Evaluating the Feasibility of Item Response Theory Pre-Equating for a Large-Scale Certification Program

• In Event: Factors Impacting Equating Results

Abstract

Preequating is possible if the items in the new form have been previously administered and calibrated. However, if the behavior of the items in the previously administered forms and the new form are not consistent due to factors such as position and context effects, then pre-equating may not produce accurate results. This study examined whether IRT pre-equating produced results that are comparable to post equating (i.e., IRT true score and chained equipercentile equating) using data from two large scale certification tests. Preliminary results suggest that the difference in the pre- and post-equated scores was larger than 0.5 (Dorans and Feigenbaum ,1994) for a large portion of the raw score scale, which may make a practical difference for test takers.

Authors

• Peng Lin, ETS

• Gautam Puhan, ETS

• Jiyun Zu, ETS