Is cheating the kind of thing we do?

Or is it the kind of thing those other people do?

In the car yesterday, I caught a story on Marketplace that was looking for insight into why people on Wall Street cheat. In the piece, host Kai Ryssdal interviewed Duke University behavioral economist Dan Ariely about research conducted (with college students, of course) on cheating.

The set-up was that the students were asked to solve a set of math problems, and that they'd be paid (50 cents) for each one they got correct. While it sounds like some of the students were only paid after they completed the set of problems, other groups of students were given an envelope of money up front and told after completing the problems (and checking the answers) to return the money in excess of what they earned for their correct answers.

When it was a matter of returning the money they hadn't earned, some of the students cheated. (Or at least, they didn't return as much money as they should have. Maybe the weak math skills that resulted in mistakes with the math problems also resulted in a miscalculation of how much they had earned? Although if that were the case, you'd figure there would be students returning too much money also.)

But here's the part that got my attention:

Ariely: The most interesting thing is that we had a student that we hired, an acting student at Carnegie Mellon, who stood up after a few seconds and said "I solved everything," -- clearly cheating because nobody can solve all these questions in 30 seconds -- "What should I do?" And the experimenter said, "If you finished, go home." Now, what will this create? Will people cheat more or cheat less? Well, it turns out it depended on what kind of sweatshirt he was wearing.

Ryssdal: I'm sorry, what kind of sweatshirt?

Ariely: Yes, what kind of sweatshirt.

Ryssdal: OK.

Ariely: Here's the story. We ran this at Carnegie Mellon, in Pittsburgh. And in Pittsburgh there are two universities: Carnegie Mellon, University of Pittsburgh. All the students who participated were Carnegie Mellon students. If the cheating student, the acting student, was wearing a Carnegie Mellon sweatshirt, he basically got people to cheat more. But if he was wearing a University of Pittsburgh sweatshirt, he got people to cheat less. What is basically happening here is that when he stood there with a Carnegie Mellon sweatshirt, he gave a social justification for a new social norm to emerge about cheating. But when he was wearing a University of Pittsburgh sweatshirt, all of the sudden people said, "This is cheating. This is what the other people in the bad school are doing. This is not what we're doing." And therefore cheated actually less.

Please set aside questions of how many universities Pittsburgh actually has, or of whether the University of Pittsburgh is a bad school except in the consciousness of a Carnegie Mellon undergraduate.

Rather, consider what this result suggests about how people evaluate whether cheating is permissible: If someone from my group is doing it, it's OK for me to do it. On the other hand, if someone from another group -- one that my group wants to see itself as better than -- does it, then I (and others in my group) will scrupulously avoid doing it because we're better than that.

I'm not entirely sure what a result like this says about the effectiveness of exposure and shunning of wrongdoers in the tribe of science. (Especially because I think exposure of mistakes might be the first step to rehabilitation of the scientists who made those mistakes, I worry about setting up groups of "bad actors" and "good actors" where group memberships are taken to be more or less permanent.)

However, I do think these results suggest that a lot of the bad behavior we see in the tribe of science may have been "learned at home" -- if not at the knee of the mentor, then from some other scientist identified as an important player or a rising star. If you see misbehavior from someone who everyone in your group tells you is a serious scientific talent, doesn't it seem natural to conclude that this behavior is really OK?

This is yet another reason for scientists, especially those involved in training new generations of scientists, to be very clear about how misbehavior can hurt the shared body of scientific knowledge and the scientific community that needs to work together to produce and test that body of knowledge.

Scientists are humans, subject to temptations to cut corners or to gain an advantage over the competition, and even people with lots of scientific talent sometimes display bad judgment. But if we're serious about getting the scientific job done, we need to try to be better than that.

More like this

As an up-and-coming scientist, if I knew that a "successful" scientist at my University had consciously cheated to get there, I would not consider that person a success, especially not a success in science. It's about getting closer to the truth, not about having the longest CV.

Also, ping!


We're assuming that the acting student was incapable of modifying how they said "I solved everything" in a subtly more or less obnoxious fashion depending on what sweatshirt they were wearing. Maybe the actor wanted the researchers to come to the conclusion that Carnegie Mellon students were snobs.

It is my belief that cheaters start early and get better with time. The longer they succeed with their cheating without being caught, the more confident they become and are willing to take bigger risks. They will do everything in their power to hide their cheating and will deny ever cheating if caught.

By S. Rivlin (not verified) on 06 Jan 2009 #permalink

I agree in 100%.

It is important to realize that these CMU students were not bad people -- they were just people and they learned from their social environment was the appropriate behavior in this case.

The same would go with all kinds of things in science -- such as not reporting "failed" experiments, ignoring observations, etc.

Irrationally yours


This sounds like by experimental design. What was the test hypothesis? Is the sweatshirt variable acting as a confounding variable? What kind of cheating is going on: repaying money or finishing the exam? The actor stated "I solved everything" which does not mean s/he got correctly solved every problem although it may imply as such. The experimenter said it was impossible (on average) to complete the exam.

Actually, this is a simple bit of social psychology that has been repeated many, many times over. We obey two masters -- ethics and social norms and the latter is much stronger. People, more often than not, adjust their behavior to what they determine to be normal. We happen to be very, very bad about determining what actually is normal and therefore a few anecdotes about an extreme will lead us to think that the extreme is normal and as a result, the norm shifts.

Universities have been using the same effect for a decade with posters proclaiming how much "normal" students drink. Students often believe the mean drinking rate to be significantly higher than it actually is and thereby adjust their own alcohol intake accordingly. When informed about the actual average, that average tends to fall because of the same sort of adjustment -- much more than any sort of preaching about the correlation between alcohol consumption and health or alcohol consumption and grades.

And, yes, anytime a philosopher of science sees the University of Pittsburgh denigrated as a "bad school" something must be said -- they don't call it the Cathedral of Learning for nothing.

Maybe "bad school" is a poor choice of words, but it is the competition, and presumably the test students want to perceived themselves as better than the competition.