[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
October 4, 2010

The Effect of Perceived and Structural Neighborhood Conditions on Adolescents' Physical Activity and Sedentary Behaviors

Author Affiliations

Author Affiliations: Department of Social Welfare, Seoul Women's University, Seoul, Korea (Dr Kim); and Departments of Epidemiology and Biostatistics (Drs Liu and Colabianchi) and Exercise Science (Dr Pate), Arnold School of Public Health, University of South Carolina, Columbia.

Arch Pediatr Adolesc Med. 2010;164(10):935-942. doi:10.1001/archpediatrics.2010.167

Objective  To test whether physical activity and sedentary behaviors are associated with perceived and structural neighborhood conditions.

Design  Secondary analysis of data from wave I (1994-1995) of the National Longitudinal Study of Adolescent Health in the United States.

Setting  In-home interviews of adolescents and their parents.

Participants  Thirteen thousand six hundred sixty-eight adolescents (6794 adolescent boys and 6874 adolescent girls; grades 7-12).

Main Exposures  Perceived neighborhood conditions (public nuisance, acquaintanceship, and informal control measured at both individual and census tract levels) and structural neighborhood characteristics (concentrated poverty, immigrant concentration, and residential stability at the census tract level).

Outcome Measures  Failure to meet moderate to vigorous physical activity (MVPA) guidelines (<5 bouts of MVPA per week) and excessive screen time (>2 hours per day).

Results  Individual-level perception of intermediate (adjusted odds ratio [OR], 0.76; 95% confidence interval [CI], 0.68-0.84) or high (OR, 0.56; 95% CI, 0.50-0.61) acquaintanceship and high informal control (OR, 0.90; 95% CI, 0.82-0.99) in the neighborhood were negatively associated with failure to meet the MVPA guidelines compared with lower levels. At the census tract level, high informal control (OR, 1.16; 95% CI, 1.04-1.29) and intermediate (OR, 1.13; 95% CI, 1.01-1.26) and high (OR, 1.22; 95% CI, 1.06-1.40) concentrated poverty compared with lower levels were also associated with failing to meet the MVPA guidelines (when adjusted for sociodemographic characteristics). Compared with the lowest levels, high public nuisance (OR, 1.19; 95% CI, 1.04-1.36) and immigrant concentration (OR, 1.18; 95% CI, 1.01-1.38) at the census tract level were associated with excessive screen time.

Conclusion  These findings suggest that both perceptual and structural neighborhood factors should be considered to understand adolescents' physical activity and sedentary behaviors.

Obesity continues to be a growing public health concern in the United States.1 Physical activity and sedentary behaviors are among the key behavioral targets of efforts to address this public health concern.2 Owing to a recognition of the multifaceted causes of the obesity epidemic,3 there has been an increased focus on understanding contextual effects on obesity, physical activity, and sedentary behaviors among children and adolescents. Neighborhood conditions can affect where and how adolescents spend their time.4 Because parents may be influenced by hazardous neighborhood conditions, they may want to keep their younger children inside of the home,5 fearing that their children may get into trouble if they spend time outside without proper supervision. Thus, children and adolescents may not be able to spend their time on active outdoor activities even if they want to unless their neighborhood provides appropriate environments, such as parks, gyms, playgrounds, recreational facilities, and/or sidewalks.6-8

Several theoretical frameworks have been proposed to understand the association between neighborhood conditions and adolescents' behaviors in general. The institutional resource model for contextual effects assumes that various types of resources in the community (eg, schools, child care, parks, and libraries) could influence neighborhood residents' outcomes.9 Inner-city neighborhood structural conditions, defined as concentrated poverty, immigrant concentration, unemployment, and/or residential mobility, were used to explain residents' behavioral changes.10 These neighborhood structural characteristics are also shown to be associated with various adolescents' developmental and behavioral outcomes, including childhood intelligence, school readiness and achievement, internalized and externalized behavioral problems, sexual onset, and teenage births.9,11-13 In contrast, collective efficacy theorists postulate that perceptual neighborhood factors, including social network, informal control, mutual trust, and shared norms among residents in the neighborhood, may capture measures of neighborhood-level capacity to monitor local youths' behavior and act as regulatory mechanisms.14,15 These theoretical approaches were used to understand adolescents' problematic behaviors such as juvenile delinquency, gang activity, and/or alcohol and drug use and have not often been used to understand the direct associations between neighborhood safety and obesity status.16,17 However, one can argue that these problematic behaviors in the neighborhood may influence parents' perception of the neighborhood and thus may indirectly influence adolescents' physical activities and sedentary behaviors.

