In interpreting the current results, several caveats should be noted. First, the sample was relatively small which may have obscured finding additional characteristics with small main effects and interactions. A related issue was that not all possible interactions were tested due to the a priori decision to limit such evaluations to covariates with significant main effect in order to minimize problems with Type I error, a step that may have increased Type II error resulting in unidentified significant interaction effects. Third, as noted above, the unique attributes of the SDPS generates caution regarding generalizing these findings to other populations. Finally, a larger sample of subjects would have greater power to detect smaller effect sizes that were not detected in the current study. Furthermore, information on additional predictors of substance use (e.g., drinking among peers, alcohol expectancies) was not available in the younger group reported here since such items were not added to the protocol until the most recent assessments of these offspring and their families.
