(2006) 212–217www.elsevier.com/locate/ypmed

Preventive Medicine 43

Parents' perceptions of neighborhood safety and children's physical activity

Lori A. Weir a,⁎, Debra Etelson b, Donald A. Brand b,c,d

a Department of Family Medicine, New York Medical College, Munger Pavilion, Room 306, Valhalla, NY 10595, USAb Department of Pediatrics, New York Medical College, Valhalla, NY 10595, USAc Department of Medicine, New York Medical College, Valhalla, NY 10595, USA

d Primary Care Research Unit, New York Medical College, Valhalla, NY 10595, USA

Available online 19 May 2006

Abstract

Objective. The obesity epidemic disproportionately affects minority and poor children. Negative perceptions of neighborhood safety in poorcommunities may affect overweight by inhibiting children's physical activity. This study investigates the degree to which parents in a poor innercity vs. a middle-class suburban community limit their children's outdoor activity because of neighborhood safety concerns.

Method. Parents of children aged 5–10 years from an inner city family practice in a poor community and from a suburban pediatric practice ina middle-class community completed a 20-item questionnaire. Parents estimated the amount of their child's activity in various situations andindicated their level of anxiety concerning gangs, child aggression, crime, traffic, and personal safety in their neighborhood.

Results. Inner city children (n = 204) engaged in less physical activity than suburban children (N = 103) (P b 0.001). Inner city parentsexpressed much greater anxiety about neighborhood safety than suburban parents (P b 0.0001). In the inner city population, children's physicalactivity levels were negatively correlated with parental anxiety about neighborhood safety (r = −0.18, P b 0.05).

Conclusions. Inner city parents have high levels of anxiety about neighborhood safety. While these concerns may not entirely explain thediscrepancy in activity levels between inner city and suburban children, a safe environment is crucial to increasing opportunities for physicalactivity.© 2006 Elsevier Inc. All rights reserved.

Keywords: Exercise; Child; Poverty areas; Social problems; Residence characteristics; Socioeconomic factors

Introduction

Comparedwith the general pediatric population, children fromdisadvantaged minority families have lower levels of physicalactivity (Lee and Cubbin, 2002; Gorden-Larsen et al., 2000) andhigher rates of obesity (FIFCFS, 2004; CDCNCHS, 1999–2000).These disparities may reflect differences in access to safe playareas, availability of after-school programs, and environmentalfactors such as traffic and street safety (Giles-Corti and Donovan,2002; Saelens et al., 2003; Timperio et al., 2005).

A few studies have examined the relationship between percep-tions of neighborhood safety and physical activity in selectedpediatric populations (Romero et al., 2001). Minority adolescentgirls but not boys are less active if they worry about neighborhoodsafety (Gomez et al., 2004). School-age children are less likely towalk or bicycle as a mode of transportation if their parents have

⁎ Corresponding author. Fax: +1 914 594 4698.E-mail address: lori_weir@nymc.edu (L.A. Weir).

0091-7435/$ - see front matter © 2006 Elsevier Inc. All rights reserved.doi:10.1016/j.ypmed.2006.03.024

safety concerns (Timperio et al., 2004). Inner city minority parentsand parents of chronically ill children may limit children's activitydue to environmental safety concerns (Goodway and Smith, 2005;Holoday et al., 1997). In preschool children, parental perceptionsof neighborhood safety increase the amount of time children spendwatching television but do not appear to influence outdoor activity(Burdette and Whitaker, 2005).

This study further explored this relationship with respect toschool-aged children. Specifically, we compared the degree towhich parents in a poor inner city vs. a middle-class suburbancommunity limit their children's outdoor activity because ofneighborhood safety concerns.

Methods

Study design

This cross-sectional study used a self-administered questionnaire to askparents of children 5 to 10 years old about their perceptions of neighborhoodsafety and about their child's physical activities. The study included parents of

Table 1Demographics of survey respondents and their children

Inner city (N = 204) Suburban (N = 103)

Child's age, in years(mean ± SD)

7.4 ± 1.9 6.9 ± 1.6

Child's sexBoy 109 (54%) 50 (49%)Girl 94 (46%) 49 (48)Not answered 1 (b1%) 4 (4%)Respondent's relationship to childMother 175 (83%) 89 (86%)Father 21 (10%) 13 (13%)Grandparent 3 (1%) 1 (1%)Other 11 (5%) 0 (0%)Respondent's educationLess than high school 53 (26%) 0 (0%)High school 60 (29%) 9 (9%)Some college 38 (19%) 22 (21%)Bachelor degree 12 (6%) 36 (35%)Master/Doctoral degree 4 (2%) 28 (27%)Not answered 37 (18%) 8 (8%)Respondent's raceHispanic 155 (76%) 16 (16%)Black 23 (11%) 18 (17%)White 11 (5%) 52 (50%)Other 4 (2%) 7 (7%)Not answered 11 (5%) 10 (10%)

213L.A. Weir et al. / Preventive Medicine 43 (2006) 212–217

children seen in a community health center located in a low-income inner citycommunity and, for comparison, parents of children seen in a private pediatricpractice in a middle-class suburban community.

