Hidden Trajectories of Influence: A Model Integrating Developmental Assets and Common Predictors of Re-Arrest Among Older Adolescents

Background and Purpose. To address recidivism in the United States, advocates suggest intervention across multiple ecologies (Cervantes et al, 2004; Hamilton et al., 2007; Sullivan et al., 2007; Quinn & Van Dyke, 2004). Health Link, a community-based case management approach, purported that access to services and supportive resources could help ease the transition from detention and prevent re-arrest. This study examines data from the Health Link program to better understand the potential benefits of incorporating a developmental assets framework into strategies to assist young people before, during and after prison. Due to high rates of detention nationally, African American and Latino youth were the focus of the study. Specifically, a theory-driven conceptual model was tested that focused on the extent that developmental asset constructs were associated with a reduced likelihood of re-arrest as well as lower propensity for mental health and substance abuse issues.

Methods. This study used data collected in the 1997-2002 evaluation of the Health Link Program. The study sample included 529 African-American and Latino males. The constructs used in this research came from data collected over two time points (baseline ad 12 month post-discharge). Mixtures of latent, composite and observed variables were used to construct the simple and full complex models. Because this study utilized secondary data, attention was made to create robust proxies (both latent and composite variables) for each model construct. The full complex model was a structural equation model with two measurement models for latent constructs, 10 composite variables (i.e., developmental asset constructs, mental health issues, mental health/substance use services and association with delinquent peers), and four observed variables (i.e., prior arrest, health insurance coverage, treatment/control status and re-arrest). The composite variables were created because scale development was inhibited by the variability of response categories across observed variables linked to each construct. Structural equation modeling (SEM) was used to test the conceptual model examining the relationships among study constructs.

Results. Results suggested the developmental assets model fit well (chi-square=91, df=51; RMSEA=.039, CFI=.957, TLI=.947), with connection being the most important asset construct at both time points. Two divergent model pathways to re-arrest were prominent. The first, a mental health issues (MH) path began with Time 1 connection and included pathways through Time 2 connection and competence. A second, substance use issues (SU) path was also linked to pre-incarceration problems with connection. Both paths were linked to re-arrest, but the MH path was more significantly linked to receipt of services and less likelihood of re-arrest. The SU path was not likely to receive services and more likely to be arrested.

Conclusion and Implications. Results of this study suggest that for social workers, a strengths-based assessment and understanding of prison admission, incarceration and reentry services through a developmental assets framework may be influential to a comprehensive and meaningful reentry strategy for older adolescents. Additionally, the findings highlight the importance of strengthening reentry strategies that focus on addressing mental health and substance use issues while also inhibiting re-arrest.