PREVENTING CHRONIC DISEASE PUBLIC HEALTH RESEARCH, PRACTICE, AND POLICY Volume 17, E73 JULY 2020 ORIGINAL RESEARCH Effects of Crime Type and Location on Park Use Behavior Oriol Marquet, PhD1; S. Scott Ogletree, PhD2; J. Aaron Hipp, PhD2; Luis J. Suau, PhD2; Candice B. Horvath, PhD2; Alexander Sinykin3; Myron F. Floyd, PhD2 Accessible Version: www.cdc.gov/pcd/issues/2020/19_0434.htm shortest route between home and park. By using ordered and bin- ary logistic modeling, we examined the relationships between the Suggested citation for this article: Marquet O, Ogletree SS, locations of crime and park use and duration of park visit, park rat- Hipp JA, Suau LJ, Horvath CB, Sinykin A, et al. Effects of Crime ing, and never visiting parks. Additional models included distance Type and Location on Park Use Behavior. Prev Chronic Dis 2020; to the closest park. 17:190434. DOI: https://doi.org/10.5888/pcd17.190434. Results PEER REVIEWED Increased crime in parks and near home was associated with few- er park visits. Greater violent crime in all locations was related to fewer park visits. Park ratings were lower for parks with high viol- Summary ent crime rates. What is already known about this topic? Crime and the fear of crime can be a barrier to park use by adults, hinder- ing the benefits parks can have for physical activity and health. Conclusion What is added by this report? Given the importance of parks as settings for outdoor recreation and physical activity, crime may have a detrimental effect on Using crime incident data, we examined the role crime locations (near home, near parks, and along routes to parks) play in park use. physical activity and, therefore, public health. What are the implications for public health practice? Higher levels of crime near home and parks limit park use and residents’ Introduction rating of parks. Sources of crime must be considered if parks are to be a means of promoting physical activity in adults. Real and perceived crime Fear of crime is a barrier to outdoor physical activity and park use can be a barrier to park use. among US adults (1,2). Given the importance of urban parks as settings for outdoor recreation and physical activity (3), the pres- ence of crime may negatively affect the public’s health. Research Abstract has shown that violence is usually spatially clustered and tends to remain stable over time (4). Living in areas exposed to crime and Introduction violence is associated with trauma and mental health issues and is Crime and the fear of crime can be a barrier to park use, and loca- linked to psychological symptoms such as depression and anxiety tions of crimes can have varied effects. Unsafe areas in or around (5,6). In addition, the presence of crime and violence may deter the park, around the residence, or along the route to the park can people’s use of parks, reducing the likelihood of their engaging in alter park use behavior. Our study aimed to examine associations healthy behaviors. between objective measures of types and location of crimes and park use behaviors. Strong evidence exists that high-quality green spaces in residen- tial environments are important for public health promotion (7), Methods because both availability and quality of green spaces near the res- In 2013 we surveyed a sample (N = 230) of residents in Greens- idence have positive and significant associations with park use and boro, North Carolina, about park use, with responses matched to physical activity (8). However, having a park available does not objective crime and spatial measures. We measured all crimes and necessarily increase park use (9), because this relationship may be violent crimes near home, near the closest park, and along the affected by the presence of crime and perceptions of safety. Fear of crime may restrict physical activities and lead people to avoid The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors’ affiliated institutions. www.cdc.gov/pcd/issues/2020/19_0434.htm • Centers for Disease Control and Prevention 1 This publication is in the public domain and is therefore without copyright. All text from this work may be reprinted freely. Use of these materials should be properly cited. PREVENTING CHRONIC DISEASE VOLUME 17, E73 PUBLIC HEALTH RESEARCH, PRACTICE, AND POLICY JULY 2020 places or situations they perceive to be unsafe. Crime can affect Methods park use through 3 different pathways. First, living in areas with high crime may pose a barrier to engaging in outdoor