Sex differences and the overlap in youths' risk factors for onset of violence and gang involvement

Abstract

AbstractRisk factors for general delinquency have been examined in numerous studies, but fewer studies have sought to identify factors specifically related to youth violence and an even smaller number of studies have focused upon youths' gang involvement or comparing patterns by sex. Comparison of findings from studies of violence risk factors to studies of gang risk factors suggests that a number of the same factors underlie both, but the extent of overlap has not often been examined in the same study. If the two behaviors share a common set of factors, more general prevention efforts may prove fruitful in reducing youths' involvement in both behaviors, while more specific programs or components may be necessary if more unique than shared predictors are found. In this study, we therefore examine the extent to which youths' early-adolescent onset of involvement in gangs and in violent behavior are related to the same proximal risk factors and the extent to which risk factors for these behaviors are shared between the sexes. Consistent with prior studies, our bivariate analyses show that a large number of risk factors were predictive of youths' onset of involvement in both violence and gangs, with some factors unique to each behavior; patterns were similar by sex. These findings suggest that early-adolescence programs targeting these common factors may reduce youths' likelihood of involvement in both gangs and violence, and that some additional behavior-specific or sex-specific components may also be helpful. Given the scant research on these issues, these findings are only suggestive, and further research is required for validation.Keywords:: youth violenceyouth gangsgang membershipgang involvementrisk factorssex differencesgender differencesviolence preventiongang preventiononset of violenceonset of gang membership AcknowledgementThis research was made possible, in part, by the support and participation of seven school districts, including the School District of Philadelphia. This project was supported by Award No. 2006-JV-FX-0011 awarded by the National Institute of Justice, Office of Justice Programs, US Department of Justice. Our thanks are extended to the numerous students, teachers, school administrators, and law enforcement officers for their involvement and assistance in this study. The opinions, findings, and conclusions or recommendations expressed in this paper are those of the authors and do not necessarily reflect the views of the Department of Justice or of the seven participating school districts.Notes 1. While it is not a focus of the current study to disentangle whether violence precedes gang membership or vice versa, youths' prior involvement in these behaviors is examined to determine the extent to which one serves as a risk factor for the other behavior. 2. This is not to say that programs intended to affect the facilitative group processes within gangs that produce violence are not necessary; indeed, extant research highlights the importance of such interventions to reduce levels of violence among gangs and their members. 3. Additional considerations include similarities and differences related to such characteristics as gender identity, race/ethnicity, and culture of origin, which are equally important but beyond the scope of this paper. 4. Two principals who were contacted declined their schools' participation, and a third school was unable to adhere to the evaluation design. In each of these instances, other schools were selected as replacements. 5. There were102 classrooms randomly assigned to receive G.R.E.A.T. and 93 control classes. 6. Esbensen et al. (Citation2008) reported a 79% consent rate, but the inclusion of two additional schools in the evaluation after the publication of that article resulted in a 78% consent rate overall. 7. Research shows that youth with early onset (prior to age 12) of delinquency, particularly violence, have greater frequency, seriousness, and persistence of later offending (e.g., Elliott et al.Citation1985, Huizinga et al.Citation1994, Hawkins et al.Citation1998); preventing early initiation of such behavior, then, should be beneficial. 8. Although the present study most closely follows analyses in Esbensen et al. (Citation2009, Citation2010), there are a few key differences. First, Esbensen et al. used data from the first national evaluation of the G.R.E.A.T. program ('G.R.E.A.T. I,' conducted between 1995–2000), while our study uses data from the second national evaluation ('G.R.E.A.T. II,' conducted between 2006–2012). Some measures from the first evaluation were excluded from the second evaluation and, therefore, from this study. Finally, Esbensen and colleagues used cross-sectional data, while we chose to use two waves of data for correct temporal ordering of risk factors and behavioral measures of interest (Krohn and Thornberry Citation2010). We therefore limit our analyses to gang joiners and violence initiators, while Esbensen et al. used current gang members (not specifically gang joiners) and youths who had engaged in violence in the past 12 months (i.e., annual prevalence, not specifically those engaging in violence for the first time). These methodological necessities and choices mean that our results may not be directly comparable to Esbensen and his colleagues' findings (or, as we discuss later, to some other studies of gang or violence involvement). 