The Black-White Difference On the K-ABC: Implications for Future Tests

Abstract

Claims that the Kaufman Assessment Battery for Children (K-ABC) is less culturally biased than other standard tests of intelligence and therefore shows a much smaller average difference between black and white children are critically examined in terms of the psychometric properties of the K-ABC. It is concluded that the apparently reduced difference between black and white samples, as compared with the one standard deviation difference typically found on other IQ tests, is not the result of greater validity or of less biased measurement of children's intelligence by the K-ABC. The diminished black-white difference on the K-ABC seems to be largely the result of psychometric and statistical artifacts: lower g loadings of the mental processing scales and greater heterogeneity of the standardization sample, which causes mean group differences to be smaller when they are expressed in standard score units. The general factor measured by the K-ABC is essentially the same g as that of the Stanford-Binet and Wechsler scales. But the K-ABC yields a more diluted and less valid measure of g than do the other tests. The K-ABC factors of successive and simultaneous mental processing, independent of the g factor, constitute only a small fraction of the total variance in K-ABC scores, and the predictive validity of these small factors per se is probably nil. The present criticisms of the K-ABC suggest new means for improving the design of future tests for measuring general intelligence.