The Bradley effect

Exit polls are sometimes wrong, but they can’t always be wrong. Last night, early exit polls showed Obama leading by 5 points; Clinton won the state by 10. Sound familiar? Same story in Texas, Ohio, Rhode Island, California, and most famously New Hampshire. That’s not just a list of states Clinton carried — that’s a list of states that showed Obama winning in the exit polls (or at least, in the case of Rhode Island, staying much closer than the ultimate spread). At some point it stops being a series of flukes, and it becomes something a bit more predictable.