0.8)
21e25 30 (31.9) 67 (54.0)
26e30 21 (22.3) 20 (16.1)
31e35 8 (8.5) 10 (8.1)
36e40 7 (7.4) 5 (4.0)
41e45 3 (3.2) 1 (0.8)
46e50 5 (5.3) 5 (4.0)
51e55 6 (6.4) 7 (5.6)
56e60 1 (1.1) 2 (1.6)
61e65 2 (2.1) 3 (2.4)
66e70 1 (1.1) 2 (1.6)
71 or above 5 (5.3) 1 (0.8)
Ethnicity
Chinese (Hong Kong SAR, China) 81 (86.2) 111 (89.5)
Chinese (Mainland China) 10 (10.6) 11 (8.9)
Persian 2 (2.1) 0 (0)
Filipino, Nepalese, Indian, or Indonesian 1 (1.1) 1 (0.8)
Black 0 (0) 1 (0.8)
White 0 (0) 0 (0)
Others 0 (0) 0 (0)
Marital Status
Married 25 (26.6) 28 (22.6)
Widowed 1(1.1) 1 (0.8)
Divorced 1 (1.1) 1 (0.8)
Single 67(71.3) 91 (73.4)
Crime or violence experience happened on respondent or respondent's family or friends in the last two years
Yes 4 (4.3) 6 (4.8)
No 87 (92.6) 114 (91.9)
Missing data 3 (3.2) 4 (3.2)

Note: A row named as missing data is presented if there are missing data for a question.
124 B. Jiang et al. / Journal of Environmental Psychology 51 (2017) 117e131
3. Results
3.1. Gender differences in reported attitudes and behaviors about
existing alley conditions
In this section, we present ndings from comparisons of histo-
grams and One-Way ANOVA analysis. First, according to the back-
ground survey (see Appendix Table A and Fig. 5), men and women
reported a similar level of familiarity with the alleys shown in the
pictures: 41.5% of men and 37.1% of women reported they are
familiar with the alleys. 6.4% of men and 5.6% of women reported
they are very familiar with the alleys. ANOVA analysis shows there
is no signicant difference in familiarity, F (1, 216) 2.04, p 0.16.
Men and women reported that they had similar alleys in their
neighborhoods: men reported there are some (31.9%), many
(30.9%), or a lot (8.5%) of similar alleys, while women reported
there are some (39.5%), many (25.0%), or a lot (4.8%) of similar
alleys in their neighborhoods. ANOVA analysis shows there is no
signicant difference in the existence of alleys in participants'
neighborhoods, F (1, 216) 1.61, p 0.21.
In terms of time spent in back alleys, there are some differences.
49.2% of women reported that they used the alleys zero times per
week on average, while the percentage for men was 37.2%. 30.3% of
women and 18.0% of men reported they never use back alleys.
However, ANOVA analysis shows the gender difference for time
spent in back alleys is only marginally signicant, F (1, 216) 2.79,
p < 0.10. Among the ve time periods, men reported a higher rate of
usage than women except in the early morning (5:00 a.m.e7:59
a.m.), when neither men nor women used the back alleys much.
ANOVA analysis shows the gender difference for these time periods
is not signicant, F (1, 216) 2.48, p < 0.12.
Women reported a higher level of hesitation to walk into a back
alley alone. 26.6% of women very often and 16.1% of women
always hesitated while only 12.8% and 8.5% of men reported
similar feelings. ANOVA analysis shows women reported signi-
cantly greater hesitation for using alleys, F (1, 216) 26.17,
p < 0.001. Women reported a higher level of insecurity or threat
walking alone in a back ally. 32.3% and 12.9% of women reported
very often and always experiencing hesitation while only 16.0%
and 7.4% of men reported similar feelings. ANOVA analysis shows
women reported a signicantly greater feeling of insecurity and
threat, F (1, 216) 25.12, p < 0.001.
When asked how they respond when they walk in an alley
alone, 52.4% and 24.2% of women reported they very often and
always walk faster, while 31.9% and 11.7% of men reported those
responses. ANOVA analysis shows women reported a signicantly
faster walking pace in alleys, F (1, 216) 35.13, p < 0.001. 15.3% and
4.8% of women reported that they very often and always have a
faster heartbeat, while 11.7% and 0% of men reported those re-
sponses. ANOVA analysis shows women reported a signicantly
faster heartbeat in alleys, F (1, 216) 14.76, p < 0.001.
Women also indicated a greater desire for social support when
they were walking through back alleys: 30.6% and 13.7% of women
very often or always wish[ed] to be accompanied by family or
friends while only 8.5% and 7.4% of men reported very often or
always having those responses. ANOVA analysis shows women
reported a signicantly greater desire for walking with others, F (1,
216) 36.72, p < 0.001.
11.3% and 2.4% of women very often or always call family or
friends and keep talking to them while walking through back alleys.
In contrast, only 5.3% and 2.1% of men reported that they very
often or always had those responses (see Fig. 5). ANOVA analysis
shows women reported a signicantly greater desire to talk to
others, F (1, 216) 12.67, p < 0.001.
In all, women reported signicantly greater feelings of fear and
more anxious behaviors regarding existing back alleys, although
they had similar familiarity with them as men.
3.2. Gender differences in safety perceptions for the four categories
of scenes
3.2.1. General gender difference
To see whether there is a general gender difference in perceived
safety for all categories, we conducted a paired t-test between men
and women and found there is a signicant difference, mean
difference 0.10, SD 0.16, t (1, 15) 2.49, p 0.03. This implies
that men reported higher perceived safety than women for all
categories.
We further looked at the gender differences for the baseline
category and the three categories of interventions. ANOVA results
conrmed that gender is a signicant factor in the tests of within-
subjects contrasts [F (1, 3) 14.20, p 0.003] while the interactive
effect between gender and the categorization is insignicant [F (1,
3) 2.74, p 0.09]. This further conrms gender disparity is a
signicant issue in this study.
3.2.2. Difference among four categories for each gender
In order to reinforce the validity of the 4 categories, we
employed an F-test to examine to what extent there are signicant
differences among the 4 categories of scenes for both men and
women. The F-test procedure is similar to the procedure in the rst
section:
RSS0 RSSa
M1
F RSSa
 FM1;NM
MN
where RSS_0 is the sum of residuals for the two groups of data;
RSS_a is the sum of residuals if group effects are considered; M is
the total number of groups, and N is the total number of cases. This
results in a p-value of <0.05 or <0.001 for all six tests. We conclude
that there is a signicant difference among the four categories
(Table 2).
Next we explored whether the gender difference is signicant
for each of the four categories. We conducted a summary t-test for
the Baseline and Cleaning intervention. For the Baseline, men re-
ported a signicantly higher sense of safety than women, mean
difference 0.23, t 2.17, df 305, p 0.03. For the Cleaning
intervention, men reported a signicant higher sense of safety than
women: mean difference 0.34, t 3.21, df 293, p 0.001.
Both tests passed the Hartley test for equal variance. We conducted
paired t-tests for the other two categories. For the Vegetation in-
terventions, men reported a signicantly higher sense of safety
than women, mean difference 0.17, t 3.35, df 3, p 0.04. For
the Urban Function & Vegetation category, the gender difference is
insignicant, mean difference 0.03, t - 0.64, df 9, p 0.54. See
Fig. 6.
3.3. Gender differences in perceived safety of 16 scenes
Next, we tested the gender differences of safety perceptions for
all 16 scenes. We conducted summary t-tests; results are shown in
Table 3 and Fig. 7.
For the Baseline scene, both women and men reported negative
safety perceptions (2.33 vs 2.56, where 3 is the cutoff value be-
tween positive and negative perception). For the Cleaning scene,
women reported negative perceived safety (2.77) but men reported
positive perceived safety (3.11), a signicantly higher sense of
safety than women. Although both men and women reported in-
creases in perceived safety for the Cleaning scene compared to the
 B. Jiang et al. / Journal of Environmental Psychology 51 (2017) 117e131 125

