I think there are other alternatives that need to be considered.

One of the key concepts is whether or not we are actually involved in evaluating our behavior. If we don't pause to think about the concept of "cheating" but instead are just having "fun", then of course we'll have fun like everyone around us.

If a rival is doing something, we get into an us vs them situation and may have to analyze the situation, leading us to examine our feelings about the rival, perhaps due to their cheating, and then having to consider the impact of our cheating on how we think of ourselves or others think of us.

For this reason, I think the anonymity of a person wearing a white shirt may also be significant.

But what if your competitor is behaving badly and doing well? As we learned from this mortgage crisis, people are all too happy to join in on the fun if it means they get rewards (McMansions) even if the path is wrong and leads to disaster eventually.

This study is also explained in Dan Ariely's book "Predictably Irrational".

Either of my children is more likely to behave well when the other is behaving badly. It looks like we humans don't grow up. In the case of children though, it's probably motivated by competition for parental attention. I'm not sure what the motivation might be for Carnegie-Mellon students.

For groups of students it's basic tribal identity.

If our tribe decides that, say, downloading mp3s is alright, then we're more likely to download mp3s. If we decide that 'only those clowns from public universities do that' then we are less likely.

What would be interesting, is to see what would happen if the actor had merely proclaimed 'we don't cheat here' while wearing a white shirt. i.e. does the reiteration of a value alter behavior more or less than a specimen of the other tribe?

Except on Wall Street.