ABSTRACT
In addition to the A-and K-type questions, the X<SUB>2</SUB> type question has been introduced into, and the K'-type question has been eliminated from the Japanese Medical Licensure Examination since 1997. The expected correct answer ratio (ECAR) without any knowledge is 10% for the X<SUB>2</SUB>-type in contrast to 20% for the K2-type. However it still becomes 100% with an 80% knowledge. We designed a formula to study the relation between the knowledge quantity (KQ) and ECAR. The ECAR is 7%, 13%, 25%, 43%, 70%, and 100% for the X<SUB>1, 2</SUB>-type in which the number of answer (NA) is indicated as one or two, while it is 5%, 10%, 20%, 40%, 70%, and 100% for the X<SUB>2, 3</SUB>-type in which NA is indicated as two or three. The KQ difference between 80% and 100% can be reflected to ECAR in these types and considered to be the better question type.
ABSTRACT
X type questions have been used for the national medical licensing examination since 1997. At Tsukuba University, X type questions have been used since 1996. We compared X typeand K type questions on the basis of the percentage of correct answers and discrimination power. The average percentage of correct answers was 68.2% for K type questions and 53.1% for X type questions. However, the average discrimination power was +0.227 for K type questions and +0.257 for X type questions. These results indicate that X type questions are more difficult and are suitable for achievement tests. The estimated knowledge quantity was 2.04 for K type questions and 2.32 for X type questions. This suggests that the person writing the questions decreased the essential difficulty of X type questions.