Willsull.net

ENVIRONMENT AND
CRIME IN THE INNER CITY
Does Vegetation Reduce Crime?
FRANCES E. KUO is an assistant professor and codirector of the Human-Environ-
ment Research Laboratory at the University of Illinois, Urbana-Champaign. Her re-
search focuses on attention, defensible space, and novice-friendly information.
WILLIAM C. SULLIVAN is an associate professor and codirector of the Human-
Environment Research Laboratory at the University of Illinois, Urbana-Champaign.
His research focuses on the psychological and social benefits of urban nature and citi-
zen participation in environmental decision making.
ABSTRACT: Although vegetation has been positively linked to fear of crime andcrime in a number of settings, recent findings in urban residential areas have hinted ata possible negative relationship: Residents living in “greener” surroundings reportlower levels of fear, fewer incivilities, and less aggressive and violent behavior. Thisstudy used police crime reports to examine the relationship between vegetation andcrime in an inner-city neighborhood. Crime rates for 98 apartment buildings withvarying levels of nearby vegetation were compared. Results indicate that althoughresidents were randomly assigned to different levels of nearby vegetation, the greenera building’s surroundings were, the fewer crimes reported. Furthermore, this patternheld for both property crimes and violent crimes. The relationship of vegetation tocrime held after the number of apartments per building, building height, vacancy rate,and number of occupied units per building were accounted for.
The highway from one merchant town to another shall be cleared so that nocover for malefactors should be allowed for a width of two hundred feet on ei-ther side; landlords who do not effect this clearance will be answerable for rob-beries committed in consequence of their default, and in case of murder theywill be in the king’s mercy.
—Statute of Winchester of 1285, Chapter V, King Edward I AUTHORS’ NOTE: A portion of these findings was presented in invited testimony tothe National Urban and Community Forestry Advisory Council (NUCFAC). This ENVIRONMENT AND BEHAVIOR, Vol. 33 No. 3, May 2001 343-367 2001 Sage Publications, Inc.
There is a long tradition of addressing crime in problem areas by removing vegetation. As early as 1285, the English King Edward I sought to reducehighway robbery by forcing property owners to clear highway edges of treesand shrubs (Pluncknett, 1960). Today, that tradition continues as park author-ities, universities, and municipalities across North America engage in activeprograms to remove vegetation because it is thought to conceal and facilitatecriminal acts (Michael & Hull, 1994; Nasar & Fisher, 1993; Weisel, Gouvis,& Harrell, 1994).
One of the settings in which crime is of greatest concern today is the inner-city neighborhood. To combat crime in this setting, should vegetationbe removed? This article suggests the opposite. We present theory and evi-dence to suggest that far from abetting crime, high-canopy trees and grassmay actually work to deter crime in poor inner-city neighborhoods.
COULD THERE BE EXCEPTIONS TO THE RULE?
As a rule, the belief is that vegetation facilitates crime because it hides per- petrators and criminal activity from view. Here, we review the evidence insupport of this “rule” and suggest conditions under which it might not apply.
Although no studies to date have examined whether crime rates are actu- ally higher in the presence of dense vegetation, a variety of evidence linksdense vegetation with fear, fear of crime, and possibly crime itself.
It is certainly the case that many people fear densely vegetated areas. In research on urban parks, densely wooded areas have consistently been asso-ciated with fear. In one study, safety ratings for 180 scenes of urban parksshowed that individuals felt most vulnerable in densely forested areas andsafest in open, mowed areas (Schroeder & Anderson, 1984). And in anotherstudy, individuals who were asked for their open-ended responses to photo- work was also supported by the Cooperative State Research, Education and Exten-sion Service, U.S. Department of Agriculture, under Project No. ILLU-65-0387.
Weare grateful for the assistance of many individuals and other institutions as well.
John Potter and Liesette Brunson assisted in data entry and data analysis in the initialstages of this project. A reviewer’s suggestion substantially strengthened the analysespresented here. The Chicago Housing Authority and the management of Ida B. Wellswere helpful in many ways, and the Chicago Police Department graciously gave usaccess to their year-end crime reports. Jerry Barrett helped produce the figures, andHelicopter Transport of Chicago donated the helicopter flight over Ida B. Wells. Cor-respondence concerning this article should be addressed to Frances E. Kuo, Human-Environment Research Laboratory, University of Illinois, 1103 S. Dorner, Urbana,IL, 61801; e-mail: f-kuo@uiuc.edu. graphs of urban parks indicated that heavily vegetated areas seemed danger-ous (Talbot & Kaplan, 1984). Although neither of these studies specificallyprobed fear of crime (as opposed to more general fear), it was clear that atleast some participants had crime in mind; one respondent specifically sug-gested that weedy areas gave muggers good hiding places (Talbot & Kaplan,1984).
Dense vegetation has also been linked specifically to fear of crime. In safety ratings for 180 scenes of parking lots, the more a photo was covered byvegetation, the lower the perceived security (Shaffer & Anderson, 1985).
And in research examining fear of crime on a university campus, denseunderstories that reduced views into areas where criminals might hide wereassociated with fear of crime (Nasar & Fisher, 1993). In these and other stud-ies, view distance seems to be an important factor. Fear of crime is higherwhere vegetation blocks views (Fisher & Nasar, 1992; Kuo, Bacaicoa, &Sullivan, 1998; Michael & Hull, 1994).
Not only has dense vegetation been linked to general fears and to fear of crime in particular, but two studies have pointed more directly at a facilitativerole of vegetation in crime. In the first study, park managers and park policeindicated that dense vegetation is regularly used by criminals to conceal theiractivities (Michael & Hull, 1994). In the second, burglars themselves lentsupport to this notion. In this study, automobile burglars described how theyused dense vegetation in a variety of ways, including to conceal their selec-tion of a target and their escape from the scene, to shield their examination ofstolen goods, and finally, in the disposal of unwanted goods (Michael, Hull,& Zahm, 1999). At the same time, Michael and his coauthors made it clearthat vegetation was neither necessary nor sufficient for a crime to take place.
The clear theme in all these studies is that dense vegetation provides potential cover for criminal activities, possibly increasing the likelihood ofcrime and certainly increasing the fear of crime. Large shrubs, underbrush,and dense woods all substantially diminish visibility and therefore are capa-ble of supporting criminal activity.
But, not all vegetation blocks views. A well-maintained grassy area cer- tainly does not block views; widely spaced, high-canopy trees have minimaleffect on visibility; and flowers and low-growing shrubs seem unlikely toprovide cover for criminal activities. We suggest that although the rule thatvegetation aids crime may hold for visibility-decreasing forms of vegetation,there are systematic exceptions to this rule. To wit, we propose that widelyspaced, high-canopy trees and other visibility-preserving forms of vegetationdo not promote crime.
