Understanding Gang Joining from a Cross Classified Multi-Level Perspective

ABSTRACT

The present study seeks to address the gap in our understanding of the role of community- and school-level factors on individual-level gang joining. We assess our research question by bringing together data from a statewide survey of more than 35,000 school youth who were enrolled in more than 250 schools and resided in more than 300 communities. We first use multilevel modeling techniques to examine the independent relationship of community and school on the likelihood of individual youth gang joining. Second, while controlling for individual-level risk and protective factors, we examine the simultaneous relationship of community- and school-level conditions on gang joining. Overall, the results of the present study suggest that both community- and school-level characteristics play a role in gang joining. The results of the cross-classified model suggests that community structural characteristics may serve as protection against gang joining. Individual-level risk and protective factors remain important for understanding gang joining. A variety of factors are related to gang joining. Though individual-level factors are important, community- and school-level factors are meaningful in the understanding of gang joining.

KEYWORDS:

• Gangs
• gang joining
• risk and protective factors
• cross-classified multilevel modelling

Acknowledgment

The authors would like to thank the Arizona Criminal Justice Commission for providing data used for this study.

Notes

¹ Furthermore, individual level perceptions of school and neighborhood can be impacted by psychological reductionism, simultaneity bias, and selection bias (Raudenbush and Sampson 1999).

² There is also a growing body of literature on the relationship between community structure and gang activity (e.g., Curry and Spergel 1988; Papachristos and Kirk 2006) as well as social structure and its relation to gang affiliation (Tolle 2016). Due to space limitations we keep our review of the literature focused on concentrations of gang membership and its relationship with community structure.

³ See Biggar Jr., Forsyth, Chen, and Richard (2016) for example of recent research using these data.

⁴ Fourteen schools were excluded from the present study because of missing data at the school level.

⁵ For a more thorough description of the sampling protocol see the 2004 Arizona Youth Survey, Arizona Criminal Justice Commission.

⁶ There has been substantial debate about how neighborhoods impact crime, and along with that debate has come detailed discussion around the operationalization of “neighborhood.” While the definitional problem surrounding neighborhood has not been solved it is clear that various geographical units have served as a proxy for neighborhood for example, blocks, block groups, census tracts, two tracts, and zip codes. We considered whether to use the term neighborhood or community in reference to our geographic unit of analysis—zip code. To be sure, zip code has been used frequently as a proxy for neighborhood, but we chose to use the term community given the geographic size of many of the zip codes in the more rural regions of the state. Zip codes can result in some loss of variation in some respects but as Baumer and South (2001) point out they are not without their advantages in that they have been found to produce strong effects, wide variation, and are used in tribal areas. Related, but somewhat different, Small and McDermott (2006: 1702) pointed out that census tracts, while perhaps appropriate for understanding neighborhood socialization, are probably too small for measuring economic issues and that “zip codes may be more appropriate because regions of economic activity are wider than those of social neighborhood activity.”

⁷ Due to space limitations these items are not presented. A detailed description of the items in tabled form can be obtained from the authors upon request.

⁸ Our measures of school crime only include incidents that occurred at school, and do not include incidents in the neighborhood or near the school.

⁹ We occasionally use the word “effects” in the present study out of deference to prior literature employing HLM. We emphasize that the present study is cross-sectional/relational and we do not mean to imply causality.

¹⁰ Home zip code was used as the community clustering variable.

¹¹ The school entity unique identifier was used as the school clustering variable.

¹² The variance partition coefficient (VPC) is estimated for non-linear models (Goldstein, Browne, and Rasbash 2002). Due to the dichotomous nature of the outcome variable, the first step was to obtain model linearization, then the VPC was estimated.

¹³ As shown in Table 3, the school-level violent crime rates are indeed higher than the school level property crime rates. This same phenomenon is observed at the national level (U.S. Department of Justice 2014) and we believe is a result of student-level reporting patterns.

¹⁴ Because our analysis involved a generalized model to predict a binary outcome, no level-1 residuals or associated variance is computed.

¹⁵ Variance explained (versus variance partition components) were not computed for the cross-classified model because they do not produce residuals.

Additional information

Notes on contributors

Lidia E. Nuño

LIDIA E. NUÑO is an Assistant Professor in the Division of Politics, Administration, and Justice at California State University Fullerton. She holds undergraduate and graduate degrees from Arizona State University. She has participated in various research projects, including international research in multiple nations in Central America and the Eastern Caribbean. Her research interests include gangs and youth violence, police legitimacy, the relationship between immigration status and criminal involvement, and methodological and quantitative methods.

Charles M. Katz

CHARLES M. KATZ is the Watts Family Director of the Center for Violence Prevention and Community Safety and is a Professor in the School of Criminology and Criminal Justice at Arizona State University. He received his PhD from the University of Nebraska at Omaha. His research interest includes gangs and gang control, police organizational theory, and crime control policy. Over the year, he has collaborated with numerous agencies to increase their organizational capacity to identify and strategically respond to crime. He is the (co) author of many peer-reviewed articles, monographs, and books including Policing Gangs in America (Cambridge University Press: 2006) and The Police in America (McGraw Hill: 2013).