​How "Freakonomics" author got college students to admit cheating

Cheating at American universities may be reaching epidemic levels, with reports of scandals at colleges ranging from elite schools such as Harvard to public universities like Florida State.

There are two reasons why some students turn to cheating, according to a new paper from "Freakonomics" author and University of Chicago economics professor Steven Levitt and fellow researcher Ming-Jen Lin at National Taiwan University's economics department. The students have "strong incentives" to cheat -- such as getting good grades and securing internships or jobs as a result -- while the likelihood of getting caught is low, they noted.

The pair, however, has come up with a clever way to catch the cheaters, and put their plan into practice. The result? Several students admitted to cheating during an exam at one university.

The question of how to catch cheaters was brought to Levitt and Lin's attention by a professor at a top school who suspected that some of his students were cheating (Neither the professor nor the university were identified.)

Even before the economists studied the professor's classroom, they noted that cheating is endemic in colleges, with a 2005 study finding that 11 percent of 8,000 college students surveyed in the U.S. and Canada admitted to copying from another student's test.

In the case of the professor who was helped by Levitt and Lin, his suspicions were raised after a student alerted the single teaching assistant who had proctored a midterm exam that another student may have been cheating. The professor emailed his 242 students and told them that "cheating is morally wrong," and encouraged the cheaters to admit their guilt. Not surprisingly, none came forward.

After another student said they had witnessed cheating, the professor asked for student confessions, and threatened to contact Levitt and Lin to help him catch the wrongdoers. Still, no one came forward, the authors noted.

That's when the economists jumped in and began their analysis of the situation. By examining the midterm results and comparing how students performed in the final exam, when they were randomly assigned seats with the idea of breaking up pairs of students who might be planning to cheat together, they found evidence that at least 10 percent of the 242 students had cheated on the midterm.

"When seating positions were randomly assigned and monitoring was increased for the final exam, almost all evidence of cheating disappears," they wrote.

Levitt and Lin also employed a clever trick at the final exam. They allowed students to seat themselves voluntarily -- and noted who the students chose to sit next to -- but then randomly moved them into other seats. They also increased the number of teaching assistants who monitored the exam room.

"Thus, we are able to observe the patterns in correlations among students who wanted to sit together, but then were not allowed to," they noted. They added, "When seating positions were randomly assigned and monitoring was increased for the final exam, almost all evidence of cheating disappears."

So what's the best way to detect if students have cheated? Don't look at the students who provided the same correct answers, they noted, since that can simply mean they studied together. The trick is to focus on when students share incorrect answers, with the authors finding that the students who sat next to each other have about twice as many shared incorrect answers as would be expected by chance. Once the students were randomly seated, the incidence of correlated incorrect answers disappeared.

"Perhaps the best supporting evidence for our claims of cheating (and also, perhaps a powerful explanation as to why so little effort is invested in detecting cheaters), comes from what happened after we carried out our analysis," Levitt and Lin write.

The professor forwarded the names of the 12 students who were "most suspicious" to the dean, who started an investigation. Four of the 12 students confessed, they noted. Unfortunately, pressure from parents on the dean's office prompted the cancellation of the investigation, the added.

"While this precluded any further admissions of guilt, the professor withheld grades of the presumptive guilty pairs until the first day of the next semester which resulted in scholarship disqualification," the paper said. "Notwithstanding this punitive action, none of the twelve accused students complained or sought redress."