WISC-III Verbal Item Invariance across Samples of Deaf and Hearing Children of Similar Measured Ability

Publisher Website
Google Scholar
Abstract
Invariance of item difficulty estimates on the Wechsler Intelligence Scale for Children-Third Edition (WISC-III; Wechsler, 1991) Verbal scale was investigated across samples of deaf children and younger hearing children of similar ability. The samples consisted of 110 deaf children who use sign language and subsamples of younger hearing children from the WISC-III standardization sample. The samples were of similar ability as measured by total raw scores and Rasch mean logit ability estimates. Items were calibrated using a Rasch Model. Methods for detecting differential item functioning (DIF), including standardized differences and anchor item fit statistics, were used to examine invariance. Numerous items were found to function differently across groups, indicating that the item level performance of the deaf children was not characteristic of younger hearing children. Thus, Verbal subtest scores should not be used as an indication of delayed verbal ability, and the subtest scores of deaf and hearing children should not be compared.