Improved estimation of academic cheating behavior using the randomized response technique

Abstract

Academic cheating behavior by university students was surveyed using the randomized response technique (RRT) and by conventional anonymous questionnaire methods. RRT is a survey method that permits sensitive information to be collected but that precludes associating the respondent with a particular response to a survey item. The estimated proportions of students who have engaged in cheating behaviors were, in general, larger using RRT. Moreover, this result is consistent with earlier findings for other sensitive behaviors. That underreporting is a serious problem with anonymous questionnaires is supported by the fact that the anonymous questionnaire estimates ranged from 39% to 83% below the RRT estimates. Furthermore, using a covariate modification of RRT, there was a distinct inverse relation between students' estimated grade-point average and the tendency to engage in cheating behavior. While these results have direct implications for estimating cheating behavior in higher education, more broadly, they raise serious concerns about the use of anonymous questionnaires when survey topics are sensitive.