activity and, Our study was part of a data collection effort undertaken in 2013 instead, favor safe, interior environments. Second, researchers in Greensboro, North Carolina. Greensboro, the third largest city have explored how the presence of crime can hinder walking be- in North Carolina, has a population of 269,666 (19) and covers an havior, with the potential to reduce access to neighborhood parks area of 127 square miles. In 2013 the median household income in (10). If walking is deemed dangerous, then access to a park would Greensboro was $41,050 (20), and the racial/ethnic composition of require driving or public transit, creating other types of con- the population was 47.8% white, 41.6% African American, 7.5% straints such as car availability and the availability of parking both Hispanic of any race, 4.7% Asian, and 5.9% other or more than 2 at the beginning and end of the trip. Third, crimes that occur in- races (19). The city was highly segregated and had a large propor- side or near the park deter people from engaging in normal park tion of minority neighborhoods (21). The park system dates to activities. Parks can be gathering areas for gang members (11), 1933 and comprises 170 properties (parks, gardens, and special fa- and some park designs — with obstructed view-lines and places to cilities). hide — facilitate crime (12). Study sample Evidence regarding crime and park use is limited; however, re- search in North Carolina found that parks located in neighbor- We conducted a mail survey to obtain data on park use and on rat- hoods with high crime indexes had less park use by all subpopula- ing of parks, physical activity, self-rated health, height and weight, tions except adolescents (13). Shinew et al found that adolescents and demographics from a sample of adults randomly drawn from avoided gang territories and other high-crime areas for organized residential addresses located within a quarter mile of one of 21 sports and outdoor recreation (11). Han et al found that violent Greensboro parks. Parks were purposely selected to capture both crimes had a stronger negative effect on park use than property low-income neighborhoods (n = 7) and middle-to-high income crimes and observed that adults were more likely to change their neighborhoods (n = 14) with only 1 park within a quarter-mile ra- park visiting habits than younger people (14). Marquet et al used dius. The selected parks were described by the city as neighbor- crime reports to associate crime changes and park attendance in 3 hood parks from 5 to 15 acres in size that served residences in a different temporal thresholds (more than 1 week, 1 month, or 3 half-mile radius (a 10-minute walk) (22). Residential addresses in months), and found that violent crime affected park use and the the study areas were obtained from county GIS parcel data number of children that used parks (15). provided by the local government. We generated 893 residential addresses from street addresses rather than from owners’ ad- Most studies investigating crime and park use measured self- dresses to target people who either rent or own a home in a assessed perceptions regarding safety (10). Using objective meas- quarter-mile of each park. Areas surrounding parks in low-income ures of crime prevents self-reporting bias (8). Understanding how areas were oversampled by 10% to better capture low-income res- safety affects park use can facilitate translation of research find- idents. The protocol for the mail survey followed a modified ver- ings into interventions (16). Previous studies have been limited by sion of Dillman’s Tailored Design Method (23). We sent a letter to the use of aggregate measures of safety perception or crime in- 893 residences explaining the study and including a questionnaire dexes, which may mask true associations between crime and park and a postage-paid return envelope. A reminder post card was sent use (17). Using disaggregated measures of crime that account for to nonrespondents approximately 2 weeks after the first mailing. different levels of violence can help to overcome some of these Approximately 2 weeks after that reminder post card, we mailed a limitations. second letter, questionnaire, and postage-paid return envelope to nonrespondents. As an incentive, respondents had a chance to win Similarly, most studies use measures of crime indexes taken at an one of four $50 gift cards in a raffle. Of the 893 questionnaires aggregated level and do not differentiate