Previous studies linking neighborhood characteristics to obesity, physical activity, and sedentary behaviors have mainly used the institutional resource theory and used built environment and other available resources, existence, and access thereof as primary indicators of neighborhood condition.6-8,18,19 However, growing evidences suggest that residents' perception of their neighborhood conditions such as neighborhood safety may be a critical indicator of their behavior. For example, Burdette and Whitaker3 found that, after adjustment for sociodemographic factors, mothers' perception of higher neighborhood safety was related to lower television viewing hours in their children. To our knowledge, no studies have simultaneously included a comprehensive set of both structural and perceptual neighborhood factors in the model of adolescents' physical activities and sedentary behaviors.

In this cross-sectional study using the data from the National Longitudinal Study of Adolescent Health (Add Health), we aim to fill this gap in the literature by examining the associations of both structural and perceptual neighborhood factors with adolescents' physical activities and sedentary behaviors. This study will shed new light on understanding additive and independent contributions of the 2 different facets of neighborhood factors on adolescents' physical activity and sedentary behaviors. Furthermore, individual perception of neighborhood factors were aggregated in their neighborhood and included in the model to capture macroscopic collective neighborhood perceptions. We hypothesized that, after adjustment for individual-level sociodemographic factors and individual- and census tract–level perception of neighborhood conditions, adolescents' odds of not meeting physical activity guidelines and of excessive screen time are positively associated with structural neighborhood characteristics (eg, higher poverty, higher immigrant concentration, and lower residential stability). We also hypothesized that these 2 behaviors are positively associated with individual perception about the neighborhood conditions and average neighborhood perception about neighborhood conditions (eg, lower informal control, lower acquaintanceship, and higher public nuisance) after adjusting for structural neighborhood characteristics.


Add Health is a nationally representative, school-based study of adolescents (grades 7-12) in 1994 and 1995. This study used only wave I data of Add Health. Of 20 745 adolescents who were enrolled, 18 924 had valid sampling weights. We excluded 110 adolescents who had any type of physical disability and 2946 adolescents with missing values in any neighborhood variables. In addition, we eliminated 2200 adolescents owing to listwise deletion of missing values in the covariates used in the models. Finally, we excluded 2 adolescents who did not provide valid answers to the sedentary behavior questions, resulting in a final analytic sample of 13 668 adolescents (6794 boys and 6874 girls).

Measures of physical activity and sedentary behaviors

In the in-home interviews, Add Health researchers used a 7-day recall questionnaire that is commonly used in other large-scale epidemiologic studies to assess daily physical activities and sedentary behaviors.20-22 Based on self-reports, we developed a measure of meeting physical activity recommendations, defined as participating in 5 or more bouts of moderate to vigorous physical activity (MVPA) per week.23 In our analysis, MVPA referred to activities that cost 5 to 8 metabolic equivalents. Metabolic equivalents were assigned using a compendium24 in which 1 metabolic equivalent is defined as the energy expenditure associated with quiet sitting. Three of 5 questions were related to MVPA and used to calculate each adolescent's bouts of MVPA. They were worded as “During the past week, how many times did you . . . ” and followed by a description of a specific MVPA activity: (1) roller/bicycling (eg, rollerblading, rollerskating, skateboarding, or bicycling); (2) active sports (eg, baseball, softball, basketball, soccer, swimming, or football), and (3) exercise (eg, jogging, walking, karate, jumping rope, gymnastics, or dancing). The following scores were assigned to each answer category: 0 (not at all), 1.5 (1 or 2 times), 3.5 (3 or 4 times), and 5 (≥5 times). We then further dichotomized the sum of MVPA scores into those meeting the MVPA guideline (≥5 bouts of MVPA per week) and those failing to meet the MVPA guideline (<5 bouts of MVPA per week).

