On election night my roommate kept remarking how useless exit polls were, and how he didn’t even see why TV stations bothered to report them. I still stand by my counterargument on the CNN/MSNBC/etc. front, which is that exit polls are something to report. Not perfect, but better than nothing.
But I was a little skeptical of his assessment of the polls’ accuracy, and I was wrong. Over at NRO’s Corner, JPod has the scoop on how telephone polls were actually more accurate than exit polls were. The latter keep skewing Democrat by significant margins, 5 to 8 percent. In close elections, of course, that can affect the predicted outcomes bigtime.
My theory is the “privacy buff” effect. More conservatives than liberals refuse to take random polls. In telephone polls they correct for this bias, but exit polls are reported raw. I would think that, over time, pollsters could introduce a correction to their exit poll data that would make it more useful. At the expense, of course, of the “raw data” appeal exit polls now have.
JPod hopes for an end to exit polling. I doubt this will happen, just because it’s such a news circus. But we can hope for more accurate exit polling.
Robert VerBruggen blogs at http://robertsrationale.blogspot.com.