by locations of the mailed, 145 were returned as undeliverable. Of the 748 deliver- crimes. Just as the effects of different domains of safety and types able surveys, 235 were completed and returned and 5 were elimin- of crime vary, different locations of crimes can have different ef- ated as incomplete or as duplicate questionnaires from the same fects (11,18). Although aggregated crime around the residence address. The final count of 230 usable questionnaires yielded a re- might be the main factor setting the perception of insecurity, the sponse rate of 31%. Early respondents were compared with late re- presence of unsafe areas around the park or along the route to the spondents to investigate nonresponse bias. No significant differ- park can also alter park use. Our study aimed to examine associ- ence in demographic characteristics was found between these 2 ations between objective measures of types and location of crimes groups. By using the respondent addresses obtained from county and park use behaviors. parcel GIS files and a survey question asking the name of the park The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors’ affiliated institutions. 2 Centers for Disease Control and Prevention • www.cdc.gov/pcd/issues/2020/19_0434.htm PREVENTING CHRONIC DISEASE VOLUME 17, E73 PUBLIC HEALTH RESEARCH, PRACTICE, AND POLICY JULY 2020 closest to the home, we were able to add contextual census-based centroid of the park, and distance from home to the closest park information at the block group level regarding average population feature (such as playground or sports court). Correlation was density and median household income for each participating strongest for distance to the park boundary (r[230] = −0.153, P = household. We used ArcGIS v10.3.1 (Esri) for all spatial computa- .02); therefore, we used distance to the park boundary as the main tions. measure of distance. Measures Statistical analysis Park use. Four survey questions measured park use: frequency of We used ordered logistic and binary logistic models to assess how park visits (How often do you usually visit the park closest to your type and location of crimes were associated with self-reported home?), duration of park visits (On a typical day when you visit measures of park use frequency, time spent in the park, park rat- the park closest to your home, how long do you usually stay?), and ing, and never going to the park. We entered presence of crimes as park rating (Overall how would you rate the parks in your neigh- a continuous dependent variable. To account for differences in borhood?). Responses were all ordered variables with no multiple neighborhood characteristics, all models were adjusted for the re- responses allowed. We also included a bivariate item that asked spondent’s age, sex, and race/ethnicity; the presence or absence of whether respondents never went to their closest parks at all. children in the household; and median household income (calcu- lated on the basis of census data). Separate models were estimated Crime. All offenses classified under the Summary Reporting Sys- for each type of crime (all crimes and Part I-only crimes) and for tem of the Federal Bureau of Investigation’s Criminal Justice In- each location’s criteria (crimes around the home, crimes around formation Service and the Uniform Crime Reporting System are the park, crimes along the shortest route from home to the park). classified either as Part I or Part II offenses (24). Part I offenses We then used distance to the park as an additional independent are criminal homicide, rape, robbery, aggravated assault, burglary, variable. For the distance to park variable we built 3 separate mod- larceny, motor vehicle theft, arson, and human trafficking; all els: a baseline model without adjusting for crime, a second model crimes are classified as Part II offenses. We used Greensboro Po- adjusting for all crimes, and a third model adjusting for Part I lice Department public data on all offenses reported in the city in crimes only. All crime variables and distance variables were stand- 2012, provided as geolocated incident points. A total of 31,845 of- ardized. We used odds ratios instead of coefficients for clarity in fenses were reported by the department, of which 14,789 (46.4%) interpretation. were classified as Part 1 offenses. We used ArcGIS to count the number of all offenses and Part I offenses only that occurred with- Because the measures of frequency of park use, duration