the same direction. There was reduced power to find effects in FinnTwin12 because there were a fewer number of total observations relative to CDP owing to the CDP’s longitudinal repeated measures design. As expected given the reduced degrees of freedom, our G × E effects were attenuated to some degree in the analyses that included all covariate × environment and covariate × genotype interaction effects. However, the effects were not removed, which adds confidence that these effects are not due to these potential statistical confounds.