MIGHT VEGETATION DETER CRIME? THEORY
Furthermore, we propose that in some settings, visibility-preserving forms of vegetation may actually deter crime. Specifically, we propose that inpoor inner-city neighborhoods, vegetation can inhibit crime through the fol-lowing two mechanisms: by increasing surveillance and by mitigating someof the psychological precursors to violence. Let’s look at each of these inturn.
Increasing surveillance. Surveillance is a well-established factor in crimi- nal activity. Jane Jacobs (1961) suggested that the simple presence of more“eyes on the street” would deter crime, and this concept was prominent inOscar Newman’s (1972) classic Defensible Space and appeared in Jeffery’s(1971) Crime Prevention Through Environmental Design. Since then, manystudies have shown that perpetrators avoid areas with greater surveillanceand greater likelihood of intervention (e.g., Bennett, 1989; Bennett &Wright, 1984; Cromwell, Olson, & Avary, 1991; Poyner & Webb, 1992).
And, substantial research has shown that criminals avoid well-used residen-tial areas where their activities might easily be observed (Coleman, 1987;Macdonald & Gifford, 1989; Merry, 1981; Rhodes & Conley, 1981).
There is some evidence to suggest that in inner-city neighborhoods, vege- tation might introduce more eyes on the street by increasing residents’ use ofneighborhood outdoor spaces. A series of studies conducted in inner-cityneighborhoods has shown that treed outdoor spaces are consistently morewell used by youth, adults, and mixed-age groups than are treeless spaces;moreover, the more trees in a space, the greater the number of simultaneoususers (Coley, Kuo, & Sullivan, 1997; Kuo, Sullivan, Coley, & Brunson, 1998;W. C. Sullivan, Kuo, & DePooter, 2001). Not surprisingly then, a recent studyfound that children were twice as likely to have adult supervision in greeninner-city neighborhood spaces than in similar but barren spaces (A. F. Tay-lor, Wiley, Kuo, & Sullivan, 1998). Thus, in these settings, higher levels ofvegetation not only preserve visibility but may also increase surveillance.
Perhaps just as important as actual surveillance in deterring crime is implied surveillance. Newman (1972) suggested that criminals might bedeterred by environmental cues suggesting that surveillance is likely evenwhen no observers are present (also see Jeffery, 1971; R. B. Taylor, 1988).
Consistent with this, territorial markers have been empirically linked to lowerrates of incivilities and crime (Brown & Altman, 1983; Perkins, Brown, &Taylor, 1996; Perkins, Wandersman, Rich, & Taylor, 1993; R. B. Taylor,1988). (And even those E&B readers who are not criminals may have experienced the power of implied surveillance—on the highway after pass-ing an empty police car.) There is some evidence to suggest that residential vegetation can act as a territorial marker. Chaudhury (1994) showed front views of houses to studentsand examined how a host of environmental features affected their ratings ofterritorial personalization. He found that the presence and maintenance ofvegetative features was the strongest predictor of territorial personalization,with an R-squared of .65. Similarly, Brown and colleagues (Brown &Altman, 1983; Brown & Bentley, 1993) found evidence suggesting thatplants and other territorial markers make properties less attractive for bur-glary. We suggest that well-maintained vegetation may constitute a particu-larly effective territorial marker. Well-maintained vegetation outside a homeserves as one of the cues to care (Nassauer, 1988), suggesting that the inhabit-ants actively care about their home territory and potentially implying that anintruder would be noticed and confronted.
Mitigating psychological precursors to violence. Another mechanism by which vegetation might inhibit crime is through mitigating mental fatigue. S.
Kaplan (1987) suggested that one of the costs of mental fatigue may be aheightened propensity for “outbursts of anger and potentially . . . violence”(p. 57), and three proposed symptoms of mental fatigue—irritability, inatten-tiveness, and decreased control over impulses—are each well-establishedpsychological precursors to violence. Irritability is linked with aggression innumerous studies (e.g., Caprara & Renzi, 1981; Coccaro, Bergeman,Kavoussi, & Seroczynski, 1997; Kant, Smith-Seemiller, & Zeiler, 1998;Kavoussi & Coccaro, 1998; Stanford, Greve, & Dickens, 1995). Inattentive-ness has been closely tied to aggression in both children (Stewart, 1985) andadolescents (Scholte, van Aken, & van Leishout, 1997). And, impulsivity isassociated with aggression and violence in a variety of populations (forreviews, see Brady, Myrick & McElroy, 1998; Markovitz, 1995; Tuinier,Verhoeven, & Van Praag, 1996).
A considerable body of studies indicates that vegetation aids in the recov- ery from mental fatigue. Contact with nature in a variety of forms—wilder-ness areas, prairie, community parks, window views, and interior plants—issystematically linked with enhanced cognitive functioning as measured byboth self-report and performance on objective tests (e.g., Canin, 1991;Cimprich, 1993; Hartig, Mang, & Evans, 1991; R. Kaplan, 1984; Lohr,Pearson-Mimms, & Goodwin, 1996; Miles, Sullivan, & Kuo, 1998; Ovitt,1996; Tennessen & Cimprich, 1995). To the extent that irritability, inatten-tiveness, and impulsivity are symptoms of mental fatigue, as first proposed in S. Kaplan (1987) and recently elucidated in Kuo and Sullivan (in press),reductions in mental fatigue should decrease violent behavior.
In sum, we propose that vegetation can deter crime in poor urban neigh- borhoods in any or all of the following ways: by increasing residents’ infor-mal surveillance of neighborhood spaces, by increasing the implied sur-veillance of these spaces, and by mitigating residents’ mental fatigue,thereby reducing the potential for violence. Next, we review empirical workpointing at a negative relationship between vegetation and crime.
MIGHT VEGETATION DETER CRIME? CIRCUMSTANTIAL EVIDENCE
There are a number of scattered hints in the empirical literature that vege- tation might have a negative relationship to crime in residential settings.
A few studies have used images to examine the relationship between vege- tation and sense of safety in residential settings. The findings from residentialsettings are in direct contrast to those obtained in studies of nonresidentialsettings: In residential settings, the more vegetation there is, the less fear ofcrime. One study used photographs of residential sites to examine effects ofarchitectural and landscape features on fear of crime and found that higherlevels of vegetation were associated with less fear of crime (Nasar, 1982).
Another study used drawings of residences and found that propertiesappeared safer when trees and shrubs were included than when they were not(Brower, Dockett, & Taylor, 1983). And, similar results were obtained froman experiment using computer-based photo simulations. In that study, aninner-city courtyard was depicted with varying densities of trees: The moredense the tree planting was, the greater the sense of safety (Kuo, Bacaicoa,et al., 1998).
One study used controlled comparisons of real residential settings to examine the relationship between vegetation and sense of safety. In a publichousing development where residents were randomly assigned to architec-turally identical apartment buildings with varying levels of vegetation imme-diately outside, those residents who lived in buildings with more trees andgrass gave systematically higher endorsements to the statement “I feel safeliving here” than did their counterparts living in relatively barren buildings(Kuo, Sullivan, et al., 1998). That is, not only do images of green residentialsettings evoke a greater sense of safety, but individuals living in such settingsreport a greater sense of safety as well.