Fig. 5. Comparison of men's and women's attitudes about and behaviors in existing back alleys. The left (blue) bar is for men and the right (orange) bar is for women. The Y axis
suggests percentage of respondents. Signicant differences (p value) are marked as ns not signicant, < 0.1, or *** < 0.001. (For interpretation of the references to colour in this
gure legend, the reader is referred to the web version of this article.)
126 B. Jiang et al. / Journal of Environmental Psychology 51 (2017) 117e131

Table 2 scenes yielded a signicantly lower sense of safety for women than
Results of F test for differences among four groups. men: the Geometric Green Platform (t 2.11, p < 0.01) and
Men df rss_0 rss_a F_val Naturalistic Green Art Gallery (t 3.00, p < 0.05). Compared to
1v2 1, 240 217.09 185.56 40.61***
men, women reported signicantly higher perceived safety for the
1v3 4, 668 523.89 424.04 39.27*** Naturalistic Green Cafe (t 2.63, p < 0.01). No signicant gender
1v4 10, 1451 1380.67 996.43 55.95*** disparity was found for the other eight types.
2v3 4, 656 441.17 417.71 9.21***
2v4 10, 1440 1148.49 990.10 23.04***
3v4 13, 1868 1412.56 1228.57 21.52*** 3.4. Gender differences in perceived safety of geometric or
Women df rss_0 rss_a F_val naturalistic vegetation