Settings

The inner city family health center cares for a medically underserved,predominately minority population in a community adjacent to New York City.Over a quarter of children in this community live below the poverty line and40% of residents do not speak English (USCB, 2005). The private pediatricfaculty practice is located 13 miles from the family health center, in the samecounty. This practice serves a predominately middle-class, Caucasian populationfrom surrounding suburbs. Both practices are affiliated with and staffed byphysicians from New York Medical College.

Survey instrument

Since no existing instrument included all of the variables needed to addressour study objective, we developed a survey instrument by adapting content andmethods from previously tested and validated instruments. To assess physicalactivity, parents estimated the average weekly time their child engaged inphysical activity in various contexts. Parental estimates strongly correlate withdirect measures of moderate to vigorous physical activity in children (Welk etal., 2000). Weekly recall produces acceptable test–retest reliability (Sallis et al.,1993; Sallis and Saelens, 2000) and – for children aged 5 to 10 – parental reportyields greater reliability and validity than self-report (Sallis and Saelens, 2000;Koo and Rohan, 1999). For questions about parental perceptions ofneighborhood safety, we adapted items from the Neighborhood EnvironmentWalkability Scale (Sallis, 2004), International Physical Activity PrevalenceStudy Self-administered Environmental Module (IPS, 2006) and from a studyabout parental perceptions of local neighborhood (Timperio et al., 2004).

We adjusted the wording and format based on parents' feedback obtainedduring pilot testing. The final 20-question survey was a reversible bookletcontaining English and Spanish versions. Reverse translation confirmed theequivalency of the English and Spanish wording.

After requesting the child's age and sex and the respondent's relationship to thechild, the questionnaire asked a series of multiple-choice questions about theactivities of the respondent's youngest child between the ages of 5 and 10.To assessphysical activity, the survey asked howmany hours per week (none, 1–6, 7–13, or14+) the child spent in each of four activities (school-sponsored after-school sportsor dance programs, other after-school sports or dance programs, outdoor play, andoutdoor play with an adult). To assess the extent of outside play, respondentsestimated the frequency of play (never, sometimes, frequently, always) in each offour situations (outside the house; in the neighborhood; in a neighborhood park; ina neighborhood park with an adult). To assess anxiety about gangs, crime,aggression by other children, traffic, and neighborhood safety in general,respondents indicated their agreement with seven statements using a 4-pointscale ranging from “strongly agree” to “strongly disagree.” “Anxiety” refers to fearor worry about neighborhood safety. Finally, the questionnaire requested therespondent's zip code, level of education, race (Black,White, Hispanic, Other), andprovided space for comments.

Survey administration

Parents of appropriately aged children with a scheduled appointment at theinner city health center between July 28 and October 22, 2004, or at the suburbanprivate practice between October 22, 2004, and February 11, 2005, were invited tocomplete the questionnaire. Surveys attached to the patient's medical record beforethe appointment were distributed to the parent in the examination room by theoffice staff. Completed surveys were returned to a drop box in the waiting area. Noincentives were offered to complete this 5-minute, anonymous, voluntary survey.The institutional review board of New York Medical College approved the study.

Statistical analysis

For each question, we tabulated the number and percentage of parents whochose each response. We created a composite activity score by assigning the codes

0, 1, 2, and 3 to response categories “none”, “1–6”, “7–13”, and “14+” h/week,respectively, and calculating the sum of the coded responses to the four activityquestions, giving a potential range of 0 to 12. When calculating the activity score,we imputed missing values by replacing missing answers with the mean value ofthe coded responses of all respondents from the site who answered the question.However, if more than half the activity questions were blank, we did not computean activity score for that respondent. Imputed values comprised 2.8% of responsesfrom the inner city site and 1.0% of responses from the suburban site. We usedanalogous methods to derive a composite anxiety score based on the seven anxietyquestions (potential range 0–21) and to impute missing values (2.6% of responsesat the inner city site and 0.8% at the suburban site).

