by

A critical review of systems of scoring multiple choice tests is presented and the superiority of a system based upon elimination method over one based upon the best answer mode is hypothesized. This is discussed in terms of the capacity of the mode to reveal the relationships among decoy options and the effects of partial information, misinformation and guessing. Tests were administered to two groups of subjects and scored according to each of the three treatment modes, classical, weighted choice, and elimination. In addition, subjects were asked to indicate their confidence in the correctness of each answer. Thus, treatment, confidence, and knowledge scores were computed for each subject and a Gulliksen-Wilks regression test was performed on the data to compare the validity and reliability of the three scoring modes. The results generally support the hypothesized superiority of the elimination scores. Elimination produced higher validities and reliabilities and less guessing than either of the other two treatments, Although the design did not permit a definitive comparison of elimination and confidence scores, there was some evidence that elimination scores were at least as valid as confidence scores. (Author/PR