An Evaluation of Procedures for Computing an Ability—Achievement Discrepancy Score

Abstract

The most common strategy for identifying children with learning disabilities has been to search for a discrepancy between ability and achievement. The need to determine a severe discrepancy has prompted the development of more than a dozen procedures for assessing the magnitude or statistical significance of ability—achievement discrepancy scores. Nine such procedures dealing only with magnitude were evaluated and found to possess one or more serious statistical deficiencies. The approach with the greatest potential involves first assessing the statistical significance (reliability) of a discrepancy score and then estimating the discriminative efficiency (validity) of that score. The utility of this approach, however, has not yet been tested by local and state educational agencies.