Sedentary behaviors were assessed via 3 questions. Each was worded as “How many hours a week do you . . . ” and involved activities like watching television and videos and playing video or computer games. Answers to these 3 questions were summed to create a total time (hours per week) that adolescents spent on sedentary activities. Twenty-five respondents who had impossible values (>168 hours/week) were recoded as 168 hours. Using national guidelines,25,26 we defined exceeding recommended total screen time as more than 2 hours per day.

Measures of perceived neighborhood condition

Perceptual domain of neighborhood factors was captured by 3 summary variables (ie, public nuisance, acquaintanceship, and informal control) measured at the individual level. These factors were used in previous neighborhood studies to examine adolescent behavioral outcomes such as sexual onset12 and delinquency.27 In this study, the acquaintanceship measure was based on adolescents' responses, while informal control and public nuisance were based on parents' responses. Acquaintanceship,28 or how well neighborhood residents knew each other, was assessed using 3 items from the in-home questionnaire: (1) “You know most of the people in your neighborhood”; (2) “In the past month, you have stopped on the street to talk with someone who lives in your neighborhood”; and (3) “People in this neighborhood look out for each other.” All answer options were scored as 1 = true and 2 = false. A summary score of answers to these questions was created, with higher scores indicating a higher level of acquaintanceship. Informal control and public nuisance were measured using responses to questions in the Parent Questionnaire. Informal control, or “shared expectation for the supervision and protection of local youths,”12(p699) was measured based on 2 questions: (1) “If you saw a neighbor's child getting into trouble, would you tell your neighbor about it?”; and (2) “If a neighbor saw your child getting into trouble, would your neighbor tell you about it?” Parents answered these questions using a 5-point Likert scale (1 = definitely would; 5 = definitely would not); answers were reverse-coded and then aggregated into a mean scale. Public nuisance was based on parents' responses to 3 questions: (1) “In this neighborhood, how big a problem is litter or trash on the streets and sidewalks?” (2) “In this neighborhood, how big a problem are drug dealers and drug users?” and (3) “How much would you like to move away from this neighborhood?” Three-point Likert scales (1 = no problem at all; 2 = a small problem; 3 = a big problem for the first 2 questions and 1 = not at all; 2 = a little; 3 = very much for the last) were used for parents' responses, which were aggregated into a mean scale of public nuisance.

To capture the contextual effects of these neighborhood perceptions,28 an average score of individual-level measures of perceptual neighborhood factors for individual participants living in the same census tract based on the 1990 census was calculated to create neighborhood-level measures of perceptual domain. These aggregated variables measure macroscopic and collective contextual aspects of the neighborhood perceptions measured at the individual level. Social scientists have widely used this approach to capture contextual effects based on individual-level characteristics.29 This resulted in 3 census tract–level perceptual domain neighborhood factors (ie, acquaintanceship, informal control, and public nuisance variables). We calculated the reliability of these aggregate variables based on individual-level responses using the procedure suggested by O’Brien.29 We chose the 1990 census tract as the geographical boundary of a neighborhood unit because census tracts are large enough to provide a reliable and homogeneous aggregated measure of characteristics30 and small enough to be considered a neighborhood unit. On average, 10.1 respondents (range, 1-279) lived in a census tract, and 697 (5.1%) respondents lived in a tract with 1 respondent to Add Health. In the final analysis sample, 13 668 respondents were from 1974 different census tracts.

Measures of structural neighborhood condition

Drawing on Wilson10,31 and collective efficacy theory,14,15,32 Browning and colleagues12 identified concentrated poverty, immigrant concentration, and residential stability as key structural neighborhood factors associated with adolescent outcomes. Nine census measures were selected based on theories and previous studies and subject to principal components analysis to create the 3 structural neighborhood factor scores.12 Concentrated poverty was measured by the proportion of households with an income below poverty level, public assistance, single female–headed households, and unemployment rate. Immigrant concentration score was calculated using the proportion of Hispanic and foreign-born individuals and those with limited English skills. Residential stability score was assessed by the proportion of households living in the same house for 5 years or longer and the proportion of housing occupied by owners. Because our neighborhood contextual measures were created based on multiple items, each measure used a different scale. To facilitate the interpretation of results, both the perceptual and structural domain neighborhood factor scores were divided into tertiles to provide for an equal number of participants in each response category.

