Recently developed statistical models for understanding development and change will likely prove useful in studying gene-environment interplay. For instance, bivariate growth curve models can be used to study direction and mutuality of influence between candidate endophenotype and externalizing behaviors. This is particularly relevant in the area of substance abuse, where use of the substance is prerequisite for the disorder but might itself, as Spear’s review (This Issue) highlights, alter brain function and maturation, which in turn could affect the endophenotype. Biometric models of gene–environment interaction in relation to candidate endophenotypes are useful in his regard because they can evaluate how the heritability of an endophenotype is moderated by the ingestion of a substance. Perlman et al. (2009) employed this approach in a longitudinal investigation that showed that the amount of alcohol ingested from age 11 to 18 did not moderate the heritability of P300 amplitude measured at age 18, thus showing that whatever the neurotoxic effects of alcohol, they did not alter the genetic influence on P300 at this stage of late adolescence.
