Latent Clusters of Peer Behaviors in Childhood and Relevance for Understanding Substance Use in Early Adolescence

Peers have been found to be an important influence on adolescent behavior. However, there remains limited research examining childhood peer behaviors as predictors of later substance use. Examinations have not included full behavior profiles comprising both antisocial and prosocial peer behaviors. This study sought to identify latent clusters of peer behaviors during childhood and their relevance for understanding substance use. The Adolescent Brain Cognitive Development (ABCD) dataset was analyzed. These data are composed of the full ABCD sample of 11,880 United States children aged 9-10 years at baseline and surveyed annually, with data collection beginning in 2017. A multistage cluster sampling methodology was used to collect data from catchment areas surrounding the 21 study sites. Latent class analysis was used to identify peer behavior clusters. Logistic regression was used to examine the relevance of childhood latent peer class assignment for predicting substance use in early adolescence. A three-class solution was identified (neutral, prosocial and antisocial peers, prosocial peers). Assignment to the prosocial and pntisocial peers peer class predicted increased substance use risk. Findings indicate that youth with peers involved in both prosocial and antisocial behaviors in childhood present the highest risk for substance use in early adolescence.