1v2 1, 359 308.01 302.63 6.38*

1v3 4, 935 818.39 675.82 49.31***
For the 16 types of scenes, ve pairs had similar content but
1v4 10, 2154 2108.06 1414.16 105.69*** different styles of vegetation: naturalistic or geometric vegetation.
2v3 4, 934 728.76 655.66 26.03*** We conducted summary t-tests to compare perceived safety on
2v4 10, 2154 1901.06 1394.00 78.31*** each pair of scenes for women and men to see whether there is any
3v4 13, 2729 2204.50 1767.18 51.95***
gender difference. Table 4 shows the results.
Note: 1 Baseline, 2 Cleaning, 3 Vegetation, 4 Urban Function & Vegetation In general, we found the geometric scenes yielded signicantly
Interventions. *< 0.05, **< 0.01. ***< 0.001. higher perceived safety in two out of ve conditions for each
gender; for the other three conditions for each gender, none of the
naturalistic scenes yielded signicantly higher perceived safety.
Baseline scene, the gender difference (the mean difference in men's
This suggests the geometric scene in general is better than the
and women's scores) is wider for the Cleaning scene than for the
naturalistic scenes at promoting perceived safety for both genders.
Baseline scene.
Specically, men reported signicantly lower perceived safety
For the four Vegetation intervention scenes, both genders show
for the naturalistic green wall than for the geometric green wall,
a slightly positive level of perceived safety. However, women report
t 2.30, p 0.02. However, there was no signicant difference for
signicantly lower perceived safety than men for three of the four
women. Women reported signicantly lower perceived safety for
scenes: naturalistic green wall, geometric green wall, and natural-
the naturalistic green wall and ground than for the geometric wall
istic green wall & ground (p < 0.05 or p < 0.01). The gender dif-
and ground, t 2.90, p 0.004. However, there was no signicant
ference in perceived safety is clearly smaller for the Vegetation
difference for men. For the two park scenes, both women
scenes (Dif. 5.3%) compared to the Baseline (Dif. 9%) and
(t 2.01, p 0.04) and men (t 3.27, p 0. 001) reported a
Cleaning scenes (Dif. 11%) (see Table 3).
lower sense of safety for the naturalistic green park than for the
For Urban Function & Vegetation intervention scenes, both men
geometric green park. For the two cafe  scenes, the differences in
and women reported positive levels of perceived safety for the
men's and women's perceived safety were not signicant.
majority of scenes, ranging from slightly positive to highly positive
(3.17e4.23 for women; 3.31 to 4.29 for men). In general, the levels
of perceived safety are higher for these interventions than the 3.5. Statistical clustering analysis
levels of perceived safety for the Vegetation intervention scenes.
For most of the Urban Function and Vegetation intervention scenes, Although we found the four theoretical categories reliable, there
there is no signicant gender difference between men's and were some mixture effects among several Vegetation Only and
women's scores. Among the 10 types of interventions, only two Urban Function and Vegetation Interventions. We wondered if a
more detailed categorization would be more helpful to

Fig. 6. The extent of gender difference across four categories of scenes. The theorygroup4 indicates four categories of scenes: 1. Baseline; 2. Cleaning; 3. Vegetation; 4. Urban
Function & Vegetation. The gender indicates two genders: 1. Men; 2. Women.
 B. Jiang et al. / Journal of Environmental Psychology 51 (2017) 117e131 127

Fig. 7. Gender disparity in average perceived safety of intervention scenes. Signicant differences are marked as * < 0.05 or ** < 0.01. The * is highlighted by red if men reported a
signicantly higher level of perceived safety. The * is green if women reported a signicantly higher level of perceived safety. (For interpretation of the references to colour in this
gure legend, the reader is referred to the web version of this article.)

professionals in choosing specic safety-promoting interventions. belongs to the cluster with the nearest mean, serving as a prototype
We grouped the 16 scenes using a statistical clustering method. of the cluster (Nowakowski, Myn  czak, Jodejko-Pietruczuk, &
We conducted a K-Means clustering to compare theoretical Werbin ska-Wojciechowska, 2014).
clustering and statistical clustering. K-means clustering aims to To see how many clusters are appropriate, we rst conducted a
partition n observations into k clusters in which each observation within-group sum of squares test for different group sizes. The

Table 3
Gender differences for the baseline and 15 intervention scenes.

Women Men Dif. (%) t

N Mean SD SE N Mean SD SE

Baseline 181 2.33 0.94 0.17 126 2.56 0.87 0.23 9 2.17*
Cleaning 180 2.77 0.89 0.21 115 3.11 0.88 0.29 11 3.21***
Naturalistic Green Wall 181 3.06 0.83 0.23 154 3.25 0.84 0.26 6 2.11*
Geometric Green Wall 165 3.19 0.77 0.25 128 3.47 0.75 0.31 8 3.11**
Naturalist Green Wall& Ground 217 3.16 0.83 0.21 131 3.35 0.73 0.29 6 2.11*
Geometric Green Wall& Ground 196 3.40 0.85 0.24 133 3.44 0.77 0.30 1 0.44ns

Green Art Gallery 200 3.20 0.79 0.27 132 3.39 0.74 0.33 6 2.11*
Green Shop 207 3.91 0.71 0.26 138 3.86 0.87 0.30 2 0.56ns
Green Cycling 187 3.51 0.88 0.26 140 3.61 0.84 0.31 3 1.00ns
Green Gym 206 3.67 0.87 0.29 143 3.67 0.82 0.36 0 0.00ns
Naturalistic Green Platform 193 3.31 0.89 0.29 131 3.31 0.98 0.35 0 0.00ns
Geometric Green Platform 183 3.17 0.95 0.23 126 3.47 0.82 0.29 9 3.00**

Naturalist Green Cafe 210 4.19 0.66 0.30 125 3.98 0.78 0.33 5 2.63**
Geometric Green Cafe  202 4.07 0.71 0.29 134 4.07 0.80 0.39 0 0.00ns
Naturalistic Green Park 186 4.08 0.76 0.24 146 3.99 0.83 0.29 2 1.00ns
Geometric Green Park 210 4.23 0.72 0.23 121 4.29 0.67 0.31 1 0.75ns

