Analyses of the association between each SNP in turn and our endophenotypes were conducted by means of the R package for rapid feasible generalized least squares (RFGLS; Li, Basu, Miller, Iacono, & McGue, 2011), a computationally efficient form of generalized least squares (GLS) developed for this purpose. GLS is useful with correlated data, such as the correlation that exists when subjects are nested in families (see Iacono et al., 2014, for details). Our sample comprised MZ and DZ twin families. In addition, the 74 stepparents in the sample (70 of them male) were treated as families of one. The conventional genome-wide significance threshold of 5 × 10−8 was used.
