Structural twin models were used to estimate the extent to which SLE, drinking onset, and their interaction accounted for genetic and/or environmental variance in PYDD. We estimated 3 sources of variance: additive genetic (A) (i.e., variation because of allelic effects that combine additively), common environment (C) (i.e., variance attributable to experiences shared by twins), and individual-specific environment (E) (i.e., variance unique to 1 twin in a pair, including measurement error) (see Prescott et al., 2004). MZ pairs share 100% of additive genetic variance, whereas DZ pairs share 50%. Higher similarity for twins in MZ pairs than DZ pairs suggests a genetic contribution to the phenotype, and equal similarity implies common environmental contributions. To the degree that twins in a pair are not similar, this implicates individual-specific environmental factors and measurement error as contributing to the outcome. We estimated twin models by maximum likelihood using the programMplus v5.2 (Muthén and Muthén, 2007) based on raw data from individuals from complete and incomplete twin pairs.
