A Comparison of Severe Discrepancy Formulae: Implications for Policy Consultation

Publisher Website
Google Scholar
Abstract
The number of youngsters identified as learning disabled (LD) has increased 160% over the past 15 years (U.S. Department of Education, 1991), amid charges of overidentification. As a result, methods for determining a severe discrepancy between ability and achievement have been subjected to scrutiny. This study compares two different approaches (a Z-score method and a regression procedure) to the determination of LD eligibility with 236 LD referrals. Those results were compared with those emanating from a more economically feasible method, namely, a policy stating that the lowest achieving referred students be adjudged LD. The results indicate the regression approach is the method of choice over the Z-score method if policy makers want to decrease the number of students identified as LD. However, these two methods are very comparable with respect to the types of decision-making errors (false positives and false negatives) when the percent of students is held constant. Interestingly, singling out the lowest achieving individuals as LD would have produced about the same decision-making results as did the two discrepancy approaches. Policy implications center around the financial, educational, and affective costs associated with LD identification.