Evidence-based decision about test scoring rules in clinical anatomy multiple-choice examinations.

Abstract

In theory the formula scoring methods increase the reliability of multiple-choice tests in comparison with number-right scoring. This study aimed to evaluate the impact of the formula scoring method in clinical anatomy multiple-choice examinations, and to compare it with that from the number-right scoring method, hoping to achieve an evidence-based decision about test scoring rules. Two hundred and ninety-eight students completed an examination in clinical anatomy which included 40 multiple-choice questions with five response options each. Among these, 245 (82.2%) examinees were assessed according to the number-right scoring method (group A) while 53 (17.8%) were assessed according to the formula scoring method (group B). The prevalence of passing was significantly higher in group A than in group B, after correction of the pass and fail cutoffs for guessing (84.9% vs. 62.3%, P = 0.005), keeping a similar reliability in both groups (0.7 vs. 0.8, P = 0.094). Pearson Correlation coefficients between practical and theoretical examination scores were 0.66 [95%CI = (0.58-0.73)] and 0.72 [95%CI = (0.56-0.83)] for groups A and B, respectively. Based solely on the reliability and validity assessments, the test-maker could therefore use either scoring rules; however, if the test-maker also takes into account the students' ability to deduce answers with partial knowledge, then the number-right score rule is most appropriate.