There is some indication that this greater sense of safety is warranted. A few studies have examined the relationship between vegetation and “incivili-ties.” R. B. Taylor, Gottfredson, and Brower (as cited in R. B. Taylor, 1988)compared street blocks with higher and lower levels of high-maintenance gardening and found fewer problems reported on street blocks with higherlevels of high-maintenance gardening. And in another study, Stamen (1993)surveyed landscaped and nonlandscaped areas in a community and foundthat the incidence of vandalism or graffiti in sites without plantings was 90%as compared to 10% in sites with plantings. Similarly, Brunson (1999) exam-ined both physical and social incivilities in public housing outdoor spaceswith trees and grass versus in similar spaces without vegetation. Residentreports indicated that graffiti, vandalism, and littering were systematicallylower in outdoor spaces with trees and grass than in comparable, more barrenspaces (Brunson, 1999). Furthermore, resident reports indicated that socialincivilities, such as the presence of noisy, disruptive individuals, strangers,and illegal activity, were also systematically lower in the greener outdoorspaces (Brunson, 1999).
Additional evidence that vegetation may reduce crime comes from two studies that examined the relationship between residential vegetation andresidents’ levels of aggression and violence. Mooney and Nicell (1992) com-pared violent assaults by Alzheimer patients during two consecutive sum-mers in five long-term care facilities—three without gardens and two inwhich exterior gardens were installed. In Alzheimer patients, increases in thenumber of aggressive assaults each year are typical because of the progres-sive deterioration of cognitive faculties; and indeed, in the facilities withoutgardens, the incidence of violent assaults increased dramatically over time.
By contrast, the incidence of violent assaults in the other facilities stayed thesame or decreased slightly after gardens were installed.
Another study compared levels of aggression and violence in an urban public housing neighborhood where residents played no role in planting ormaintaining the vegetation outside their apartments and were randomlyassigned to levels of greenness. Levels of aggression and violence were sys-tematically lower for individuals living in green surroundings than for indi-viduals living in barren surroundings; moreover, lack of nature significantlypredicted levels of mental fatigue, which in turn significantly predictedaggression. Mediation testing indicated that the relationship between vegeta-tion and aggression was fully mediated through attention (Kuo & Sullivan,in press).
In sum, there is a variety of evidence suggesting that vegetation may be linked to lower levels of crime in residential neighborhoods, particularlypoor inner-city neighborhoods. Residential vegetation has been linked with agreater sense of safety, fewer incivilities, and less aggressive and violentbehavior. Of these findings, the most direct evidence of a negative linkbetween vegetation and crime comes from residents’ reports of illegal activities in the space outside their apartment building and from residents’self-reports of (criminally) aggressive behavior.
The study presented here is the first to examine the relationship between vegetation and crime in an inner-city neighborhood using police crimereports. Although police crime reports are far from infallible (O’Brien,1990), one advantage of such reports is that they are based on actual counts ofcrimes reported over the course of a year and thus are less subject to the dis-tortions introduced by having residents estimate the frequencies of suchevents from memory. Thus, the convergence of findings from resident reportsand police reports would lend confidence to a negative link between vegeta-tion and crime. In this study, we examined the relationship between the vege-tation outside of apartment buildings and the number of police crime reportsfor those buildings over a 2-year period. We collected police data on propertycrimes, violent crimes, and total crimes for 98 apartment buildings in oneinner-city neighborhood and used the amount of tree and grass cover outsideeach building to predict crime.
Data presented here were collected as part of the Vital Neighborhood Common Spaces archive, a multistudy research effort examining the effectsof the physical environment on the functioning of individuals, families, andcommunities residing in urban public housing.
POPULATION, SETTING, AND DESIGN
Ida B. Wells is a large public housing development in Chicago. Wells pro- vides housing for approximately 5,700 individuals, of which 65% are female,97% are African American, and 44% are children younger than 14 years old(Chicago Housing Authority, 1995). Ida B. Wells is one of the 12 poorestneighborhoods in the United States (Ihejirika, 1995). At the time of thisstudy, approximately 93% of the people living at Wells were officially unem-ployed, and roughly 50% of the families received Aid to Families withDependent Children (Chicago Housing Authority, 1995).
The amount of nature outside apartment buildings at Ida B. Wells varies considerably. When the development was originally built in the 1940s, treesand grass were planted around each of the low-rise buildings. Over time,many of these green spaces have been paved in an effort to keep dust downand maintenance costs low; this paving has killed many of the original trees, Figure 1: Ground Level View at Ida B. Wells Showing Apartment Buildings With
Varying Amounts of Tree and Grass Cover
leaving some areas completely barren, others with small trees or some grass,and still others with mature high-canopy trees (see Figure 1). Because shrubswere relatively rare, vegetation at Ida B. Wells was essentially the amount oftree and grass cover around each building.
A number of apartment buildings at Wells were excluded from this study.
First, the high-rise and midrise (seven-story) buildings were excluded to keepthe buildings sampled similar in size, number of residents, and amount ofoutdoor common space. Second, of the 124 low-rise (one to four stories)apartment buildings, those buildings adjacent or nearly adjacent to the policestation within the development were excluded because the presence of policeofficers would be expected to be a significant deterrent to crime. And finally,a small cluster of low-rise buildings was excluded because the buildings’irregular placement with respect to each other and the street made it unclearwhere the common space associated with one building ended and the nextbegan. The final sample included 98 buildings.
Ida B. Wells offers a number of rare methodological advantages for inves- tigating the relationship between residential vegetation and crime. Althoughlevels of vegetation outside the apartment buildings vary considerably, theresidents are strikingly homogeneous with respect to many of the individualcharacteristics that have been shown to increase vulnerability to crime—income, education, and life circumstances. This similarity among residentscoupled with the consistent low-rise architecture decreases the sources ofextraneous variability in crime. This increases the power to detect differencesin the amount of crime associated with differences in the level of vegetationoutside each apartment building.
Perhaps more important, the apartment assignment procedures and land- scaping policies of public housing work to ensure that there are no systematic relationships between the vegetation outside an apartment building and thecharacteristics of its residents. Applicants for public housing at Ida B. Wells(and elsewhere in Chicago public housing) are assigned to individual apart-ments without regard for the level of nearby vegetation. And although resi-dents have some choice in accepting or rejecting a particular apartment intheory, in practice the level of nearby vegetation is not a significant factor inresidents’ choices, and most residents simply accept the first available apart-ment (Kuo, Sullivan, et al., 1998). Moreover, residents play little or no role indecisions to introduce or remove trees. Thus, in this study, there were no a pri-ori reasons to expect a relationship between the level of vegetation outside anapartment building and the characteristics of its inhabitants—more “respon-sible” residents might just as likely live in barren buildings as in greenbuildings.
