Confirmatory factor analysis of the Kuwaiti adaptation of the WISC-R

Date of Award




Degree Name

Doctor of Philosophy (Ph.D.)


Educational Research

First Committee Member

Maria M. Llabre - Committee Chair


The purpose of this study was to examine the psychometric properties of the Kuwait Wechsler Intelligent Scale for Children (WISC-K) and compare its factorial composition to that reported by Kaufman (1975) on the original English version. The WISC-K 1986 standardization sample, which includes 200 children (100 males and 100 females) at each of the eleven age groups provided the data source for this study (N = 2200). Both descriptive analyses and confirmatory factor analyses were performed. The confirmatory factor analyses tested the three-factor model proposed by Kaufman (1975) and the two-factor model proposed by Wechsler (1974) with various levels of equality constraints. Both individual and simultaneous confirmatory factor analyses were performed using LISREL VI (Joreskog & Sorbom, 1986). The results of the descriptive analyses indicated that the reliability coefficients for the subtests of the Verbal Scale, which were not directly translated, were satisfactory; whereas the reliability coefficients for the Performance Scale subtests, which are exactly the same as in the original WISC-R, were low at some age levels. The results of the confirmatory factor analyses indicated that neither the three-factor model nor the two-factor model explained the variation in the WISC-K for all age groups simultaneously. However, the three and two-factor models were able to explain the variation in the WISC-K subtests at ages 6$1\over2$, 10$1\over2$, 11$1\over2$, 14$1\over2$, and 15$1\over2$. Of the two models, the three-factor model with no constraints appeared most appropriate for these five age groups. The findings of this study supported only in part the factorial validity for the WISC-K. The results also pointed out certain limitations in the WISC-K in need of modification, particularly in the Picture Arrangement subtest. Finally, the findings support previous recommendations for the construction of tests patterned after a model test of another culture provided all the items are constructed to fit the new culture, rather than based on a direct translation of the model test items.


Education, Tests and Measurements

Link to Full Text


Link to Full Text