A comparison of equating methods based on item response theory for the purpose of constructing an item bank

Abstract

In this study we compared the concurrent calibration method and the separate calibration method for item parameter equating for the type of test design in which both the set of anchor items and the set of newly written items are used for scoring. It is also assumed that the number of anchor items per form is less than the newly written items and the latter will be added to the item bank after each of the test administration.

A simulation study showed that, when the item discrimination parameter values are increasing with time, the separate calibration method performed better than the concurrent method in terms of the recovery of the true parameter values.