“Assessment Of Factor Models Underlying The WISC-III In White, Black, And Hispanic Subgroups Of The Standardization Sample”, 1997-06-23 (; similar):
Structural and measurement invariance of the WISC-III was examined across White (n = 1542), Black (n = 338), and Hispanic (n = 242) subgroups of the standardization sample. Data analyzed were separate subtest scaled score and raw score variance-covariance matrices for each subgroup. Both sets of scores were analyzed as scaled scores may mask unique response patterns within each subgroup.
Within groups and simultaneous maximum likelihood confirmatory factor analyses were performed to fit data to four competing correlated factor models: (a) model consisting of all 13 WISC-III subtests; (b) a verbal-performance factor model; (c) a Verbal Comprehension, Perceptual Organization, and Processing Speed model; and (d) a Verbal Comprehension, Perceptual Organization, Freedom From Distractibility, and Processing Speed model. Freedom It was hypothesized that verbal and performance factors would fit the data best for each group. It was further hypothesized that factor loading patterns would differ across groups and that analyses of raw score data would reveal idiosyncratic response patterns across groups.
The chi-square/df ratio, Tucker-Lewis Index, and Adjusted Goodness of Fit Index indicated that the 4-factor model fit both sets of data best within each racial-ethnic group. These indices and an incremental fit index demonstrated that the 4-factor model exhibited structural and measurement invariance across groups. The same 4 factors explained the variance-covariance matrices of each group and WISC-III subtests are measured with the same reliability. Differences in rank order of subtest factor loadings were observed when scaled score data were analyzed which was not expected.
Test development procedures safeguarding against bias and acculturation factors may account for the structural and measurement invariance of the 4-factor model. Item content of the WISC-III was meant to appeal to a multicultural society. Geographic proximity and intermixing between racial-ethnic groups may also account for the results. The 4-factor model may be used clinically in assessing children from White, Black, and Hispanic groups. Since the groups studied were heterogeneous with respect to cultural practices and socioeconomic status, practitioners should not disregard racial-ethnic group membership when assessing children from diverse backgrounds.