Ask me anything: Assessing academic dishonesty

Abstract

We provide a method for assessing self-reported rates of cheating among students. The method is both i) privacy-preserving in the sense that one cannot use answers as evidence that any particular student cheated and ii) non-anonymous in the sense that one can record each student's answer for use in future correlative studies. Because accuracy relies on students' willful participation, we describe how to convince students that they take no risk by taking the survey. This method showed that 42% of 847 students willfully cheated in an Algorithms course. Surveying 181 CS Theory students showed no difference in cheating rates on written vs. coding assignments.