Measures of covariates

We considered a number of sociodemographic variables in our analyses. They were adolescents' age, sex, race/ethnicity, in-school status, region and nativity from the in-home questionnaire, and the highest parental education and household income from the Parent Questionnaire.

Statistical analysis

Multilevel logistic regression models were used to examine the associations between these neighborhood factors and adolescents' physical activity (failure to meet MVPA guideline: <5 bouts per week) and sedentary behavior (excessive screen time: >2 hours per day), controlling for all covariates. We tested the interaction terms between sex and each neighborhood factor. None of them were statistically significant at P < .05. Thus, we did not provide sex-stratified results. Stata,33 version 10, was used for the statistical analyses.


Sample and neighborhood characteristics

Our final analysis sample was 49% female and 51% male; and 68% white, 14% black, 11% Hispanic, 3% Asian/Pacific Islander, and 4% of other racial/ethnic origin. Among them, 95% were born in the United States. The mean age at wave I was 15.8 years. About 31% of parents had at least a college education, 59% were high school graduates or had some college education, and 11% were not high school graduates. Mean household income was $45 476 per year (Table 1).

Table 1. 
Sociodemographic Characteristics in the Total Sample and by MVPA and Screen Time Guidelines
Sociodemographic Characteristics in the Total Sample and by MVPA and Screen Time Guidelines

On average, parents answered that public nuisance (ie, litter or trash on the streets; drug dealers and drug users) was not a problem at all or was a small problem in their neighborhood (mean [standard error (SE)] score, 1.55 [0.02]). Three out of 4 times, adolescents indicated that they knew most of the people in their neighborhood, they had stopped on the street to talk to someone in their neighborhood, and their neighbors looked out for each other (mean [SE] score, 0.75 [0.01]). Parents responded that they definitely or probably would tell their neighbor if they saw the neighbor's child getting into trouble and believed that their neighbors would do the same for their children (mean [SE] score, 4.12 [0.02]). These measures were practically identical after aggregating at the census tract level (Table 2).

Table 2. 
Individual- and Census Tract–Level Perceptual and Structural Neighborhood Characteristics
Individual- and Census Tract–Level Perceptual and Structural Neighborhood Characteristics

Neighborhood characteristics and physical activity

Table 3 shows the bivariate relationship between neighborhood characteristics and meeting the MVPA guideline. Adolescents' perception of neighborhood acquaintanceship was associated with failing to meet the MVPA guideline outcome (44%, 36%, and 29% from the lowest to the highest tertile; P < .001). The proportion of adolescents failing to meet the MVPA guideline decreased significantly from the lowest to the highest tertiles of perceived informal control, from 38% to 33% to 32%, respectively (P < .001).

Table 3. 
Prevalence of Meeting MVPA Guideline and Exceeding Screen Time Guideline by Neighborhood Factors
Prevalence of Meeting MVPA Guideline and Exceeding Screen Time Guideline by Neighborhood Factors

In census tract–level neighborhood measures, the proportion of adolescents who failed to meet the MVPA guideline was significantly associated with average acquaintanceship (38%, 33%, and 32%, respectively, from the lowest to the highest tertile; P = .004) and concentrated poverty (30%, 35%, and 38%, respectively, from the lowest to the highest tertile; P = .001).

Table 4 presents the adjusted odds ratios (ORs) of failing to meet the MVPA guideline for each neighborhood factor in multilevel logistic regression analyses. After controlling for the effects of sex, race/ethnicity, age, parental education, in-school status, nativity status, region, and household income, adolescents in the middle (adjusted OR, 0.76; P < .001) or the highest (adjusted OR, 0.56; P < .001) tertile of neighborhood acquaintanceship were less likely to fail to meet the MVPA guideline than those in the lowest tertile. Children of parents with the highest tertile of perceived informal neighborhood control compared with the lowest were less likely (adjusted OR, 0.90; P = .03) to fail to meet the MVPA guideline.

Table 4. 
Multilevel Logistic Regression Analysis of Meeting MVPA Guideline and Exceeding Screen Time Guideline
Multilevel Logistic Regression Analysis of Meeting MVPA Guideline and Exceeding Screen Time Guideline

At the census tract level, adolescents living in the neighborhood with the highest tertile of informal control compared with the lowest tertile were more likely to fail to meet the MVPA guideline (adjusted OR, 1.16; P = .01). For structural domain neighborhood characteristics, adolescents in the middle (adjusted OR, 1.13; P = .03) or highest (adjusted OR, 1.22; P = .006) tertile of concentrated poverty compared with the lowest tertile were significantly more likely to fail to meet the MVPA guideline.