MEASURES
Crime reports. Chicago Police Department year-end Uniform Crime Reports were analyzed for this study. These crime reports summarize foreach address at Ida B. Wells the specific crimes (e.g., aggravated assault andstrong-armed robbery) that were reported during the year. These reportsinclude both citizen-initiated complaints and those filed by an officer withouta citizen complaint.
When a crime is reported to the police, an officer is dispatched to interview the victim or victims and any witnesses. The officer then files a report aboutthe incident describing the specific crime or crimes, the date, the addresswhere the crime(s) occurred, and other pertinent information. Details fromthis report are then summarized in the year-end crime reports.
From 2 years of crime reports, we created three summary variables index- ing crime for each low-rise apartment building at Ida B. Wells, following theclassification scheme used by the Department of Justice (Bureau of JusticeStatistics, 1999). In this scheme, property crime is the sum of simple thefts,vehicle thefts, burglaries, and arson; violent crime includes assaults, batter-ies, robberies, and homicides; and total crimes is the sum of all crimesreported.
Vegetation. To assess the density of trees and grass around each of the low-rise buildings, we took dozens of 35mm slide photographs of the devel-opment by helicopter, passing over each cluster of buildings from a numberof vantages (see Figure 2). We also took ground-level photographs of many ofthe outdoor spaces. All the slides were taken in June when the tree canopy Figure 2: Aerial View of a Portion of Ida B. Wells Showing Buildings With Varying
Amounts of Tree and Grass Cover
was full and the grass was green. For each building, the aerial slides were puttogether with slides taken at ground level; there were at minimum three dif-ferent views from aerial and ground-level photos of each space (front, back,left side, and right side) around each building. Five students in landscapearchitecture and horticulture then independently rated the level of vegetationin each space. Each of the individuals rating the spaces received a map of thedevelopment that defined the boundaries of the specific spaces under study.
The raters viewed the slides and recorded their ratings on the maps. A total of220 spaces was rated, each on a 5-point scale (0 = no trees or grass, 4 = a spacecompletely covered with tree canopy). Interrater reliability for these ratingswas .94.1 The five ratings were averaged to give a mean nature rating for eachspace. The nature ratings for the front, back, and side spaces around eachbuilding were then averaged to produce a summary vegetation rating. Ratingsof vegetation for the 98 buildings ranged from 0.6 to 3.0.
Other factors likely to affect crime. Four additional variables possibly related to vegetation and the number of crimes reported per building wereassessed through (a) on-site analysis, (b) Chicago Housing Authority floor Simple Ordinary Least Squares Regressions
Using Vegetation to Predict Crimes Per Building
plans of each building type in the development, and (c) Chicago HousingAuthority apartment vacancy records.
Number of units is the number of apartment units in a building; the range Number of occupied units is the average number of units rented in a partic- ular building during the 2 years of the study; the mean was 7.8, and the rangewas from 0.5 to 15. We were able to obtain data on 84 of the 98 buildings inthis sample.
Vacancy is the 2-year average of the number of vacant apartments divided by the number of units in the building; the mean was 13%, and the range wasfrom 0% to 92%. We were able to obtain data on 84 of the 98 buildings in thissample.
Building height is the number of floors in a building; the range was from 1 If vegetation reduces crime, then we would expect to find that the greener a building’s surroundings are, the fewer crimes reported. Perhaps the moststraightforward test of this possibility is to conduct simple regressions withvegetation as the independent variable and the three summary crime indicesas dependent variables (see Table 1). Results from these ordinary leastsquares regressions indicate that vegetation is significantly and negativelyrelated to each of the measures of crime. The greener a building’s surround-ings are, the fewer total crimes; this pattern holds for both property crimesand violent crimes. For each of the three indices, vegetation accounts for 7%to 8% of the variance in the number of crimes reported per building.
Figure 3 provides a more concrete sense of the amount of crime associated with different levels of vegetation. For this figure, the continuous vegetationvariable was recoded into the following three categories: low (ratings from Figure 3: Mean Number of Crimes Reported Per Building for Apartment Build-
ings With Different Amounts of Vegetation (each icon represents one
reported crime)
0.0 up to 1.0), medium (from 1.0 up to 2.0), and high (from 2.0 up to 3.0,inclusive). Figure 3 shows the average number of total, property, and violentcrimes reported for buildings with low, medium, and high levels of vegeta-tion. Compared to buildings with low levels of vegetation, those withmedium levels had 42% fewer total crimes, 40% fewer property crimes, and44% fewer violent crimes. The comparison between low and high levels ofvegetation was even more striking: Buildings with high levels of vegetationhad 52% fewer total crimes, 48% fewer property crimes, and 56% fewer vio-lent crimes than buildings with low levels of vegetation. Fisher’s protectedleast significant difference analyses indicate that for each measure of crime,low and medium buildings were significantly different at p < .05. The samepattern held for comparisons between low and high buildings. Althoughbuildings with high levels of vegetation had 17% fewer total crimes, 13%fewer property crimes, and 21% fewer violent crimes than buildings withmedium levels of vegetation, these differences were not statisticallysignificant.
These data reveal a clear negative relationship between vegetation and crime and hint that this relationship is strongest when comparing buildingswith low levels of vegetation to buildings with either medium or high levels.
Although these findings are exciting and intriguing, they do not control forother important variables. The analyses that follow provide a closer look at Multiple Regressions Using Number of Units
and Vegetation to Predict Crimes Per Building
NOTE: The multiple regressions for total crimes: adjusted R 2 = .52 (N = 98, p < .0001); for propertycrime: adjusted R 2 = .45 (N = 98, p < .0001); for violent crime: adjusted R 2 = .44 (N = 98, p < .0001).
the relationship between vegetation and crime, taking into account other fac-tors likely to affect the number of crimes per building.
TESTING POTENTIAL CONFOUNDS
Controlling for number of apartments. Perhaps one of the most important variables to control for in predicting the amount of crime in a setting (e.g., abuilding, neighborhood, or city) is the number of people in that setting.
Because more apartments per building mean more potential perpetrators andmore potential victims, one would expect more crimes in buildings with moreapartments. Indeed, previous research has shown the number of units in abuilding to be related to the number of reported crimes (Newman & Franck,1980). Thus, it is not surprising that in this sample, strong positive linear rela-tionships exist between the number of units and the number of propertycrimes (r = .62, p < .0001), violent crimes (r = .63, p < .0001), and total crimes(r = .67, p < .0001). That is, the more apartments in a building, the morecrimes reported for that building.
To examine whether the relationship between vegetation and crime still held when the number of apartments in a building was controlled, a series ofmultiple regressions were conducted in which both vegetation and number ofunits were used to predict the number of crimes reported per building. AsTable 2 shows, when the number of units per building is controlled, vegeta-tion continues to be a significant negative predictor of total crime, propertycrime, and violent crime. In other words, the level of greenness around abuilding at Ida B. Wells predicts the number of crimes that have occurred inthat building even after the number of apartments in the building has beenaccounted for.
