A second unique feature of our sample is that the family structure of the data allows us to estimate heritability using both biometric as well as GCTA methods. The comparison of the two sets of heritability estimates can be informative in terms of the missing heritability problem (Manolio et al., 2009). Specifically, GCTA estimates the extent to which additive genetic effects captured by variants included on the GWAS array account for phenotypic differences (Yang et al., 2011). In principle, GCTA estimates the magnitude of heritable variance detectable from markers from the genotype array when statistical power is perfect (i.e., sample size is infinite) (Visscher, Yang, & Goddard, 2010). Alternatively, biometric estimates of heritability index all additive genetic effects, regardless of whether they are captured by the SNP array. GCTA estimates for height (Yang et al., 2010) and general intelligence (Davies et al., 2011) indicate that somewhat more than half the heritable variance for these traits can be accounted for by common SNPs. GCTA estimates for personality (Vinkhuyzen et al., 2012) and schizophrenia (Lee et al., 2012) suggest that less than
