Research That Matters (January 17 - 20, 2008)

Background: Legal involvement among older adolescents leaving the foster care system is a barrier for reaching successful transition to adulthood. Knowledge of the reasons for legal involvement may eventually lend itself to the development of targeted system policy changes to prevent or ameliorate its harmful consequences. Therefore, we explored qualitative differences in legal involvement effects by employing a latent class analysis in a large NIMH funded study of older adolescents in foster care. This analytic strategy provided the basis for empirically deriving common or shared legal involvement for subgroups in relation to a range of covariates. Methods: Youth in foster care (N=404) were interviewed in person and alone at their place of residence by trained professional interviewers. The study sample was 57% female and ethnically composed of predominately African-Americans (51%) and Whites (43%). Indicators of legal involvement such as number of arrests, illegally making money, illicit drug sales, gun carrying, and assaults on others were used to identify subgroups across the pool of participants. Covariates used were demographics, employment status, maltreatment history and DSM-IV mental health and substance use disorder diagnoses. Additional measures of family support, neighborhood problems, deviant peer affiliations, and religiosity were also employed. Latent class analysis was used to identify the optimal number of homogeneous subgroups. External validation of latent classes was executed using aforementioned variables with Chi-Square and Analysis of variance (ANOVA) tests. Finally, multinomial logistic regression served to assess the likelihood of class membership based on relevant risk variables. Results: Overall, the four class solution exhibited the best empirical and conceptual fit. Latent classes consisted of a low risk (N = 278), moderate risk (N=66), and two high risk classes characterized by aggression (N =32) and drug culture involvement (N = 28). Chi-square tests revealed significant differences by gender (X2 [3] = 150.64, p<.001), ethnicity (X2 [3] = 19.28, p<.001), and employment (X2 [3] = 18.49, p = .005). There were also statistically significant proportional differences across classes with respect to maltreatment, mental and behavioral health and substance abuse (both dependence and polysubstance use). One-way analysis of variance (ANOVA) with Dunnett's T3 and Bonferroni post-hoc comparisons revealed significant mean differences across number of caseworkers (F [3] = 4.84, p =.002), family support (F [3] = 4.84, p<.001), neighborhood problems (F [3] = 4.84, p <.001) and deviant peer affiliations (F [3] = 4.84, p <.001). Multinomial logistic regression showed that low levels of family support, living in highly distressed neighborhoods, affiliating deviant peers, and a range of co-occurring disorders including substance dependence, ADHD, and APD were significant predictors of risk for legal involvement. Implications for policy and practice: Results indicate that older adolescents differed substantially in their risk for legal involvement and that variables operating at both the dispositional and contextual levels are important predictors of risk for legal involvement. Findings strongly suggest that stronger family supports, mental health services, extended care and attention to employment would attenuate the risk for legal involvement for many youth aging out of the foster care system.