Intercorrelations Among Possible Predictors
of Crime and Three Crime Scales
Other potential confounds. To identify other potential confounds between vegetation and crime, correlations were conducted between vegetation andthe following three factors that have been shown in other studies to be associ-ated with crime: vacancy rate (R. B. Taylor, Shumaker, & Gottfredson, 1985),the number of occupied apartments per building (Newman & Franck, 1980),and building height (Newman, 1972; Newman & Franck, 1980). As the firstcolumn in Table 3 shows, vegetation is not related to either vacancy rate ornumber of occupied units but is strongly and negatively related to buildingheight; the taller the building is, the lower the level of vegetation. The fourthcolumn in Table 3 indicates that building height has a strong positive relation-ship to total crime, property crime, and violent crime. Thus, the relationshipbetween vegetation and crime is confounded by building height: Taller build-ings are both less green and have more reported crimes than shorter buildings.
These findings raise the possibility that vegetation predicts crime only by vir-tue of its shared variance with building height.
To test for this possibility, we examined whether vegetation still predicts crime when building height and number of units are controlled. Table 4 pro-vides the results of a series of multiple regressions in which vegetation, build-ing height, and number of units were used to predict crime. If vegetationpredicts crime by virtue of its relationship with building height, then vegeta-tion should no longer predict crime when building height is controlled, andbuilding height should predict crime with vegetation controlled. As Table 4 Multiple Regression Using Three Independent Variables (number of
units, vegetation, and building height) to Predict Crimes Per Building
NOTE: The multiple regressions for total crimes: adjusted R 2 = .51 (N = 98, p < .0001); for propertycrime: adjusted R 2 = .44 (N = 98, p < .0001); for violent crime: adjusted R 2 = .43 (N = 98, p < .0001).
shows, however, this is not the case; vegetation remains a significant or mar-ginally significant predictor of crime with building height and number ofunits controlled. Moreover, building height has no predictive power whenvegetation and number of units are controlled. These findings indicate thatalthough building height is confounded with vegetation, it cannot account forthe link between vegetation and crime.
Thus far, the analyses have established that (a) there is a reliable associa- tion between the amount of vegetation outside a building and the number ofcrimes recorded for that building by the police, (b) these relationships areindependent of the number of units in a building, and (c) these relationshipsare independent of building height. These analyses show that vegetation pre-dicts crime and that this relationship cannot be accounted for by these otherconfounding variables.
DOES ADDING VEGETATION IMPROVE THE
CURRENT ARSENAL OF CRIME PREDICTORS?
To determine whether vegetation makes any unique, additional contribu- tion to the current arsenal of predictors, we conducted a multiple regressionin which all available significant predictors of crime were entered (i.e., vege-tation, other predictors that were confounded with vegetation, and other pre-dictors that were not confounded with vegetation). This kitchen-sinkmultiple regression, in which vegetation and number of units, buildingheight, vacancy rate, and number of occupied units were entered as predic-tors, indicated that vegetation does make a unique contribution to the currentarsenal of predictors. Vegetation was a significant predictor of total crime (β= –1.1, p = .05) even when all other crime predictors have been accounted for.
Moreover, the relatively low variance inflation factor for vegetation in thisregression (1.31) indicates that vegetation is relatively independent of the other predictors. In addition, comparison of the adjusted R2s of the kitchen-sink multiple regressions with and without vegetation indicated that the addi-tional predictive power gained by adding vegetation outweighs the loss ofdegrees of freedom incurred in increasing the total number of predictors. Theadjusted R2 for the model with only the current arsenal of predictors was .23;the adjusted R2 for the model with the current arsenal of predictors plus vege-tation was .26. Although this increase represents only 3% of the total variancein crime, it represents a sizable proportion of the current predictive power(13%). Together, these findings indicate that adding vegetation improves thecurrent arsenal of predictors, adding unique explanatory power.
A Cuthbert plot (Cp) analysis yielded additional evidence of the predic- tive power of vegetation. Cp analysis is a technique for determining the mostpowerful, most parsimonious model out of a set of multiple predictors (SASInstitute, 1998). Essentially, given a set of predictors, Cp analysis tests allpossible combinations of predictors and selects the best model. An alterna-tive to comparing adjusted R2s, Cp analysis is particularly helpful when thereis multicollinearity between predictors, as was the case here. Cp analysisindicated that the best model for predicting total crime, selecting from theentire set of available predictors (number of units, building height, vacancyrate, number of occupied units, and vegetation), comprises only two predic-tors—number of units and vegetation (Cp = 1.32). Thus, in these data, thebest possible model of crime comprises only vegetation and one otherpredictor.
DISCUSSION
This study examined the relationship between vegetation and crime for 98 apartment buildings in an inner-city neighborhood. Analyses revealed con-sistent, systematically negative relationships between the density of trees andgrass around the buildings and the number of crimes per building reported tothe police. The greener a building’s surroundings are, the fewer total crimes;moreover, this relationship extended to both property crimes and violentcrimes. Levels of nearby vegetation explained 7% to 8% of the variance in thenumber of crimes reported per building. The link between vegetation andcrime could not be accounted for by either of the two confounding variablesidentified. Vegetation contributed significant additional predictive powerabove and beyond four other classic environmental predictors of crime. Andout of all possible combinations of available predictors, vegetation was iden-tified as one of the two predictors in the best possible model of crime.
The findings contribute to our understanding of the relationship between vegetation and crime and suggest opportunities for intervention and futureresearch.
CONTRIBUTIONS TO THE UNDERSTANDING OF VEGETATION AND CRIME
One contribution of this work is to propose a systematic exception to the rule that vegetation promotes crime. The rule in both folk theory and environ-mental criminology has been that vegetation promotes crime by providingconcealment for criminals and criminal activities. If the mechanism by whichvegetation affects crime is indeed concealment, then one implication of thisrule is that vegetation should not promote crime when it preserves visibility.
The contribution here is simply to point out that many forms of vegetationpreserve visibility and therefore ought not promote crime. Indeed, we foundthat in this sample of inner-city apartment buildings, buildings with widelyspaced, high-canopy trees and grassy areas did not experience higher rates ofcrime. These findings suggest that at the very least, crime prevention con-cerns do not justify removing high-canopy vegetation in inner-city neighbor-hoods. They demonstrate that one of the classic suspects in environmentalcriminology does not always promote crime.
Moreover, the findings indicate a large and systematically negative link between levels of vegetation and police reports of crime in this setting.
Although this is the first study to demonstrate such a link, the findings areconsistent with previous work linking vegetation with lower levels of incivil-ities (Brunson, 1999; Stamen, Yates, & Cline, as cited in S. Sullivan, 1993) aswell as previous work linking vegetation with lower levels of aggression andviolence (Kuo & Sullivan, in press). The results obtained here were based onpolice crime reports, whereas the Brunson (1999) and the Kuo and Sullivan(in press) findings were based on residents’ memories and self- reports. Theconvergence of findings from such different measures lends confidence thatin inner-city residential settings, the relationship between vegetation andcrime is negative—the more vegetation, the less crime.