We used Chi-squared tests to determine whether responses to individualquestions varied by study site, Mann–Whitney tests to compare activity andanxiety scores across sites, and Kruskal–Wallis tests with Dunn's posttesting tocompare these scores across racial groups. To test for an association betweenindividual parents' anxiety and their child's overall activity, we calculated theSpearman correlation coefficient from the anxiety and activity scores ofrespondents with both scores.

Results

Overview

Of the 564 eligible visits to the inner city site, 283 parentsreceived the survey. The remaining 281 parents did not receivethe survey because they failed to keep their scheduledappointments or, in a few cases, because the medical recordwas not available to attach the survey. Altogether, 215 of the 283parents who received surveys returned them, of which 204 (94English and 110 Spanish) were sufficiently complete for analysisgiving a response rate of 72% (204/283).

Of the 165 eligible visits to the suburban site, 143 parentsreceived the survey, 111 of which were returned and 103 suffi-ciently complete for analysis (101 English and 2 Spanish),giving a response rate of 72% (103/143). The same reasons

Fig. 1. Numbers above the bars indicate percents.P values refer to differences between sites in the distribution of responses. n/a = not answered. NS = not significant (P N 0.05)

214 L.A. Weir et al. / Preventive Medicine 43 (2006) 212–217

.

215L.A. Weir et al. / Preventive Medicine 43 (2006) 212–217

(cancellations, no-shows, or unavailability of the medicalrecord) prevented 22 parents from receiving the survey.

Respondent demographics are summarized in Table 1. Asexpected, inner city parents had higher minority representationand lower education levels than suburban parents.

Children's physical activity

Fig. 1 summarizes respondents' answers to the activity andoutside questions. The activity score, which had a potentialrange of 0 to 12, summarized responses to the four activityquestions. The mean activity scores were 2.8 (1.8) and 3.4 (1.4)

Fig. 2. Numbers above the bars indicate percents. P values refer to differen

for the inner city and suburban children, respectively, indicatingless physical activity in the former group (P b 0.001 by Mann–Whitney test). Analysis of responses to combinations of activityquestions (data not shown in Fig. 1) revealed that 58% (119/204) of inner city children vs. 30% (31/103) of suburbanchildren do not participate in any organized sports or danceprograms (P b 0.0001 by Chi-squared test) and that 21% (42/204) vs. 4% (4/103) neither participate in these programs norplay outside their house, except when accompanied by an adult(P b 0.0001).

To evaluate whether physical activity varied by respondents'race, we computed mean activity scores for each racial group. At

ces between sites in the distribution of responses. n/a = not answered.

216 L.A. Weir et al. / Preventive Medicine 43 (2006) 212–217

the inner city site, the activity scores were 2.8 for Hispanic, 2.7 forBlack, 2.5 for White, and 2.8 for Other/Unknown (not signi-ficant). At the suburban site, the scores were 3.6 for Hispanic, 2.8for Black, 3.7 forWhite, and 3.2 for Other/Unknown (P b 0.05 byKruskal–Wallis test). The difference between Black and Whitegroups was statistically significant at the suburban site (P b 0.05by Dunn's posttest).

Perceived neighborhood safety

Fig. 2 summarizes responses to the neighborhood safetyquestions. The anxiety score, a composite of responses to the 7questions, had a potential range of 0 to 21. The anxiety scoreswere12.4 (4.6) and 4.9 (3.3) for the inner city and suburban parentsrespectively, indicating much higher anxiety about neighborhoodsafety in the former group (P b 0.0001 byMann–Whitney test). Inevery question, inner city parents expressed markedly greaterconcern than suburban parents. By combining “strongly agree”and “agree” responses to yield the percent of parents whoworried;70% of inner city parents vs. 12% of suburban parents worriedabout their child being threatened by gangs, 62% vs. 14%worriedthat other childrenmight hurt their child, 36%vs. 9% felt that therewas no safe play area in their neighborhood, 58% vs. 8% believedit was dangerous to let a child play outside, 60% vs. 27% felt thattraffic was a problem, 50% vs. 3% believed that neighborhoodcrime made it unsafe to play outdoors, and 48% vs. 3% feltpersonally unsafe in their own neighborhood (P b 0.0001 for eachcomparison, by Chi-squared tests).