Note. Dif. (%) (perceived safety by women e perceived safety by men)/perceived safety by women  100. N s not signicant, * < 0.05, ** < 0.01. *** < 0.001.
128 B. Jiang et al. / Journal of Environmental Psychology 51 (2017) 117e131

Table 4 show a greater level of hesitation, avoidance, psycho-physiological

Comparison between naturalistic and geometric scenes for both genders. responses of fear, and need of social support. Second, levels of
Naturalistic vs Geometric Women Men gender disparity for the four categories of scenes are clearly
Green Wall df 344, t 1.51 df 280, t 2.30*
different: For the Baseline and Cleaning scenes, perceived safety
Green Wall & Ground df 411, t 2.90** df 262, t 0.97 remained low for both genders, but there was a signicant gender
Green Platform df 374, t 1.48 df 255, t 1.42 disparity in scores (men's ratings were higher). The gender differ-
Green Park df 394, t 2.01* df 264, t 3.27*** ence did not narrow with the Cleaning intervention, and in fact
Green 
Cafe df 410, t 1.78 df 257, t 0.92
gender difference was greater for the Cleaning intervention than for
Note: * < 0.05 or ** < 0.01. the Baseline scene. The Vegetation interventions were perceived as
moderately safe for both genders, and gender difference decreased,
but there were still signicant differences between the genders
results in Fig. 8 show that the clustering quality indicated by
(men's ratings were higher). When Urban Functions were added to
within-group sum of square stays consistent after six or more
the Vegetation interventions, perceived safety scores were higher
groups. Thus we decided to group the data into 6-group clusters. 6-
for both genders, and the gender disparity largely disappeared.
group clustering showed a better performance, indicated by a
Third, geometric scenes generally yielded higher perceived safety
higher F value and a clearer separation between Vegetation and
than naturalistic scenes for both genders.
Urban Function & Vegetation interventions (Table 5). In general,
either 4-group or 6-group clustering showed similar clusters as the
4 theoretical grouping, but the results show there are similarities 4.1. Interpretation of key ndings and implications
among a few of the high-end Vegetation Only interventions and the
low-end Urban Function and Vegetation interventions. The 6-group We found a strong gender disparity in sense of safety for the
clusters, shown in Table 5, are more ne-tuned than the 4-group existing disorderly scenes (Baseline scenes). Both women and men
clusters and can be more useful for practitioners to select the in- perceived them as unsafe but women reported signicantly lower
terventions that best promote perceived safety for each gender. perceived safety. The background survey of existing alleys also
showed that women have greater fear and avoidance of using back
4. Discussion alleys than men. This nding corresponds to previous studies that
indicate women are often perceived as and perceive themselves as
This study yielded three major ndings. First, we found signif- more vulnerable than men in public spaces, especially in disorderly
icant gender disparity in the attitudes and perceptions of safety for public spaces (Sreetheran & van den Bosch, 2014; Steinmetz &
existing back alleys. Women were much more likely than men to Austin, 2014).

