A limitation of this study is that our sample may not be representative of all psychiatrists practicing in the United States, because members of the APA and psychiatrists who attend APA meetings may be systematically different from other psychiatrists. Our response rate was adequate for this type of study, but response bias may limit the generalizability of our results—that is, we cannot determine whether psychiatrists who elected to participate in the study are systematically different from those who did not respond. Moreover, all data were collected by self-report, which may make comparisons less reliable. We attempted to standardize responses by defining all key terms, but our findings must be interpreted as stemming from respondents’ subjective beliefs. Finally, the cross-sectional study design prevents us from drawing conclusions as to causation. Longitudinal research is needed to clarify the direction of the relationships that we have reported. Conclusions