A survey of 1,340 students in grades six through twelve was conducted to test, in the context of structural equation models, the predictive validity of a theoretical model of antecedents of adolescent alcohol use and misuse. Constructs including parents' alcohol use, older siblings alcohol use, parents' approval of students alcohol use, older sibling approval of students alcohol use, peer use and approval of alcohol use (PUA), parental nurturance, parental permissiveness, child's grade in school, susceptibility to peer pressure (SPP), and deviant self-image, were included. The final iterations of the models accounted for more than half of the variance in both alcohol use and alcohol misuse. In a standardized solution, the two largest direct effects on both adolescent alcohol use and misuse were from SPP and PUA. When a seven item measure of SPP, including three items specific to alcohol use was used, the SPP latent variable accounted for a somewhat greater percentage of the variance in adolescent alcohol use and misuse than did the latent variable of PUA. When the three alcohol-specific items were deleted from SPP, however, PUA accounted for more variance than SPP. Other predictors, including parental behaviors, proved to be significant when their indirect effects were evaluated. Models predicting alcohol use and alcohol misuse were similar. SPP, PUA, and other significant predictors should be included in future models predicting adolescent alcohol use and misuse, as well as in future interventions targeting these behaviors. Longitudinal studies should be used to test these findings. Evaluation of prevention should include examination of possible interactions of these predictors with each other and with subject subgroup classifications.

• AA 06324/AA/NIAAA NIH HHS/United States