The intergenerational transmission of "intelligence": Down the slippery slopes of The Bell Curve

Herrnstein and Murray report that conditional on maternal "intelligence" (AFQT scores), child test scores are little affected by variations in socioeconomic status. Using the same data, we demonstrate that their finding is very fragile. We explore the effect of adopting a more representative sample of children, including blacks and Latinos, allowing nonlinearities in the relationships, and incorporating richer measures of socioeconomic status. Making any one of these changes overturns their finding: Socioeconomic status and child test scores are positively and significantly related. Evidence is presented suggesting AFQT scores are likely better markers for family background than "intelligence".