Twins reared together: minimizing shared environmental effects.

The assumption that genetic variance is primarily (if not all) additive is usually made in biometric-genetic analyses of data collected on twins raised together. It is known amongst those familiar with twin methods that this assumption may lead to overestimates of heritability and under-estimates of shared environmental variance (E2), although this limitation is not always made clear to genetically native readers of such applications. The concept of "emergenic" genetic mechanisms (a potentially extreme epistatic or nonadditive mechanism) discussed by Lykken (1982) raises the possibility that genetic variance may be substantially nonadditive in some applications. The aims of the present paper are to investigate the potential size of such nonadditivity and such misestimations and to provoke discussion on the empirical plausibility (or otherwise) of epistatic effects. For if substantially present, the results of conventional twin analyses are substantially biased.