I'd think that normally the problem is you don't have the data, and just have anecdotes.

Particularly for admittance to class, I think you've got a problem. Generally disparate selection outcomes are treated as prima facie evidence of prejudice. Something as well controlled as this - when I picture the scenario, it's either the guy in trouble, or everyone circling the wagons and chanting the tribal truth because they don't want to open the door to routine measurement of what they do.

If a teacher is incompetent and can't teach anyone he won't get fired, but if he only can't teach blacks? Maybe they'll circle the wagons there too. I could see it going either way, but relatively more likely trouble for the teacher compared to most non sexual transgressions.