Abstract
Seven formulas for scoring two-option test items are considered. Five of the formulas take into account variations in the number of items on which each of the two options is keyed as correct. The results of an experiment involving the seven scoring formulas are then presented. The dependent variables in the experiment were the scores obtained by applying the seven formulas to the responses of 180 students to a 100-item general psychology examination. The four independent variables were: proportion of total number of items on which option "a" was keyed as correct (P); whether or not examinees were told the proportion of "a" options keyed as correct (T); whether or not examinees were instructed to guess when in doubt about the correct answer (G); level of examinee's knowledge of the subject-matter of the test (K). Statistical analysis revealed that variable P had a significant effect with some formulas but not with others, and that the effects of variables T and G were greater in the low-knowledge than in the high-knowledge groups. In those instances in which variable T was significant, examinees who had been told the number of correct options in each category and to guess when in doubt usually scored higher than examinees who had not been told these numbers but to guess when in doubt. For a given P × T combination, examinees who were told to guess when in doubt usually scored lower than those given no instructions concerning guessing. The effects of the experimental conditions varied somewhat with the particular scoring formula employed.
Get full access to this article
View all access options for this article.