Academic Dishonesty and Online Examinations

ABSTRACT The widespread and growing use of online courses coupled with concerns about academic dishonesty in the Internet era warrants an evaluation of how online courses are assessed. We ran an experiment in two phases to evaluate the effect of different versions of online multiple-choice examinations and examination proctoring on examination performance. We find that examination results are significantly higher in unproctored than in proctored examinations and lower in versions of the unproctored online examination modified to mitigate cheating. We also find a larger performance drop off (compared with midterm examination performance) for unproctored than proctored students, and within the unproctored group, the final examination performance drop off varied by version, as expected. Lastly, student learning, as reflected in final examination performance, differs between the proctored and unproctored midterms. Lower student learning, grade inflation, and unfair grade distribution are concerns for online courses with unproctored examinations.