of park in a 0.5 mile (10-minute walk) buffer of each participant resid- use, and park rating were ordinal, we used an ordered logistic re- ence (street network buffer) (25,26). Because the police report all gression to model them, whereas we used a binary logistic model crimes and offenses that occur within parks at the nearest address, for never going to the park. We used Stata SE v14.2 (StataCorp) we created a buffer of 15 meters around each park boundary and for all statistical analyses. counted how many offenses and Part I-only offenses occurred im- mediately around or within the park boundaries. Finally, we used Results the ArcGIS network analyst function to calculate the shortest routes from each respondent’s residence to their closest-to-home Our sample (N = 230) was predominantly women (73.9%) and park, assuming this to be the walking route from home to park. All consisted of older adults (mean age, 54.5 y; standard deviation offenses and Part I-only offenses that fell along this route were [SD], 16.4); 57.8% identified themselves as African American, then counted. Overall we used 3 measures of crime location and 68.3% reported children in the home (Table 1). Most parti- (around home, around park, and along the route from home to cipants had at least some college, a college degree, or a profes- park) and 2 crime classifications (all crimes and Part I crimes sional degree (78.9%) and reported their health as very good or only) as independent variables and to examine associations excellent (60.8%). Participants lived in areas slightly more densely between crime type, crime locations, and park use behavior. populated than the Greensboro average (4,030 people/mi2 vs 2,131 people/mi2) (Table 2) and in areas with higher median household Distance to park. Distance to the closest park was calculated by income than the overall Greensboro median ($61,859 vs $40,361). using ArcGIS network analyst functions. We performed a sensitiv- On average, participants had a park within a quarter mile of their ity analysis correlating park use frequency with distance from home (mean, 0.28 mi; SD, 0.16 mi). Participants also lived in home to the boundary of the park, distance from home to the areas with more crime than the city average (173 crimes per 1,000 population vs 49.28 crimes per 1,000 population for the city). The areas around parks had slightly higher crime rates than the resid- The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors’ affiliated institutions. www.cdc.gov/pcd/issues/2020/19_0434.htm • Centers for Disease Control and Prevention 3 PREVENTING CHRONIC DISEASE VOLUME 17, E73 PUBLIC HEALTH RESEARCH, PRACTICE, AND POLICY JULY 2020 ential neighborhoods of respondents, both in terms of all crimes Perception of safety is a key element for using parks (27). Violent and in terms of Part I crimes. Despite these numbers, 43.9% of our or egregious crimes are likely to affect safety perception more than sample perceived parks in their neighborhood to be very good or minor offenses. As recently investigated by Han et al, (14) violent excellent (Table 1). crimes tend to attract greater media coverage and thus have a greater effect on safety perception. By using the Federal Bureau of The relationship between categories of crime rates and crime loca- Investigation’s Uniform Crime Reporting categories, we were able tions varied with park behavior (Table 3). After accounting for to focus on violent crimes (Part I offenses) and explore their ef- age, sex, race, area income, and the presence of children at home, fects separately. However, the type of crime is likely not the only we found that the number of crimes located around the park was factor affecting park use. The actual location of the criminal of- associated with visiting the parks less often (OR = 0.51; 95% CI, fense is also relevant to the decision making of potential park 0.34–0.77). Similarly, crimes around the home were also associ- users. To date, previous studies have not analyzed the exact loca- ated with less frequent park visits (OR = 0.56; 95% CI, tion of crimes in relation to park use. 0.31–0.82). The number of crimes counted along the route to the park was associated with less frequent park visits to a smaller de- By using objective measures of crime and accounting both for the gree (OR = 0.73; 95% CI, 0.54–1.00) although results were bor- type of crime and the location of the offense, we observed that derline significant. Crime type or crime presence had no