In order to test whether the association between parental knowledge and adolescent externalizing behavior arises because of evocative rGE, passive rGE, or environmental effects free from genetic confounds, we employed the ECOT model (Figure 1). The lower box on the left of Figure 1 represents the adolescent twin and sibling sample, while the larger box on the right represents the parent twin sample. The ECOT model was run using Mx (Neale, 1999). Missing data was accommodated using full information maximum likelihood data estimation procedures. All analyses were conducted after controlling for age, sex, and age difference (for non-twin siblings in NEAD and cousin pairs in TOSS). Measurement error was estimated, but constrained to be equal across siblings, studies, and phenotypes in order to preserve model identification (ε1 and ε2 in Figure 1).
