Models were fitted using −2log likelihood simultaneously in 5 groups; MZ male, DZ male, MZ female, DZ female, and DZ mixed-sex pairs. Nested models were tested using the χ 2 fit statistic to test for significant change in model fit. Akaike’s Information Criterion (AIC) was also used in tandem with the −2LL. Broadly speaking, when comparing models a smaller AIC value suggests a better (or more parsimonious) fit. The AIC can be particularly useful when comparing models that are not nested (e.g. ACE models vs. saturated models). Prior to genetic analyses square-root transformations were applied to all variables to normalise their distributions. Residuals were taken to account for the effects of sex and age on all variables.