signific- crimes committed in or around the park were the strongest negat- ant effect on duration of park visits. Lower park ratings, however, ive predictor of frequency of park visits, followed closely by the were associated with higher crime presence around the park (OR = number of crimes around the residence. The number of all crimes 0.33; 95% CI, 0.22–0.50) and more Part I crimes committed located within the shortest route from the residence to the park had around the park (OR = 0.48; 95% CI, 0.33–0.71). Higher crime no significant effect on park use. The subset of only violent crimes rates around the home (OR = 0.54; 95% CI, 0.37–0.80) and Part I (Part I crimes) located either around the home, around the park, or crimes around the home were also associated with lower park rat- along the route from home to the park, were all significant predict- ings (OR = 0.52; 95% CI, 0.35–0.76). Finally, crime locations ors of decreased frequency of park use, with violent crimes around were also associated with those who reported they never went to the residence having the strongest effects. the park. Having high levels of crime around the park was associ- ated with more participants reporting that they never go to the park Having either more crimes or more violent crimes around the park (OR = 2.81; 95% CI, 1.69–4.68), and the same resulted with high- was most strongly associated with low park ratings by residents. er numbers of Part I crimes committed along the route to the park This is important given that previous research found park rating to (OR = 1.54; 95% CI, 1.03–2.32) and around the park (OR = 1.85; be a stronger predictor of park use than other factors, such as dis- 95% CI, 1.17–2.91). tance or accessibility (9,28). Our findings suggest that having more crimes in a park might decrease not only how often resid- The distance to the park influenced frequency and duration of park ents go to the park but also how they view and rate that park, po- visits, park rating, and the number of participants who reported tentially lowering a park’s appeal. Although we found no crime never going to the park (Table 4). After adjusting for age, sex, measure that affected duration of park visits, we did find that both race, area income, and the presence of children at home, the only violent and nonviolent crimes that occurred around the park and significant associations were distance to the park and duration of along the route to the park were positively associated with the park visits, with more distant parks being associated with longer number of people who said they never went to the park. Finally, park visits (OR = 1.46; 95% CI, 1.09–1.96). This relationship was our results also suggest that distance to the park affected only the unchanged when we controlled for the presence of all crime (OR = amount of time spent at the park and not the decision whether to 1.46; 95% CI, 1.08–1.96) and the presence of Part I crimes only go to the park. Interestingly, this association between distance and around the park (OR = 1.46; 95% CI, 1.08–1.96). length of stay appeared to be unchanged when we accounted for crime or violent crimes. Discussion Our sample included older African American women, a group Given the important health benefits of park use, park and land use most vulnerable to crime of the populations we studied, and re- organizations in the United States are advocating for more and search has shown that the most vulnerable groups tend to take pro- better parks within a 10-minute walk from home. However, creat- tective action and change behaviors when a threat is perceived (1). ing greater access may not lead to greater use, because use also Although previous studies showed a strong correlation between can be affected by the incidence of crime or by perceptions that crime rates and poverty (13), our sample’s median household in- neighborhoods are unsafe. The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors’ affiliated institutions. 4 Centers for Disease Control and Prevention • www.cdc.gov/pcd/issues/2020/19_0434.htm PREVENTING CHRONIC DISEASE VOLUME 17, E73 PUBLIC HEALTH RESEARCH, PRACTICE, AND POLICY JULY 2020 come was 28% higher than the Greensboro median. Our finding References that proximity to the nearest park was not associated with in- creased park use is similar to that of previous studies that found no 1. Roman CG, Chalfin A. Fear of walking outdoors. A multilevel association between park proximity and physical activity in the ecologic analysis of crime and disorder. Am J Prev Med 2008; park (9,17). 