Estimating the number of loci determining skin colour in a hybrid population

A model developed by Stern to estimate the number of loci determining human skin colour in African-European hybrid populations is tested on recently collected reflectance data. Despite improved measurement scales and admixture estimates, and the use of statistical goodness-of-fit tests, contradictory results are obtained. It is shown that previous authors' use of visual inspection to compare expected and observed phenotype distributions introduced considerable bias into their results.