Detecting Student Cooperation on LMS Exams

Abstract

Instructors develop courses to promote student learning and a multitude of strategies to measure if learning occurs through exams, projects, and other assignments. However, the measurements may also be an indication of how well the students cheat. In the classroom, proctors and multiple versions of an exam help to deter cheating, and comparing right and wrong answers among students helps to detect it. Video proctors for Online exams prove effective at preventing cheating, but they pose privacy and other issues. Comparing right and wrong answers between students remains a useful tool to detect online cheating. We propose analyzing learning management system (LMS) data to support and possibly replace other detection methods. LMS data have the potential to reveal cooperating students through a statistical analysis of student answer times. The proposed methodology is accessible to instructors without involving administrators in data collection.