Twin studies in behavioral research: a skeptical view

We review in detail two major ongoing research projects that employ samples of twins reared apart (and in one case, twins reared together). The studies attempt, via model fitting, to estimate proportions of genetic and environmental variance for many human traits. We discuss problems concerning the representativeness of samples, the accuracy and reliability of the data, the extent of contact of nominally separated twins, the measurement of selective placement effects, and the particular model-fitting procedures. The two studies agree in their conclusions, but we do not find the conclusions to be convincing. We suggest that no scientific purpose is served by the flood of heritability estimates generated by these studies.