Fig. 8. The within groups sum of square reaches a stable low level as the number of clusters reaches 5 for women (up graph) and men (bottom graph, which indicates 5 is an
optimal amount for k-means clusters.
 B. Jiang et al. / Journal of Environmental Psychology 51 (2017) 117e131
Table 5
Results of the six-group K-means cluster analysis.
Clusters Men
Sub-groups
1st Baseline
2nd Cleaning
3rd Naturalistic Green Wall
Geometric Green Wall
Naturalist Green Wall& Ground
Geometric Green Wall& Ground
Green Art Gallery
Naturalistic Green Platform Geometric Green Platform
4th Green Shop
Green Cycling
Green Gym
5th 
Naturalist Green Cafe
Geometric Green Cafe 
Naturalistic Green Park
6th Geometric Green Park
Summary F(5, 10) 64.55, p < 0.001
Although we found that the Cleaning intervention promoted a
greater perceived safety for both men and women than the Baseline
scene, women's ratings were signicantly lower than men's ratings.
Women reported a negative rating and men reported a slightly
positive rating for the Cleaning intervention. Surprisingly, the
gender disparity was greater for the Cleaning interventions than for
the Baseline scenes. For the Vegetation interventions, women re-
ported a signicantly lower sense of safety than men for three of
four interventions. These ndings indicate that interventions
grounded in the broken windows theory (those that change the
appearance rather than the function of the space) fail to minimize
the gender difference in perceived safety. These ndings are sup-
ported by previous studies showing that women are more fearful of
urban green spaces than men (Sreetheran & van den Bosch, 2014).
In general, men worry about non-sexual assault while women have
serious fears of sexual assault, and that fear can increase their
overall fear of a space (Ferraro, 1996). The Cleaning and Vegetation
Only interventions change the visual appearance of alleys without
increasing public surveillance and visible escape routes. These in-
terventions may not adequately alleviate women's safety concerns
(Cozens, Saville, & Hillier, 2005; Sreetheran & van den Bosch, 2014).
For the Urban Function & Vegetation interventions, however, we
found that, in general, both genders' perceived safety ratings are
similarly positive. This nding suggests that interventions mainly
grounded in the routine activity theory are effective at reducing the
gender difference between men's and women's perceived safety of
urban alleys. Adding facilities or accommodating legitimate urban
functions supports social activity and discourages the presence of
criminals in alleys (Cozens et al., 2005). These interventions may
have a signicant preventative effect on sexual assault, alleviating
women's biggest safety concern. In addition, women tend to feel
safer when they are in the presence of other legitimate citizens
(Koskela & Pain, 2000). This nding is supported by a recent study,
which reported that participating in public activities on campus can
reduce female students' fears (Steinmetz & Austin, 2014).
Nevertheless, the results of the urban functions were mixed.
Women reported a signicantly lower sense of safety than men for
two urban function interventions: Green Art Gallery and Geometric
Green Platform. These interventions received lower ratings than
the other Urban Function interventions from both men and women.
One possible reason is that the green art gallery is a mild, passive
intervention that may not attract many legitimate users and
129
Women
Mean Sub-groups Mean
2.56 Baseline 2.33
3.11 Cleaning 2.77
3.38 Naturalistic Green Wall 3.06
Geometric Green Wall
Naturalist Green Wall& Ground
Green Art Gallery
Geometric Green Platform
3.71 Naturalistic Green Platform 3.40
Green Cycling
Geometric Green Wall& Ground
4.01 Green Shop 3.67
Green Gym
4.29 Naturalist Green Cafe 4.23
Geometric Green Cafe 
Naturalistic Green Park
Geometric Green Park
F (5,10) 106.52, p < 0.001
informal guardians in the alley space, which is one important factor
inuencing women's perceived safety (Cohen & Felson, 1979;
Krenichyn, 2004; Steinmetz & Austin, 2014). The overhead plat-
form may be less preferred by women because it provides an ideal
offensive space for potential offenders to observe, follow, and
attack women (Jacques & Reynald, 2012). Females might feel more
vulnerable and sense a loss of control and privacy with the over-
head platforms (Andrade & Devlin, 2015; Ulrich, 1991).
We found that both women and men perceived interventions
with geometric vegetation as safer than interventions with natu-
ralistic vegetation. A few studies have compared men's and
women's safety perceptions for various styles of urban landscapes
(Mahidin & Maulan, 2012; Parsons, 1995; Yang, Li, Elder, & Wang,
2013). These studies argue that geometric landscapes imply a
stronger sense of stewardship, greater visual accessibility, and less
opportunity for ambush than naturalistic landscapes. Our ndings
can help designers make evidence-based decisions about spatial
forms of landscapes that will help people feel safe. Geometric
vegetation would be a better choice than naturalistic vegetation for
urban public spaces such as back alleys where personal safety is a
primary concern. The style of vegetation may matter less in other
urban spaces, such as parks and plazas, where safety is less of a
concern.
To increase men's and women's sense of safety and to minimize
the gender difference in perceived safety, designers should trans-
form disorderly and functionless alleys into ones that are clean,
physically attractive and that have a clear function that will invite
people's daily use. Cleaning the alley or adding green plants brings