A third contribution of the work here is to help resolve a puzzle in previous work on residential vegetation and sense of safety. A number of studies havefound that residential vegetation is associated with greater sense of safety(Brower et al., 1983; Kuo, Bacaicoa, et al., 1998; Kuo, Sullivan, et al., 1998;Nasar, 1982). In combination with the old rule that vegetation promotescrime, such findings raised the disturbing possibility that residents systemati-cally misperceive green areas as safe. And yet other research has found goodconcurrent validity between measures of fear, perceptions of disorder, andmedia reports of crime (e.g., Perkins & Taylor, 1996). The finding here that vegetation is systematically linked with lower levels of crime suggests thatindividuals are accurate in their perception of green areas as safer.
A final contribution of this work is to propose two mechanisms by which vegetation may deter crime in inner-city neighborhoods. Specifically, wepropose that vegetation may deter crime both by increasing informal surveil-lance and by mitigating some of the psychological precursors to violence.
Although neither of these mechanisms—nor the more general question ofcausality—can be addressed in these data, there is clear empirical support forthese mechanisms in other work. Substantial previous research has shownthat surveillance deters crime and that in inner-city neighborhoods, greeneroutdoor spaces receive greater use, thereby increasing informal surveillance.
Moreover, Kuo and Sullivan’s (in press) work showed that for residents ran-domly assigned to apartment buildings with different levels of vegetation,higher levels of vegetation systematically predicted lower levels of aggres-sion, and mediation analyses indicated that this link was mediated viaattentional functioning. In addition, we can address a number of alternativeinterpretations for the findings here. Public housing policies in this setting aresuch that levels of income, education, and employment among residents arelargely held constant; residents are randomly assigned to varying levels ofvegetation; and the amount of trees and grass outside an apartment is notunder residents’ control. And the confound analyses conducted here indicatethat the link between vegetation and lower crime could not be explained by anumber of classic environmental predictors of crime—vacancy rates, build-ing height, the number of apartments, and the number of occupied apartmentsin a building.
POSSIBILITIES FOR INTERVENTION AND FUTURE RESEARCH
The findings in this study set the stage for more ambitious explorations of the relationship between urban residential vegetation and crime. Now thatthere is good reason to think that visibility-preserving vegetation does notnecessarily promote crime and may even inhibit crime in inner-city neighbor-hoods, it seems appropriate to attempt an intervention study or two. Interven-tion studies employing true experimental designs might be used to answer anumber of important questions with regard to the effects of vegetation oncrime. Urban public housing communities might be especially amenablesites for such research as housing authorities tend to have centralized controlover landscaping for dozens and even hundreds of identical buildings.
A study in which identical or matched apartment buildings in a poor urban area were randomly assigned to receive different levels of vegetation couldhelp address the question of causality and the question of the shape of the relationship between vegetation and crime. Would crime rates decrease lin-early or curvilinearly with increasing vegetation? In this sample, the differ-ence between low and moderate green cover buildings was 3.1 crimes, but thedifference between moderate and high green cover buildings was only 0.7crimes. One possible interpretation of this pattern is that the relationshipbetween vegetation and crime is nonlinear with diminishing returns. Anotheris that the 0.7 crime difference between the moderate and high vegetationconditions is a poor estimate because of the relatively low number ofhigh-vegetation buildings in the sample, and the relationship between vege-tation and crime is actually linear across the entire range of vegetation.
Future studies might systematically vary the arrangement and mainte- nance of vegetation and examine the rates of crime associated with these fac-tors. The vegetation in this study was not configured to provide symbolicbarriers or to mark the territory of particular apartment buildings. Wouldarrangements that create symbolic barriers and delineate the territory of par-ticular residences (e.g., with small hedges) be more effective in decreasingcrime than other arrangements? Brown and colleagues (Brown & Altman,1983; Brown & Bentley, 1993) found evidence suggesting that plants andother territorial markers may make a property less attractive for burglary, butno study has yet randomly assigned different planting arrangements to differ-ent buildings and compared the resulting rates of property crime. Analo-gously, well-maintained vegetation seems to be a particularly effectiveterritorial marker (Chaudhury, 1994), but research has yet to systematicallyexamine the effect of different levels of maintenance on crime.
Future research might also look more closely—and more broadly—at the outcomes of planting interventions. In this sample, vegetation predicted lev-els of both property crime and violent crime. This is noteworthy given thatstudies in environmental criminology often find that the relationship betweenthe physical environment and crime depends on the specific category ofcrime (e.g., Brantingham & Brantingham, 1993). It would be interesting anduseful to examine the relationships between vegetation and more specific cat-egories of crime or other categories altogether. For instance, does vegetationhave more of an effect on impulsive crimes than on “rational” crimes? Wemight expect impulsive crimes committed out of frustration or rage to bereduced through the beneficial effects of vegetation on mental fatigue. And tothe extent that perpetrators consciously calculate risks in selecting their tar-gets, more “rational,” premeditated crimes might be reduced through thebeneficial effects of vegetation on informal surveillance.
In examining the outcomes of planting interventions, it will be important to address the possible displacement of crime. One of the standard concernsin efforts to combat crime is that although interventions may reduce crime in targeted locations, the effect may be to simply displace crime to other areas,yielding no overall decrease in crime (Gabor, 1981). Would adding vegeta-tion and decreasing crime in one part of an inner-city neighborhood simplyincrease crime in another part of the neighborhood? The answer may dependon the type of crime in question. By reducing the irritability, impulsivity, andcognitive deficits associated with mental fatigue and hence preventing minorconflicts from spiraling out of control, vegetation might inhibit violentcrimes in some residences without increasing violent crimes in others. On theother hand, by increasing informal surveillance of some outdoor spaces with-out reducing the actual impetus for burglary and other premeditated crimes,vegetation might serve to simply shift such crimes to more vulnerable targets.
Future research should examine rates of crime both in and around the inter-vention areas.
Such comparisons might shed light on the mechanisms by which vegeta- tion affects crime. To further address the question of mechanism, levels ofinformal surveillance and mental fatigue might be measured in buildingsreceiving the planting intervention and in matched buildings selected as con-trols. Mediation analyses could then be conducted to examine the joint linksbetween vegetation, crime, and the proposed mediators. Does vegetationaffect crime only when it increases residents’ use of outdoor spaces and lev-els of informal surveillance? Finally, one exciting possibility for future work would be to compare the outcomes from intervention studies in which residents were either involvedor uninvolved in the greening process. The question here would be whetherthe process of tree planting could enhance residents’ territoriality, therebydeterring crime over and above the direct effect of the presence of vegetation.
