Contents lists available at ScienceDirect

Urban Forestry & Urban Greening

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

An exploratory study of perceived safety in a neighborhood park usingimmersive virtual environments

Perver K. Barana,b,⁎, Payam Tabrizianc, Yujia Zhaid, Jordan W. Smithe, Myron F. Floydf

a Center for Geospatial Analytics, North Carolina State University, Jordan Hall 5106, 2800 Faucette Dr., Raleigh, NC, 27695, USAb College of Design, North Carolina State University, Jordan Hall 5106, 2800 Faucette Dr., Raleigh, NC, 27695, USAc College of Design, North Carolina State University, Brooks Hall, Campus Box 7701, Raleigh, NC, 27695, USAd College of Architecture and Urban Planning, Big Data and Urban Spatial Analytics LAB, Tongji University, No. 1239 Siping Road, Yangpu District, Shanghai, 200092,Chinae Institute of Outdoor Recreation and Tourism, Department of Environment and Society, Utah State University, BNR 289, 5215 Old Main Hill, Logan, UT, USAfDepartment of Parks, Recreation and Tourism Management, North Carolina State University, Biltmore Hall 4008C, 2820 Faucette Dr., Raleigh, NC, 27695, USA

A R T I C L E I N F O

Keywords:EnclosureImmersive virtual environmentsPerceived safetyUrban parkVegetation

A B S T R A C T

Spatial configuration and physical characteristics of landscape features can strongly influence perceptions of fearand danger. This study examined how situational characteristics, particularly spatial enclosure shaped by sur-rounding vegetation, are related to perception of safety in a park setting. Study stimuli involved eight 360°immersive virtual environments (IVE) representing low, medium, and high spatial enclosure situations based onthe degree of visual and locomotive permeability shaped by the physical arrangement of vegetation. Forty-eightstudents experienced the IVEs wearing a head mounted display device and then indicated on a 5-point scale howsafe they would feel walking alone in that location during the day. Immediately after rating each IVE, partici-pants indicated the main reasons they would feel either safe or unsafe in that particular location. Analysis resultsindicated that subjects perceived high enclosure environments as significantly less safe than medium and lowenclosure environments. In addition to enclosure formed by vegetation, attributes that contributed to perceptionof safety were presence of non-threatening people and paths. Results indicated that gender differences in per-ceived safety were significant for the high and medium enclosed environments only. Study findings would allowurban planners and park managers to better understand how the spatial characteristics of existing or plannedurban greenspace are likely to influence perceived safety and consequently use patterns and the attainment ofsocial and psychological benefits provided by urban parks. Such an understanding can help generate evidence-based guidelines for improving safety while preserving desired aesthetic and ecological properties of the land-scape.

1. Introduction

Urban parks provide space to engage in leisure-time (Giles-Corti andDonovan, 2002) and utilitarian physical activities (Zlot and Schmid,2005). They support social well-being (Tinsley et al., 2002) and enableusers to have positive psychological experiences (Nordh et al., 2009).Typically, urban parks are available without charge to individual usersand thus are particularly important in enabling “active living” acrossdiverse population groups. In addition to size, attraction, and accessi-bility (Baran et al., 2014), use of urban parks is highly dependent uponhow safe users feel there. Perception of danger or feelings of fear likelyinfluence individuals’ preferences (Herzog and Kutzli, 2002) and

discourage use (Madge, 1997; Molnar et al., 2004). This may compro-mise the ability of parks to facilitate positive experiences and limit theiroptimal utilization (Giles-Corti and Donovan, 2002; Gatersleben andAndrews, 2013).

A number of studies have focused on perceived safety in relation tospatial attributes of landscapes in a variety of open spaces, specificallyin forest and urban settings (e.g., alleys, college campus) (Fisher andNasar, 1992; Herzog and Miller, 1998; Herzog and Kutzli, 2002; Herzogand Kropscott, 2004; Chiang et al., 2014). This body of research in-dicates that, in general, people prefer landscapes that are open and offera wide view of the surroundings. Enclosed spaces tend to evoke feelingsof insecurity and fear (Herzog and Kutzli, 2002; Stamps, 2005a; Skår,

https://doi.org/10.1016/j.ufug.2018.08.009Received 17 December 2017; Received in revised form 2 August 2018; Accepted 13 August 2018

⁎ Corresponding author.E-mail addresses: kperver@ncsu.edu (P.K. Baran), ptabriz@ncsu.edu (P. Tabrizian), zhai@tongji.edu.cn (Y. Zhai), jordan.smith@usu.edu (J.W. Smith),

mffloyd@ncsu.edu (M.F. Floyd).

Urban Forestry & Urban Greening 35 (2018) 72–81

Available online 16 August 20181618-8667/ © 2018 Elsevier GmbH. All rights reserved.

T

2010). In enclosed spaces, obstructed views are associated with higherlevels of anticipated threats, while the lack of visible horizontal area(i.e., limited ground surface that can been seen around an observer),can be interpreted as restricting movement and limiting individuals’ability to escape in case of danger (Stamps, 2010a 2012). For urbanlandscapes (e.g., college campus), a small body of research suggestsbuildings, walls, or vegetation that could conceal potential offenders(Fisher and Nasar, 1992,1995) may be perceived as threatening. Re-search on safety in urban parks is limited (Madge, 1997; Jorgensenet al., 2012). Indeed, Jansson and colleagues (Jansson et al., 2013)highlighted the need for further research to understand the role ofphysical environment, particularly vegetation’s appearance in a spatialcontext, on perceived safety in urban green spaces such as parks andresidential areas. To address this knowledge gap, the present studyoffers an examination of how different situational characteristics, par-ticularly spatial enclosure shaped by the arrangement of vegetation,play a role in perception of safety in a park setting. We utilized Im-mersive Virtual Environment (IVE) technology to create realistic re-presentations of a range of park settings and compared perceptions ofsafety across varying levels of spatial enclosure.

1.1. Landscape and perception of safety

The majority of studies focusing on a physical environment’s role inperceived safety have been informed by Appleton’s (1975) prospect-refuge theory (Fisher and Nasar, 1992; Loewen et al., 1993; Herzog andKirk, 2005) and Stamps’ (2005b, 2010a) permeability theory (Stamps,2005a). According to Appleton’s theory, people naturally prefer pro-spect and refuge, where prospect is defined as an open view that enablesa person to see what is ahead and refuge is the presence of features thatafford protection (Appleton, 1975). Fisher and Nasar (1992) com-plemented Appleton’s theory by developing a general typology forevaluating individuals’ perceptions of safety. They found people favor asecond level of prospect and refuge, which occurs when a person ob-serves a place ahead offering prospect (an open view) or refuge (a placeof protection). In other words, the degree to which a space affordsfleeing a potential attack plays a key role in individuals’ perceptions ofsafety.

Permeability theory suggests that judgments of safety and dangerrun parallel to perceptions of spaciousness and enclosure (Stamps,2005b, 2010a). Research on permeability theory rests on the notionthat higher degrees of enclosure imply possible threats and conceal-ment opportunities for an offender; therefore, highly enclosed locationsevoke a sense of fear (Nasar et al., 1993). Among the multiple spatialindicators used to test permeability theory, visual and locomotivepermeability are believed to be the most important (Stamps, 2005b,2010a, 2011b, 2012). Visual permeability refers to the degree to whichan individual can see the features of an environment without obstruc-tion, while locomotive permeability refers to the ability to movethrough an environment (Stamps, 2010a).