34(4):306–12. 2. Roman CG, Stodolska M, Yahner J, Shinew K. Pathways to Although our study provides some new insights, it has limitations. outdoor recreation, physical activity, and delinquency among We used self-reported measures of park use, which could cause re- urban Latino adolescents. Ann Behav Med 2013;45(Suppl call bias. Our research was cross-sectional in design and had a lim- 1):S151–61. ited sample, which did not allow for generalization. Additionally, 3. Evenson KR, Jones SA, Holliday KM, Cohen DA, McKenzie crime data from law enforcement were taken at face value despite TL. Park characteristics, use, and physical activity: a review of known limitations of such data (29). Significant strengths of our studies using SOPARC (System for Observing Play and study were our use of objective measures of crime and crime types Recreation in Communities). Prev Med 2016;86:153–66. rather than aggregate indexes and our relating the exact location of 4. Barboza G. A secondary spatial analysis of gun violence near crimes to 3 dimensions of neighborhood park use that can affect Boston schools: a public health approach. J Urban Health physical activity (park rating, frequency of visits, and duration of 2018;95(3):344–60. visit). 5. Berman SL, Silverman WK, Kurtines WM. Youth exposure to Future research is needed, given our study’s limitations. In partic- crime and violence: its effects and implications for ular, researchers are encouraged to explore whether our findings intervention. J Cogn Psychother 2000;14(1):37–50. can be duplicated with larger samples in more diverse geographic 6. Wilson WC, Rosenthal BS. The relationship between exposure regions and whether interactions between variables yield more nu- to community violence and psychological distress among anced results between crime, sex, or age and park use. Parks can adolescents: a meta-analysis. Violence Vict 2003; be an important component of quality of life, but crime can be a 18(3):335–52. barrier. How to reduce crime in and around parks will need to be 7. Weimann H, Rylander L, van den Bosch MA, Albin M, considered if parks and green spaces are to be a means of promot- Skärbäck E, Grahn P, et al. Perception of safety is a ing physical activity. prerequisite for the association between neighbourhood green qualities and physical activity: results from a cross-sectional Acknowledgments study in Sweden. Health Place 2017;45:124–30. 8. van Dillen SME, de Vries S, Groenewegen PP, Spreeuwenberg This research received partial funding from the US Forest Service, P. Greenspace in urban neighbourhoods and residents’ health: Southern Research Station (agreement 16-JV-11330144-065). We adding quality to quantity. J Epidemiol Community Health thank the Greensboro Police Department for providing the crime 2012;66(6):e8. data for the analysis. No copyrighted material was used in this art- 9. Kaczynski AT, Potwarka LR, Saelens BE. Association of park icle. size, distance, and features with physical activity in neighborhood parks. Am J Public Health 2008;98(8):1451–6. Author Information 10. Foster S, Giles-Corti B, Knuiman M. Does fear of crime discourage walkers? A social-ecological exploration of fear as Corresponding Author: Oriol Marquet, University of Barcelona, a deterrent to walking. Environ Behav 2014;46(6):698–717. Department of Geografía, Campus de Bellaterra, Edifici B, Carrer 11. Shinew KJ, Stodolska M, Roman CG, Yahner J. Crime, de la Fortuna, s/n, 08193, Bellaterra, Barcelona, Spain. Email: physical activity and outdoor recreation among Latino oriol.marquet@uab.cat. adolescents in Chicago. Prev Med 2013;57(5):541–4. 12. Iqbal A, Ceccato V. Does crime in parks affect apartment Author Affiliations: 1 Universitat Autònoma de Barcelona, prices? J Scand Stud Criminol Crime Prev 2015;16(1):97–121. Barcelona, Spain. 2 North Carolina State University, Raleigh, 13. Baran PK, Smith WR, Moore RC, Floyd MF, Bocarro JN, North Carolina. 