The following is a guest post by Ruben Lopez, school psychologist with the Moreno Valley Unified School District, CA and member of the IQs Corner Virtual Community of Scholars. Rueben reviewed the following article and has provided his comments below.
Watkins, M.W., Glutting, J. J. & Lei, P. (2007). Validity of the Full-Scale IQ when there is significant variability among WISC-III and WISC-IV factor scores. Applied Neuropsychology, 14, 13-20. (click here to view/download)
- For many years now—I think I read it over 25 years ago—school psychology textbooks have told psychologists that significant scatter (“variability”) among component subscales decreased the strength of the composite/full scale score to predict academic achievement . When applied to the Wechsler Intelligence Scale for Children, Fourth Edition (WISC-IV), this would mean that you should not use the Full Scale to predict academic achievement and use a subscale, for instance, the Perceptual Reasoning Index, instead. In their article referenced above, Watkins, Glutting, and Lei “directly addressed” the question of scatter and the WISC by analyzing the scores of a substantial number of students. Although the article contains a technical argument in opposition to the use of multiple regression to address the question—an important issue, even for practitioners, nonetheless--, I’ll just talk about the direct study; yet I note that they used a “moderated multiple regression.”
- The tests studied were the third (III) (1991) and the fourth (IV) (2003) editions of the WISC and the first (1992) and the second (2) (2002) editions of the Wechsler Individual Achievement Test (WIAT). Scores on these tests from three final samples consisting of 412, 412, and 136 students obtained during standardization were analyzed. Two of the samples were described as “linking samples”, samples of students whose scores on the WISC and WIAT were analyzed for standardization to determine the relationship between intelligence and academic achievement. One of the samples consisted of students who had actually been evaluated and found eligible for special education under specific learning disability, emotional disturbance, or mental retardation. The data appear to be substantial.
- The statistical analysis and interpretation of the data appear reasonable. All the samples were divided into two groups, a flat profile group and a variable profile group. The variable profile group consisted of students who had “at least one statistically significant factor score difference.” About the statistical analysis, the authors said, “moderated multiple regression was used to detect any bias in the predictive validity of FSIQ scores [Full Scale scores] between participants with and without significant factor score variability.” Regarding the analysis, the researchers reported, “For all samples, FSIQs were significant predictors of performance on reading and math tests, but neither factor profile group (flat versus variable) nor the interaction between the FSIQ and factor profile group significantly added to the prediction.” From this, Watkins, Glutting, and Lei concluded, “Our results challenge the practice of discounting the global IQ as a predictor of academic achievement when factor scores significantly vary.”
- At least for now, for me, the application is to not disregard the full scale score if there is too much scatter. So, I won’t disregard the full scale solely because of scatter. On the other hand, I note that recent Woodcock-Johnson-III studies provide strong evidence that some component scale scores—reflecting broad CHC factors—are significant predictors of academic achievement that definitely should be considered.
Powered by ScribeFire.