A number of studies have examined the relationships betweenspatial enclosure and perceptions of safety and/or danger (e.g., Herzogand Miller, 1998; Herzog and Chernick, 2000; Stamps, 2005a). Forinstance, Herzog and Miller (1998) exposed respondents to series ofunmodified photographs depicting 18 alleys and 18 field/forest scenesto observe variation in preference, mystery, danger, openness, andcurvature; they reported perceived danger as highly correlated withperceived openness. Similarly, Herzog and Chernick (2000) examinedperceived safety and danger using unmodified photographs of 48 urbanand field/forest settings and found a strong negative correlation(r=−.72) between perceived openness and perceived danger. Like-wise, Stamps (2005a) had participants rate 21 slides of artificial scenesrepresenting different environments in three Greek cities and found astrong correlation (r= .82) between perceived enclosure and feelingsafe in an environment. Relying on responses to site plans, on-siteratings and observed behavior in a college campus, Fisher and Nasar

(1992) found that fear of crime was highest in enclosed areas withrefuge potential for offenders.

Stamps (2012) expanded permeability theory by studying the effectof proximate and distal boundaries on perceived enclosure. He arguedthat if perceived enclosure mediates safety judgments by indicating thedistance to possible threats, the proximate boundary should havestronger effects on perceived spaciousness or perceived enclosure thandistal boundaries. In a specific location, the distance between an ob-server and proximate/distal boundaries is tantamount to the size ofhorizontal area that one perceives ahead (i.e., the sight distance,Troped et al., 2006). The form of landscape elements, e.g., vegetation,can influence sight distance. For instance, in the case of bending orhighly sinuous paths, although the vegetation along the approachingcurve will appear to be distant from the observer, it will form a distalboundary that may influence perceived enclosure and consequently,perceived safety.

Across both urban and more natural landscapes, empirical evidencesuggests strong associations between landscape attributes and per-ceived safety (A. Jorgensen et al., 2002; Jansson et al., 2013; Chianget al., 2014). The scale, type, and density and foliage of vegetation, aswell as the form of landscape elements, can moderate the sense of en-closure or spaciousness. Height of vegetation defines the proximate anddistal boundaries in a setting and influences sense of enclosure (Stamps,2012). A row of shrubs or trees with dense lower limbs can form a solidboundary and highly influence perceived enclosure, whereas trees withhigher limbs can form more permeable boundaries (Stamps, 2012). Forinstance, Nasar et al. (1993) asked college students to mark on a mapareas they avoid because they felt them to be unsafe. They found apositive relationship between fear of crime across a university campusand dense growth of shrubs and trees with low hanging limbs. Utilizinga mail questionnaire and semi-structured interviews, Jorgensen et al.(2007) examined sense of fear in a residential UK neighborhood. Theyfound shrubs were considered to be visibility barriers, potential hidingplaces for assailants, and sanctuaries for incivilities. Similarly, studieson forest trails (Chiang et al., 2014) and field/forest settings (Herzogand Kutzli, 2002) suggest shrubs and other types of ground cover mayhinder locomotive permeability and influence perception of safety.

The form of paths and trails within a park may influence perceivedspaciousness and perception of safety, especially when considered inrelation to other elements that form visual boundaries (e.g., trees,fences, walls, etc.). Paths and trails define horizontal areas that arehighly correlated with perceived spaciousness (Stamps, 2011a) andlocomotive permeability (Stamps, 2010a). Wide paths surrounded bypermeable vegetation are likely to be perceived as more spacious thannarrow paths with elevated boundaries. In addition, the elongation of apath, defined as the ratio of the visible length to its width, may benegatively associated with perceived spaciousness (Stamps, 2011a).Winding park paths and natural trails can increase the sense of fear,especially when surrounded by dense vegetation. Moreover, the pathdemarcation and surface material influence the landscape legibility,another factor likely to influence perceived safety. Landscape legibilityis defined as the extent to which the environment provide cues for or-ientation and way finding (Kaplan and Kaplan, 1989). Unpaved trails,poorly defined paths, or undefined areas within urban parks may con-fuse and stress people. In addition, research indicates the presence ofother people can be an important factor in mediating perceptions ofsafety (L. J. Jorgensen et al., 2012). In general, the presence of othernon-threatening people increases perception of safety. Finally, researchon gender’s role in safety perception indicates that, in general, womentend to perceive greater safety risk relative to men (Madge, 1997; W. R.Smith et al., 2001), and that they are more fearful in green spaces thantheir male counterparts (Maruthaveeran and van den Bosch, 2014).Since, a higher degree of enclosure leads to higher perceived risk(Stamps, 2005b 2010a) we expected there might be some differences inperceived safety between men and women relative to level of enclosure.However, extant research has not examined if differences in perceived

P.K. Baran et al. Urban Forestry & Urban Greening 35 (2018) 72–81

73

safety between men and women are consistent across different levels ofenclosure.

1.2. Immersive virtual environments (IVE) as mode of presentation

Research examining mechanisms and correlates of perceived safetyhas utilized a variety of methods (Maruthaveeran and van den Bosch,2014). Studies utilizing correlational design have employed surveysand/or interviews to elicit perceptions of safety of a particular area(Madge, 1997; A. Jorgensen et al., 2007) or have asked respondents tomark on a map areas that are perceived as unsafe (Nasar et al., 1993).Surveys are potent for elucidating the multiple expected and un-expected determinants of perceived safety. However, since evaluationsare highly moderated by individual’s knowledge or impressions ofcrime events or other incivilities associated with environments(Garofalo, 1981; Nasar and Jones, 1997; Jorgensen et al., 2012), it isoften challenging to draw inferences about specific spatial attributesthat contribute to perceived safety via a survey. Although studies takingparticipants to a site or to a variety of sites for on-site experience andmeasurement (Fisher and Nasar, 1992; Nasar and Jones, 1997) addressthis shortcoming by providing high external and ecological validity,this method is relatively difficult and expensive to employ(Maruthaveeran and van den Bosch, 2014). Alternatively, simulationmethods have proven to be effective in creating environmental stimuliin research on perceived safety.

Most research using simulation has relied on unmodified images ofreal settings presented as two-dimensional photographs/slides (Herzogand Miller, 1998; Herzog and Chernick, 2000; Herzog and Kutzli, 2002;Herzog and Kirk, 2005; Jorgensen et al., 2012; Chiang et al., 2014). Ahandful of studies has also utilized manipulated real-world photographs(A. Jorgensen et al., 2002) or artificial scenes (Stamps, 2005a). Al-though static two-dimensional images appear to have high validity inrepresenting environments and capturing aesthetic judgments (Stamps,2010b), forming an aesthetic evaluation from a single image is verydifferent from evaluating entire landscapes as they are actually ex-perienced in real life (Pearson and Craig, 2014). In relation to psy-chological processes, and particularly evaluations of safety, J. W. Smith(2015) argued there might be a considerable gap between “psycholo-gical processes of fear-elicitation” (p. 11,491) in a human’s day-to-dayexperience in comparison to feelings induced by still images and/orcomputer-generated slides. Heft (2007) also maintains that affectiveresponses to landscape experience may be highly influenced by thedegree of ‘presence’ individuals feel when evaluating a landscape.Presence is the subjective experience of being in one place or en-vironment, while being physically situated in another (Slater, 2009).

In response to these shortcomings, research considering environ-mental cues in fear elicitation has begun to explore more interactiveand simulated modes of representation (L. J. Jorgensen et al., 2012).For example, in a study examining the impact of prospect, refuge, andescape features on fear in alleys, Wang and Taylor (2006) took picturesof what lay straight ahead, 45° to the left, and 45° to the right at dif-ferent points along an alley. Then they stitched the pictures together tocreate a roughly 180° panorama at each point. The panoramas wereexperienced in a sequence as two-dimensional slides to simulatewalking through an alley. Similarly, Cozens and his colleagues ex-amined personal safety at railway stations by simulating a walk-through using virtual reality (VR) technology (Cozens et al., 2003).They stitched together a number of 360° panoramas taken at variouspoints around the railway stations to create a VR “walk-through” pre-sentation. Then, the VR presentation was used with focus groups tounderstand individuals’ personal safety concerns.