3 Greensboro Police Department, Greensboro, Cosco NG, et al. Park use among youth and adults: North Carolina. examination of individual, social, and urban form factors. Environ Behav 2014;46(6):768–800. The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors’ affiliated institutions. www.cdc.gov/pcd/issues/2020/19_0434.htm • Centers for Disease Control and Prevention 5 PREVENTING CHRONIC DISEASE VOLUME 17, E73 PUBLIC HEALTH RESEARCH, PRACTICE, AND POLICY JULY 2020 14. Han B, Cohen DA, Derose KP, Li J, Williamson S. Violent 28. Giles-Corti B, Broomhall MH, Knuiman M, Collins C, crime and park use in low-income urban neighborhoods. Am J Douglas K, Ng K, et al. Increasing walking: how important is Prev Med 2018;54(3):352–8. distance to, attractiveness, and size of public open space? Am J 15. Marquet O, Hipp JA, Alberico C, Huang J-H, Fry D, Mazak E, Prev Med 2005;28(2,Suppl 2):169–76. et al. Short-term associations between objective crime, park- 29. Pepper J, Petrie C, Sullivan S. Measurement error in criminal use, and park-based physical activity in low-income justice data. In: Piquero AR, Weisburd D, editors. Handbook neighborhoods. Prev Med 2019;126:105735. of quantitative criminology. New York (NY): Springer; 2010, 16. Ribeiro AI, Pires A, Carvalho MS, Pina MF. Distance to parks p. 353–74. and non-residential destinations influences physical activity of older people, but crime doesn’t: a cross-sectional study in a southern European city. BMC Public Health 2015;15(1):593. 17. Ou JY, Levy JI, Peters JL, Bongiovanni R, Garcia-Soto J, Medina R, et al. A walk in the park: the influence of urban parks and community violence on physical activity in Chelsea, MA. Int J Environ Res Public Health 2016;13(1):E97. 18. Ding D, Sallis JF, Kerr J, Lee S, Rosenberg DE. Neighborhood environment and physical activity among youth a review. Am J Prev Med 2011;41(4):442–55. 19. US Census Bureau. State and county QuickFacts. Greensboro, North Carolina. 2010. https://www.census.gov/quickfacts/fact/ table/greensborocitynorthcarolina/POP010210. Accessed December 5, 2019. 20. City of Greensboro. Growth and development trends. 2012. http://www.greensboro-nc.gov/Modules/ ShowDocument.aspx?documentID= 13596. Accessed December 2, 2019. 21. Frey WH, Meyers D. CensusScope.org. North Carolina. Segregation: dissimilarity indices. 2013. http:// www.censusscope.org/us/s37/rank_dissimilarity_white_ black.html. Accessed December 5, 2019. 22. Harnik P, Simms J. How far to your nearest park? The Trust for Public Land. 2004. http://cloud.tpl.org/pubs/ccpe_ Distance_from_a_Park.pdf. Accessed December 3, 2019. 23. Dillman DA, Smyth JD, Christian LM. Internet, mail, and mixed-mode surveys: the tailored design method. Hoboken (NJ): John Wiley and Sons, Inc; 2009. 24. US Department of Justice. Summary reporting system user manual. 2013. www.fbi.gov/file-repository/ucr/ucr-srs-user- manual-v1.pdf/view. Accessed December 2, 2019. 25. Cohen DA, McKenzie TL, Sehgal A, Williamson S, Golinelli D, Lurie N. Contribution of public parks to physical activity. Am J Public Health 2007;97(3):509–14. 26. Harnik P, Martin A. Close-to-home parks: a half-mile or less. Center for City Park Excellence, Trust for Public Land. 2016. https://web.archive.org/web/20181003164620/https:// parkscore.tpl.org/Methodology/TPL_10MinWalk.pdf. Accessed December 3, 2019. 27. Weir LA, Etelson D, Brand DA. Parents’ perceptions of neighborhood safety and children’s physical activity. Prev Med 2006;43(3):212–7. The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors’ affiliated institutions. 6 Centers for Disease Control and Prevention • www.cdc.gov/pcd/issues/2020/19_0434.htm PREVENTING CHRONIC DISEASE VOLUME 17, E73 PUBLIC HEALTH RESEARCH, PRACTICE, AND POLICY JULY 2020 Tables Table 1. Characteristics, Survey of Respondents (N = 230)a on Relationship Between Park Use and Crime, Greensboro, North Carolina, 2013 Characteristic No. (%) Demographics Sex, n = 218 Female 161 (73.9) Male 57 (26.1) Age, y, n = 217 18–34 27 (12.4) 35–49 53 (24.4) 50–64 78 (35.9) ≥65 59 (27.2) Race/ethnicity, n = 217 Non-Hispanic white 83 (38.1) African American 126 (57.8) Other 9 (4.1) Children in household, n = 218 Yes 149 (68.3) No 69 (31.7) Education, n = 218 High school diploma or less 46 (21.1) Some college or associate's degree 69 (31.7) Bachelor degree/professional license 103 (47.2) Self-reported general health, n = 222 Poor or fair 30 (13.5) Good 57 (25.7) Very good or excellent 135 (60.8) 2 Body mass index (weight in kg/height in m ), n = 213 Normal (18.5 to <25) 74 (34.7) Overweight (25 to <30) 76 (35.7) Obese (≥30) 63 (29.6) Park Behavior Visit frequency, n = 229 Never 27 (11.8) Rarely 69 (30.1) Couple times a month 41 (17.7) Once a month 31 (13.5) Few times a week 47 (20.5) Daily 14 (6.1) a Number of respondents in each category varies by the number of valid responses to the question. (continued on next page) The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors’ affiliated