I would like to comment on the article regarding genetic locus associations with cognitive ability (Reference Plomin, DeFries and CraigPlomin & Craig, 2001). The use of g as a measure of ‘intelligence’ or ‘cognitive function’ is controversial and far from universally accepted. Significant criticisms ofg have been put forth (Reference Neisser, Boodoo and BouchardGould, 1996) and g has been used to promote some rather objectionable eugenic views in the past (Reference Plomin, Hill and CraigHernstein & Murray, 1994). Admittedly, that still leaves the task of explaining the alleged positive correlation between a high g score and some genetic loci.

First, I would hope that any common racial/cultural prevalence found in the ‘high’ g group compared with the ‘average’g group has been controlled for, as this would otherwise tend to produce a number of false positives simply due to genetic homogeneity.

Second, despite the claim that the expected number of chance false positives in such a DNA pooling study were exceeded, I challenge the authors to use a real-world control rather than a mathematical calculation to prove this assertion. This could be calculated quite simply by randomising the initial sample without regard to g scores and determining the number of positive linkages found. The same randomisation could then be performed for the second sample and a replication attempted. Presumably, any replicated linkages would be false positives unless you wanted to track down the study subjects and find something that they had in common (e.g. finding raisins to be tasty or some such thing). This could be repeated several times to give a true expected false positive rate. I suspect that, on average, the randomised groups will have as many positive linkages as those found in the initial study.