Active involvement in tree-planting programs has been claimed to enhance acommunity’s sense of territoriality (Dwyer, McPherson, Schroeder, &Rowntree, 1992), and the community greening lore is replete with stories inwhich greening efforts have been accompanied by dramatic decreases incrime and incivilities (e.g., Hynes, 1996; Lewis, 1980; Littman, 1996; Trustfor Public Lands, 1996). Previous research in inner-city neighborhoods sug-gests that residents would be willing to help plant and care for trees (Kuo,Bacaicoa, et al., 1998). As planting is the single largest cost associated withthe care and maintenance of the urban forest (McPherson, Nowak, &Rowntree, 1994), involving residents would substantially defray the alreadylow costs associated with a planting intervention.
Ultimately, the largest reductions in crime will come from strategies that address the factors underlying crime (e.g., intense poverty and the availabil-ity of guns). In the meantime, this study offers a ray of hope by identifying aneasily manipulable environmental feature that has a systematic, negative relationship with property crimes, violent crime, and total crimes. The workpresented here suggests the exciting possibility that in barren inner-cityneighborhoods, planting a few trees may work to inhibit crime, creating saferneighborhoods for poor families and their children.
1. In these data, agreement between raters is analogous to the reliability of items in a scale; the hope is that different raters will respond to a particular building in a similar fashion. Thus, toassess interrater agreement, a Cronbach’s alpha was calculated with individual raters treated likeindividual items in a scale and individual buildings treated like individual respondents.
REFERENCES
Bennett, T. (1989). Burglars’ choice of targets. In D. Evans & D. Herbert (Eds.), The geography of crime (pp. 176-192). New York: Routledge.
Bennett, T., & Wright, R. (1984). Burglars on burglary: Prevention and the offender. Brookfield, Brady, K. T., Myrick, H., & McElroy, S. (1998). The relationship between substance use disor- ders, impulse control disorders, and pathological aggression. American Journal on Addic-tions, 7, 221-230.
Brantingham, P. L., & Brantingham, P. J. (1993). Nodes, paths and edges: Considerations on the complexity of crime and the physical environment. Journal of Environmental Psychology,13, 3-28.
Brower, S., Dockett, K., & Taylor, R. B. (1983). Residents’ perceptions of territorial features and perceived local threat. Environment and Behavior, 15, 419-437.
Brown, B. B., & Altman, I. (1983). Territoriality, defensible space and residential burglary: An environmental analysis. Journal of Environmental Psychology, 3, 203-220.
Brown, B. B., & Bentley, D. L. (1993). Residential burglars judge risk: The role of territoriality.
Journal of Environmental Psychology, 13, 51-61.
Brunson, L. (1999). Resident appropriation of defensible space in public housing: Implications for safety and community. Unpublished doctoral dissertation, University of Illinois, Urbana-Champaign.
Bureau of Justice Statistics. (1999). National crime victimization survey: Criminal victimization 1998, changes 1997-98 with trends 1993-98 (NCJ 176353). Washington, DC: Department ofJustice Office of Justice Programs.
Canin, L. H. (1991). Psychological restoration among AIDS caregivers: Maintaining self-care.
Unpublished doctoral dissertation, University of Michigan.
Caprara, G. V., & Renzi, P. (1981). The frustration-aggression hypothesis vs. irritability.
Recherches de Psychologie Sociale, 3, 75-80.
Chaudhury, H. (1994). Territorial personalization and place-identity: A case study in Rio Grande Valley, Texas. In A. D. Seidel (Ed.), Banking on design (pp. 46-54). Oklahoma City, OK:EDRA.
Chicago Housing Authority. (1995). Statistical profile: The Chicago Housing Authority 1994 to Cimprich, B. (1993). Development of an intervention to restore attention in cancer patients.
Cancer Nursing, 16, 83-92.
Coccaro, E. F., Bergeman, C. S., Kavoussi, R. J., & Seroczynski, A. D. (1997). Heritability of aggression and irritability: A twin study of the Buss-Durkee aggression scales in adult malesubjects. Biological Psychiatry, 41, 273-284.
Coleman, A. (1987). Utopia on trial: Vision and reality in planned housing. London: Shipman.
Coley, R. L., Kuo, F. E., & Sullivan, W. C. (1997). Where does community grow? The social con- text created by nature in urban public housing. Environment and Behavior, 29, 468-492.
Cromwell, P. F., Olson, J. N., & Avary, D. W. (1991). Breaking and entering: An ethnographic analysis of burglary. Hillsdale, NJ: Lawrence Erlbaum.
Dwyer, J. F., McPherson, E. G., Schroeder, H. W., & Rowntree, R. A. (1992). Assessing the ben- efits and costs of the urban forest. Journal of Aboriculture, 18, 227-234.
Fisher, B. S., & Nasar, J. L. (1992). Fear of crime in relation to three exterior site features: Pros- pect, refuge, and escape. Environment and Behavior, 24, 35-65.
Gabor, T. (1981). The crime displacement hypothesis: An empirical examination. Crime and Delinquency, 27, 390-404.
Hartig, T., Mang, M., & Evans, G. W. (1991). Restorative effects of natural environment experi- ence. Environment and Behavior, 23, 3-26.
Hynes, H. P. (1996). A patch of Eden: America’s inner-city gardeners. White River Junction, Jacobs, J. (1961). The death and life of great American cities. New York: Random House.
Jeffery, R. C. (1971). Crime prevention through environmental design. Beverly Hills, CA: Sage.
Kant, R., Smith-Seemiller, L., & Zeiler, D. (1998). Treatment of aggression and irritability after head injury. Brain Injury, 12, 661-666.
Kaplan, R. (1984). Wilderness perception and psychological benefits: An analysis of a continu- ing program. Leisure Sciences, 6, 271-290.
Kaplan, S. (1987). Mental fatigue and the designed environment. In J. Harvey & D. Henning (Eds.), Public environments (pp. 55-60). Washington, DC: Environmental Design ResearchAssociation.
Kavoussi, R. J., & Coccaro, E. F. (1998). Divalproex sodium for impulsive aggressive behavior in patients with personality disorder. Journal of Clinical Psychiatry, 59, 676-680.
Kuo, F. E., Bacaicoa, M., & Sullivan, W. C. (1998). Transforming inner-city landscapes: Trees, sense of safety, and preference. Environment and Behavior, 30, 28-59.
Kuo, F. E., & Sullivan, W. C. (in press). Aggression and violence in the inner city: Impacts of environment and mental fatigue. Environment & Behavior, 33(4).
Kuo, F. E., Sullivan, W. C., Coley, R. L., & Brunson, L. (1998). Fertile ground for community: Inner-city neighborhood common spaces. American Journal of Community Psychology, 26,823-851.
Lewis, C. A. (1980). Gardening programs promote improved maintenance and community rela- tions in public housing developments. Journal of Housing, 37, 614-617.
Littman, M. (1996). Green city. The neighborhood works: Building alternative visions for the city [Online]. Available: www:http://cnt.org/tnw/193grnci.htm Lohr, V. I., Pearson-Mimms, C. H., & Goodwin, G. K. (1996). Interior plants may improve worker productivity and reduce stress in a windowless environment. Journal of Environmen-tal Horticulture, 14, 97-100.
