Limitations of the study include the fact that the BAC estimation procedure was not as precise as that obtained in the laboratory and that these results may not generalize to treatment-seeking samples and/or social drinkers. In addition, selection bias in the group of consenters to the DNA analyses resulted in a greater representation of female participants among consenters. Although gender was controlled for in all genetic models, selection bias cannot be completely ruled out. Similarly, non-consenters tended to be heavier drinkers than consenters, and as such, the selection of individuals of very heavy drinking patterns may bias the sample in terms of genetic and phenotypic characteristics related to responses to alcohol. In this study only the initial two drinks of the day were assessed, which may not generalize to the subjective effects of alcohol observed at higher levels of BAC. Nevertheless, the subjective effects of alcohol after the first two drinks may be especially relevant to whether or not individuals escalate their drinking within a given episode, and more generally, in drinking situations. Lastly, the gender by eBAC interaction was
