Applying the Scoring-Rule Method to Multiple Choice Tests

Abstract

The author proposes the application of the scoring-rule method for measuring the subjective probability of multiple choice tests. In doing this, he generalized the scoring-rule method in order to derive general score functions under both free and constrained conditions. As a means of determining score function characteristics, he also introduces a measure for the "sensitivity of the score function." Further, the author explores the advantages and disadvantages of each score function when utilizing educational multiple choice tests and reports on the consequences of using the resulting tests. He discovered that there is a high correlation between expectation and actual student scores. Finally, the author extends the use of the scoring-rule method to multiple branching types of multiple choice tests.