Academic Integrity in On-line Exams: Evidence From a Randomized Field Experiment

Two common approaches to minimize cheating are to (1) remind students of the honor code, and (2) randomize the question ordering to students. This research study suggests that neither approach is very affective at eliminating cheating.

Research Summary

This study conducted a field experiment to investigate cheating in the context of a sudden transition from an in-person final exam to a remote, unproctored, online final exam. The exam was part of a required undergraduate class in game theory at a major Spanish university. The course included four sections with nearly 500 students. The study design included 1) introducing an honor code reminder during the exam to the treatment group, and 2) changing the ordering of questions between groups. The authors hypothesized that reminding students of the university's honor code during the exam would reduce subsequent cheating behavior. They also hypothesized that students receiving questions in later stages of the exam compared to students who received the same questions earlier in the exam, would have higher average correctness and also answer the questions quicker.

The authors found no support for reminding students of the honor code as it no effect on cheating levels. Results did indicate that students receiving questions in later stages of the exam resulted in significantly higher average correctness (7.7% higher) and a significantly shorter average completion time (18.1% faster). Based on the authors' empirical techniques, they estimated that between 13.4 and 22.4% of students engaged in cheating.

‍Key Takeaways

1. Honor Code Reminders Don't Work

Reminding students before or during the exam of the honor code does not reducing cheating. Other research suggests that middle of exam warnings of cheating behavior are the best way to reduce cheating behavior.

2. Unproctored Exams Are Prone to Information Sharing and Collaborative  Cheating

Instructors and universities need to be aware of how online assessments are more vulnerable to these forms of cheating compared to traditional in-person assessments.

3. Randomizing The Question Order Does Not Eliminate Cheating

Randomizing the question order is a good practice but it is not sufficient by itself to eliminate cheating. At EXAMIND, we have put together a beginner's guide on randomization to help instructors learn how to apply the best randomization techniques.