Learned about this in undergrad. My problem with it is the inherent issue in quantifying "nice". I know the theory relies more on evolutionary fitness than it does the actual act of being nice, but quantifying something with so many influences is impossible.
And, using humans as a model for this sort of study is flawed. We have entirely too many influences on the nice phenotype to even begin to remotely account for all the variables. In essence, studies such as his are more an exercise in proving a hypothesis rather than coming up with any tangible results.