The Use of Multiple Regression in Evaluating Alternative Methods of Scoring Multiple Choice Tests

Abstract

The purpose of this paper is to provide specific examples of how multiple regression analysis could be used to analyze item discrimination, item validity, and test validity when confidence testing is employed. Current practices tend to utilize apriori scoring formulas rather than maximize the predictiveness possible with the obtained data. We will also suggest that the application of these methods may require the development of multivariate techniques for assessing test reliability.