Power analyses were run to explore the sample size necessary to detect a significant difference between the correlation in DZ same-sex (DZss) twin pairs and DZ opposite-sex (DZos) twin pairs. Using an MX-script [24], we tested the difference between DZss and DZos correlations by likelihood-ratio tests. The required sample sizes for statistical power ranging from .75 to .99 (with significance level 0.05 and 1 degree of freedom) are shown in Table S6.
