Testing the Social Interaction Learning Model with Adolescent Substance Use Outcomes in a Prospective Australian Sample

Christopher Mehus, Jennifer L. Doty, Gary C. K. Chan, Adrian B. Kelly, Richard F. Catalano, John Toumbourou, Barbara McMorris

Introduction: Parents and peer contexts both impact the development of adolescent substance misuse. The Social Interaction Learning (SIL) model provides a theoretical explanation of relationships between parenting, peers, and youth outcomes. The SIL model posits that parenting has a direct effect on youth behavioral outcomes but also that this relationship is partially mediated through children’s anti-social behavior and anti-social peers. In other words, poor parenting contributes to increased anti-social behavior, which contributes to anti-social peer associations, which contribute to poor outcomes. The SIL model has been tested primarily with samples from the United States with conduct-disorder outcomes. Here, we add to the field by testing the full SIL model with substance-use outcomes in a longitudinal sample of Australian youth.

Methods: Three waves of data (grade 5, 6 & 10) from students in Victoria, Australia participating in the International Youth Development Study (IYDS) were utilized. After excluding dishonest respondents, the analytic sample consisted of n=907 youth. We created latent variables of parenting and anti-social behavior in 5th grade and anti-social peer associations in 6th grade. Substance use outcomes were 10th grade recent alcohol, marijuana, and tobacco use, and binge drinking. We controlled for substance use in 5thgrade. Using Structural Equation Modeling, we evaluated model fit when including the direct paths from parenting to substance use outcomes, as well as the mediational paths from parenting to anti-social behavior, to anti-social peers, and then to substance use outcomes.

Results: The model fit was good (model results: χ2 = 249.79, df = 95, p< .001; RMSEA = .04 [90% CI: .036 - .049]; CFI = .94; SRMR = .04). The normed chi-square also suggested adequate fit (χ2/df = 2.63). Path estimates from parenting to anti-social behavior (β = .42) and anti-social behavior to anti-social peers (β = .44) were significant (p < .001). In turn, having anti-social peers was significantly related (p < .001) to alcohol use (β = .23), binge drinking β = .28), tobacco use (β = .32), and marijuana use (β = .18). From parenting, only the direct paths to binge drinking (β = .11) and marijuana use (β = .10) were significant (p < .05), but indirect effects were significant (p < .001), suggesting mediation through the anti-social behavior and peer pathways.

Conclusion: Findings support the usefulness of the SIL model in a cultural and policy context that differs from the US with regard to underage drinking. Replication of this model in diverse societies suggests these paths may have universal application and, therefore, preventive interventions grounded in this model may be easily adaptable to a range of communities.

This abstract was submitted to the 2017 Society for Prevention Research Annual Meeting.