Factor and subtest discrepancies on the differential ability scales: examining prevalence and validity in predicting...

Past literature has largely ignored the population frequency of multivariate factor and subtest score discrepancies. Another limitation has been that statistical models imperfectly model the clinical assessment process, whereby significant discrepancies between both factors and subtests are included in predictions about an individual's academic achievement. The present study examined these issues using a nationally representative sample (N = 1,185) completing the Differential Ability Scales. Results indicate that approximately 80% of children in a nonreferred sample show at least one statistically significant ability discrepancy. In addition, the global estimate of cognitive ability was the most parsimonious predictor of academic achievement, whereas information about ability discrepancies did not significantly improve prediction. Findings suggest that when predicting academic achievement, there is little value in interpreting cognitive scores beyond the global ability estimate.