Long-Term Crime Patterns: Does Later Adult Criminal Justice System Involvement Vary over Time Based on Social Bonds at Age 17?

Background and Purpose:  Social control theory suggests that youth are less likely to engage in crime when they have stronger social ties (Hirschi, 1969). However, research on salient social bonds and criminal justice involvement among foster youth is limited (Cusick, Havlicek, & Courtney, 2012). Additionally, while research has shown that crime rates peak around age 17 and then decline throughout adulthood (Laub & Sampson, 2003), age-crime trends has not been well-studied in foster youth. To better understand the association between social ties and criminal justice involvement (CJI) among transition-aged foster youth, this study seeks to address two questions: Are there distinct subgroups based on social bonds at age 17? If so, do risks of later CJI vary over time by these subgroups?

Methods:  The analytic sample included 540 respondents who completed all three interview waves of the CalYOUTH Study. Predictors were measured at Wave 1 (age 17), and the outcome was measured at Waves 2 and 3 (age 19 and 21). The outcome variable, CJI, is a binary self-report measure capturing arrest, incarceration, or conviction of crime since last interview. Control variables include youths’ demographic characteristics, maltreatment history, pre-care family problems, behavioral health problems, foster care characteristics, and juvenile justice involvement prior to age 18. The main predictor, social bonds, used 10 indicators capturing two domains of social bonds: interpersonal bonds (birth family, new family) and institutional bonds (professionals, school, employment, religiosity). First, latent class analysis (LCA) was used to derive subclasses of foster youth based on social bonds at age 17.  Second, binary logistic regression examined associations between latent classes and CJI, controlling for other covariates.  All analyses used weights to account for the survey design.

Results: CJI declined across ages (40% at age 17, 17% at age 19, 15% at age 21; p<.001). In estimating LCA models, a two-class model was accepted based on greater statistical model fit (BLRT p<.001, BIC=6208.9, entropy=.644) than other alternative models. Results show that youths in Class 1 are distinguished by high interpersonal bonds (particularly higher on “romantic partnership” and “cohabitation”, p<.001) whereas youths in Class 2 possess high institutional bonds (particularly higher on “education”, p<.01). Regression results found a significant association between social bond classes and CJI only at age 21. At age 21, the estimated odds of CJI were significantly higher for youths in Class 1 (high interpersonal bonds) than those in Class 2 (high institutional bonds) (OR=2.3, p<.05).

Conclusions and Implications:  Similar to research on youth in the general population, a downward trend in CJI was observed for foster youth as they entered adulthood. Results suggest that social bonds are associated with CJI for young people transitioning to adulthood. The LCA provides nuanced findings indicating there may be specific types of social bonds that help to mitigate CJI. Institutional bonds may be particularly important to emphasize and promote during the critical developmental period before adulthood. Further attention is needed to understand the increased risk of CJI among youths in the certain interpersonal bonds (e.g., parenthood, cohabitation) group.