An Approach to Test Equating under the Latent D-scoring Method

Abstract

This study offers an approach to test equating under the latent D-scoring method (DSM-L) using the nonequivalent groups with anchor tests (NEAT) design. The accuracy of the test equating was examined via a simulation study under a 3 × 3 design by two conditions: group ability at three levels and test difficulty at three levels. The results for bias (systematic error) and root mean square error (RMSE; total error) indicated very high accuracy in the recovery of true equated D-scores across all simulation conditions, providing strong support of the proposed approach to test equating under the DSM-L.