A comparison of ability-achievement discrepancy models for identifying learning disabilities

Three ability-achievement discrepancy methods for identifying learning disabilities were compared. The first method uses a simple standard score difference calculation; the second method a regression equation; and the third, a variation of the second, takes the standard error of estimate into account. These three methods were examined using varying significance criteria, producing five individual models: the simple difference models (IA and lB) used 16 and 23-point discrepancy criteria, respectively; the basic regression models (2A and 2B) used 16 and 23 points; and the regression variation model (3) used a 95% confidence level. The five models were applied to 145 students' IQ (Wechsler Intelligence Scale for Children-Third Edition) and achievement (Woodcock-Johnson Achievement Tests-Revised) scores; all students had been referred for psychoeducational testing.