Another reason I'd be doing the e-mail verification thing is so professors can't rate themselves highly, which I've heard is rampant on RateMyProfessors.
My point is that requiring a .edu address is artificially restricting your potential, and most certainly legitimate, users.
I read RateMyProfessors and the information given is less detailed than the information I get talking to people in the halls of my school. The site definitely has a use, but a professor rating himself artificially high will probably contradict the actual widely held student view that the reader already knows.
The idea is that an .edu address is important and necessary in more places than you think (and beyond the likes of facebook as well) and provides one way of verifying you are in some way affiliated with the .edu in question, but you don't have to check separate inboxes to do so. I know I haven't logged in to my .edu accounts in possibly years and just have them forwarded to my main email address. I can't imagine there aren't a lot of people out there doing the same.
I think this restriction is a valid one provided you have a provision for manually checking users that don't have such an email address (probably a small minority).