Joshua Gans draws my attention to a cute US field experiment, which involved dropping 100 wallets, and seeing whether the finders returned them. While the sample clearly isn’t random,* it’s big enough that the age, gender and racial differences are all (by my estimates) statistically significant at the 95% level.
While young people, men and African-Americans are all less likely to return wallets, the thing that really stands out from the data is that if you drop your wallet on the ground, you really really don’t want a young man to find it. This is consistent with the fact that most crime is committed by young men.
* What I mean by this is that those who happen to be walking in public places aren’t a random sample of the population. To get a more random sample, they could have pickedÂ addresses off the electoral roll, and then left a wallet outside each house. But it might have been harder to get away with videotaping the finders.