E analyses, one theme that has emerged is that genetic variance for conduct disorder is higher in less restrictive environments (e.g., those characterized by higher peer deviance, less parental monitoring, or urban residency), which is consistent with the social opportunity mechanism of G × E described by Shanahan and Hofer (2005). The reverse also holds, such that genetic variance is minimized in more restrictive environments characterized by lower peer deviance, higher parental monitoring, or rural residency, which is consistent with the social control mechanism of G × E (Shanahan and Hofer, 2005). In the time since Purcell (2002) popularized biometric models for probing latent G × E effects, others have raised concerns about Type I error and proposed alternative parameterizations (Rathouz et al., 2008; van der Sluis et al., 2012). The robustness of G × E effects for conduct disorder related phenotypes to these alternative specifications has not been examined systematically.
