Understanding how genetic, environmental, and developmental influences impact complex behavioral health outcomes is a challenging task. Large samples are necessary, both to utilize increasingly sophisticated molecular genetic methods (because of the small effect sizes of most risk variants) and because of the heterogeneity and subtlety of many developmental pathways. However, characterizing large numbers of subjects phenotypically, across time, along with the needed assessments of present and past environmental risk factors can be cost-prohibitive. Accordingly, achieving the requisite sample size has often required combining samples collected by different groups, having different characteristics, and using varied assessment measures. This introduces noise into any analysis and often means that only the most basic phenotypes are available for study once data are harmonized.