a moderate improvement in safety perceptions for both men and
women, but these interventions are not enough, especially for
women. Legitimate and active urban functions, such as parks,
shops, cycling paths and parking, gyms, and cafe s, should be added
to alleys to make both women and men feel safe. Urban functions
that are less active and socially inviting, such as art gallery and
platform interventions, should not be prioritized because they are
perceived as less safe, especially by women. Vegetation in alleys
should be geometric and formal instead of naturalistic and wild to
increase the sense of safety for both men and women. Ideally, alley
revitalization efforts should add both urban functions and attrac-
tive, geometric vegetation, thereby improving both the appearance
and the daily use of the space.
130 B. Jiang et al. / Journal of Environmental Psychology 51 (2017) 117e131
4.2. Ideas for future research
Participants for this study were mainly young and middle-aged
adults; seniors and children were under-represented. Age may
signicantly affect people's sense of safety. Teenage and young
women in their 20s and 30s are more likely to be victims of sexual
assault than older women (Felson & Cundiff, 2012). Seniors may
lack the physical strength to protect themselves from crime or
attack (Yavuz & Welch, 2010). Getting older might be negatively
associated with women's fear, but positively associated with men's
fear (Felson & Cundiff, 2012; Smith & Torstensson, 1997). In addi-
tion, children may have unique safety concerns in back alleys that
designers should address. Future research should include older
participants and children in order to better represent the entire
populations' safety needs.
We measured perceived safety but didn't include multiple
measures of perceived safety and fear, which limits our under-
standing of this issue. Future research should use a more compre-
hensive list of survey questions. Because participants self-reported
their safety perceptions, it is possible that some participants under-
report or over-report their fear of an environment. Measuring
people's emotional, neurological, and physiological reactions to
back alleys would give a more complete understanding of people's
sense of safety in an environment (Jiang, Chang, & Sullivan, 2014;
Kim, Jeong, Kim et al., 2010; Kim, Jeong, Baek et al., 2010).
Only a small portion of the participants we recruited were vic-
tims of crimes in back alleys, though it is possible that some par-
ticipants chose not to report the crimes. Understanding the safety
perceptions of victims of crimes may help designers choose the
safest types of interventions. It is possible that experience would
have different impacts on women and men as women tend to
ruminate more on personal experience (Johnson & Whisman, 2013)
and generalize fear across time, space, and type of victimization
(Smith & Torstensson, 1997). Future research should pay more
attention to the perceptions of victims of crimes.
The percentage of Chinese and native Hong Kong residents in
the sample is slightly higher than the census population, which
suggests an underpresentation of minorities in this study. Some
ethnic minorities refused to participate in the survey for personal
reasons. Others cannot speak uent English or Chinese, which
made approaching them difcult. Future research should make a
greater effort to recruit more minorities, especially illegal immi-
grants or unemployed workers who are often invisible, and
whose safety concerns also need to be addressed.
Another issue that needs attention is social or cultural differ-
ences in landscape designs and safety perceptions. We carefully
avoided adding landscape features or amenities with distinct cul-
tural meanings, though adding such features or amenities might
increase or decrease perceived safety in some circumstances.
Future research should test this possibility. In addition, social
constructions of gender and social norms may have inuenced our
participants' responses. Similar studies should be conducted in
other countries or regions with different gender and social norms.
5. Conclusion
This study is an initial effort to identify the extent of gender
difference of perceived safety for a wide variety of back alley scenes
and interventions. We seek to minimize this gender difference in
safety perception by proposing interventions that both genders
nd safe. Our ndings demonstrate that interventions grounded in
the broken windows theory, those that clean and green the alleys,
are insufcient in helping people, especially women, feel safe.
However, urban function interventions grounded in the routine
activity theory, interventions that invite social interaction and daily
use, can largely reduce the gender difference by improving
perceived safety for both males and females. In addition, alleys with
geometric vegetation can elicit a greater sense of safety for both
women and men than alleys with naturalistic vegetation. In all,
these ndings provide clear evidence to support the efforts of city
managers, urban planners, landscape architects, and community
associations seeking to create safe and vital back alley environ-
ments for men and women in high-density cities across the world.
Acknowledgements
The study was nancially supported by the Faculty of Archi-
tecture in The University of Hong Kong (HKU). Thanks for advice
from two anonymous reviewers. Thanks for advice from three
scholars in the Division of Landscape Architecture from HKU:
Matthew Pryor, Scott Melbourne, and Suvarna Apte. Thanks for
advice from Dr. William C. Sullivan from the University of Illinois at
Urbana-Champaign and Dr. Chun-yen Chang from the National
