Adolescent Substance Use and Aggressive Behaviours in Multiple Structural Peer Contexts (SRCD)

Abstract

Many changes in health-related behaviours occur during adolescence (Williams, Holmbeck, & Greenley, 2002). Peers play a critical role in such behaviour changes, since they can substantially influence youths’ health behaviours (Ryan, 2001). Adolescents tend to adhere to the behavioural norms (i.e., the average level of behaviour by peers) conveyed by peers in relevant peer contexts as this will likely enhance their status, which is of increasing concern to adolescents (Brown & Larson, 2009). As such, adolescents tend to be similar in behaviour to their peers (homophily; Lazarsfeld & Merton, 1954). Since adolescents interact with peers in multiple contexts, adolescent are exposed to multiple, potentially opposing behavioural norms. Therefore, various peer contexts may account for behavioural similarity (Kiesner, Kerr, & Stattin, 2004) and behavioural similarity may vary across different peer contexts as a function of the specific behaviour. Typically, earlier peer homophily studies examined behavioural similarity in one peer context. The current study extended earlier research by examining homophily in three structural peer contexts (classroom, school, age-cohort). Structural peer contexts are formal, involuntary peer contexts to which adolescents belong without having any choice over their membership. Further, contrary to most other peer homophily studies, the current study investigated homophily in multiple adolescent health-risk behaviours including substance use (tobacco, alcohol, cannabis) and aggressive behaviours (bullying, physical fights) in a sample of 5.642 12- to 16-year-old Dutch adolescents (264 classes, 68 schools) drawn from the 2009/2010 Health Behaviour in School-aged Children (HBSC) study (Roberts et al. 2009). Given the three-level hierarchical structure of the data (participants nested within classrooms nested within schools), multilevel analyses were run. Descriptive and correlational statistics are presented in Table 1.The distribution of the behaviours were positively skewed (most adolescents did not engage in them). Table 1 show that most individual scores were positively associated with the peer context profile scores (PS; i.e., the behavioural norm). Results of the final multilevel Model 5 (including all fixed and random effects) are presented in Table 2. Analyses revealed significant homophily effects of age-cohort and classroom norm on substance use, and significant homophily effects of classroom and school norm on aggressive behaviours. Cannabis use, however, was related to school norm and not classroom norm. Furthermore, age-cohort’s effect on individual alcohol use was moderated by classroom norm; age-cohort alcohol use was more strongly associated with individual alcohol use for adolescents whose classmates scored high on alcohol use (B = 1.00, SEB = .27), compared to medium classroom alcohol use (B =.50, SEB = .08) or low classroom alcohol use (B = .39, SEB = .07, all ps < .001). Thus, low classroom alcohol use weakened the effect of high age-cohort alcohol use on individual alcohol use. To conclude, substance use appears to be a more age-related phenomenon, where aggressive behaviours appear to be more affected by direct interactions (classmates) or shared environments (schoolmates). Similarly, shared environments may also play an important role in cannabis use. These results help policy makers to determine which peer context to target first to promote adolescent health.