FALLACIOUS ANALYSES CANNOT TEST FOR RACIAL-DIFFERENCES - A REPLY TO GOREY AND CRYNS

Although insufficient information was provided to allow an examination of the actual calculations in Gorey and Cryns' meta-analysis, their r esults and conclusions are demonstrably false. For example, they erron eously reported the point-biserial effect size r of the white-black di fference in IQ alternately as 0.226 and 0.022, even though it is known that the actual value is 0.50 (100 - 85/15 = z score of 1.00 which tr ansforms to the point-biserial r of 0.50). Gorey and Cryns' analysis a lso failed to detect widely acknowledged black-white differences in ou t-of-wedlock births, crime, and numerous other indicators of 'social o rganization'. Much of this failure to detect reality was due to inclus ion of 'matched samples' at distributional extremes. Regardless, their conclusions do not follow from their analyses. First, their only stat istically significant results were consistent with those reported by R ushton (Personality and Individual Differences, 9, 1009-1024, 1988). S econd, their 'per cent variance, accounted for' argument is statistica lly correct but substantively erroneous. Finally, Gorey and Cryns excl uded from discussion macrophysiological variables like testosterone, r ate of two-egg twinning, and brain size (which, like IQ, also shows a block-white r = 0.50). The racial gradient on all these variables is f ound worldwide and is directly relevant to causal analysis.