In Online Courses, Catching Cheaters May be More Effective Than Warning Them

Posted June 10, 2022

Telling students you have a method to detect cheating doesn't prevent it, but identifying the perpetrators after they cheat does, finds new research by Assistant Professor Daniel Dench in Georgia Tech’s School of Economics. 

In the experiment, published in the Journal of Economic Behavior and Organization, Dench and co-author Theodore Joyce report that warning undergraduate students after their first offense reduced subsequent cheating attempts by at least 65%.

Their findings are important as online classes and degrees expand and courses remain virtual amid the Covid-19 pandemic. Students may be tempted to cheat in an environment with less oversight than a traditional classroom, but "before you go nuclear, you can deter them from trying to do it in the future," Dench said. "If there's knowledge out there that we actually do know who's cheating and when, then it's a lot less likely that people will do it."

To conduct the experiment, Dench and Joyce collected data from four finance, management, and accounting classes at a large public university outside of Georgia in 2019. Students completed their homework in Microsoft Excel, where the researchers embedded a unique identifier code in each student’s software that could flag work copied and pasted from another’s file. Unlike classes with papers or exams where plagiarism may be harder to detect, Dench said, this was a hard measure of cheating that was difficult for the students to deny. 

Still, the researchers found that reminding students when the course began that the software could detect cheating had little effect on cheating rates. Only flagging plagiarism after it happened by emailing the student directly and informing them they were on a watch list significantly reduced cheating — not only in that class, but in the following classes as well. 

For example, in the Finance and Management courses, cheating fell by 80% to 90% in the ensuing semesters. Another notable aspect, according to Dench? “The rates dropped in future semesters before we even started policing," he said. "That's an indication that we've changed the culture around cheating."

The experiment has some shortcomings, the researchers reported. The software couldn’t identify students who cheated from an outside source rather than copying from another student in the class. Dench and Joyce also could not determine how much cheating was curtailed by the email warning versus word of mouth from a friend or fellow student in the class. Finally, although instructors sent every student the first warning that the software could detect plagiarism, they were unable to confirm whether or not students had read it. 

Even so, "the takeaways appear clear," the researchers wrote. "Boilerplate messaging in syllabi regarding academic integrity appears to be largely ignored. Even email messages delivered a week before an assignment is due are ineffective. Not until students are caught and at risk for serious disciplinary action does cheating decline on subsequent work." 

The article, “Information and Credible Sanctions in Curbing Online Cheating Among Undergraduates: A Field Experiment,” appeared in the March 2022 issue of the Journal of Economic Behavior and Organization. It is available at https://doi.org/10.1016/j.jebo.2022.01.018.

Let’s connect! Follow the School of Economics on Facebook, Twitter, Instagram, and LinkedIn to keep up with our students, school news, and upcoming events.