Prediction of Item Difficulty on a Reading Comprehension Test

Abstract

This study investigated the relative contributions of item variables under the three categories of question, text, and content features to item difficulty by performing regression analysis, among 2666 high school seniors who took a multiple-choice reading comprehension test to enter a university in Korea. The results of regression analyses showed that two variables, vocabulary difficulty as determined by the number of difficult vocabulary items in a passage and syntactic complexity as measured by the number of vocabulary in a sentence, in tandem, significantly accounted for variance in item difficulty for the less skilled readers (R2=.335, F=8.058, p<.01). For the skilled readers, only one variable, vocabulary difficulty, accounted for a significant amount of variance in item difficulty (R2=.183, F=7.380, p<.01). The findings of this study were discussed with the focus on the crucial roles of lexical and grammatical knowledge in reading comprehension in general and a reading comprehension test performance in particular, followed by future study areas that warrant to shed further light on the contributions of various item variables to item difficulty.