In this symposium we discuss applications of four advanced statistical techniques to allow for a deeper assessment of contextual factors: Structural Equation Modeling; Differential Item Functioning; Discrete-Time Event History Analysis; and Propensity Scores. Each technique is demonstrated in a separate paper and provides unique information regarding adolescent substance use using data from the Santiago Longitudinal Study, a large NIDA-funded study of adolescent health, mental health, and substance use in Chile.
To understand the influence of family and peers and the role that they each play in protecting or promoting risky behavior, presentation #1 discusses the use of Structural Equations Modeling to study common paths shared by mental health problems and cigarette, alcohol, and marijuana consumption. To assess differences in the age at initiation for alcohol, cigarette and marijuana consumption, presentation #2 uses Discrete-Time Event History Analysis to show that each substance has a different age and trajectory of initiation. To explore the specific effects of peers on adolescent alcohol use, presentation #3 uses Propensity Score Stratification to examine heterogeneous peer drinking effects across groups of youth with different levels of probabilities of socializing with drinking peers. To examine individual attitudinal differences towards substances, presentation #4 uses Differential Item Functioning to examine whether gender influences how adolescents construct perceptions and attitudes towards cigarettes and marijuana.
Each paper will describe the particular statistical technique employed, the results, and the implications of their findings as they relate to the multi-level discoveries that are possible using various statistical techniques. Overall, the social work implications of presentation #1 include that multiple factors are involved in the consumption of substance use and mental health problems which imply that prevention and promotion efforts have to simultaneously target multiple systems for any real changes to be made in the lives of adolescents. In addition, presentation #2 suggests that early intervention efforts should continue to prioritize preventing the onset of alcohol and tobacco consumption while at the same time continue with activities aimed at preventing youth from initiating marijuana use The results of presentation #3 indicate that having peers who drink exert considerable influences on adolescent alcohol use suggesting that school-based interventions that focus on groups of adolescents, rather than at the individual level, may be more effective for curbing the harmful effects of peer influence on alcohol use. The results of presentation #4 suggest that gender influences how adolescents endorse the individual items measuring drug attitudes which translate to having social work interventions focused specifically on targeted messages that are sensitive to clients' gender. In conclusion, using different statistical techniques we can tell a richer story that takes into account the multiple social contexts of adolescents and how they each distinctly play a role in the various aspects of their drug use.