IVEs–as an alternative to conventional presentation methods–evokea high degree of realism and/or presence (Slater, 2009; Kim et al.,2012; Heydarian et al., 2015; J. W. Smith, 2015). Especially whendisplayed through a Head Mounted Display (HMD) device (e.g., OculusRift) or a CAVE™ system, IVEs ‘surround’ the user with a continuous

stream of visual stimuli to create the perception they are present withinand interacting with an environment (Sanchez-Vives and Slater, 2005).Compared to 2D representations, IVEs enable the user to explore theenvironment from different angles and perspectives (Passig et al.,2016), not only what is seen straight ahead. The more similar an en-vironmental representation becomes to the real-world environment itmimics, the more realistic users' responses are expected to be (Kuligaet al., 2015). Research has found that participants have similar per-ception of a real target-environment and its simulated IVE reproduc-tion, implying high ecological validity of IVEs in representing buildingsand urban spaces (Heydarian et al., 2015; Luigi et al., 2015).

IVEs have already been utilized in a variety of research areas, in-cluding in the study of pedestrians’ route choice (Natapov and Fisher-Gewirtzman, 2016), psychological and physiological responses tostressful situations (Crescentini et al., 2016), effects of sounds of natureon stress recovery (Annerstedt et al., 2013), the effect of the designcharacteristics of an indoor environment (open vs. closed rooms) on thephysiological stress response (Fich et al., 2014), and perception of re-storative potential of natural and built environments (Tabrizian et al.,2018), and as a medium in psychological treatment for mental healthproblems (Valmaggia et al., 2016). IVE technology has also been in-creasingly utilized in architecture and urban planning practice andeducation (Drettakis et al., 2007; Schnabel, 2014).

The continuous experience of an environment offered by IVEs maybe particularly helpful in studying spatial enclosure. Stamps (2005a)defines enclosure as a “…region of three dimensional space throughwhich the difficulty of passage by something is affected, and the degreeof that difficulty will depend on what is passing through…” (p.106).IVEs displayed through HMDs enable respondents to view all facets of a360° environment (e.g., rearview, overhead) (Naceri et al., 2010; Passiget al., 2016), which is important in examining perceived safety.

1.3. Study overview

In the present study, we examine how different situational char-acteristics, particularly spatial enclosure shaped by surrounding vege-tation, and gender relate to individuals’ perception of safety usingrealistic representations of actual park settings displayed to study par-ticipants through IVEs via a HMD device. Following Stamps(2005b,2010a,2012) permeability theory, we used visual and locomo-tive permeability to develop operational measures for level of spatialenclosure relevant for 360° panoramas used in IVEs.

In view of the extant research discussed above, we address thefollowing hypotheses: H1. Spatial enclosure has a negative impact onperceived safety; i.e., less enclosed environments are perceived as saferthan more highly enclosed environments. H2. Perceived safety is dif-ferent for males and females; i.e., females perceive environments as lesssafe than males. H3. There are differences in perceived safety betweenmales and females relative to level of enclosure, i.e., enclosure levelinteracts with gender to affect perceived safety. The difference in per-ceived safety between males and females is larger in more highly en-closed environments than in less enclosed environments.

2. Methods

2.1. Participants

The participants in the study were 48 undergraduate students fromthe Department of Parks, Recreation and Tourism Management at auniversity in the southeastern United States: 19 (40%) male and 29(60%) female students. Participation in the study was voluntary.

2.2. Study Site and Stimuli

The study site for this research is Fred Fletcher Park in Raleigh, NC.Built in 1981, this urban park has an area of 8.66 ha. The park houses

P.K. Baran et al. Urban Forestry & Urban Greening 35 (2018) 72–81

74

multiple facilities including a water garden, a picnic shelter, a basket-ball court, tennis courts, a softball field, a multipurpose field and achildren’s playground. The study focuses on the central area of the park(6.03 ha), which includes walking paths cutting across various vegeta-tion typologies and densities offering different degrees of spatial en-closure (Fig. 1). Because of its location in relation to the universitycampus, we assumed participants were not familiar with either thestudy park or the park’s neighborhood.

Study stimuli included eight 360° IVEs representing environmentswith different levels of enclosure within the study area. First, through aseries of site audits, eight locations exhibiting various levels of spatialenclosure were identified. We developed a method to measure spatialenclosure using the concepts of visual permeability and locomotivepermeability (Stamps, 2005b, 2010a, 2012). The first step in this pro-cess involved delineating visual barriers. Through three rounds of fieldaudits (conducted by two Ph.D. students and a faculty member) guidedby aerial photography, visual barriers across the entire study area wereidentified. Visual barriers were defined as vegetation and built featuresthat entirely blocked the vision of a 1.76m tall human (average U.S.male) observer situated at the center of the location. Vegetationmeeting this criterion was mostly elevated shrubs and evergreens withdense low limbs. These barriers were then digitized over the aerialphotography in ArcGIS.

The next step involved measuring level of enclosure for each of theeight locations. Since IVEs present individuals with an entirely im-mersive experience, measurements required a protocol that accountedfor the full 360° view. Guided by Stamps’ (2011b) work this was doneby measuring the length of visible arc (representing both visual andlocomotive permeability) and the size of visible horizontal area (re-presenting locomotive permeability) of a circular boundary around thecenter of each location. Preliminary analysis revealed that a 15-meterradius would adequately reflect the enclosure differences between theeight locations. To calculate level of enclosure, the analysis boundarywas intersected with the lines of sight (360° Isovists) radiating from thecenter of each location and constrained by the visibility barriers. Visiblearc was then assessed by calculating the visible proportion of theboundary’s circumference and the visible area was assessed by calcu-lating the proportion of the visible horizontal area (Fig. 2). Thesemeasures produced three distinct enclosure levels for the eight loca-tions. Four locations were classified as low-enclosed, two locations weremedium-enclosed, and two locations were high-enclosed (Table 1).

For each location, an array of 54 images (9 images in the row, 6images in column) were acquired using a digital SLR camera fitted

within a GigaPan EPIC Pro robotic controller mounted atop a tripod. Allimages were taken in December 2013 at noontime (12:00 pm) undercloudy conditions to minimize shading effects. The images were stit-ched together to create a full 360° equirectangular image and convertedinto IVEs through a process known as cube mapping (J. W. Smith,2015). Finally, the virtual environments were rendered and displayedon HMDs (Oculus Rift DK1) through virtual reality software (WorldVizard software toolkit 4, 2012).

2.3. Experimental procedure

After consenting to participate, subjects received instructions forviewing the IVEs. The HMD device and the IVE software were cali-brated to adjust for each participant’s head size, field of view, andheight. Each participant viewed the eight IVEs (Fig. 3) via the HMDdevice. To control for order effect (Brown and Daniel, 1987), the IVEswere displayed in a random order for each participant (Cheng, 2007).The randomization of IVEs was achieved through a “shuffling” process,which was programmed into the experiment’s Python script executedby the virtual reality software (World Vizard software toolkit 4, 2012).Participants were instructed to turn around in place to experience theentire environment. After viewing each IVE, participants were asked torate how safe they would feel walking alone in the place during daytimeusing a 5-point bipolar rating scale, with 1 being “very unsafe,” 5 being“very safe,” and 3 being a neutral midpoint. In addition, immediatelyafter rating each IVE, participants were asked to indicate the mainreasons they would feel either safe or unsafe in that particular location.The participants were immersed in the virtual environments throughoutthe entire data collection phase, allowing them to explore the en-vironment while responding to the questions. The duration of viewingeach environment was not limited and respondents spent about 1.5 minto view each IVE and respond to the questions. A research assistantrecorded responses and the participant’s gender on a hard copy form.The entire procedure took about 20min. Data were collected over threeweeks in April 2014. The study protocol was approved by the uni-versity’s Institutional Review Board.

2.4. Analyses

The analysis was performed in four steps. First, descriptive statisticsfor perceived safety scores were calculated. Second, a two-way repeatedmeasures ANOVA was used to assess the effects of enclosure level (high,medium, and low) and gender (male, female) on perceived safety as

Fig. 1. Study park and study site boundaries.

