About those exit polls

Larry Sabato has another great post, this one in part about the Wisconsin recall election. I love Sabato's stuff, it's very informative and insightful while quite nonpartisan at the same time. Not an easy trick in this day and age. (You'll have to scroll down at Sabato's blog to see the posts mentioned here)

I've been noting with dismay the amount of conspiracy theorizing and delusional ranting by some on the left who question the results of the Wisconsin recall election--in part because early exit polls showed a 50-50 split, rather than the healthy margin of victory that Walker ended up with. I've remarked to others that some Dems sound just like Republicans--after they lose--with their cries of voter fraud. But Sabato has a very reasonable explanation:

"The Wisconsin recall is yet another example of the limitations of exit poll results, which are not always accurate, at least in their top-line, horse race numbers. The first waves in the Wisconsin exit poll showed a 50-50 race, so the early reporting and projections on TV bore little relation to the reality of the tabulated vote. Could we all make a note to discount the top-line results of the November 6 exit poll -- and the news media's breathless projections derived from it? They were wrong on election night 1992, forecasting a big Clinton victory when it turned out to be a quite modest five percentage point (43%) victory over President George H.W. Bush. Then there was that not-so-minor snafu back in 2000 with the exit poll in Florida. And exit polling in 2004 created the Kerry administration for several hours; the unadjusted results showed Kerry winning handily in almost all the swing states, to the chagrin of President George W. Bush's campaign staffers.

"The truth is, Republicans disproportionately distrust the media and pollsters, and many won't be interviewed coming out of the ballot stations. Apparently, there's no good way to correct for this."

For the presidential race, Sabato does some neat number-crunching on independent voters and notes that many are women who are quite unhappy with Obama--but who say they are unlikely to vote. On the other hand, there are significantly large numbers of unregistered voters out there--who tend to be young and nonwhite--who would vote strongly for Obama if his team could get them registered and to the polls.

OK candidates, I think you know what you have to do.