Did they take class into account at all? That could explain some of the results. Or even if they did, and all of these stats have been adjusted for class, carryover from class stereotypes could still explain the results.
I imagine you could, by using fake profiles and fake messages, do a controlled experiment which took all these factors out. My guess would be that you'd still see a vaguely similar effect but that it would be much less pronounced.
I didn't even think about the money thing, but we do know that women care an awful lot about it. Who knows; a lot of the blatant racism towards black men (i.e. in the self report section) could be because women generalize that black men are poor. It could also explain why everyone seems to want to get with a white dude; he's more likely to be rich/have higher status.