A New Procedure in Large-Scale Assessment Academic Growth Detection as Mixture Group Academic Growths Vary

• In Event: Approaches to Scaling, Equating, and Linking

Abstract

The purpose of this study is to, as different levels of growth occur in various size subgroups, investigate the practical consequences of the newly proposed academic growth detection procedure and the traditional methods in terms of accuracy and consistency of proficiency classifications. Grade Response Model (GRM; Samejima, 1969) and 3PL- logistic model are used. Stocking-Lord linking approach is applied. Findings of this study is to indicate the extent to which differential growth is captured through the newly proposed multi-linking subgroup academic growth detection procedure and the traditional IRT linking approach, as different levels of growth occur in various size subgroups.

Authors

• Ou Zhang, Pearson

• M. David Miller, University of Florida