Sex differences on the WISC-R

Abstract

Sex differences on 13 subtests of the Wechsler Intelligence Scale for Children Revised (WISC-R), and in the factor structure of the 13 subtests, are examined in the white subsample (N = 944 males, 924 females) of the total national standardization sample of the WISC-R (Kaufman and Doppelt, 1976; Wechsler, 1974). This sample of children, whose ages range from 6 to 16.5 years, were chosen by a stratified, random-sampling procedure to be representative of the population of the United States, in accord with demographic features revealed in the 1970 census. All of the WISC-R subtests, listed in Table 1, are highly familiar, except Tapping Span, which was not included in the published version of the test. It is ostensibly a non-verbal test of visual imitative memory. A straight row of four l wooden blocks, spaced apart, is placed before the S. The examiner, holding a block between his thumb and index finger taps out a pattern on the row of four blocks, say, 1-4-2-3, if we imagine the blocks are consecutively numbered from left to right. The S's task is to imitate the same pattern of taps on the row of blocks, tapping out the pattern just as the examiner had done. Task difficulty is increased by tapping out longer and more complex series. The use of subtest scaled scores, with a mean of 10 and SD of 3 at every age in the total standardization sample, in effect obviates age differences in the test scores. All the analyses in this study are based on the age-standardized scaled scores. Table 1 shows the means and SDs of the scaled scores for males and females, and the difference between the sexes, in scaled score units and in SD units. Despite a multivariate statistical test which shows the overall differences between the sexes to be significant beyond the 0.001 level, the differences are generally quite small, with the marked exception of the Coding test, on which females exceed males by about half of a SD. This one quite extreme sex difference in the WISC-R battery has also been noted by Samuel 11983) in large samples both of whites and blacks (independent of the present standardization sample). The Coding test (matching geometric symbols with numbers) seems to involve perceptual speed and accuracy, dexterity in copying symbols, and rote memory. The only other subtest which shows an appreciable difference (0.37 SD), in favor of males, is Information. The sex differences can also be expressed in the form of the point-biserial correlation coefficient, rr~, with males coded 0, females 1. The rph, in the range of low values found here, is virtually a linear function of the mean difference. Full-Scale IQ (FSIQ) can be partialed out of each rph, revealing the relative size of the sex differences on the various subtests when the sexes are statistically equated on FSIQ. These results are shown in Fig. 1. Partialing out FSIQ has only a slight