Abstract
The purpose of this research was to obtain statistical information on test components included in English-language dentistry tests administered over a two year period. Two different types of test items were analyzed in order to creat more appropriate test questions. Both types consisted of multiple-choice questions. Type A was composed of one English-language term as a stem and four Japanese-language equivalents as options. A distractor could be selected in the case of an exam sitter not knowing the correct answer. On the other hand, type B included one Japanese term as a stem and four English equivalents as options. All items in both tests were analyzed using classical test theory (CTT) and item response theory (IRT). Results demonstrated a negatively skewed distribution pattern and the curve was sharply peaked (leptokurtic distribution) in type A. In type B, however, the data spread out in a flat curve (platykurtic distribution). It was statistically proved that easy items in type A were more numerous and the number of considerably difficult items increased in type B.
