In addition, we examined genetic factor scores (Boomsma et al., 1990) derived from a biometric common pathway (Kendler, Heath, Martin, & Eaves, 1987) or psychometric factors (McArdle & Goldsmith, 1990) model, with P3 amplitude at the three parietal locations as indicators of a common factor, representing what is shared by the three P3 amplitude measures. Unique factors captured electrode-specific influences, including noise. Preliminary moderated factor analyses (D. J. Bauer & Hussong, 2009) indicated that the four primary covariates (age, gender, generation, and recording system) were significant influences on the common factor mean. In addition, gender and generation influenced both the unique and common factor variances. (Table S1 provides parameter estimates from the preliminary factor analyses, and the supporting information provides additional detail concerning the model.) Thus, our final biometric model allowed for gender and generation effects on common factor and unique variances, with effects on the factor mean accommodated through adjusting P3 measures for the four covariates (age, gender, generation, and recording system). Figure 1 illustrates the model for one individual, although data from all family members are used in estimating the model.
