Development of a Multidimensional Approach to Understanding Youthful Offenders: The Influence of Psychosocial and Personality Risk Factors

This study employed a multivariate, multidimensional approach to understanding psychosocial and personality variables associated with institutional maladjustment and recidivism among youthful offenders. Participants included nine hundred serious and chronic male youthful offenders incarcerated in the Texas Youth Commission (TYC); sample sizes varied by analysis. Empirically-validated psychosocial factors (e.g., intelligence, home approval status), past criminal history variables, and two self-report personality measures of empathy and hostility were entered into hierarchical regression and structural equation modeling (SEM) analyses to predict institutional behavior and recidivism at one- and three-year intervals after release from the TYC. Confirmatory factor analysis of the personality measures revealed one underlying factor indicative of their theoretical constructs of empathy and hostility. Some differences were noted between youth in the specialized treatment programs; however, effect sizes were small to moderate. Overall, regression and SEM results indicated the variables accounted for a meaningful proportion of the variance in the outcomes. Specifically, although length of stay in the TYC was associated with institutional behavior, younger age of onset, higher hostility, and greater home disapproval also contributed significantly. Past criminal behavior was predictive of future reoffending, but lower empathy, greater home disapproval, and younger age of onset accounted for a substantial portion of the variance in recidivism. Institutional maladjustment served as a mediator between the psychosocial and personality variables and the recidivism outcomes. Treatment implications are provided, including a discussion of the tenuous association between length of sentence and recidivism and an emphasis on the importance of evaluating dynamic personality and psychosocial variables beyond static measures of past behavior.