국가수준 학업성취도 평가의 최적의 선택지 개수 분석

Abstract

This study investigated the effects of reducing the number of options in multiple-choice items on the psychometric characteristics of the English section of the NAEA (National Assessment of Educational Achievement). The purpose of the NAEA is to measure educational progress and achievement and it is administered to all sixth, ninth, and eleventh grade Korean students annually. The English section for the ninth graders is the focus of this study; it is made up of thirty-four five-option multiple-choice items and six constructed-response items. A five-option, multiple-choice English test was converted to a four- and a three-option version by eliminating the least and the second least frequently chosen options. Item difficulty, item discrimination, guessing, reliability, and information function were computed as a function of the number of options. This study employed the 3-parameter IRT model. Results showed that there were no significant differences among the three means of item difficulty as well as item discrimination indices; however, the inspection of the test information function indicated that the five-option version generally yielded the highest amount of test information function over the ability scale. Results indicate that streamlined three- or four-option versions are not likely to function as well as the five-option version at the test level.