As indicated above, a variety of secondary, interpretive analyses are planned. In addition to the alternate genotype codings and covariates listed above, tests of heterogeneity between datasets are expected to lead to secondary meta-analyses of results from more homogeneous datasets. In particular, it has been argued that many studies that have failed to replicate the original studies lacked in-depth environmental measures [30]. We will investigate potential differences in results between brief, self-report questionnaire measures of stressful life events and face-to-face interviews as well as differences due to the type of study and the diagnostic system used for assessment. Therefore, in addition to subsets identified through tests for heterogeneity, stratified meta-analyses based on the following factors are planned:
