The scoring of multiple-choice questions.

Despite the increased use of multiple-choice questions in medical examinations over the last few years, there is little agreement on methods suitable for scoring them. Despite this, the theory required is adequately covered in books on educational statistics (Lord and Novik, 1968). The present paper discusses some of the scoring methods used, or proposed, by British medical examiners in the general context of the theory of scoring. The constraints imposed by various formats of question, and some of the constraints imposed by particular procedures of data collection will also be considered. In the United States, methods have been introduced by the National Board of Medical Examiners which have been regarded as the model for much subsequent research (Hubbard and Clemans, 1961). However, the emphasis on questions being presented in such a way that only one answer out of five should be selected by the candidate has coloured thought beyond its convenience, and other methods of presentation have been devised. These include the indeterminate question, in which any number of the offered alternatives may be correct, and the use of true/false questions. The only constant distinction between the one-from-five type of question and these other types of question is that in the latter the answer to any alternative is not constrained by the form, or answers, to the other alternatives. Since so much of the theory of scoring rests on the onefrom-five principle (Hubbard and Clemans, 1961), confusion has sometimes arisen when that theory is translated to the indeterminate