Empirical Option Weights Improve the Validity of a Multiple-Choice Knowledge Test

Standard dichotomous scoring of multiple-choice test items grants no partial credit for partial knowledge. Empirical option weighting is an alternative, polychotomous scoring method that uses the point-biserial correlation between option choices and total score as a weight for each answer alternative. Extant studies demonstrate that the method increases reliability of multiple-choice tests in comparison to conventional scoring. Most previous studies employed a correlational validation approach, however, and provided mixed findings with regard to the validity of empirical option weighting. The present study is the first investigation using an experimental approach to determine the reliability and validity of empirical option weighting. To obtain an external validation criterion, we experimentally induced various degrees of knowledge in a domain of which participants had no knowledge. We found that in comparison to dichotomous scoring, empirical option weighting increased both reliability and validity of a multiple-choice knowledge test employing distractors that were appealing to test takers with different levels of knowledge. A potential application of the present results is the computation and publication of empirical option weights for existing multiple-choice knowledge tests that have previously been scored dichotomously.