A brief report on a comparison of six scoring methods for multiple true-false items

Six methods of scoring multiple true-false items were compared in terms of the reliabilities, difficulties, and discrimination. It was found that the differences in reliabilities and discrimination values were not statistically significant. After adjustments for scoring metrics were made, the multiple true-false, the correction-for-guessing, and the let-omit method were found to yield higher item means (i.e., easier), whereas the multiple response method yielded the lowest item mean. Results of this study suggest that, for norm-referenced score interpretations, there is insufficient evidence to support any one of the six methods as superior to others psychometrically. For criterion-referenced score interpretations, however, the effects of the scoring method on score interpretation and the determination of passing scores need to be taken into consideration.