Direct and indirect IQ heritability estimates: a puzzle

Direct estimates of IQ heritability based on a single family relationship such as adopted-apart relatives are often 50% greater than indirect estimates that rely on differences in correlations such as the classical twin method or nonadoptive-adoptive comparisons. Factors such as nonadditive genetic variance, assortative mating, selective placement, measurement error, age differences, and genotype-environment correlation and interaction do not obviously explain the difference between direct and indirect IQ heritability estimates. Because direct estimates are derived from separated individuals and indirect estimates are derived from individuals reared together in families, some aspect of the within-family environment seems a likely candidate but its exact nature remains to be understood.