Profiling Cheaters in College

In a recent issue of the Journal of Experimental Psychology: Applied, Williams, Nathanson and Paulhus (2010)1 examined the personality characteristics of scholastic cheaters.

This is a critical issue for me because its connection to applicant cheating behavior.  How can we predict which job applicants will lie on their applications, distort their responses, or outright cheat on knowledge and ability tests?  And perhaps even more critically, who will act dishonestly on the job?  While the subjects of these studies are college students, it does shed some light on this parallel issue.

In this paper, Williams et al. report the results of three studies.  In Study 1, they conducted a literature review to identify personality traits likely to predict dishonest behavior.  They settled on 1) three Big Five traits (Agreeableness, Conscientiousness, and Openness to Experience) and 2) the Dark Triad: a combination of narcissism (entitled and grandiose), Machiavellianism (willing to manipulate others for personal goals), and psychopathy (erratic and antisocial).  They then correlated these with self-reported cheating, finding significant and reasonably large relationships with the Dark Triad, Agreeableness and Conscientiousness.

In Study 2, they repeated this process, but instead of using self-reported cheating, used Turn-It-In scores.  If you’re unfamiliar with Turn-It-In, the basic principle is that student essay submissions are compared with a database of previously-turned-in papers and published sources.  This produces a percentage score that indicates to what extent the content is plagiarized from other sources.  As of the data collection for this study, Turn-It-In scores also included legitimate citations.  This was corrected by recoding the Turn-It-In scores into a “plagiarism index” based on scores derived by two research assistants.  The specific mechanism by which this index was created is not clear; I wonder if the researchers did not find significant results with the original Turn-It-In scores and decided they needed to recode.  In any case, once again, correlations are reported.  The Dark Triad all predict plagiarism scores, as does Agreeableness (though not Conscientiousness) and verbal ability.

This makes sense – students who have poorer ability would have a harder time writing essays and would be more likely to turn to plagiarism, regardless of their personality.  Though I wonder if there are any interactions.

In Study 3, the authors examined the mediators of psychopathy, i.e. determining what is the motivation for psychopathic individuals to cheat?  The procedure for this study was a little odd.  The authors took a previously existing self-reported reasons-for-cheating scale, conducted a factor analysis to condense it into three categories (unrestrained achievement, fear of punishment, and moral inhibition), and then examined a mediational model using bootstrapping.  They determined from this that the three motivation-to-cheat categories partially mediated the relationship between psychopathy and self-reported cheating.

Again, Study 3 was a little odd to me.  They clearly had the potential to examine the relationship between all of the Dark Triad and Turn-It-In scores in addition to the single trait they looked at, and the less interesting outcome.  It seems a little peculiar to choose this particular set, and no explanation is given for choosing this one potentially mediated relationship instead of one of the other 5.

This makes me wonder if there is something about psychopathy in particular that makes it unusual – perhaps a low base rate?  Means and SDs are not given for study variables, so is a possibility.  The specific questions used for motivation-to-cheat also seem like they would have low base rates, for example: “I cheat because I’m not honest/moral.”  I can’t imagine many individuals answer positively to this.  So perhaps we are just capturing, in a sense, the worst of the worst?

In any case, Study 1 and Study 2 are interesting in and of themselves, though I wish they had included both self-reported cheating and observed cheating in the same study.  We can say pretty confidently that the Dark Triad and some of the Big Five predict cheating.

The next logical question, though: what next?  Even if we can profile cheaters (either college students or applicants), what can we do with this information?

The authors suggest something that strikes me as a bit odd, especially shortly after a warning about causal inference:

More generally, educators should benefit from awareness that the most probable cheaters are those low in scholastic preparedness
and high in psychopathy. Attention to the first group requires redoubling efforts to prevent students from falling behind.

There is nothing here to suggest that this approach would work.  Verbal ability was captured using a vocabulary test, which has little to do with course success.  If anything, this suggests that college students with lower intelligence or are unprepared for college are more susceptible to the lure of easy high grades, which isn’t all that surprising. And arguably, the highly intelligent cheaters are the ones we should be watching most closely.

1. Williams, K., Nathanson, C., & Paulhus, D. (2010). Identifying and profiling scholastic cheaters: Their personality, cognitive ability, and motivation. Journal of Experimental Psychology: Applied, 16 (3), 293-307 DOI: 10.1037/a0020773 [↩]