Findings from the current study should be interpreted with certain limitations in mind. First, as indicated by the wide confidence intervals on variance component estimates, the model derived from analyses conducted with the AA subsample is somewhat unstable and should therefore be treated as suggestive rather than definitive. Second, the absence of statistical significance between the equated vs. the separate AA and EA models should be considered when comparing findings across ethnic groups, but not without weighing in results of the additional univariate analyses of problem alcohol use suggesting the absence of significance is likely due at least in part to low power. Third, given our interest in examining the association of initiation of alcohol use with problem drinking, which is contingent on having consumed alcohol, we excluded lifetime abstainers, and African Americans were overrepresented in this group (13.2% of African-American vs. 5.5% of European-American MOAFTS participants). Our conclusions regarding the relative contributions of genetic and environmental factors to the timing of first drink therefore may not generalize to the phenotype of any lifetime use. Fourth, although minimized by the