Taiwan University. Thanks for enormous support from Dean and
Professor Christopher John Webster of the Faculty of Architecture in
HKU.
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.jenvp.2017.03.012.
References
Andrade, C. C., & Devlin, A. S. (2015). Stress reduction in the hospital room:
Applying Ulrich's theory of supportive design. Journal of Environmental Psy-
chology, 41, 125e134. http://dx.doi.org/10.1016/j.jenvp.2014.12.001.
Cohen, L. E., & Felson, M. (1979). Social change and crime rate trends: A routine
activity approach. American Sociological Review, 44(4), 588e608.
Cozens, P. M., Saville, G., & Hillier, D. (2005). Crime prevention through environ-
mental design (CPTED): A review and modern bibliography. Property Manage-
ment, 23(5), 328e356.
Crime Concern. (2004). People's conceptions of personal security and their concerns
about crime on public transport: Research ndings. London: Department of
Transport.
Ekinsmyth, C. (2002). 16 Feminist methodology Carol Ekinsmyth. In P. Shurmer-
Smith (Ed.), Doing cultural geography (p. 177). Sage Publications.
Felson, R. B. (2014). Back to basics: Gender and the social psychology of agression. In
R. Gartner, & B. Mccarthy (Eds.), The Oxford handbook of gender, sex, and crime
(pp. 78e98). New York: Oxford University Press.
Felson, R. B., & Cundiff, P. R. (2012). Age and sexual assault during robberies. Evo-
lution and Human Behavior, 33(1), 10e16. http://dx.doi.org/10.1016/j.
evolhumbehav.2011.04.002.
Ferraro, K. F. (1996). Women's fear of victimization: Shadow of sexual assault?
Social Forces, 75(2), 667e690. http://dx.doi.org/10.1093/sf/75.2.667.
Foster, C., Hillsdon, M., & Thorogood, M. (2004). Environmental perceptions and
walking in English adults. Journal of Epidemiology and Community Health, 58(11),
924e928 (1979-).
Fujita, K., Fujiwara, Y., Kumagai, S., Watanabe, S., Yoshida, Y., Motohashi, Y., et al.
(2004). The frequency of going outdoors, and physical, psychological and social
functioning among community-dwelling older adults. [Nihon koshu eisei zasshi]
Japanese Journal of Public Health, 51(3), 168e180.
Gilchrist, E., Bannister, J., Ditton, J., & Farral, S. (2008). Women and the fear of crime
challenge the accepted stereitype. In K. Evans, & J. Jamieson (Eds.), Gender and
crime: A reader (pp. 77e86). New York: Open University Press.
Gover, A. R., Tomsich, E. A., Jennings, W. G., & Higgins, G. E. (2011). An exploratory
study on perceptions of safety, fear of crime, and victimization experiences
among faculty and staff at an urban university: A focus on gender. Criminal
Justice Studies, 24(1), 37e55. http://dx.doi.org/10.1080/1478601X.2011.544193.
Ho, C.-h., Sasidharan, V., Elmendorf, W., Willits, F. K., Graefe, A., & Godbey, G. (2005).
Gender and ethnic variations in urban park preferences, visitation, and
perceived benets. Journal of Leisure Research, 37(3), 281e306.
Hong Kong Police Force. (2014). Hong Kong police review, 2014. Retrieved 8/16, 2016,
from http://www.police.gov.hk/info/review/2014/tc/index.html#111/z.
Jacques, S., & Reynald, D. M. (2012). The offenders' perspective on Prevention:
Guarding against victimization and law enforcement. Journal of Research in
Crime and Delinquency, 49(2), 269e294. http://dx.doi.org/10.1177/
0022427811408433.
Jiang, B., Chang, C.-Y., & Sullivan, W. C. (2014). A dose of nature: Tree cover, stress
reduction, and gender differences. Landscape and Urban Planning, 132(0),
26e36. http://dx.doi.org/10.1016/j.landurbplan.2014.08.005.
 B. Jiang et al. / Journal of Environmental Psychology 51 (2017) 117e131
Jiang, B., Larsen, L., Deal, B., & Sullivan, W. C. (2015). A doseeresponse curve
describing the relationship between tree cover density and landscape prefer-
ence. Landscape and Urban Planning, 139(0), 16e25. http://dx.doi.org/10.1016/j.
landurbplan.2015.02.018.
Johnson, D. P., & Whisman, M. A. (2013). Gender differences in rumination: A meta-
analysis. Personality and Individual Differences, 55(4), 367e374. http://dx.doi.
org/10.1016/j.paid.2013.03.019.
Jorgensen, L. J., Ellis, G. D., & Ruddell, E. (2013). Fear perceptions in public parks.
Environment and Behavior, 45(7), 803e820. http://dx.doi.org/10.1177/
0013916512446334.
Junker, B., & Buchecker, M. (2008). Aesthetic preferences versus ecological objec-
tives in river restorations. Landscape and Urban Planning, 85(3e4), 141e154.
http://dx.doi.org/10.1016/j.landurbplan.2007.11.002.
Keizer, K., Lindenberg, S., & Steg, L. (2008). The spreading of disorder. Science,
322(5908), 1681e1685.
Kim, T.-H., Jeong, G.-W., Baek, H.-S., Kim, G.-W., Sundaram, T., Kang, H.-K., et al.
(2010). Human brain activation in response to visual stimulation with rural and
urban scenery pictures: A functional magnetic resonance imaging study. Science
of The Total Environment, 408(12), 2600e2607. http://dx.doi.org/10.1016/j.
scitotenv.2010.02.025.
Kim, G. W., Jeong, G. W., Kim, T. H., Baek, H. S., Oh, S. K.Kang, H. K., et al., (2010).
Functional neuroanatomy associated with natural and urban scenic views in the
human brain: 3.0T functional MR imaging. Korean Journal of Radiology, 11(5),
507e513. http://dx.doi.org/10.3348/kjr.2010.11.5.507.
Koskela, H., & Pain, R. (2000). Revisiting fear and place: women's fear of attack and
the built environment. Geoforum, 31(2), 269e280. http://dx.doi.org/10.1016/
S0016-7185(99)00033-0.
Krenichyn, K. L. (2004). Women, physical activity, and the urban outdoor environment
a study of Prospect Park, Brooklyn, New York. Thesis (Ph.D.). City University of
New York, 2004.
Kuo, F. E., Bacaicoa, M., & Sullivan, W. C. (1998). Transforming inner-city landscapes
- trees, sense of safety, and preference. Environment and Behavior, 30(1), 28e59.
