WISC-III SUBTEST COMPOSITE RELIABILITIES AND SPECIFICITIES - INTERPRETIVE AIDS

Using data from the standardization sample reported in the WISC-III ma nual (Wechsler, 1991), two studies were conducted. The first study det ermined WISC-III subtest composite reliabilities; in the second, subte st specificities were calculated. These data were compared to comparab le subtest composites and subtest specificities derived from the WISC- R. WISC-R and WISC-III reliability composites and specificity estimate s also were averaged across age levels and rank ordered. WISC-III reli abilities are slightly higher than those from the WISC-R. Typically, W ISC-III subtest composites are sufficiently reliable for screening, di agnostic, and hypothesis-generation purposes. Average WISC-III composi te reliabilities range from .80 to .94 and tend to improve with age ac ross the 6- to 16-year age span. Also, larger reliability coefficients were obtained for those composites comprised of larger numbers of con tributing subtests and for composites composed of verbal subtests. Sub test specificities from the WISC-III subtests are similar to those obt ained from the WISC-R. Seventy-eight percent of the subtest scores acr oss all ages can be considered to reflect measures of some unique cons truct, as compared to 73% from the WISC-R. In general, performance sub tests yielded higher specificity estimates than verbal subtests.