body mass and therefore require more drinks to feel the effects of alcohol, we statistically controlled for sex and BMI. It will be important for future work to examine the role of biological sex as a potential moderator. Third, given that the sample came from a high-risk community study and all participants had experience with alcohol, the results may not generalize to low-risk and/or healthy community populations. At the same time, because drinking levels would be anticipated to be substantially lower in such populations, it is also likely that any findings would be considerably attenuated because of lack of variance. Finally, some methodological caveats should be noted. Our assessments were largely self-report measures completed by the adolescent/young adult participants, introducing the possibility that shared method variance played a role in the results. Additionally, measuring early (i.e., innate) sensitivity to alcohol in a sample with substantial variability in drinking age initiation meant that there was large variability in assessment ages, though we statistically controlled for age. Finally, though a sample size of 218 would be considered large for many types of studies, this represents a small sample for testing a conditional process genetics model; despite being consistent with much of the