9. This is distinguished from youth who simply report 'current gang membership,' as these youths may also have been gang members prior to Wave 2. For example, 'current gang members' at Wave 2 = 177, while 'gang-joiners' = 151. In addition, Wave 2 was selected in part because this wave had more gang joiners in the sample than later waves (W3 = 131, W4 = 96, W5 = 63, W6 = 27).10. We include 'gang fights' in our violence index because the nearly half of the respondents in our sample who have engaged in this behavior are nongang (never gang) members. For example, of those who had been in gang fights at Wave 1, 48% were never gang members and 53% were Wave 2 gang-joiners. These findings are consistent with analyses of the same dataset in a study by Melde and Esbensen (Citation2013), who report that depending on the wave, between 58% and 69% of respondents who engaged in gang fights were nongang members. It is possible that nongang members either are, in fact, involved in gang fights despite not being gang-involved or that they interpret the behavior as including 'group fighting.'11. This percentage is comparable to other studies using self-report data from a general sample of youths in the USA. For example, in Hill et al.'s (Citation1999) study, 6% of their sample was gang-involved at age 15 and 5% was gang-involved at ages 14, 16, and 18; in Esbensen et al.'s (Citation2001) study, 8.8% of the eighth-grade sample was gang-involved. It should be noted that these studies used 'current gang membership' measures, which produces a higher prevalence than does restricting the sample to 'gang-joiners.'12. Additional exploratory analyses with other combinations of risk factors (e.g., peer commitment variables in place of peer association measures) revealed no differences in statistically significant predictors. Importantly, low commitment to prosocial peers, a unique predictor of gang-joining in the bivariate analyses, was not a statistically significant predictor in any of the exploratory multivariate analyses.13. Multicollinearity does not appear to be a problem in these analyses; there are no Variance Inflation Factors (VIFs) over 4, and examination of the Condition Index shows none over 30. Further, the largest correlation coefficient was 0.378 (peer delinquency and neutralizations).14. Tests for equality of coefficients across both sets of models revealed no statistically significant differences.15. Probably most comparable to our analyses is Thornberry et al. (Citation2003), in which the dependent variable was joining a gang at Wave 3 or later. For the risk factors common to both studies, our bivariate results are more different from than similar to theirs. Delinquent beliefs predicted gang involvement for both sexes in both studies. In Thornberry et al.'s study, parental supervision, school commitment, delinquent peers, and prior violence were predictive for males but not females, while in our study, they were predictive for both sexes. And, in Thornberry et al.'s study, neighborhood disorganization predicted females' but not males' involvement, while the opposite was true in our study. Finally, unsupervised time with friends was not a statistically significant predictor in their study, while it predicted both sexes' gang involvement in ours. We suggest that our results may differ in part due to the samples—Thornberry et al.'s sample pools youth aged 14.5 to 17.5, while our sample is limited to youth whose average age is 12. There is some suggestion in the literature (e.g., Jang and Krohn Citation1995, Espiritu Citation1998) of more sex-variance in explanations for deviance at older (mid-late adolescence) than at younger (late childhood/early adolescence) ages. Thus, developmental issues may play a role in differentiating predictive factors by sex. Huizinga et al. (Citation2003) also reported analyses examining gang-joiners at ages 13–14, 15–16, and 17–18, with a number of predictors consistent with our findings for the overall sample, including delinquent peers, few prosocial peers, weak beliefs about the wrongfulness of delinquent behaviors, and use of neutralizations.16. Some scholars, including those associated with the Rochester Youth Development Study and the Seattle Social Development Study, have conducted risk factor analyses on both violence and gang membership using those data, but not specifically with the same waves or subset of the sample.Additional informationNotes on contributorsDana PetersonDana Peterson, Ph.D., is Associate Dean and Associate Professor in the School of Criminal Justice at the University at Albany, State University of New York. She teaches and conducts research primarily in the areas of youth violence, youth gangs and gang prevention, and how sex and gender structure these experiences.Kirstin A. MorganKirstin A. Morgan, MA, is a doctoral student in the School of Criminal Justice at the University at Albany, State University of New York. Her research interests are youth gangs, gang prevention and intervention, juvenile delinquency and the juvenile justice system, and research methodology.