Causal Inference of Adolescent Antisocial Behavior and Young Adult Alcohol Use On Adult Criminal Justice Involvement

Life events such as marriage and stable employment have been found to facilitate the desistence of antisocial behavior during the transition from adolescence to young adulthood (Laub & Sampson, 1993; Le Blanc & Loeber, 1998), whereas alcohol use has been identified as a risk factor during this process, the effect of which is moderated by gender (Huang et al., 2001; Mason & Windle, 2002). However, many previous studies failed to control potential confounders, or failed to consider the time-varying nature of many confounders that could potentially be influenced by previous antisocial behaviors. The current study aims to examine the effect of adolescent and young adult antisocial behavior, and the mediation of young adult alcohol use, on adult criminal involvement. The secondary aim is to investigate the desistence of different subtypes of antisocial behavior, non-aggressive delinquency and aggressive delinquency, as well as gender’s moderation effect.

A national representative sample with approximately 14,000 participants were selected from the 1^st wave (94-95, adolescents in grades 7-12), 3^rd wave (01-02, young adults age 18-26), and 4^th wave (07-08, adults age 24-32) of Add Health data (Harris et al., 2009). Participants self-reported their past year’s involvement in non-aggressive delinquency, aggressive delinquency and alcohol use in both 1^st and 3^rd wave. In addition to demographics including age, gender, ethnicity, participants also reported other confounders, such as parent-child relationship, self-esteem, social support, school achievement, future uncertainty, marriage status, current employment status, many of which were time-varying. Self-reported criminal justice involvement in 4^thwave served as outcome. Marginal structural models with inverse-probability-of-treatment weighting was employed to account for time-varying confounders and estimate causal direct and mediation effects in the presence of moderation (Coffman & Zhong, 2012; Robins et al., 2000).

Results suggest that involvement in adolescent antisocial behavior, no matter non-aggressive or aggressive delinquency, predicted young adult alcohol use, but in opposite direction (B = 0.68 and -0.12). They also positively predicted involvement in antisocial behavior in young adulthood (B = 1.26 and 1.73), which were moderated by gender. Females demonstrated greater susceptibility to adolescent antisocial behavior. However, while both subtypes committed in adolescence and young adulthood predicted criminal justice involvement in adulthood, young adult alcohol use only mediated adolescent aggressive delinquency (B = 0.27), not non-aggressive delinquency (B = -0.01, ns). The results highlight the potential distinct mechanisms of alcohol use on predicting desistence of different subtypes of antisocial behavior and emphasize the moderation of gender.