Refining of Kanji Placement Test Using the Rasch Model

Abstract

This study examines and refines a kanji placement test using the Rasch model. The participants were international students who were studying Japanese at a college in Japan. One hundred eighty seven students took a non-revised multiple-choice kanji placement test and 176 students took a revised placement test. The results of the analysis revealed some items and distracters which did not function well in the original version, therefore the following revisions were made: 1) revision or elimination of malfunctioned items, 2) elimination of a distracter from each question, and 3) elimination of an item which had the same item difficulty as another. After revisions were carried out twice on the test, the final version was found to be more efficient than the original one, since it had less items and distracters while keeping the same reliability level as before. Items that can measure the candidates' ability properly also increased in the final version.