To evaluate whether perceived neighborhood safety variedby respondents' race, we computed the mean anxiety scores foreach racial group. At the inner city site, the anxiety scores were13.1 for Hispanic, 8.4 for Black, 11.6 for White, and 12.6 forOther/Unknown (P b 0.0001 by Kruskal–Wallis test). At thesuburban site, the scores were 7.1 for Hispanic, 5.6 for Black,3.7 for White, and 5.6 for Other/Unknown (P b 0.05). One pair-wise difference (Hispanic vs. Black) was significant at the innercity site (P b 0.001 by Dunn's posttest) and one pair-wisedifference (Hispanic vs. White) was significant at the suburbansite (P b 0.01).

Forty inner city parents wrote comments. The most commonof these included 13 that expressed a need for greater security inthe community, noting slow police response times, racism, ab-sence of a neighborhood crime watch, drugs, and prostitution; 7that expressed worry about children's safety; and 6 thatmentioned the need for improved access to child care andafter-school programs. Of the 13 suburban parents who wrotecomments, 10 mentioned that weather influences their child'soutdoor activity.

Correlation analysis

Joint analysis of the activity and anxiety scores revealed that,in the inner city environment, the overall amount of children'sphysical activity had a weak negative correlation with parentalanxiety about neighborhood safety (Spearman r = −0.18; P =0.01; N = 188). No correlation was found in the suburbanenvironment (P = 0.35; N = 97).

Discussion

Although it is not surprising to find high anxiety aboutneighborhood safety in an inner city community but not in asuburban community, the degree of contrast is unsettling. Innercity parents worried not only about the safety of their children'splay environment but also about their own personal safety.Actual crime statistics justify the differing perceptions in thetwo communities: 25% more violent crimes per capita occur inthe inner city vs. the suburban community (Area Connect2005a,b). Our findings suggest that safety concerns play a keyrole in limiting children's outdoor activities and underscore thebasic need for security to create an environment that supportsthese activities.

Our survey revealed a higher level of parental anxiety aboutneighborhood safety and a lower level of children's physicalactivity in a poor inner city community as compared with amiddle-class suburban community. Although this type of cross-sectional study cannot prove a cause–effect relationship(Morgenstern, 1995), the results support the study hypothesis.Furthermore, parents' responses to the neighborhood safetyquestions in themselves strongly suggest that inner city parentsrestrict their children's activities because they believe theirneighborhoods are unsafe (Fig. 2).

If safety concerns cause parents to restrict their children'soutdoor activities, it would be reasonable to anticipate a negativecorrelation between individual parents' anxiety scores and theirchildren's activity scores. The negative correlation we found atthe inner city site demonstrates that the relationship betweenanxiety and activity leans in the hypothesized direction in thatcommunity. However, the low magnitude of the correlation(−0.18) implies that a combination of factors must determinechildren's activity levels, since parental anxiety explains only asmall portion of the variability.

Because we collected inner city data during summer and fallmonths and suburban data during the winter, weather effectswould not have been equivalent. Since adverse winter weatherwould have hindered physical activity in suburban children, thecontrast in physical activity between the two populations mayhave been more pronounced had we collected data at the twosites concurrently. It is unlikely that concurrent data collectionwould have changed the direction of our findings with respect tophysical activity.

Since our data were obtained from one inner city and onesuburban site, the data may not be representative of all suchpractices. It will be important to replicate this survey at othersites. It is also important to acknowledge that our study couldnot discern which specific characteristics of the inner city andsuburban populations were most responsible for the observeddifferences between them. It is reasonable to hypothesize that acombination of socioeconomic and demographic characteris-tics, population density, living arrangements, attitudes towardpolice, and actual crime statistics might explain the differentperceptions, but further studies are needed to examine theseissues in more detail.

Our study targeted an underserved inner city population, whileusing a more affluent, suburban population for comparison. The

217L.A. Weir et al. / Preventive Medicine 43 (2006) 212–217

study suggests that parental concerns about neighborhood safetyplay a role in limiting children's outdoor activities in the formerpopulation, but not in the latter. Physicians should be sensitive toparents' safety concernswhen discussing opportunities for outdooractivity. In addition, by communicating with public officials andcommunity activists – especially in inner city areas – physicianscould become advocates for enhanced neighborhood safety andimproved options for physical activity.

Conclusions

This study found high levels of worry about neighborhoodsafety in a poor inner city community and relatively low levelsof concern in a neighboring suburban community. The studyalso found that inner city children were less physically activethan suburban children. These results, combined with anegative correlation between parents' concern and children'sactivity in the inner city population, suggest that inner cityparents restrict their children's physical activity due to safetyconcerns.