Neighborhood characteristics and sedentary behaviors

As Table 3 shows, adolescents' perception of neighborhood acquaintanceship was not associated with excessive screen time. The proportion of adolescents with excessive screen hours increased significantly from the lowest to the highest tertiles of perceived informal control from 53% to 54% to 57%, respectively (P = .02). Across the 3 tertiles of perceived public nuisance, from the lowest to the highest, the proportion of adolescents with excessive screen time went up from 52% to 55% to 59%, respectively (P < .001).

In census tract–level neighborhood measures, the proportion of adolescents with excessive screen time was positively associated with changes in average informal control (52%, 54%, and 60% from the lowest to the highest tertiles, respectively; P = .002), average public nuisance (49%, 55%, and 62% from the lowest to the highest tertiles, respectively; P < .001), and concentrated poverty (50%, 55%, and 63% from the lowest to the highest tertiles, respectively; P < .001).

Results from multilevel logistic regression of excessive screen time were summarized in Table 4. At the individual level, none of the neighborhood perceptual domain variables were associated with odds of excessive screen time after adjusting for covariates. At the neighborhood level, adolescents living in the neighborhood with the highest average public nuisance (adjusted OR, 1.19; P = .01) or the highest immigrant concentration (adjusted OR, 1.18; P = .04) were more likely to have excessive screen time compared with the lowest tertile of the corresponding variables.


In this national sample of adolescents, we found that failing to meet MVPA guidelines was negatively associated with parents' perception of higher neighborhood informal control and adolescents' perception of higher neighborhood acquaintanceship at the individual level and higher census tract average perception of acquaintanceship and lower concentrated poverty at the neighborhood level. All of the relationships except that of census tract–level average acquaintanceship were significant even after adjusting for other covariates. Interestingly, the association of census tract–level average informal control was significant in the final model, whereas it was not in the unadjusted bivariate analyses.

We also found that parents' perception of higher public nuisance and informal control in individual- and census tract–level and higher census tract–level concentrated poverty were associated with increased odds of adolescents having excessive screen time in a bivariate relationship. After the adjustment for covariates and other neighborhood factors, none of the individual-level perception variables were significant. The results also indicate that, after controlling for individual-level socioeconomic factors, neighborhood-level measures of perceptual and structural factors are more important correlates of excessive screen time than individual-level perception of their neighborhood.

Some findings are inconsistent with previous studies. In our study, individual-level parental perception of public nuisance, which is related to neighborhood safety, was not significantly associated with any outcomes in the final model. However, in another study using the data from the Fragile Families and Children's Well-Being Study, authors found that mothers' perception of neighborhood safety, measured by items similar to our measure of public nuisance, was related to their children's television watching time but not to their outdoor physical activities.3 This discrepancy may be partly explained by the children's age differences between the 2 studies. It is plausible that adolescents in Add Health were less influenced than preschool-aged children in Fragile Families and Children's Well-Being Study by their parents' control over how and where they spent their time, which may be related to parental assessment about their neighborhood safety. Together, these findings indicate that the average perception of neighborhood condition across residents in the neighborhood may be a more important predictor of adolescents' television viewing hours than it is for younger children.

This study is unique because we used both parents' and adolescents' perceptions of their neighborhood conditions. After adjustment for sociodemographic covariates, we found that adolescents' perception of neighborhood acquaintanceship was more strongly associated with odds of meeting the MVPA guideline than any of the parental perception variables (ie, public nuisance and informal control) or other neighborhood-level variables. However, we do not know to what extent the apparently stronger association of adolescents' perception of acquaintanceship should be attributed to the fact that this measure was based on their own responses; the other 2 perception measures were based on parental responses. Future studies should ask the same sets of questions to both parents and adolescents to disentangle relative contribution of these 2 factors.

