These results should be interpreted in the context of several important limitations. First, hierarchical linear modeling was used rather than structural equations modeling in our analyses. We chose this method because it allowed us to easily incorporate and interpret data on parental psychopathology in our measures of genetic risk. Second, although the present study uses developmentally-appropriate problematic drinking measures across time, there were differences in how the question was posed to the subject and how the response options were made available, and both introduce the potential for measurement invariance. Further, each measure is confounded with age and/or period of development. Therefore, it is possible that non-specific genetic risk (EXT-GR) is important for initiation of drinking regardless of age/period of development. Likewise, it is possible that alcohol-specific genetic risk (AUD-GR) is important for AD regardless of age/developmental period, however risk for AD peaks later in development. We note that Kendler et al. (2011) examined the same maximal drinking (drinking frequency × drinking quantity) measure across development. Therefore, comparisons between the current study and the previous study (Kendler et al., 2011) should