P.K. Baran et al. Urban Forestry & Urban Greening 35 (2018) 72–81

75

well as test for interaction effects. Tukey’s HSD multiple comparisontests were performed to further evaluate pairwise differences for per-ceived safety scores across enclosure levels and between males andfemales. Third, participants’ responses regarding reasons for feeling safeor unsafe in a particular location were transcribed, coded, and classifiedinto themes or categories. Finally, Pearson correlation coefficients werecalculated to examine associations between frequency of the attributesmentioned for each environment and perceived safety. Analyses wereperformed with R studio, with the exception of the correlation analysis,which was performed with SPSS version 22.0.

3. Results

Table 2 shows descriptive statistics for perceived safety for the eightenvironments for all participants and by gender. The lowest mean va-lues for perceived safety (M=3.08, M=3.25) were observed for thetwo most enclosed environments (Environments 7 and 8). The meanperceived safety values for the two medium enclosed environmentswere 4.27 and 4.35 and ranged from 4.25 to 4.79 for the four lowenclosed environments. One of the four environments with the lowestlevel of enclosure (Environment 5) was perceived as the safest(M=4.79). This was the only environment that had a view of a play-ground where a woman and a child were playing, as well as the onlyenvironment where mean perceived safety was the same for both malesand females. For all other environments, mean perceived safety scoreswere lower for females relative to males.

ANOVA results indicated a main effect of enclosure, F(1.75,80.72)= 133.62, p < .001, on perceived safety (Table 3). Tukey’s HSDpost-hoc analysis (Table 4) indicated perceived safety for high

enclosure environments was significantly lower than for medium en-closure environments, t(92) = −13.28, p < .001, and for low en-closure environments, t(92) =−14.90, p < .001. However, perceivedsafety did not differ significantly between low and medium enclosureenvironments, t(92)= 1.63, p= .240. The main effect of genderyielded an F-ratio (146) of 7.80 which was significant (p= .008)(Table 3). Additionally, Tukey’s HSD test indicated the mean perceivedsafety score was significantly higher for males than for females, t(46) =−2.80, p< .001 (Table 4). The interaction effect of gender and en-closure level was non-significant, F(1.75, 80.72)= 2.28, p= .120(Table 3). However, further analysis of pairwise differences for theenclosure and gender interaction (Table 5) indicated significant per-ceived safety differences between genders in high enclosure environ-ments, t(84.36) = −2.99, p= .004, and medium enclosure environ-ments, t(84.36) = −2.86, p= .005, whereas the gender difference forthe low enclosure environments was not significant, t(84.36) = −1.20,p= .232.

Qualitative analyses of respondents’ main reasons for feeling safe orunsafe in a particular environment yielded 16 themes or categories.Three of these categories were related to the spatial attributes (i.e.,“open/empty/see everything,” “enclosed/not see,” and “trees/bushes”).The remaining 13 categories reflected ancillary themes, such as“benches,” “no benches,” “next to road,” etc. The most commonlymentioned reasons for feeling either safe or unsafe were "open/empty/see everything” (109 times), presence of “house/building” (103 times),presence of “people” or potential for presence of people (e.g., “peoplemight be around,” “people may come”, etc.) (81 times), “enclosed/cannot see” (66 times), presence of “pathways/trails” (65 times), andpresence of "trees/bushes" (54 times) (see Table 6).

Fig. 2. Mapped indicators of degree of enclosure: Examples for low enclosure, medium enclosure and high enclosure environments.

Table 1Descriptive Statistics for Enclosure Indicators including Visible Area and Visible Arc by Each Environment.

Enclosure level Environment number Visible area Visible arc

Area (sq m) Proportion (%) Length (m) Proportion (%)

Low 1 692.7 98 86.7 923 657.4 93 82.0 875 706.9 100 94.3 1006 706.9 100 94.3 100

Medium 2 388.8 55 50.0 534 367.6 52 34.9 37

High 7 268.6 38 17.0 188 162.6 23 16.0 17

P.K. Baran et al. Urban Forestry & Urban Greening 35 (2018) 72–81

76

Table 6 shows the correlations between the 16 categories and themean perceived safety scores. Out of the 16 categories, all three cate-gories related to spatial enclosure were strongly related to perceivedsafety. In particular, two of these themes showed the strongest sig-nificant associations with perceived safety. Namely, both the “enclosed/cannot see” and the “trees/bushes” themes had the same and verystrong negative correlation with perceived safety (r=−.970, p <.001). Similarly, the “open/empty/see everything” theme had a veryhigh positive correlation with perceived safety (r= .765, p= .027).The association between perceived safety and “trees/bushes” was more

pronounced in females (r=−.931) than in males (r=−.891) (notshown). A strong negative association was observed between perceivedaccessibility (no pathways/no pavement) and perceived safety(r=−0.950, p < 0.001). Absence of roads was also negatively asso-ciated with perceived safety (r=−.964, p < .001). Strong

Fig. 3. Study stimuli displayed as panoramic stitched images. Because the 360° images are condensed into 2-dimensuional space and the size of the images, certainfeatures are not visible in the images.

Table 2Mean Values and Standard Deviations for Perceived Safety Scores for Male,Female and All Participants by Environment and by Enclosure Level.

Malesn=19

Femalesn=29

TotalN=48

Enclosurelevel

Environment number M SD M SD M SD

Low 1 4.58 0.61 4.34 0.77 4.44 0.713 4.58 0.61 4.03 0.82 4.25 0.795 4.79 0.54 4.79 0.41 4.79 0.466 4.53 0.84 4.41 0.78 4.46 0.80

Medium 2 4.53 0.61 4.10 0.72 4.27 0.714 4.74 0.45 4.10 0.82 4.35 0.76

High 7 3.63 1.12 3.00 1.23 3.25 1.218 3.37 0.90 2.90 0.82 3.08 0.87

Total 4.34 0.88 3.96 1.03 4.11 0.99

Note. Safety score rates are between 1 and 5; 1 = very unsafe, 5 = very safe.

Table 3Two-Way Repeated Measures Analysis of Variance for Main and InteractionEffects of Enclosure Level and Gender on Perceived Safety.

Source df MSE F gesa p

Enclosure levelb 1.75, 80.72 0.19 133.62 .11 < .001

Gender 1, 46 0.83 7.80 .46 .008

Gender × Enclosure level 1.75, 80.72 0.19 2.28 .01 .120

Notes: Dependent variable is perceived safety. Sphericity correction method:GG.

a Generalized Eta-Squared measure of effect size (Bakeman, 2005).b Enclosure levels: Low, Medium, High.

Table 4Tukey’s HSD Multiple Comparisons of Perceived Safety Mean Scores ShowingPairwise Differences between Enclosure Levels and between Genders.

Contrast Estimate SE df t p

High - Low −1.28 0.09 92 −14.90 < .001Enclosure High - Medium −1.14 0.09 92 −13.28 < .001

Low - Medium 0.14 0.09 92 1.63 .240

Gender Female - Male −0.43 0.15 46 −2.80 < .001

P.K. Baran et al. Urban Forestry & Urban Greening 35 (2018) 72–81

77

correlations emerged between social themes and perceived safety.Presence of “people” (existing or potential for presence of people) in thescene was positively associated with perceived safety (r= .873,p= .005), whereas “no people” was negatively associated (r= .909,p= .002) with perceived safety. It should be noted that although onlyone of the scenes (Environment 5) explicitly depicted people (a childand a woman playing in the playground) respondents repeatedly re-ferred to the implied presence of people for the other scenes, perhapsdue to the cues they gathered from each scene (e.g., “There is an openamphitheater, people may be around for playing music”). No significantcorrelations were observed between the other categories (built features,park furniture, cars, and maintenance) and perceived safety.