institutions. www.cdc.gov/pcd/issues/2020/19_0434.htm • Centers for Disease Control and Prevention 7 PREVENTING CHRONIC DISEASE VOLUME 17, E73 PUBLIC HEALTH RESEARCH, PRACTICE, AND POLICY JULY 2020 (continued) Table 1. Characteristics, Survey of Respondents (N = 230)a on Relationship Between Park Use and Crime, Greensboro, North Carolina, 2013 Characteristic No. (%) Length of stay, n = 185 <15 min 29 (15.7) 15–30 min 70 (37.8) 31–60 min 49 (26.5) >1h–<2 h 26 (14.1) 2–3 h 11 (5.9) Park rating, n = 228 Poor 22 (9.6) Fair 47 (20.6) Good 59 (25.9) Very good 74 (32.5) Excellent 26 (11.4) a Number of respondents in each category varies by the number of valid responses to the question. The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors’ affiliated institutions. 8 Centers for Disease Control and Prevention • www.cdc.gov/pcd/issues/2020/19_0434.htm PREVENTING CHRONIC DISEASE VOLUME 17, E73 PUBLIC HEALTH RESEARCH, PRACTICE, AND POLICY JULY 2020 Table 2. Spatial Characteristics of Sample Households and Neighborhoods, Survey of Respondents (N = 230) on Relationship Between Park Use and Crime, Greensboro, North Carolina, 2013a Variable Neighborhood Around Residence Neighborhood Around Park 2 Density (no. people/mi ) 4,030.7 (3,698.0) 4,358.3 (4,337.9) Median household income, $ 61,858.6 (32,464.5) 64,770.2 (33,443.6) All crimes/1,000 people within block groupb 173.0 (185.8) 191.1 (206.1) Part 1 crimesc/1,000 people within block groupb 89.7 (128.0) 99.1 (150.5) d Count of all crimes within 0.5 mile buffer 322.7 (266.1) 550.2 (407.8) Count of Part 1 crimes in 0.5 mile bufferd 141.5 (122.8) 265.7 (244.1) a Values are mean (standard deviation). b Within the block group where the residence or park is located. c Criminal homicide, rape, robbery, aggravated assault, burglary, larceny, motor vehicle theft, arson, and human trafficking. d Unweighted count of crimes committed within a 0.5 miles street-network buffer of the residence or park. The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors’ affiliated institutions. www.cdc.gov/pcd/issues/2020/19_0434.htm • Centers for Disease Control and Prevention 9 PREVENTING CHRONIC DISEASE VOLUME 17, E73 PUBLIC HEALTH RESEARCH, PRACTICE, AND POLICY JULY 2020 Table 3. Effect of Type and Location of Crimes on Park Use, Survey of Respondents (N = 230) on Relationship Between Park Use and Crime, Greensboro, North Car- olina, 2013a Variable Frequency of Park Visits Duration of Park Visits Park Rating Never Goes to Park All crimes Home 0.56 (0.38–0.82) [.003] 0.99 (0.68–1.44) [.96] 0.54 (0.37– 0.80) [.002] 1.39 (0.76–1.87) [.44] Route 0.73 (0.54–1.00) [.05] 1.02 (0.74–1.41) [.89] 0.79 (0.58–1.07) [.13] 1.59 (1.07–2.36) [.02] Park 0.51 (0.34–0.77) [.001] 0.81 (0.52–1.24) [.33] 0.33 (0.22–0.50) [<.001] 2.81 (1.69–4.68) [<.001] Part I crimesb Home 0.60 (0.41–0.87) [.01] 0.94 (0.64–1.38) [.76] 0.52 (0.35–0.76) [.001] 1.40 (0.76–1.89) [.45] Route 0.71 (0.51–0.97) [.03] 0.96 (0.69–1.33) [.79] 0.71 (0.51–0.99) [.045] 1.54 (1.03–2.32)[.04] Park 0.64 (0.44–0.94) [.02] 0.72 (0.48–1.07)[.10] 0.48 (0.33–0.71) [<.001] 1.85 (1.17–2.91)[.01] a Values are odds ratio (95% confidence interval) [P value]. b Criminal homicide, rape, robbery, aggravated assault, burglary, larceny, motor vehicle theft, arson, and human trafficking. The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors’ affiliated institutions. 10 Centers for Disease Control and Prevention • www.cdc.gov/pcd/issues/2020/19_0434.htm PREVENTING CHRONIC DISEASE VOLUME 17, E73 PUBLIC HEALTH RESEARCH, PRACTICE, AND POLICY JULY 2020 Table 4. Effect of Distance to Park on Park Use, Survey of Respondents (N = 230) on Relationship Between Park Use and Crime, Greensboro, North Carolina, 2013a Variable Frequency of Park Visits Duration of Parks Visits Park Rating Never Goes to Park Without crime 1.07 (0.82–1.42) [.60] 1.46 (1.09–1.96) [.01] 1.01 (0.78–1.29) [.97] 0.82 (0.54–1.24) [.35] With all crime 1.04 (0.79–1.37) [.78] 1.46 (1.08–1.96) [.01] 0.95 (0.74–1.23) [.70] 0.84 (0.56–1.28) [.42] b Without Part I crime 1.05 (0.80–1.39) [.72] 1.46 (1.08–1.96) [.01] 0.96 (0.74–1.24) [.74] 0.84 (0.55–1.27) [.41] a Values are odds ratio (95% confidence interval) [P value]. b Criminal homicide, rape, robbery, aggravated assault, burglary, larceny, motor vehicle theft, arson, and human trafficking. The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors’ affiliated institutions. www.cdc.gov/pcd/issues/2020/19_0434.htm • Centers for Disease Control and Prevention 11