Acknowledgments

We thank Marybeth Radley and Will Sheldon for their assis-tance with data collection and Patricia A. Patrick, MPH, forassistance with the analysis.

This study was supported by Award 2D54HP00022 from theHealth Resources and Services Administration.

References

Area Connect White Plains Crime Statistics and Data Resources. Accessed onlineat: http://whiteplains.areaconnect.com/crime1.htm. Accessed October 31, 2005.

Area Connect: Yonkers New York Crime Statistics and Data Resources. Accessedonline at: http://yonkers.areaconnect.com/crime1.htm. Accessed October 26,2005.

Burdette, H.L., Whitaker, R.C., 2005. A national study of neighborhood safety,outdoor play, television viewing, and obesity in preschool children. Pediatrics116, 657–662.

Centers for Disease Control and Prevention, National Center forHealth Statistics. National Health and Nutrition Examination Survey1999–2000.

Federal Interagency Forum on Child and Family Statistics. America's Children

2004, Table POP8.C. Available at: http://childstats.gov/ac2004. AccessedJanuary 19, 2005.

Giles-Corti, B., Donovan, R.J., 2002. Socioeconomic status differences inrecreational physical activity levels and real and perceived access to asupportive physical environment. Prev. Med. 35, 601–611.

Gomez, J.E., Johnson, B.A., Selva, M., Sallis, J.F., 2004. Violent crime andoutdoor physical activity among inner-city youth. Prev. Med. 39,876–881.

Goodway, J.D., Smith, D.W., 2005. Keeping all children healthy: challenges toleading an active lifestyle for preschool children qualifying for at-riskprograms. Fam. Commun. Health 28 (2), 142–155.

Gorden-Larsen, P., McMurray, R.G., Popkin, B.M., 2000. Determinants ofadolescent physical activity and inactivity patterns. Pediatrics 105, E83.

Holoday, B., Swan, J.H., Turner-Henson, A., 1997. Images of the neighborhoodand activity patterns of chronically ill school age children. Environ. Behav.29, 348–370.

International Physical Activity Prevalence Study Self-Administered Environ-mental Module. Available at: http://www.drjamessallis.sdsu.edu/IPAQIPS.pdf. Accessed March 13, 2006.

Koo, M.M., Rohan, T.E., 1999. Comparison of four habitual physical activityquestionnaires in girls aged 7–15 yr. Med. Sci. Sports Exerc. 31, 421–427.

Lee, R.E., Cubbin, C., 2002. Neighborhood context and youth cardiovascularhealth behaviors. Am. J. Public Health 92, 428–436.

Morgenstern, H., 1995. Ecologic studies in epidemiology: concepts, principles,and methods. Annu. Rev. Public Health 16, 61–81.

Romero, A.J., Robinson, T.N., Kraemer, H.C., Erickson, S.J., Haydel, F.,Mendozo, F., Killen, J.D., 2001. Are perceived neighborhood hazards abarrier to physical activity in children? Arch. Pediatr. Adolesc. Med. 155,1143–1148.

Saelens, B.E., Sallis, J.F., Black, J.B., Chen, D., 2003. Neighborhood-baseddifferences in physical activity: an environment scale evaluation. Am. J.Public Health 93, 1552–1558.

Sallis, J.F. Neighborhood Environment Walkability Scale. Available at: http://www.drjamessallis.sdsu.edu/NEWS.pdf. Accessed February 18, 2004.

Sallis, J.F., Saelens, B.E., 2000. Assessment of physical activity by self report:status, limitations, and future directions. Res. Q. Exerc. Sport 71, 1–14.

Sallis, J.F., Buono, M.J., Roby, J.J., Micale, F.G., Nelson, J.A., 1993. Seven-dayrecall and other physical activity self-reports in children and adolescents.Med. Sci. Sports Exerc. 25, 99–108.

Timperio, A., Crawford, D., Telford, A., Salmon, J., 2004. Perceptions about thelocal neighborhood and walking and cycling among children. Prev. Med. 38,39–47.

Timperio, A., Salmon, J., Telford, A., Crawford, D., 2005. Perceptions of localneighbourhood environments and their relationship to childhood overweightand obesity. Int. J. Obes. 29 (2), 170–175.

USCensus Bureau. Available at: http://censtats.census.gov/data/NY/1603684000.pdf. Accessed January 24, 2005.

Welk, G.J., Corbin, C.B., Dale, D., 2000. Measurement issues in the assessmentof physical activity in children. Res. Q. Exerc. Sport 71 (suppl. 2), S59–S73.