토익읽기 선다형 문제의 구인타당성과 문항난이도 예측 연구

The construct vaildity of multiple choice items and the prediction of item difficulty for TOEIC reading comprehension.

The purpose of this study was to predict the difficulty of TOEIC reading comprehension items, and it was also related to the construct validity of multiple choice items. Independent variables included three variables. The first variable was a text variable. The second was a question variable. Finally. the third was an overlapping variable, using both the text and the question variable (text/item). The three independent variables had their own sub-predictor variables. There were three general hypotheses: 1) The independent variables of the text, item, and text/item are not correlated with the item difficulty. 2) There is no difference in the prediction of the item difficulty for the three independent variables. and 3) Sub-predictor variables related to the item difficulty are not different in each ability level. The result indicated that the text variables were more related to the item difficulty than the item variables in the lowest ability group, but vice versa in the highest group. This research recommends that test makers should arrange the features of the text variable in the lower groups, but the features of the question variables in the higher groups for the item difficulty, considering the ability of the testee carefully in creating a question.