Although, in many ways, we improved the design of previous candidate gene studies, our report still has limitations. First, as noted previously, a proper correction for multiple testing has been problematic in these studies. Using too liberal an approach will maximize power but is likely to lead to false positives. A correction method that is too conservative will decrease power to detect true results. We attempted to strike a balance between the two approaches by using gene-based correction. However, we did not correct for testing multiple phenotypes because there is a priori evidence of association of at least some of the 10 DA system genes with each of the phenotypes examined. Furthermore, many of these phenotypes are highly inter-correlated, making an appropriate correction problematic. It can be argued that our approach is still too liberal. The possibility that some proportion of our findings represents false positives is plausible. Therefore, our findings should be considered tentative, pending the outcome of attempted replications. Secondly, quantitative alcohol-related traits and disinhibitory symptoms were measured only among cases but not controls, so the values for
