The overall objectives of this research are 1) to improve our understanding of the causal processes involved in becoming a user or abuser of alcohol; and 2) to improve upon methods of statistical analysis of longitudinal data in exploring causal processes. This work will be conducted on seven existing longitudinal data sets, each with four or more waves of data collected from youth between the ages of 10 and 25. The specific objectives and aims are as follows: 1. Test and replicate hypotheses regarding causal processes using the framework of an integrative theory of behavior onset and change developed by these investigators, the Theory of Triadic Influence (TTI). Specific hypotheses concern mediating influences, moderating effects (including of gender and ethnicity) and reciprocal causation mechanisms (both among predictors and regarding feedback effects of behavioral experience). 2. Extend methods of longitudinal data analysis in the following five areas: conceptualizing stages and levels of substance use, modeling of progression from stage to stage, dealing with time-varying predictors and endogeneity, handling missing data and attrition, and modeling and testing interactions in causal chains (moderated mediation). The proposed study is important because it: encompasses the high risk age range (10-25); enables replication of tests and contrasts of the most important of the existing theories; incorporates a wider range of predictor variables than any other single study; includes data from multiple ethnically and socioeconomically diverse samples; utilizes sophisticated analytical techniques to examine mediating, moderating and feedback relationships; has large enough combined sample sizes to enable split-half testing and replication for many analyses; provides opportunities to develop and test new methodology for modeling of multi- wave data; is cost-efficient because of the existence of the six data sets; and will have important implications for future prevention programs/research.