4. Discussion

There is a need to better understand the relationship between spa-tial enclosure shaped by vegetation and perceived safety in a parksetting. Several noteworthy findings emerged from this study about thisrelationship. First, the findings indicate that the link between enclosureand perceived safety in a park setting is similar to that found in otheropen spaces, such as college campus and forest settings. Namely, thefindings corroborate research grounded in prospect and refuge

(Appleton, 1975) as well as permeability (Stamps, 2005b, 2010a, 2012)theories indicating that lower level of enclosure is associated with po-sitive impressions of safety (Fisher and Nasar, 1992; Nasar et al., 1993;Herzog and Chernick, 2000; Stamps, 2005a). Overall, study resultssupport the hypothesis (H1) that less enclosed environments are per-ceived as safer than more highly enclosed environments. The environ-ments with the highest level of enclosure had the lowest degrees ofperceived safety, and conversely, medium and low enclosure environ-ments had significantly higher perceived safety. It is worth mentioningthat although less enclosed environments were perceived as safer, noneof the environments were actually perceived as unsafe (i.e., the lowestoverall perceived safety rating was 3.08). This may be due to the factthat the majority of the scenes included man-made elements (e.g.,buildings in the park surroundings, or features within the park loca-tions, such as open amphitheater and playground), which providedclues for the possible presence of people. In addition, none of the scenesexhibited social or physical incivilities (e.g., bench sleepers, un-maintained areas, litter, graffiti, etc.) (Robinson et al., 2003) or naturalthreats (e.g., wild animals, etc.). These associations are further sup-ported by the self-reports about the main reasons participants wouldfeel either safe or unsafe in a particular location where the strongestcorrelations were found between themes related to spatial enclosure(including reference to “being able to see” or “not see”) and safetyratings. A strong negative correlation was found between presence of“trees” and/or “bushes” and safety perceptions, reinforcing researchreporting strong association between landscape features and safetyjudgments in forest (Chiang et al., 2014) and park (A. Jorgensen et al.,2002) settings.

The study findings indicate that safety judgments in a park settingare not significantly different between medium and low enclosure en-vironments, suggesting that being able to partly view the surroundingenvironment may mediate perceptions of safety and danger (Herzogand Kutzli, 2002). This implies that as long as partial visibility is pro-vided, interventions that affect lateral boundaries, such as vegetationdensification and diversification, can be applied without compromisingperceptions of safety. However, the small number of environments usedin this study limited the ability to explore whether smaller, incrementaldifferences in enclosure level affect perceived safety. Future researchmay look at how different arrangements, heights, and visibility of lat-eral boundaries influence perceptions of safety in parks and other openpublic spaces, and whether the results generated by partial enclosurecan be replicated in different ways, using different landscape elements.

In order to quantify differences in enclosure level across the eightpark locations, the current study utilized the notions of visual and lo-comotive permeability (Stamps, 2010a, 2011b, 2012). However, thestudy was not designed to test the individual contributions of visual andlocomotive permeability on perceived safety. Between the two en-vironments with the highest enclosure level (Environments 7 and 8),the one with the smaller visible horizontal area (Environment 8) had alower safety rating; however, the difference was not statistically sig-nificant. Similarly, slight differences in visible arc did not lead to sig-nificant differences in perceived safety between the two medium en-closure environments (Environments 2 and 4). In addition, self-reportsdid not include any information that specifically relates the visiblehorizontal area available for movement to safety evaluations. This is apromising area for future research.

The self-report data show that in the case of low enclosure en-vironments, pathway characteristics and in particular, pathway leg-ibility, may play a pronounced role in moderating safety judgments. Inthe two locations with the highest degree of visibility (i.e., largestvisible arc), the one that lacks a demarcated pathway (Environment 6)was perceived as less safe than the one with a paved pathway(Environment 5). Although the difference is not significant, locomotiveaccess (Herzog and Kutzli, 2002) and the ability to easily find one’s waymay contribute to a sense of security and comfort (Kaplan and Kaplan,1989). Conversely, environments that do not have clearly demarcated

Table 5Tukey’s HSD Multiple Comparisons of Perceived Safety Mean Scores ShowingPairwise Differences between Genders by Enclosure Levels.

Enclosure level Contrast Estimate SE df t p

Low Female - Male −0.22 0.18 84 36 −1.20 .232

Medium Female - Male −0.53 0.18 84 36 −2.86 .005

High Female - Male −0.55 0.18 84 36 −2.99 .004

Note: p value adjustment: Tukey’s HSD method for comparing a family of 3 and2 estimates.

Table 6Frequency of Reasons for Safety Judgments and Pearson Correlations betweenFrequency of Reasons for Safety Judgments and Perceived Safety for the EightEnvironments.

Category for safety judgment Malesn=19

Femalesn=29

TotalN=48

r pa

Open/Empty/See everythingb 46 63 109 .765 .027

House/Building 40 63 103 .304 .463

People (present or might bepresent)

27 54 81 .873 .005

Enclosed/Cannot seeb 19 47 66 −.970 < .001

Pathways/Trails 25 40 65 .515 .192

Trees/Bushesb 14 40 54 −.970 < .001

Ball field/Other facilities 24 23 47 .612 .107

Cars 9 28 37 .377 .358

Benches 17 18 35 .527 .180

No pathways/no pavement 4 21 25 −.950 < .001

Well maintained 12 13 25 .692 .057

Poorly maintained 12 13 25 −.526 .181

Next to road 10 13 23 .605 .112

No People 5 16 21 −.909 .002

Off road 0 2 2 −.964 < .001

No benches 0 2 2 −.350 .396

a Due to small sample size statistical significances should be interpreted withcaution.

b Categories related to spatial enclosure.

P.K. Baran et al. Urban Forestry & Urban Greening 35 (2018) 72–81

78

pathways may lead to sense of feeling lost, which in turn could evokeperceptions of danger and/or feelings of fear.

The study results indicate a considerable difference in the levels ofsafety perceived by men relative to women. In support of the secondhypothesis (H2) the finding parallels research in various fields reportinglower overall safety ratings for women (Ferraro, 1996; Madge, 1997;Pain, 1997; Jorgensen et al., 2012), despite the fact that young menmay be more likely to be victims of personal violence (although this gaphas been shrinking in recent years, Lauritsen and Heimer, 2008).However, examining mean perceived safety differences across en-closure levels, the current study found that gender differences are notequal for different levels of enclosure, which supports the third hy-pothesis (H3). Differences in perceived safety between genders aresignificant for the high and medium enclosed environments, whereasfor the low enclosure environments, no significant differences wereobserved. The differences found for the high and medium enclosedenvironments may be partially due to the fact that men are more likelyto suppress the presentation of their fears than women (Sutton andFarrall, 2005). Nevertheless, by providing more nuanced analysis, ourresults contravene previous studies that generalized a gender differencefor an entire setting or a specific landscape typology, such as urbanparks (Madge, 1997; Jorgensen et al., 2012).

Another noteworthy finding of the study relates to the role of socialcontext and built environment features in moderating perceived safety.For the environments that were rated favorably, a considerable numberof comments referred to either man-made elements (buildings, street, orplayground) or presence or potential for presence of other people in theenvironment. This confirms L. J. orgensen et al.’s (2012) findings inurban parks suggesting social and environmental cues may collectivelycontribute to perceptions of safety.

Generally, perception of safety is a multifaceted phenomenon, and aplethora of individual, social, and environmental factors contribute toperceived safety and fear (Jansson et al., 2013; Maruthaveeran and vanden Bosch, 2014). However, although the reasons why people are afraidmight not have environmental origins (e.g., social construction of animage of an area, previous experience in similar places, prior in-formation about crime, etc.), as this study reveals, the physical en-vironment, including landscape elements, can increase perceptions ofsafety (Crowe, 2000; Jansson et al., 2013).

