Corrections to Twenty Years of Banding : The Necessity of Precision

Abstract

Statistically-based banding is often considered a viable method for minimizing adverse impact in test-based decisions. By utilizing the standard error of the difference (SED), scores are equated based on the assumption that there is substantial unreliability in any single observed score. However, as noted by Dudek (1979), the current procedure for calculating the standard error of measurement (SEM), which is used when calculating SED, is erroneous. Concurrently, the tendency of score to regress to the mean with multiple administrations of a test is a problem that is often overlooked in literature (Smith & Smith, 2005). This study compares the differences between banding based selection decisions when these corrections are applied. In general, results suggest that applying these corrections to methods of calculating SED can produce variations in the decisions. These potential legal and ethical implications of these discrepancies are discussed.