Are Certain People More Prone to Cheating?
Large-scale research summaries show which characteristics affect cheating.
Posted Jan 21, 2019
Given the presence of cheating in virtually all domains of life—with potentially damaging consequences for individuals and societies—researchers have tried to identify situational factors and personality characteristics that make it more or less likely that a person cheats. In order to identify such factors and characteristics, particularly in the last decade researchers have adopted experimental approaches that, surprisingly at first glance, do not record (or otherwise care about) which study participants actually cheat.
Recently, existing findings following such approaches—let’s call them cheating experiments herein—have been summarized, providing some robust answers on which personality characteristics are related to cheating. Before we turn to some of these summaries (other recent summaries can be found here and here), this is how such cheating experiments work:
The basic idea of such cheating experiments is that researchers compare participants’ reports about a favorable outcome with the statistically expected occurrence of the outcome. Let us illustrate this with one of the most straightforward examples, the die-roll task:
In the die-roll task, people are asked to roll a die in private and to report the outcome. If they report a specific outcome—say, that they rolled a “6”—they receive a profit. Obviously, because people roll the die in private, they can simply misreport the outcome, e.g., reporting a “6” although one has rolled a “4”. On an individual level, the researchers do not know who is honest and who misreports the outcome. But researchers know that rolling a “6” with a fair die happens in 1/6 of the cases (indeed, we once “asked” student assistants to roll each die used in such experiments 1,998 times to check this…). So, if one now asks several people to do the die-roll task, one can link people’s personality characteristics to whether they have reported a “6” or not. Although one does not know which of the reported “6” are due to cheating, a relation (correlation) between a personality characteristic and having reported a “6” or not allows to infer that the personality characteristic is related to cheating. In fact, the only other explanation would be that the actual outcome of a die role with a fair die depends on person characteristics (say, men rolling a "6" more often than women), which seems implausible.
Summarizing findings from this and/or similar tasks like the coin-flip task (note that the tasks differ in some aspects, but we will neglect the differences herein), Gerlach, Teodorescu, and Hertwig conducted a meta-analysis comprising results from 565 experiments with 44,050 participants overall. They found that 42% of the men and 38% of the women cheated, not only confirming a general occurrence of cheating, but also suggesting a (small) gender difference—men cheat a bit more. Explanations for this might be that men are more prone to risk-taking (i.e., fear less possible sanctions) or have a different perception about risks in general, which shines through even in cheating experiments without individual sanctions. Further, research has suggested that women cheat less because they, in general, might be more concerned about the harm that is inflicted on others. Next to this gender difference, Gerlach and colleagues found a (small) age effect, suggesting that every year of life lowers the occurrence of cheating a bit (when considering adults). Again, a potential explanation might be that the propensity for risk-taking differs, namely, in a way that younger people are more prone to risk-taking.
Gerlach and colleagues also compared the behavior of students studying something else than economics with non-students, as well as of students studying something else than economics with economics students, but found, overall, hardly any support that this affected cheating.
In another recent project using data from several single cheating experiments, Heck, Thielmann, Moshagen, and Hilbig reanalyzed data from 16 studies (with 5,002 participants overall) in which basic personality traits were linked to cheating experiments such as the described die-roll task. They considered studies assessing the Big Five traits (via one inventory) and/or the six Hexaco traits (main differences between these two personality models are described here). From the investigated traits, only Honesty-Humility from the Hexaco Model of Personality showed a medium to large effect on cheating. That is, people who describe themselves in personality questionnaires as more fair, modest, sincere, and less greedy actually cheat less in cheating experiments. Notably, this effect was robust even when there was a time gap of six months between administration of the personality questionnaire and the cheating experiment. The reanalysis also revealed anecdotal evidence for links between some of the other (either Big Five or Hexaco) traits and cheating, but these effects disappeared (or were only very weak) when Honesty-Humility was considered as well.
In summary, cheating occurs and can be studied even when researchers do not assess who actually cheats. There are small effects concerning gender and age, with men and younger people cheating a bit more. Further, individual differences in cheating are represented in a basic trait called Honesty-Humility (already), but not much in other basic personality traits. Given the comprehensive research summaries, these findings seem very robust, especially in combination, because research has also indicated that women and older people have higher levels in Honesty-Humility. Next to the considered factors, however, research has yet to robustly test whether other person characteristics are related to cheating to not.