Macdonald, J. E., & Gifford, R. (1989). Territorial cues and defensible space theory: The bur- glar’s point of view. Journal of Environmental Psychology, 9, 193-205.
Markovitz, P. (1995). Pharmacotherapy of impulsivity, aggression, and related disorders. In E. Hollander (Ed.), Impulsivity and aggression (pp. 263-287). Chichester, UK: Wiley.
McPherson, E. G., Nowak, D. J., & Rowntree, R. A. (1994). Chicago’s urban forest ecosystem: Results of the Chicago urban forest climate project (Report NE-186). Washington, DC:USDA Forest Service, Northeastern Forest Experiment Station.
Merry, S. E. (1981). Defensible space undefended. Urban Affairs Quarterly, 16, 397-422.
Michael, S. N., & Hull, R. B. (1994). Effects of vegetation on crime in urban parks. Blacksburg: Virginia Polytechnic Institute and State University, College of Forestry and WildlifeResources, Department of Forestry.
Michael, S. N., Hull, R. B., & Zahm, D. L. (1999). Environmental factors influencing auto bur- glary: A case study. Manuscript submitted for publication.
Miles, I., Sullivan, W. C., & Kuo, F. E. (1998). Prairie restoration volunteers: The benefits of par- ticipation. Urban Ecosystems, 2(1), 27-41.
Mooney, P., & Nicell, P. L. (1992). The importance of exterior environment for Alzheimer resi- dents: Effective care and risk management. Healthcare Management Forum, 5(2), 23-29.
Nasar, J. L. (1982). A model relating visual attributes in the residential environment to fear of crime. Journal of Environmental Systems, 11, 247-255.
Nasar, J. L., & Fisher, B. S. (1993). “Hot spots” of fear and crime: A multi-method investigation.
Journal of Environmental Psychology, 13, 187-206.
Nassauer, J. I. (1988). Landscape care: Perceptions of local people in landscape ecology and sus- tainable development. In Landscape and land use planning: Proceedings from the 1988International Federation of Landscape Architects World Congress (pp. 27-41). Washington,DC: American Society of Landscape Architects.
Newman, O. (1972). Defensible space: Crime prevention through urban planning. New York: Newman, O., & Franck, K. (1980). Factors influencing crime and instability in urban housing developments. Washington, DC: Law Enforcement Assistance Administration, NationalInstitute of Law Enforcement and Criminal Justice.
O’Brien, R. M. (1990). Estimating the reliability of aggregate-level variables based on individ- ual-level characteristics. Sociological Methods and Research, 18, 473-503.
Ovitt, M. A. (1996). The effect of a view of nature on performance and stress reduction of ICU nurses. Unpublished master’s thesis, University of Illinois.
Perkins, D. D., Brown, B. B., & Taylor, R. B. (1996). The ecology of empowerment: Predicting participation in community organizations. Journal of Social Issues, 52, 85-110.
Perkins, D. D., & Taylor, R. B. (1996). Ecological assessments of community disorder: Their relationship to fear of crime and theoretical implications. American Journal of CommunityPsychology, 24, 63-107.
Perkins, D. D., Wandersman, A., Rich, R. C., & Taylor, R. B. (1993). The physical environment of street crime: Defensible space, territoriality and incivilities. Journal of EnvironmentalPsychology, 13, 29-49.
Pluncknett, T.F.T. (1960). Edward I and criminal law. Cambridge, UK: Cambridge University Poyner, B., & Webb, B. (1992). Reducing theft of shopping bags in city center markets. In R. V.
Clarke (Ed.), Situational crime prevention: Successful case studies (pp. 99-107). New York:Harrow & Henston.
Rhodes, W. M., & Conley, C. (1981). Crime and mobility: An empirical study. In P. J.
Brantingham & P. L. Brantingham (Eds.), Environmental criminology (pp. 167-188).
Beverly Hills, CA: Sage.
SAS Institute. (1998). Users guide, Volume 2, Version 6 (4th ed.). Cary, NC: Author.
Scholte, R.H.J., van Aken, M.A.G., & van Leishout, C.F.M. (1997). Adolescent personality fac- tors in self-ratings and peer nominations and their prediction of peer acceptance and peerrejection. Journal of Personality Assessment, 69, 534-554.
Schroeder, H. W., & Anderson, L. M. (1984). Perception of personal safety in urban recreation sites. Journal of Leisure Research, 16, 178-194.
Shaffer, G. S., & Anderson, L. M. (1985). Perceptions of the security and attractiveness of urban parking lots. Journal of Environmental Psychology, 5, 311-323.
Stamen, T. (1993). Graffiti deterrent proposed by horticulturalist [Press release]. Riverside: University of California, Riverside.
Stanford, M. S., Greve, K. W., & Dickens, T. J. (1995). Irritability and impulsiveness: Relation- ship to self-reported impulsive aggression. Personality and Individual Differences, 19,757-760.
Stewart, M. A. (1985). Aggressive conduct disorder: A brief review. Aggressive Behavior, 11, Sullivan, S. (1993, July 21). Anti-tagger idea that grows on you. The Press Enterprise, p. 1.
Sullivan, W. C., Kuo, F. E., & DePooter, S. (2001). Tree cover and social activities in inner-city neighborhood common spaces. Manuscript in preparation.
Talbot, J., & Kaplan, R. (1984). Needs and fears: The response to trees and nature in the inner city. Journal of Arboriculture, 10, 222-228.
Taylor, A. F., Wiley, A., Kuo, F. E., & Sullivan, W. C. (1998). Growing up in the inner city: Green spaces as places to grow. Environment and Behavior, 30, 3-27.
Taylor, R. B. (1988). Human territorial functioning. New York: Cambridge University Press.
Taylor, R. B., Shumaker, S. A., & Gottfredson, S. D. (1985). Neighborhood-level links between physical features and local sentiments: Deterioration, fear of crime, and confidence. Journalof Architectural and Planning Research, 2, 261-275.
Tennessen, C., & Cimprich, B. (1995). Views to nature: Effects on attention. Journal of Environ- mental Psychology, 15, 77-85.
Trust for Public Lands. (1996). Healing America’s cities: How urban parks can make cities safe and healthy. San Francisco: Author.
Tuinier, S., Verhoeven, W.M.A., & Van Praag, H. M. (1996). Serotonin and disruptive behavior: A critical evaluation of the clinical data. Human Psychopharmacology Clinical and Experi-mental, 11, 469-482.
Weisel, D. L., Gouvis, C., & Harrell, A. V. (1994). Addressing community decay and crime: Alternative approaches and explanations (Final report submitted to the National Institute ofJustice). Washington, DC: The Urban Institute.

Source: http://willsull.net/Publications_files/Kuo,%20Sullivan%202001%20crime.pdf