Neighborhood Disadvantage, Social Capital, Street Context, and Youth Violence

Abstract

This article integrates arguments from three perspectives on the relationship between communities and crime—constrained residential choices, social capital, and street context perspectives—to specify a conceptual model of community disadvantage and the violence of individual adolescents. Specifically, we propose that status characteristics (e.g., race, poverty, female headship) restrict the residential choices of families. Residence in extremely disadvantaged communities, in turn, increases the chances of violent behavior by youths by influencing the development and maintenance of community and family social capital, and by influencing the chances that youths are exposed to a criminogenic street context. We assess our conceptual model using community contextual and individual-level data from the National Longitudinal Study of Adolescent Health. Our findings suggest that individual or family status characteristics influence violence largely because of the communities in which disadvantaged persons and families reside. Although we find that community social capital does not predict individual violence, both family social capital and measures of an alternative street milieu are strong predictors of individual violence. Moreover, our street context variables appear to be more important than the social capital variables in explaining how community disadvantage affects violence.

ACKNOWLEDGMENTS

Our research was supported by a grant from the College of Humanities and Social Science at North Carolina State University and by funding from the National Consortium on Violence Research. The authors are grateful to Joseph B. Lang for his statistical advice. The data for this study are from the Add Health project, a program project designed by J. Richard Udry (PI) and Peter Bearman, and funded by grant P01-HD31921 from the National Institute of Child Health and Human Development to the Carolina Population Center, University of North Carolina at Chapel Hill, with cooperative funding from 17 other agencies. Persons interested in obtaining data files from the National Longitudinal Study of Adolescent Health should contact Add Health, Carolina Population Center, 123 West Franklin Street, Chapel Hill, NC 27516-2524 (http://www.cpc.unc.edu/addhealth). Neither the funding agencies nor the collectors of the data bear any responsibility for the analyses and interpretation drawn here.

NOTES

Notes

¹ Although studies in the communities and crime tradition also focus on victimization or suitable targets and fear of crime (see Wilcox Rountree and Land 1996; Markowitz et al. 2001; Wilcox, Land, and Hunt 2003), we focus here on research that more pointedly targets the causes of crime.

² When studies have included variables that tap what we call criminogenic street context, they have tended to treat these as control variables entered before variables like community disadvantage (e.g., McNulty and Bellair 2003a).

³ Research on communities and crime juxtaposes composition or selection arguments with arguments focusing on emergent properties of communities (see Sampson 1985). Our theoretical discussion and empirical assessment of residential constraints take into consideration composition/selection; discussion and assessment of community effects, social capital, and street context are consistent with perspectives focusing on emergent properties.

⁴ There were a few variables with nontrivial missing data that were not addressed in the sampling weights, namely, information on network closure, parent's participation in organizations, and collective supervision. We reestimated our models using mean substitution for missing data, and found no meaningful differences in the findings reported here.

⁵ We include only ascribed characteristics and characteristics of families over which adolescents have no control (i.e., poverty status) as factors determining selection into disadvantaged communities. This is consistent with Sampson et al. (2002) argument that the strategy of entering individual, family process, and peer variables as controls alongside neighborhood characteristics confounds the “importance of both long-term community influences and mediating developmental pathways regarding children's personal traits and dispositions, learning patterns from peers, family socialization, and more” (p. 469).

⁶ Seventeen cases were deleted from the analysis because the respondents reported multiple residential mothers and/or multiple residential fathers. Although some researchers advocate the use of additional measures of family structure (see Parker and Johns 2002), we focus on female headship because it is linked most often to family disadvantage and thus is part of the nexus of factors that constrains residential choices and leads to selection into disadvantage communities.

⁷ Twenty-three cases were deleted from the analysis because the respondents' reported sex was different in wave 1 and wave 2 of the interviews.

⁸ Studies demonstrate that residents in particular areas tend to disagree about the exact boundaries of their neighborhood (see Furstenberg et al. 1999), which renders the measurement of appropriate neighborhood units problematic. Given these constraints on systematic measurement of neighborhoods and given restrictions on available data generally, researchers often have used census tracts in their analyses of urban crime (Krivo and Peterson 1996; Crutchfield, Glusker, and Bridges 1999; Peterson, Krivo, and Harris 2000; McNulty and Bellair 2003a). Census tract boundaries are drawn by census tract committees to account for natural boundaries and population characteristics in a way that creates units that are meant to represent natural social aggregates. Research on appropriate units of analysis reports that use of census tract aggregates does not produce substantially different results from aggregates such as block groups (see Gephart's 1997 review).

⁹ We also estimated models that included measures of residential stability and neighborhood crime but found that these variables were highly associated with our community disadvantage index, and we could not disentangle their separate effects. Our principle components analyses indicated that the reliability of the community disadvantage scale was decreased when these variables were added to the factor.

¹⁰ Moreover, youths' violence may influence their family relationships, relationships between their families and community members, and their exposure to violence and delinquent peers. It is much less likely that youths' prior violence leads to residence in a disadvantaged community. Thus, the ordering of variables in our models corresponds to our theoretical expectations about temporal order.

¹¹ Although these indicators do not capture all that is likely important about a criminogenic street context, they do capture many of the elements of street context discussed in literature addressing oppositional or criminal street milieus (e.g., Sampson and Wilson 1995; Anderson 1999). Future quantitative research would benefit from inclusion of more direct assessments of neighborhood culture, not only context, such as some measure of aggregated neighborhood attitudes about violence and definitions favoring using violence as a means of gaining respect. Unfortunately, such measures are currently unavailable in datasets containing macro- and microlevel information on communities and violence.

¹² While this misses some situations of violence to which youths may be exposed, especially most instances of family violence, it does capture exposure to serious violence—namely stabbing—occurring in the street context. This is the only variable representing this construct available in the data and it clearly captures exposure to extreme violence.

¹³ In a supplementary analysis, we verified that the effect of witnessing serious violence remains significant in a model in which self-reported neighborhood crime is controlled.