Finally, these applications of IDA collectively demonstrate how we can identify replication of effects across studies as well as the failure for replication. Core findings were largely replicated across studies, despite significant differences in studies due to geography, cohort, assessment, and sampling. However, this was not always the case. In attempting to reconcile differences in our findings across studies, we were able to model potential study differences (e.g., in rates of parental depression and antisociality across studies) that may account for significant interactions between our predictors of interest and study membership in these models. In most cases, these covariates did not account for significant across study differences in our findings. However, the test of whether these factors accounted for replication failures addressed the plausibility of such comorbid parental disorders as the source of between study differences, obviating the need for a new study in which this factor is systematically controlled in order to reconcile study differences.