Overall, the results of this study parallel research in other openspaces, e.g., college campus and forest settings, which has used avariety of data collection methods, including on-site experiences (Fisherand Nasar, 1992; Nasar and Jones, 1997), surveys and/or interviews(Madge, 1997; Jorgensen et al., 2007), and responses to two-dimen-sional stimuli (Herzog and Kirk, 2005; Jorgensen et al., 2012; Chianget al., 2014). The results of the current study supported our expecta-tions about the relationships between enclosure and perceived safety ina park setting, which were drawn from existing theory (i.e., prospect-refuge and permeability) and empirical research. Overall, this suggeststhat IVEs may be successfully used in collecting data on perceivedsafety. However, this study was not specifically designed to examine theeffectiveness or advantages of IVEs relative to other mediums of pre-sentation. This is a promising area for future research. The currentstudy also focused on self-reported perceived safety and did not ex-amine other related concepts, such as perceived danger or feelings offear, which may lead to different results. Examining feelings of fearacross different enclosure levels can be particularly interesting, since itis possible to compare subjective and objective measures for a morecomplete understanding of the effects of enclosure on fear. A handful ofstudies using self-reported fear ratings and physiological fear responseshave not found statistically significant correlations between these twosets of measures (Ulrich et al., 1991; Mühlberger et al., 2007). In spiteof this, though, both Ulrich et al. (1991) and Mühlberger et al. (2007)report broad synchrony between the self-reported and physiologicalfindings, in general. For example Ulrich et al. (1991) found that naturalsettings had more restorative influences than urban environments on

three affective dimensions, including perceived fear. Similarly, theirphysiological measures–pulse transit time (PTT), spontaneous skinconductance responding (SCR), and frontalis muscle tension (EMG)–also indicate greater recovery influences of nature. Likewise, in Mühl-berger et al.’s (2007) study on psychophysiological assessment ofphobic fear during virtual tunnel driving, elevated fear responsesduring tunnel drives were reflected in both self-reported fear ratingsand heart rate (HR) changes measures. These findings indicate a futureneed for broad assessment of fear responses.

4.1. Limitations

As with any study, the present study has several limitations. First,the study assessed overall perceptions of each environment and did notcontrol for variations in duration of participants’ attention to differentparts of the environment. Given that Oculus view covers about 110°, atleast three different conditions could be experienced within a 360° ro-tation in an IVE. Future studies may benefit from systematic analysis ofperceptions in relation to specific features of a given environment.Second, except for gender, individual differences and socio-demo-graphic factors were not considered in the study. Previous research hasidentified the role of individual as well as socio-cultural differences insafety perception (Skogan, 1995; Jansson et al., 2013). Factors such aspersonality, frequency of exposure to nature, or previous experiences insimilar places may affect perceptions of danger and safety. Similarly,social-cultural factors, such as acceptable behaviors, image and re-putation of an area (Sampson, 2013), collective memory, and conse-quently meanings and social construction of space (Pain, 1997) mayalso influence perceptions. The small number of participants precludedlooking at these characteristics. However, since all the environmentswere from a single park, some of these factors (such as reputation of thearea or pre-existing fear of the neighborhood) were not of concern.Finally, the participants in the study were undergraduate students inthe Department of Parks, Recreation and Tourism Management. Due tofamiliarity with parks in general, these participants may perceive parkenvironments differently than the general public. Similarly, becauseundergraduate students may perceive risks differently than a re-presentative sample of the adult population (Herzog and Kutzli, 2002)the generalizability of the findings beyond the study population islimited.

5. Conclusions

This study examined how spatial enclosure shaped by the physicalarrangement of vegetation relates to individuals’ perception of safety ina park setting. The findings suggest spatial enclosure created by vege-tation is one of the main attributes contributing to perceived safety in apark. Additional attributes that contribute to perception of safety arepresence of non-threatening people and paths. The results complementthe limited existing evidence of these relationships in the context ofurban parks.

In terms of its methodology, this study pioneered the use of IVEs tostudy perceived safety. The study suggests IVE technology could besuccessfully used for collecting data on perceived safety. This methodcan advance similar studies to represent a thorough and continuousexperience of in situ locations. To supplement this approach, the studyalso developed a method for measuring spatial enclosure for 360° pa-noramas. In addition, the methodology used in this study providesopportunities to systematically manipulate environmental stimuli forrigorous experimental studies designed to assess the effect of variousenvironmental attributes and spatial arrangements on individuals’perceptions, cognitive functioning, and affective responses. To reinforce(or contradict) the findings, future research should assess the effec-tiveness and advantages of using IVEs relative to other mediums ofpresentation in examining the relationship between physical attributesand perception of safety.

P.K. Baran et al. Urban Forestry & Urban Greening 35 (2018) 72–81

79

With regard to practical implications, such an empirically groundedunderstanding of how spatial characteristics of urban parks affect per-ceived safety would allow urban planners and park managers to de-velop spatially explicit design and management strategies. Admittedly,open landscapes offer better possibilities for social interaction andsafety. However, it would be unrealistic to suggest that because peoplefind conditions equating to the low enclosure environments safest,these conditions should prevail exclusively. Recent efforts in landscapedesign are partially focused on maintaining vegetation diversity anddensity to improve ecological relevance (Carrus et al., 2015). In addi-tion, natural landscapes are particularly critical for maintaining urbanparks’ ability to offer high-quality social and psychological benefits(Pretty et al., 2005). For example, a recent study suggests environmentsthat balance enclosed dense growth with more open views are optimalfor restoration (Stigsdotter et al., 2017). The results of this study showthat medium levels of enclosure may not significantly impair safetyevaluations. Also, perceived safety could be mediated through pro-viding minimal indications of human-made features as well as socialcues. As cities become increasingly dense, understanding how to max-imize the use and benefit of limited public green space is criticallyimportant. Finally, the method developed for this study holds strongpotential for evaluating other public spaces, such as trails, streets, andpublic squares.

Funding

This research did not receive any specific grant from fundingagencies in the public, commercial, or not-for-profit sectors.

References

Annerstedt, M., Jönsson, P., Wallergård, M., Johansson, G., Karlson, B., Grahn, P.,Hansen, Å.M., Währborg, P., 2013. Inducing physiological stress recovery withsounds of nature in a virtual reality forest - results from a pilot study. Physiol. Behav.118, 240–250. https://doi.org/10.1016/j.physbeh.2013.05.023.

Appleton, J., 1975. The Experience of Landscape. NY: Wiley, New York.Bakeman, R., 2005. Recommended effect size statistics for repeated measures designs.

Behav. Res. Methods 37 (3), 379–384.Baran, P.K., Smith, W.R., Moore, R.C., Floyd, M.F., Bocarro, J.N., Cosco, N.G., Danninger,

T.M., 2014. Park use among youth and adults: examination of individual, social, andurban form factors. Environ. Behav. 46 (6), 768–800. https://doi.org/10.1177/0013916512470134.

Brown, T.C., Daniel, T.C., 1987. Context effects in perceived environmental quality as-sessment: scene selection and landscape quality ratings. J. Environ. Psychol. 7 (3),233–250.

Carrus, G., Scopelliti, M., Lafortezza, R., Colangelo, G., Ferrini, F., Salbitano, F., et al.,2015. Go greener, feel better? The positive effects of biodiversity on the well-being ofindividuals visiting urban and peri-urban green areas. Landsc. Urban Plan. 134,221–228. https://doi.org/10.1016/j.landurbplan.2014.10.022.

Cheng, C.K., 2007. Understanding Visual Preferences for Landscapes: an Examination ofthe Relationship Between Aesthetics and Emotional Bonding (Unpublished DoctoralDissertation). Texas A&M University, College Station, TX.

Chiang, Y.C., Nasar, J.L., Ko, C.C., 2014. Influence of visibility and situational threats onforest trail evaluations. Landsc. Urban Plan. 125, 166–173. https://doi.org/10.1016/j.landurbplan.2014.02.004.

