more than twice as high as DZ twin correlations, non-additive genetic influences (D) were estimated. Non-additive genetic effects and shared environmental effects are confounded in twins reared together, respectively leading to DZ twin correlations less than one half the MZ correlations, and inflation of the DZ correlation to be greater than one half the MZ correlations (Martin et al., 1978). As such, non-additive genetic effects and shared environmental effects could not be included in the same models. Univariate threshold models were fitted to 2×2 or 3×3 contingency tables for the nicotine withdrawal measures. Likelihood-based 95 percent confidence intervals were computed for estimates of genetic and environmental parameters (Neale et al., 2002). The full ACE or ADE models were estimated first and compared against reduced models by likelihood-ratio chi-square, to determine the best-fitting model.
