[mcherm]

First of all, I believe that it DOES need to be demonstrated -- repeatedly, to lots of people, so we can change how we respond to such a situation.

But more relevantly, this experiment was not attempting to "demonstrate" the effect. It was attempting to better understand the effect. That, in particular, is research that we badly need.

(PS: I'm defending the idea of performing this sort of experiment, not the specific experimental protocol followed. In particular, I question whether applying negative judgements to people is actually a valid proxy for causing harm.)