Cognitive performance pattern underlying WJ-R test performance of Hispanic children.

Abstract

The purpose of this study was to determine whether the Woodcock-Johnson-Revised Cognitive test is biased when used with a Hispanic population of school-age children. Norming data, provided by R. Woodcock, Ph.D., for grades three, five, eight, and eleven were used for the study. Three hypotheses were explored. The first hypothesis called for a comparison by gender. The second hypothesis called for a comparison of non-Hispanics and Hispanics. The third hypothesis called for comparisons between all combinations of grade levels using only the non-Hispanic subgroup. Using the results of confirmatory factor analysis from LISREL VIII (1993), the chi-square difference test, and three goodness-of-fit indexes provided evidence of similarity in factor patterns between target groups. Hypothesis 1 stated that there were no differences between male and female factor patterns. The results of the confirmatory factor analysis supported the acceptance of hypothesis 1. Hypothesis 2 stated that there were no differences between non-Hispanic and Hispanic students. The results of the confirmatory factor analysis supported a qualified acceptance of hypothesis 2. The relationships between the latent variables are significantly different. Age, therefore, may have been a confounding variable in this study. Hypothesis 3 stated that there were no differences in patterns between grades. Only one of the six grade comparisons, 3-5, found model 1 to be the preferred model. All other comparisons found model 3 to be the preferred model. The residual or error terms were variable in matrix patterns, indicating that a factor other than age may be influencing the relationships. A fourth analysis was utilized and determined model 1 to be the preferred model. The results of the analysis indicate that differential patterns of processing, rather than age, may be the variable influencing the relationship of latent variables.