We hypothesized that if parents felt a higher level of informal control in their neighborhood, they would feel safer and would encourage their children to go out and play more and that the children would thus spend less time on screen activities. However, the individual-level parental perception of informal control conformed to our hypothesis, whereas the census tract–level aggregate of informal control perception was positively associated with odds of failing to meet the MVPA guideline. We also noted that the parental perception of informal control was not significant before the census tract–level measures were entered in the model, which indicates a suppression effect.34 Further analysis of cross-level interaction between individual and census tract–level perception of informal control (results not shown) reveals that the association between parental perception of informal control and adolescents' odds of failing to meet the MVPA guideline tends to be attenuated if they live in a neighborhood with a higher level of collective informal control, though the coefficient of cross-level interaction failed to reach significance at P < .05. In other words, the parental perception of informal control in their neighborhood becomes increasingly influential if they live in a neighborhood with a lower level of informal control.

One of the limitations of this study is that we classified an adolescent as meeting physical activity recommendations if he or she was active at least 5 days of the week. The newest federal guidelines published in late 2008 suggest that children should be active on all days.35 Furthermore, because Add Health did not contain information on the duration of the physical activity, we included any activity regardless of duration. This likely affected the estimated percentages of children who failed to meet the MVPA guideline in our sample. However, we believe that this measurement error would be independent of neighborhood characteristics of interest in our study; thus, this would attenuate the associations of interest. It should also be noted that, owing to lack of information in Add Health, we could not account for the effect of built environments, some of which have been reported to be associated with physical activities in other studies.36,37 Furthermore, the reliability measures of census tract–level perception of neighborhood characteristics were low, partly owing to small sample sizes in each census tract. Despite this, the significant findings at the census tract level indicate the robustness of the associations between neighborhood variables and physical activity and sedentary behaviors in our study.

Both research and practice fields have tried increasingly to understand the association between neighborhood factors and physical activity behaviors.4,8,38 Yet, the main focus is placed on understanding the physical, or built, environment and its association with physical activity, sedentary behaviors, and/or obesity.8,18,19,39,40 Neighborhood conditions are multifaceted and should be conceptualized across different dimensions.9 For example, perceived level of social disorder and collective efficacy may influence how and where people spend their time in relation to physical activity and sedentary behaviors.16,41 However, there is only a limited number of studies focusing on other aspects of neighborhood conditions, such as residents' perception of their neighborhood condition. This study contributes to the existing literature by placing its focus on both perceptual and structural neighborhood characteristics and their associations with adolescents' MVPA and screen time using data from a large nationally representative sample. If future studies can further replicate our findings, this would suggest that, to combat obesity and physical inactivity epidemics in the United States, public health professionals and pediatric care providers should take into account the important roles of perceived and structural neighborhood characteristics in behavioral choices above and beyond individual attributes.

Correspondence: Jinseok Kim, PhD, Department of Social Welfare, Seoul Women's University, 126 Kongneung, 2 Dong, Nowon-Ku, Seoul, Korea 139-774 (jskim@swu.ac.kr).

Accepted for Publication: March 11, 2010.

Author Contributions:Study concept and design: Kim, Liu, and Pate. Acquisition of data: Kim. Analysis and interpretation of data: Kim, Liu, and Colabianchi. Drafting of the manuscript: Kim and Liu. Critical revision of the manuscript for important intellectual content: Kim, Liu, Colabianchi, and Pate. Statistical analysis: Kim, Liu, and Colabianchi. Obtained funding: Kim and Liu. Administrative, technical, and material support: Kim and Liu. Study supervision: Liu and Pate.

Financial Disclosure: None reported.

Funding/Support: This project was sponsored by grant R40MC08958 from the Maternal and Child Health Research Program, Health Resources and Services Administration. This research uses data from Add Health, a program project directed by Kathleen Mullan Harris and designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Harris at the University of North Carolina at Chapel Hill, and funded by grant P01-HD31921 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development, with cooperative funding from 23 other federal agencies and foundations. Special acknowledgment is due Ronald R. Rindfuss and Barbara Entwisle for assistance in the original design. Information on how to obtain the Add Health data files is available on the Add Health website (http://www.cpc.unc.edu/addhealth). No direct support was received from grant P01-HD31921 for this analysis.

Additional Information: Persons interested in obtaining data files from Add Health should contact Add Health, Carolina Population Center, 123 W Franklin St, Chapel Hill, NC 27516-2524 (addhealth@unc.edu).

