Under the classical twin model assumptions, the expected MZ twin correlation is a function of the proportions of variance in a trait explained by additive (h 2) and non-additive (d 2) genetic effects: r(MZ) = h 2 + d 2. The expected DZ twin correlation is a different function of these two types of effects: r(DZ) = ½h 2 + ¼d 2. IRT-score-based twin correlations (Table 1) were used as the basis to assess both qualitative and quantitative sex effects. This was done by fitting the same model to data from all six cohorts simultaneously allowing for different estimates of h 2 and d 2 in each sex, and allowing the opposite-sex twin correlation to be different from its expectation, ½h m h f + ¼ d m d f. The estimates of parameters (h 2, e 2 and d 2 by sex) thus were constrained to be the same across cohorts. First it was tested whether the correlation in opposite-sex twins could be equated to the expectation above (i.e. testing for qualitative sex effects). Next, it was tested whether
