Abstract. This study used both exploratory (EFA) and confirmatory factor analyses (CFA) to examine the factor structure of the WISC-III among White and Black students from the WISC-III standardization sample and a sample of 348 Black students referred for psychological evaluation. Results of the EFA provided evidence of a large first principal factor as well as the expected Verbal and Performance components across all three groups. Empirical support for the Freedom from Distractibility dimension was provided only from the confirmatory factor analyses. Although the four factor confirmatory model exhibited the best overall statistical fit, inspection of specific factor loadings revealed anomalies with the third and fourth factors, especially for the Referred Black sample. Implications for school psychologists are presented and recommendations for future research are provided.
The issue of nondiscriminatory assessment is a concept of considerable legal and ethical importance to all psychologists (CNPAAEMI, 2000). The selection of test instruments that are free of test bias is paramount for school psychologists who work with ethnically diverse populations. Construct validity is perhaps the most fundamental of all types of measurement validity (Messick, 1989), and often derives from correlational studies and factor analytic research. Empirical support for comparable factor structures across ethnic groups suggests that similar constructs or latent traits are being assessed and provides preliminary support for the use of a test with those populations. Rogers (1998) emphasized the importance of establishing cross-cultural factorial similarity:
The construct validity of a test may also be a concern when the test evidences factorial invariance across racial/ethnic minority groups. Test developers and publishers need to report empirical evidence in the test manual that attests to the stability of the factor structure of a test for various majority and minority groups. When such information is not reported either in the manual or in the extant literature, the instrument is considered to be of limited practical utility because it is impossible to independently judge the factorial stability of the measure. (p. 361)
WISC-R Construct Evidence
Historically, the Wechsler Intelligence Scale for Children--Revised (WISC-R; 1974) has been the most commonly used test of intelligence for children referred for psycho-educational evaluation (Hutton, Dubes, & Muir 1992; Lutey & Copeland, 1982). Despite their popularity, Wechsler tests have been criticized for their lack of a strong theoretical foundation (Macmann & Barnett, 1992, 1994; Witt & Gresham, 1985). Although Wechsler (1939) viewed intelligence as a global capacity, he also believed that two dimensions underlie intelligence; all subsequent tests in the Wechsler family have been constructed to assess Verbal and Performance IQs. Despite a lack of change in the underlying theory this evolution continued such that factor analytic studies of the WISC-R found three factors to be present and the recently introduced Wechsler Intelligence Scale for Children--Third Edition (WISC-III; Wechsler, 1991) measures four IQ factor scores. Further, in using a Gf-Gc framework, McGrew (1997) suggests that the Wechsler t ests may actually be measuring as many as 13 broad and narrow cognitive abilities. Because it remains unclear how many "types" of intelligence are being measured by the Wechsler scales there currently exists considerable disagreement among school psychologists regarding their level of diagnostic interpretability (Kush, 1996).
Construct validity for Verbal and Performance IQs as well as for the Verbal Comprehension and Perceptual Organization dimensions of the WISC-R has been well established and shown to be invariant across age (Conger, Conger, Farrell, & Ward, 1979), gender (Reynolds & Gutkin, 1980), and ethnicity (Gutkin & Reynolds, 1981; Reschly, 1978; Taylor, Ziegler, & Partenio, 1984). …