Various bloggers have been discussing issues relating to IQ, notably in relation to Herrnstein and Murray’s very bad book The Bell Curve (bad intellectually because it was sloppy and dishonest and bad morally because it used dishonest arguments in a bad cause). Although I’m not an expert on issues specifically relating to intelligence testing and heritability, I know enough about the literature to make the following point about measurement issues (on which I can claim to be an expert). The literature shows that, in Western societies, variations in both environment and heredity make a substantial difference to measures of intellectual performance (including educational achievement and IQ test scores). Furthermore both the literature and everyday experience tell us that these things interact in a very complex fashion.

When two causal variables interact in this fashion to determine a third variable of interest, the idea that there is a meaningful sense in which proportions of the variance in the variable of interest can be allocated to the causal variables (as in statements of the form “x per cent of variance in IQ is due to heredity”) is unsustainable. Minor changes in definitions, model structure and functional form will produce major changes in results.