Longitudinal social network analysis of peer, family and school contextual influences on adolescent drinking frequency

Purpose: The aim of the study was to identify the mechanisms relating to parental control, adolescent secrecy, and school context that shape patterns of adolescent drinking frequency and appraise the implications for systems-level intervention. Methods: The Belfast Youth Development Study collected information on friendship networks in schools, alcohol use, and Stattin and Kerr's parental monitoring subscales across 5 years of postprimary school education in annual waves from age 11–15 years. Stochastic Actor-Oriented Models were fitted to 22 schools (N = 3,220) to assess friendship formation and peer influence processes related to drinking frequency and their variation by parental control or child secrecy. Meta-regressions and summary statistic ego-alter selection tables assessed how network and behavior co-evolution varied according to school gender and the proportion of weekly or more frequent drinkers in each school. Results: Adolescents tended to mimic their peers' drinking levels, and frequent drinkers befriended those who drank similarly to them. Those with high parental control were less likely to befriend low-control peers, whereas low-control pupils were more likely to befriend each other. Adolescents with low-control parents nominated fewer friends in schools with higher proportions of drinking frequently. There was a tendency toward befriending highly secretive peers in boys schools only. Conclusions: Our results suggest that the optimal strategy for selecting seed nodes in a diffusion of innovations network intervention may vary according to school context, and that targeting family interventions around parent characteristics may modify the wider school network, potentially augmenting network intervention processes