Abstract:
Very little is known about social network patterns associated with substance use among adolescent New Zealanders. Social network analysis is useful for understanding subtle and interdependent drivers of adolescent substance use in ways that are not possible using conventional statistical analysis. This thesis tests the hypotheses that: there are upstream determinants of adolescent substance use stemming from social network structure, social influence processes can be distinguished from social selection processes and that substance use cognitions can diffuse through networks at a significant rate based on exposure and susceptibility of networks actors. An important component of this thesis applies newly developed social network and agent-based models to test hypotheses and answer follow-on research questions. Social network data was collected from two Auckland co-educational high school friendship networks (n=135 students aged 16 to 18) in 2013. Cross-sectional analyses were performed on both networks and a longitudinal analysis was performed on one of the networks. Homophily according to gender and ethnicity was found in the observed networks. Smokers tended to be friends with other smokers and those friendships appeared to be driven by social selection processes. The same processes did not appear to drive drinking behaviour. Homophily among drinkers was demonstrated and appear to be driven by social influence processes. Anti-substance use cognitions also appeared to diffuse based on exposure to popular and/or male network actors who held the cognition. These findings are important because to our knowledge this is the first study to examine associations between exposures in adolescent social network structures to substance use and associated cognitions in New Zealand.