Alan Ehrenhalt is a former executive editor of GOVERNING.
E-mail: mailbox@governing.comTwitter: @governing
Maybe somebody can answer this question for me: All of us are bombarded with exit polls, and all of us know that they can be wrong about final results -- famously wrong, as with John Kerry in 2004, but also in quite a few other cases since then. We know this because it's possible to lay the official tally against the poll and see that they don't match.
Why, then, do we take it as absolute truth when an exit poll proclaims that Hillary Clinton got 60 percent of the vote among female Catholics, or that English majors went for Barack Obama 4-1? These are interesting numbers, but they're almost always unverifiable. The only way you can check on them is to run a competing poll at the same time among a similar group, and that's rarely done any more. Who knows what the real numbers might be?
The bottom line is that we're suspicious of exit polls -- properly so -- where their accuracy can be tested, and utterly gullible when it's impossible to know whether they are right or wrong.
I realize that there's a strong constituency that lives off these polls -- TV networks, bloggers, consultants, political scientists - but I also think that we are making bold assumptions that may or may not have much in common with reality.
Maybe English majors actually love John McCain. I doubt that, but I also doubt that an exit poll is the best way to find out.
Alan Ehrenhalt is a former executive editor of GOVERNING.
E-mail: mailbox@governing.com 
GOVERNING Politics is the place for news and analysis on campaigns and elections. If there's a ballot measure in California, a legislative election in Alabama, a mayoral election in Anchorage or a governor's race in Rhode Island, GOVERNING Politics probably is writing about it. We love everything about state and local politics, from polls and campaign ads to policy debates and demographic trends.