For male problem alcohol use at the second assessment, it was possible to drop the moderating effect of religiosity on all parameters without a significant reduction in the fit of the model. Consequently, the phenotypic variance for male problem alcohol use remained constant irrespective of religiosity. For females, it was possible to drop the moderating effects of religiosity on the genetic variance and the shared environment variance, but not on the non-shared environment variance. Therefore, both the non-shared environment and total phenotypic variance were attenuated at higher levels of religiosity.
