A straightforward graphical/statistical approach to help substantiate cheating on multiple-choice examinations.

Abstract

Academic dishonesty is prevalent in universities in the form of cheating on examinations, with the problem being much greater in classes that have a large number of students that require close seating arrangements for in-class exams. The scenario described below was experienced during an in-class exam that included the possibility of an Honor Code violation between two students that was observed independently by three different faculty proctors. Herein we detail an objective, statistical approach taken to maintain exam and academic integrity that is compelling and transparent to students and the University Honor Council. Using the established error-similarity analysis for multiple-choice exams, it was determined that the number of identical incorrect answers found on the exams of the two individuals in question was sufficiently greater than the number expected by chance (probability of P < 0.00001). The number of total identical incorrect answers found on the remaining exams (across 65 students, n = 89 comparisons) was plotted as function of the number of total incorrect answers found on these exams (incorrect answers ranged from 1 to 22) and clearly supported that there was an Honor Code violation between the two students in question. The techniques used herein established, beyond a reasonable doubt, that a form of cheating had occurred between these students. However, caution must be taken as further investigation is requisite to establish whether the Honor Code violation was unidirectional (one student copying off the other) or bidirectional (collusion between the two students) in nature.NEW & NOTEWORTHY Academic dishonesty is prevalent in universities, especially on examinations with a large number of students in close seating arrangements. Cheating on a multiple-choice exam was suspected by observations from proctors of the examination. Application of error-similarity analysis associated with identical incorrect answers demonstrated that the probability of cheating was confirmed (P < 0.00001) between two examinees. Further comparisons with the remaining exams provided graphic evidence that a violation of the University's Honor Code had occurred.