Cozens, P., Neale, R., Whitaker, J., Hillier, D., 2003. Investigating personal safety atrailway stations using “virtual reality” technology. Facilities 21 (7/8), 188–194.https://doi.org/10.1108/02632770310489936.

Crescentini, C., Chittaro, L., Capurso, V., Sioni, R., Fabbro, F., 2016. Psychological andphysiological responses to stressful situations in immersive virtual reality: differencesbetween users who practice mindfulness meditation and controls. Comput. HumanBehav. 59, 304–316. https://doi.org/10.1016/j.chb.2016.02.031.

Crowe, T.D., 2000. Crime Prevention Through Environmental Design: Applications ofArchitectural Design and Space Management Concepts, 2nd ed. Butterworth-Heinemann, Boston, MA.

Drettakis, G., Roussou, M., Reche, A., Tsingos, N., 2007. Design and evaluation of a real-world virtual environment for architecture and urban planning. Presence 16 (3),318–332. https://doi.org/10.1162/pres.16.3.318.

Ferraro, K.F., 1996. Women’s fear of victimization: shadow of sexual assault? Soc. Forces75 (2), 667–690. https://doi.org/10.2307/2580418.

Fich, L.B., Jönsson, P., Kirkegaard, P.H., Wallergård, M., Garde, A.H., Hansen, Å., 2014.Can architectural design alter the physiological reaction to psychosocial stress? Avirtual TSST experiment. Physiol. Behav. 135, 91–97. https://doi.org/10.1016/j.physbeh.2014.05.034.

Fisher, B.S., Nasar, J.L., 1992. Fear of crime in relation to three exterior site features:prospect, refuge, and escape. Environ. Behav. 24 (1), 35–65.

Fisher, B.S., Nasar, J.L., 1995. Fear spots in relation to microlevel physical cues: exploringthe overlooked. J. Res. Crime Delinq. 32 (2), 214–239. https://doi.org/10.1177/0022427895032002005.

Garofalo, J., 1981. The fear of crime: causes and consequences. J. Crim. Law Criminol. 72(2), 839–857. https://doi.org/10.2307/1143018.

Gatersleben, B., Andrews, M., 2013. When walking in nature is not restorative - the role ofprospect and refuge. Health Place 20, 91–101. https://doi.org/10.1016/j.healthplace.2013.01.001.

Giles-Corti, B., Donovan, R.J., 2002. The relative influence of individual, social andphysical environment determinants of physical activity. Soc. Sci. Med. 54 (12),1793–1812. https://doi.org/10.1016/S0277-9536(01)00150-2.

Heft, H., 2007. The participatory character of landscape. Proceedings from Open space,people space 2. Internation Conference of Innovative Approaches to ResearchExcellence in Landscape and Health (Pp. 18-23) Retrieved from. http://www.openspace.eca.ed.ac.uk/wp-content/uploads/2015/12/OSPS2-Conference-Proceedings-2007.pdf.

Herzog, T.R., Chernick, K.K., 2000. Tranquility and danger in urban and natural settings.J. Environ. Psychol. 20 (1), 29–39. https://doi.org/10.1006/jevp.1999.0151.

Herzog, T.R., Kirk, K.M., 2005. Pathway curvature and border visibility as predictors ofpreference and danger in forest settings. Environ. Behav. 37 (5), 620–639.

Herzog, T.R., Kropscott, L.S., 2004. Legibility, mystery, and visual access as predictors ofpreference and perceived danger in forest settings without pathways. Environ. Behav.36 (5), 659–677.

Herzog, T.R., Kutzli, G.E., 2002. Preference and perceived danger in field/forest settings.Environ. Behav. 34 (6), 819–835.

Herzog, T.R., Miller, E.J., 1998. The role of mystery in perceived danger and environ-mental preference. Environ. Behav. 30 (4), 429–449. https://doi.org/10.1177/001391659803000401.

Heydarian, A., Carneiro, J.P., Gerber, D., Becerik-Gerber, B., Hayes, T., Wood, W., 2015.Immersive virtual environments versus physical built environments: a benchmarkingstudy for building design and user-built environment explorations. Autom. Constr.54, 116–126. https://doi.org/10.1016/j.autcon.2015.03.020.

Jansson, M., Fors, H., Lindgren, T., Wiström, B., 2013. Perceived personal safety in re-lation to urban woodland vegetation – A review. Urban For. Urban Green. 12 (2),127–133. https://doi.org/10.1016/j.ufug.2013.01.005.

Jorgensen, A., Hitchmough, J., Calvert, T., 2002. Woodland spaces and edges: their im-pact on perception of safety and preference. Landsc. Urban Plan. 60 (3), 135–150.https://doi.org/10.1016/S0169-2046(02)00052-X.

Jorgensen, A., Hitchmough, J., Dunnett, N., 2007. Woodland as a setting for housing-appreciation and fear and the contribution to residential satisfaction and placeidentity in Warrington New Town. UK. Landscape and Urban Planning 79, 273–287.

Jorgensen, L.J., Ellis, G.D., Ruddell, E., 2012. Fear perceptions in public parks: interac-tions of environmental concealment, the presence of people recreating, and gender.Environ. Behav. 45 (7), 803–820. https://doi.org/10.1177/0013916512446334.

Kaplan, R., Kaplan, S., 1989. The Experience of Nature. A Psychological Perspective.Cambridge University Press, Cambridge, MA.

Kim, K., Rosenthal, M.Z., Zielinski, D., Brady, R., 2012. Comparison of desktop, headmounted display, and six wall fully immersive systems using a stressful task. 2012IEEE Virtual Reality Workshops (VRW) 143-144. https://doi.org/10.1109/VR.2012.6180922.

Kuliga, S.F., Thrash, T., Dalton, R.C., Hölscher, C., 2015. Virtual reality as an empiricalresearch tool — exploring user experience in a real building and a correspondingvirtual model. Comput. Environ. Urban Syst. 54, 363–375. https://doi.org/10.1016/j.compenvurbsys.2015.09.006.

Lauritsen, J.L., Heimer, K., 2008. The gender gap in violent victimization, 1973–2004. J.Quant. Criminol. 24, 125–147. https://doi.org/10.1007/s10940-008-9041-y.

Loewen, L.J., Steel, G.D., Suedfeld, P., 1993. Perceived safety from crime in the urbanenvironment. J. Environ. Psychol. 13 (4), 323–331. https://doi.org/10.1016/S0272-4944(05)80254-3.

Luigi, M., Massimiliano, M., Aniello, P., Gennaro, R., Virginia, P.R., 2015. On the validityof immersive virtual reality as tool for multisensory evaluation of urban spaces.Energy Procedia 78, 471–476. https://doi.org/10.1016/j.egypro.2015.11.703.

Madge, C., 1997. Public parks and geography of fear. Tijdschr. Voor Econ. En Soc. Geogr.88 (3), 237–250. https://doi.org/10.1111/j.1467-9663.1997.tb01601.x.

Maruthaveeran, S., van den Bosch, C.C.K., 2014. A socio cological exploration of fear ofcrime in urban green spaces - A systematic review. Urban For. Urban Green. 13, 1–18.https://doi.org/10.1016/j.ufug.2013.11.006.

Molnar, B.E., Gortmaker, S.L., Bull, F.C., Buka, S.L., 2004. Unsafe to play? Neighborhooddisorder and lack of safety predict reduced physical activity among urban childrenand adolescents. Am. J. Health Promot. 18 (5), 378–386. https://doi.org/10.4278/0890-1171-18.5.378.

Mühlberger, A., Bülthoff, H.H., Wiedemann, G., Pauli, P., 2007. Virtual reality for thepsychophysiological assessment of phobic fear: responses during virtual tunneldriving. Psychol. Assess. 19 (3), 340–346. https://doi.org/10.1037/1040-3590.19.3.340.

