共通項目の部分得点モデル化によるテストの等化

概要

論文の詳細を見る
The process of awarding points for a single anchor item across multiple tests on the basis of a common scale is referred to as equating. In general, the item response theory has been used for equating. This theory assumes that an item has a binary response whose score is 1 for a correct answer and 0 for an incorrect answer. This is a well-known example of a typical 2-parameter logistic model. In the present study, a new polytomous anchor item was created by calculating the sum of the scores for multiple anchor items. This was then equalized by applying Fujimori s partial test score model and was compared with the score that was equalized according to the 2-parameter logistic model. No difference was observed between the results of the two models. As compared with the conventional methods, equalization by the modeling of partial scores for anchor items requires the deduction of fewer model parameters in the anchor. Despite this disparity, equivalent results were obtained. Therefore, it is concluded that modeling partial scores is a more efficient method.
文教大学の論文
2005-12-20