Ogden  CLCarroll  MDCurtin  LR McDowell  MATabak  CJFlegal  KM Prevalence of overweight and obesity in the United States, 1999-2004.  JAMA 2006;295 (13) 1549- 1555PubMedGoogle ScholarCrossref
 US Department of Health and Human Services.  Physical Activity and Health A Report of the Surgeon General. Atlanta, GA: US Department of Health and Human Services, Center for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion1996;Google Scholar
Burdette  HLWhitaker  RC A national study of neighborhood safety, outdoor play, television viewing, and obesity in preschool children.  Pediatrics 2005;116 (3) 657- 662PubMedGoogle ScholarCrossref
MacLeod  KEGee  GCCrawford  PWang  MC Neighbourhood environment as a predictor of television watching among girls.  J Epidemiol Community Health 2008;62 (4) 288- 292PubMedGoogle ScholarCrossref
Molnar  BEGortmaker  SLBull  FCBuka  SL Unsafe to play? neighborhood disorder and lack of safety predict reduced physical activity among urban children and adolescents.  Am J Health Promot 2004;18 (5) 378- 386PubMedGoogle ScholarCrossref
Babey  SHHastert  TAYu  HBrown  ER Physical activity among adolescents: when do parks matter?  Am J Prev Med 2008;34 (4) 345- 348PubMedGoogle ScholarCrossref
Dyment  JEBell  AC Grounds for movement: green school grounds as sites for promoting physical activity.  Health Educ Res 2008;23 (6) 952- 962PubMedGoogle ScholarCrossref
Davison  KKLawson  CT Do attributes in the physical environment influence children's physical activity? a review of the literature.  Int J Behav Nutr Phys Act July2006;319PubMedGoogle ScholarCrossref
Leventhal  TBrooks-Gunn  J The neighborhoods they live in: the effects of neighborhood residence on child and adolescent outcomes.  Psychol Bull 2000;126 (2) 309- 337PubMedGoogle ScholarCrossref
Wilson  WJ The Truly Disadvantaged: The Inner City, The Underclass, and Public Policy.  Chicago, IL University of Chicago Press1987;
Brooks-Gunn  JDuncan  GJKlebanov  PKSealand  N Do neighborhoods influence child and adolescent development?  Am J Sociol 1993;99 (2) 353- 395Google ScholarCrossref
Browning  CRLeventhal  TBrooks-Gunn  J Neighborhood context and racial differences in early adolescent sexual activity.  Demography 2004;41 (4) 697- 720PubMedGoogle ScholarCrossref
Earls  FCarlson  M The social ecology of child health and well-being.  Annu Rev Public Health 2001;22143- 166PubMedGoogle ScholarCrossref
Leventhal  TBrooks-Gunn  J Diversity in developmental trajectories across adolescence: Program and neighborhood influences. Lerner  RMSteinberg  L Handbook of adolescent psychology. 2nd ed. Hoboken, NJ John Wiley and Sons2004;451- 486Google Scholar
Sampson  RJRaudenbush  SWEarls  F Neighborhoods and violent crime: a multilevel study of collective efficacy.  Science 1997;277 (5328) 918- 924PubMedGoogle ScholarCrossref
Burdette  HLWadden  TAWhitaker  RC Neighborhood safety, collective efficacy, and obesity in women with young children.  Obesity (Silver Spring) 2006;14 (3) 518- 525PubMedGoogle ScholarCrossref
Lumeng  JCAppugliese  DCabral  HJBradley  RHZuckerman  B Neighborhood safety and overweight status in children.  Arch Pediatr Adolesc Med 2006;160 (1) 25- 31PubMedGoogle ScholarCrossref
Evans  GW Child development and the physical environment.  Annu Rev Psychol 2006;57423- 451PubMedGoogle ScholarCrossref
French  SAStory  MJeffery  RW Environmental influences on eating and physical activity.  Annu Rev Public Health 2001;22309- 335PubMedGoogle ScholarCrossref
Andersen  RECrespo  CJBartlett  SJCheskin  LJPratt  M Relationship of physical activity and television watching with body weight and level of fatness among children: results from the Third National Health and Nutrition Examination Survey.  JAMA 1998;279 (12) 938- 942PubMedGoogle ScholarCrossref