Naceri, A., Chellali, R., Dionnet, F., Toma, S., 2010. Depth perception within virtualenvironments: comparison between two display technologies. International Journalon Advances in Intelligent Systems 3 (1&2), 51–64.

Nasar, J.L., Jones, K.M., 1997. Landscapes of fear and stress. Environ. Behav. 29 (3),291–323. https://doi.org/10.1177/001391659702900301.

Nasar, J.L., Fisher, B., Grannis, M., 1993. Proximate physical cues to fear of crime.Landsc. Urban Plan. 26 (1-4), 161–178. https://doi.org/10.1016/0169-2046(93)90014-5.

Natapov, A., Fisher-Gewirtzman, D., 2016. Visibility of urban activities and pedestrianroutes: an experiment in a virtual environment. Comput. Environ. Urban Syst. 58,60–70.

P.K. Baran et al. Urban Forestry & Urban Greening 35 (2018) 72–81

80

Nordh, H., Hartig, T., Hagerhall, C.M., Fry, G., 2009. Components of small urban parksthat predict the possibility for restoration. Urban For. Urban Green. 8, 225–235.https://doi.org/10.1016/j.ufug.2009.06.003.

Pain, R., 1997. Social geographies of women’s fear of crime. Transactions of the Instituteof British Geographers. New Series 22 (2), 231–244.

Passig, D., Tzuriel, D., Eshel-Kedmi, G., 2016. Improving children’s cognitive modifia-bility by dynamic assessment in 3D Immersive Virtual Reality environments. Comput.Educ. 95, 296–308. https://doi.org/10.1016/j.compedu.2016.01.009.

Pearson, D.G., Craig, T., 2014. The great outdoors? Exploring the mental health benefitsof natural environments. Front. Psychol. 5, 1178. https://doi.org/10.3389/fpsyg.2014.01178.

Pretty, J., Peacock, J., Sellens, M., Griffin, M., 2005. The mental and physical healthoutcomes of green exercise. Int. J. Environ. Health Res. 15 (5), 319–337. https://doi.org/10.1080/09603120500155963.

Robinson, J.B., Lawton, B.A., Taylor, R.B., Perkins, D.D., 2003. Multilevel longitudinalimpacts of incivilities: fear of crime, expected safety, and block satisfaction. J. Quant.Criminol. 19 (3), 237–274. https://doi.org/10.1023/A:1024956925170.

Sampson, R.J., 2013. Great American City: Chicago and the Enduring NeighborhoodEffect. The University of Chicago Press, Chicago, IL.

Sanchez-Vives, M.V., Slater, M., 2005. From presence to consciousness through virtualreality. Nat. Rev. Neurosci. 6, 332–339. https://doi.org/10.1038/nrn1651.

Schnabel, M.A., 2014. The immersive virtual environment design studio. In: Wang, X.,Tsai, J.J.-H. (Eds.), Collaborative Design in Virtual Environments. Springer,Dordrecht, pp. 177–191.

Skår, M., 2010. Forest dear forest fear: dwellers’ relationship to their neighbourhoodforest. Landsc. Urban Plan. 98, 110–116.

Skogan, W.G., 1995. Crime and racial fears of White Americans. American Academy ofPolitical and Social Science 539, 59–71.

Slater, M., 2009. Place illusion and plausibility can lead to realistic behaviour in im-mersive virtual environments. Philos. Trans. Biol. Sci. 364, 3549–3557. https://doi.org/10.1098/rstb.2009.0138.

Smith, J.W., 2015. Immersive virtual environment technology to supplement environ-mental perception, preference and behavior research: a review with applications. Int.J. Environ. Res. Public Health 12 (9), 11486–11505. https://doi.org/10.3390/ijerph120911486.

Smith, W.R., Torstensson, M., Johansson, K., 2001. Perceived risk and fear of crime:gender differences in contextual sensitivity. Int. Rev. Vict. 8 (2), 159–181. https://doi.org/10.1177/026975800100800204.

Stamps, A.E., 2005a. Enclosure and safety in urbanscapes. Environ. Behav. 37 (1),102–133. https://doi.org/10.1177/0013916504266806.

Stamps, A.E., 2005b. Isovists, enclosure, and permeability theory. Environ. Plann. BPlann. Des. 32 (5), 735–762. https://doi.org/10.1068/b31138.

Stamps, A.E., 2010a. Effects of permeability on perceived enclosure and spaciousness.

Environ. Behav. 42 (6), 864–886. https://doi.org/10.1177/0013916509337287.Stamps, A.E., 2010b. Use of static and dynamic media to simulate environments: a meta-

analysis. Percept. Mot. Skills 111 (2), 355–364. https://doi.org/10.2466/22.24.27.PMS.111.5.355-364.

Stamps, A.E., 2011a. Effects of area, height, elongation, and color on perceived spa-ciousness. Environ. Behav. 43 (2), 252–273. https://doi.org/10.1177/0013916509354696.

Stamps, A.E., 2011b. Effects of boundary height and horizontal size within boundary onperceived enclosure. Percept. Mot. Skills 113 (3), 995–998. https://doi.org/10.2466/24.27.PMS.113.6.995-998.

Stamps, A.E., 2012. Effects of multiple boundaries on perceived spaciousness and en-closure. Environ. Behav. 45 (7), 851–875.

Stigsdotter, U.K., Corazon, S.S., Sidenius, U., Refshauge, A.D., Grahn, P., 2017. Forestdesign for mental health promotion—using perceived sensory dimensions to elicitrestorative responses. Landsc. Urban Plan. 160, 1–15. https://doi.org/10.1016/j.landurbplan.2016.11.012.

Sutton, R.M., Farrall, S., 2005. Gender, socially desirable responding and the fear ofcrime: Are women really more anxious about crime? Br. J. Criminol. 45 (2), 212–224.https://doi.org/10.1093/bjc/azh084.

Tabrizian, P., Baran, P.K., Smith, W.R., Meentemeyer, R.K., 2018. Exploring perceivedrestoration potential of urban green enclosure through immersive virtual environ-ments. J. Environ. Psychol. 55, 99–109. https://doi.org/10.1016/j.jenvp.2018.01.001.

Tinsley, H.E.A., Tinsley, D.J., Croskeys, C.E., 2002. Park usage, social milieu, and psy-chosocial benefits of park use reported by older urban park users from four ethnicgroups. Leis. Sci. 24 (2), 199–218. https://doi.org/10.1080/01490400252900158.

Troped, P.J., Cromley, E.K., Fragala, M.S., Melly, S.J., Hasbrouck, H.H., Gortmaker, S.L.,Brownson, R.C., 2006. Development and reliability and validity testing of an audittool for trail/path characteristics: the path environment audit tool (PEAT). J. Phys.Act. Health 3 (Suppl1), S158–S175.

Ulrich, R.S., Simons, R.F., Losito, B.D., Fiorito, E., Miles, M.A., Zelson, M., 1991. Stressrecovery during exposure to natural and urban environments. J. Environ. Psychol. 11(3), 201–230. https://doi.org/10.1016/S0272-4944(05)80184-7.

Valmaggia, L.R., Latif, L., Kempton, M.J., Rus-Calafell, M., 2016. Virtual reality in thepsychological treatment for mental health problems: an systematic review of recentevidence. Psychiatry Res. 236, 189–195. https://doi.org/10.1016/j.psychres.2016.01.015.

Wang, K., Taylor, R.B., 2006. Simulated walks through dangerous alleys: impacts offeatures and progress on fear. J. Environ. Psychol. 26 (4), 269–283. https://doi.org/10.1016/j.jenvp.2006.07.006.

Zlot, A.I., Schmid, T.L., 2005. Relationships among community characteristics andwalking and bicycling for transportation or recreation. Am. J. Health Promot. 19 (4),314–317. https://doi.org/10.4278/0890-1171-19.4.314.

P.K. Baran et al. Urban Forestry & Urban Greening 35 (2018) 72–81

81