Laurie, N., Claire, D., Holloway, S. L., & Smith, F. (2000). Geographies of new femi-
ninities. Harlow: Addison Wesley Longman.
Li, X., Zhang, C., Li, W., Ricard, R., Meng, Q., & Zhang, W. (2015). Assessing street-
level urban greenery using google street view and a modied green view in-
dex. Urban Forestry & Urban Greening, 14(3), 675e685. http://dx.doi.org/10.
1016/j.ufug.2015.06.006.
Maas, J., Spreeuwenberg, P., Winsum-Westra, M. V., Verheij, R. A., de Vries, S., &
Groenewegen, P. P. (2009). Is green space in the living environment associated
with people's feelings of social safety? Environment and Planning, A, 41(7),
1763e1777.
Madge, C. (1997). Public parks and the geography of fear. Tijdschrift voor Econo-
mische en Sociale Geograe, 88(3), 237e250.
Mahidin, A. M. M., & Maulan, S. (2012). Understanding children preferences of
natural environment as a start for environmental sustainability. Procedia - Social
and Behavioral Sciences, 38, 324e333. http://dx.doi.org/10.1016/j.sbspro.2012.
03.354.
Mazey, M. E. (1983). Her space, her place : A geography of women. Washington, D.C.
Washington, D.C.: Association of American Geographers.
Mazey, M. E., & Lee, D. R. (1983). Her space, her place: A geography of women.
Washington: Association of American Geographers.
Meghan, E. H., Marcus, F., & Brandon, C. W. (2013). The capable guardian in routine
activities theory: A theoretical and conceptual reappraisal. Crime Prevention and
Community Safety, 15(1), 65. http://dx.doi.org/10.1057/cpcs.2012.14.
Nowakowski, T., Myn  czak, M., Jodejko-Pietruczuk, A., & Werbin  ska-
Wojciechowska, S. (Eds.). (2014). Safety and reliability: Methodology and appli-
cations. CRC Press.
131
O'Brien, E. A. (2005). Publics* and woodlands in England: Well-being, local identity,
social learning, conict and management. Forestry, 78(4), 321e336. http://
dx.doi.org/10.1093/forestry/cpi042.
Pain, R. H. (1997). Social geographies of women's fear of crime. Transactions of the
Institute of British Geographers, 22(2), 231e244.
Pain, R. (2001). Gender, race, age and fear in the city. Urban Studies (Routledge),
38(5/6), 899e913. http://dx.doi.org/10.1080/00420980120046590.
Parsons, R. (1995). Conict between ecological sustainability and environmental
aesthetics - conundrum, canard or curiosity. Landscape and Urban Planning,
32(3), 227e244.
Raanaas, R. K., Patil, G. G., & Hartig, T. (2012). Health benets of a view of nature
through the window: A quasi-experimental study of patients in a residential
rehabilitation center. Clinical Rehabilitation, 26(1), 21e32.
Richardson, E. A., & Mitchell, R. (2010). Gender differences in relationships between
urban green space and health in the United Kingdom. Social Science and Med-
icine, 71(3), 568e575.
Seymour, M., Wolch, J., Reynolds, K. D., & Bradbury, H. (2010). Resident perceptions
of urban alleys and alley greening. Applied Geography, 30(3), 380e393. http://
dx.doi.org/10.1016/j.apgeog.2009.11.002.
Skogan, W. G. (2012). Disorder and crime. In C. W. D. P. F. Brandon (Ed.), The Oxford
handbook of crime prevention (pp. 173e188). New York: Oxford University Press.
Smith, W. R., & Torstensson, M. (1997). Gender differences in risk perception and
neutralizing fear of crime. Toward Resolving the Paradoxes, 37(4), 608e634.
Snedker, K. A. (2012). Explaining the gender difference in fear of Crime: Assess-
ments of risk and vulnerability among New York city residents. Feminist
Criminology, 7(2), 75e111. http://dx.doi.org/10.1177/1557085111424405.
Sreetheran, M., & van den Bosch, C. C. K. (2014). A socio-ecological exploration of
fear of crime in urban green spaces - a systematic review. Urban Forestry and
Urban Greening, 13(1), 1e18.
Steinmetz, N. M., & Austin, D. M. (2014). Fear of criminal victimization on a college
campus: A visual and survey analysis of location and demographic factors.
American Journal of Criminal Justice : AJCJ, 39(3), 511e537. http://dx.doi.org/10.
1007/s12103-013-9227-1.
Sullivan, W. C., & Lovell, S. T. (2006). Improving the visual quality of commercial
development at the ruraleurban fringe. Landscape and Urban Planning, 77(1e2),
152e166. http://dx.doi.org/10.1016/j.landurbplan.2005.01.008.
Sweet, E. L., & Ortiz Escalante, S. (2015). Bringing bodies into planning: Visceral
methods, fear and gender violence. Urban Studies, 52(10), 1826e1845. http://
dx.doi.org/10.1177/0042098014541157.
Todorova, A., Asakawa, S., & Aikoh, T. (2004). Preferences for and attitudes towards
street owers and trees in Sapporo, Japan. Landscape and Urban Planning, 69(4),
403e416.
Ulrich, R. S. (1991). Effects of interior design on wellness: Theory and recent sci-
entic research. J Health Care Inter Des, 3, 97e109.
Welsh, B. C., Braga, A. A., & Bruinsma, G. J. N. (2015). Reimagining broken windows:
From theory to policy. Journal of Research in Crime and Delinquency, 52(4),
447e463. http://dx.doi.org/10.1177/0022427815581399.
Wilson, J. Q., & Kelling, G. L. (1982). Broken windows. Atlantic Monthly, 249(3),
29e38.
Wolch, J., Newell, J., Seymour, M., Huang, H. B., Reynolds, K., & Mapes, J. (2010). The
forgotten and the future: Reclaiming back alleys for a sustainable city. Envi-
ronment & Planning A, 42(12), 2874e2896.
Yang, B., Li, S., Elder, B. R., & Wang, Z. (2013). Community-planning approaches and
residents' perceived safety: A landscape analysis of park design in the wood-
lands, Texas. Journal of Architectural and Planning Research, 30(4), 311e327.
Yavuz, N., & Welch, E. W. (2010). Addressing fear of crime in public space: Gender
differences in reaction to safety measures in train transit. Urban studies (Edin-
burgh, Scotland), 47(12), 2491.