Heath  GWPratt  MWarren  CWKann  L Physical activity patterns in American high school students: results from the 1990 Youth Risk Behavior Survey.  Arch Pediatr Adolesc Med 1994;148 (11) 1131- 1136PubMedGoogle ScholarCrossref
Sallis  JFBuono  MJRoby  JJMicale  FGNelson  JA Seven-day recall and other physical activity self-reports in children and adolescents.  Med Sci Sports Exerc 1993;25 (1) 99- 108PubMedGoogle ScholarCrossref
Pate  RRPratt  MBlair  SN  et al.  Physical activity and public health: a recommendation from the Centers for Disease Control and Prevention and the American College of Sports Medicine.  JAMA 1995;273 (5) 402- 407PubMedGoogle ScholarCrossref
Ainsworth  BEHaskell  WLWhitt  MC  et al.  Compendium of physical activities: an update of activity codes and MET intensities.  Med Sci Sports Exerc 2000;32 (9) ((suppl)) S498- S504PubMedGoogle ScholarCrossref
Committee on Public Education, American Academy of Pediatrics: media violence.  Pediatrics 2001;108 (5) 1222- 1226PubMedGoogle ScholarCrossref
American Academy of Pediatrics. Committee on Public Education, American Academy of Pediatrics.  Pediatrics 2001;107 (2) 423- 426PubMedGoogle ScholarCrossref
Elliott  DSWilson  WJHuizinga  DSampson  RJElliott  ARankin  B The effects of neighborhood disadvantage on adolescent development.  J Res Crime Delinq 1996;33 (4) 389- 426Google ScholarCrossref
Sampson  RJ Linking the micro- and macrolevel dimensions of community social organization.  Soc Forces 1991;70 (1) 43- 64Google ScholarCrossref
O’Brien  RM Estimating the reliability of aggregate-level variables based on individual-level characteristics.  Sociol Methods Res 1990;18 (4) 473- 504Google ScholarCrossref
U.S. Census Bureau, Definition: census tract. http://ask.census.gov/cgi-bin/askcensus.cfg/php/enduser/std_adp.php?p_faqid=245. Accessed January 9, 2007
Wilson  WJ When Work Disappears.  New York, NY Alfred A. Knopf1996;
Sampson  RJMorenoff  JDEarls  F Beyond social capital: spatial dynamics of collective efficacy for children.  Am Sociol Rev 1999;64633- 660Google ScholarCrossref
 Stata Statistical Software, release 10.0 [computer program].  College Station, TX Stata Corporation2007;
MacKinnon  DPKrull  JLLockwood  CM Equivalence of the mediation, confounding and suppression effect.  Prev Sci 2000;1 (4) 173- 181PubMedGoogle ScholarCrossref
Ridley  KAinsworth  BEOlds  TS Development of a compendium of energy expenditures for youth.  Int J Behav Nutr Phys Act 2008;5 (5) 45- 52PubMedGoogle ScholarCrossref
Cohen  DAFinch  BKBower  ASastry  N Collective efficacy and obesity: the potential influence of social factors on health.  Soc Sci Med 2006;62 (3) 769- 778PubMedGoogle ScholarCrossref
Cohen  DAInagami  SFinch  B The built environment and collective efficacy.  Health Place 2008;14 (2) 198- 208PubMedGoogle ScholarCrossref
Romero  AJRobinson  TNKraemer  HC  et al.  Are perceived neighborhood hazards a barrier to physical activity in children?  Arch Pediatr Adolesc Med 2001;155 (10) 1143- 1148PubMedGoogle ScholarCrossref
Lopez  RPHynes  HP Obesity, physical activity, and the urban environment: public health research needs.  Environ Health 2006;5 (25) 10Google ScholarCrossref
Gordon-Larsen  PNelson  MCPage  PPopkin  BM Inequality in the built environment underlies key health disparities in physical activity and obesity.  Pediatrics 2006;117 (2) 417- 424PubMedGoogle ScholarCrossref
Kaplan  SKaplan  R Health, supportive environments, and the reasonable person model.  Am J Public Health 2003;93 (9) 1484- 1489PubMedGoogle ScholarCrossref