Dangerous Places: Gang Members and Neighborhood Levels of Gun Assault

Abstract

Despite attention to the role of gangs in urban gun violence, much remains to be learned about the spatial distribution and consequences of residential gang membership. This study uses data from St. Louis to examine the effects of resident gang membership on rates of gun assault. We also consider whether gun violence is conditioned by the level of gang membership in surrounding communities. As anticipated, communities with the highest number of gang members also have the highest rates of gun assault. However, much of the impact of gang membership on gun assaults extends outside of the boundaries of gang neighborhoods, especially those neighborhoods with few or no gang members. The number of gang members in surrounding neighborhoods has no discernible effect on gun assaults in communities with higher rates of gang membership. Finally, controlling for the spatial proximity of residential gang membership helps to account for some of the association between neighborhood disadvantage and gun assaults.

Keywords:

• gangs
• guns
• neighborhoods
• violence

Notes

¹ The city of St. Louis includes 113 census tracts, 110 of which contain adequate residential populations (>250 people) to ensure reliable community characteristic measures (Kubrin & Weitzer, 2003). As census boundaries may not be the best representation of neighborhood perimeters, we replicated our analyses using the 79 neighborhood boundaries defined by the City of St. Louis. Results were not substantively different and are available upon request.

² Information on the motive behind the assault was unavailable; thus, the measure captures assaults beyond gang violence (e.g. domestic disputes). Nonetheless, this remains consistent with our theoretical rationale that higher rates of residential gang membership should maintain an independent contribution to gun violence in communities.

³ This represents the number of newly documented gang members coming to police attention and verified between 1998 and 2002. A four-year measure was used to provide an estimate of community differences in residential gang affiliation that was not sensitive to short-term police crackdowns. This estimate captures a stable yet temporally proximate estimate of gang membership levels. This is consistent with the purging practices of law enforcement gang databases (e.g. CalGang, REJIS), and with research finding the vast majority of individuals exit gangs within four years (Pyrooz, Sweeten, & Piquero, 2013, p. 261). Police recorded 1,720 gang members between 1993 (the first-year police began documenting the addresses of known gang members) and 2002. Supplemental analyses were conducted with a measure of gang membership that included neighborhood counts of gang members documented from 1993 to 2002. Results were similar to those shown here and are available upon request.

⁴ Gun seizures may be entangled to the policing patterns and crime levels in a community, and it is difficult to disentangle this relationship. Research conducted in St. Louis by Burruss and Decker (2002) suggests that most guns (50%) seized by the police happen in the course of routine patrols. Only 39 percent come from calls for service, and the remainder came from warrants (11%); thus, most of the guns are seized during traditional policing operations. St. Louis was a part of the Project Safe Neighborhoods program implemented in October, 2002 (see Decker et al., 2007). Although gun assaults did decline in the intervention areas after the intervention, the decline was greater in non-intervention communities—owed largely due larger trends in crime decline.

⁵ The drug-related death measure was included as a proxy for drug markets as part of a larger goal to capture geographic variability in conditions that may provoke or facilitate armed conflict. We also estimated a model that replaced the rate of drug-positive deaths with the drug arrest rate. The results indicate that the drug arrest rate is not a significant predictor of gun assault nor does its inclusion alter the effects of gang membership on gun assault.

⁶ The outcome and its spatial lag are determined simultaneously, Wy* is endogenous to Y (Land & Deane, 1992, p. 228). Endogeneity concerns also arise because the Wy* is a function of the lagged values of the predictors in the equation. The lagged gun assault count is then correlated with the error term resulting in biased, inconsistent regression coefficients.

⁷ We explored the utility of controlling for various temporally lagged measures of violent crime. The correlation between gun assault levels from 2002 to 2004 and gun assault levels from 1995 to 1997 exceeds .90, an unacceptable level of multicollinearity. We instead use a measure of prior levels of overall violent crime (which includes all forms of gun violence). The inclusion of this measure alleviated problems associated with multicollinearity and provides a more comprehensive measure of the types of victimizations that may prompt or reduce residents’ gang membership.

⁸ We explored if the concentration of gun seizures influenced gun assault indirectly by moderating the influence of gang member prevalence on gun assault (results not shown); the relationship was not significant.

⁹ We estimated models to explore the robustness of the link between neighborhood gang membership and gun assault (results not shown). First, we regressed the rates of other crimes (non-gun assault, robbery with no gun, and robbery with a gun) on the gang measures. If gang membership reflects the presence of a deviant population generally, we would anticipate significant effects of neighborhood gang membership on a variety of violent crimes. Analyses revealed no significant association between neighborhood gang membership and other violent crimes. Finally, we explored the possibility that the effect of proximity to gang members is spurious because of propinquity to disadvantaged neighborhoods (Mears & Bhati, 2006). We included a spatial lag of concentrated disadvantage in two additional models: one with and one without a spatial lag of gang membership. Results indicate that disadvantage in surrounding neighborhoods is not a significant predictor of gun assault rates, and its inclusion does not alter the spatial lag effect of gang membership. We were unable to control for the spatial lag of gun assault in the latter model as it yielded unacceptable levels of multicollinearity and produced unstable coefficient estimates. Because the spatial lag of gun assault is not significant in any models containing the spatial lag of gang membership, we estimated the effects of proximity to gang members (controlling for the spatial lag of disadvantage) without the inclusion of the spatial lag for gun assault.

Additional information

Notes on contributors

Beth M. Huebner

Beth M. Huebner is an associate professor and graduate director in the Department of Criminology and Criminal Justice at the University of Missouri-St. Louis. Her principal research interests include the collateral consequences of incarceration, prisoner reentry, and public policy.

Kimberly Martin

Kimberly Martin is a statistician at the United States Bureau of Justice Statistics. She received her PhD from the University of Missouri-St. Louis. Her research interests are violent crime, courts and sentencing, and quantitative methodology.

Richard K. Moule

Richard K. Moule Jr. is a doctoral student in the School of Criminology and Criminal Justice at Arizona State University. He received his BS in criminology from The College of New Jersey and his MS in criminology and criminal justice from Arizona State University. His research interests include gangs and deviant networks, life course criminology, and the intersection of technology and criminology theory. His research has appeared in Social Science Research, Journal of Qualitative Criminal Justice and Criminology, and is forthcoming in the Handbook of Security. He is presently the archivist for the Walter B. Miller Library, housed in the School of Criminology and Criminal Justice.

David Pyrooz

David Pyrooz is an assistant professor in the Department of Criminal Justice and Criminology at Sam Houston State University. He received his BS and MS in Criminology from California State University, Fresno and PhD in Criminology and Criminal Justice from Arizona State University. His research interests revolve around gangs and deviant networks, developmental and life course criminology, and violent offending and victimization. He is the recipient of a Graduate Research Fellowship from the National Institute of Justice and is the author of Confronting Gangs: Crime and Community (Oxford Press, with G. David Curry and Scott H. Decker).

Scott H. Decker

Scott Decker graduated from DePauw University with a BA in Social Justice. He earned the PhD in Criminology from Florida State University in 1976. He is a foundation professor in the School of Criminology and Criminal Justice at Arizona State University. His main research interests are in the areas of gangs, violence, criminal justice policy, and the offender’s perspective. He is a fellow in the American Society of Criminology and the Academy of Criminal Justice Sciences.