Culverhouse et al. include an a priori plan to test for effects of study design features on heterogeneity in findings, and include a list of five design features to be tested. We applaud this approach. However, the list of design features to be tested omits sample size. We find this omission curious because sample size has been at the heart of debate in the literature about prior meta-anlayses of this GxE. The heterogeneity analyses proposed by Culverhouse (cross-sectional vs. longitudinal, interview vs. questionnaire, specific stressor vs. undifferentiated stressor) are important analyses to guide the field going forward. Unfortunately, because so many high-quality longitudinal, interview-based, and specific-stressor studies have been excluded by the sample-size restriction, the results of the planned analyses will be difficult to interpret. Excluding small studies instead of testing for their putative bias on findings seems a missed opportunity for the Culverhouse team. In fact, our claim is not really that smaller studies are more desirable. Our claim is that the largest studies are least desirable because they have the worst measurement technology and in many cases have
