As noted above, the majority of our interesting findings were seen in the AA part of the sample, despite our having slightly lower power in the AA part of the sample. This has been the case for all other SD traits that we studied in this sample [alcohol, cocaine, and opioid dependence (1–3)] but not for posttraumatic stress disorder, where the most interesting results were in the EA part of the sample (37). We believe that this is most likely an artifact of population history and differing selection pressures on populations, but this hypothesis remains to be tested.
