Exit Polls, Schmexit Polls

In a post yesterday about Election Day exit polls, we noted that long gone is the Voter News Service, the consortium that handled exit polls, vote tabulations and projections four years ago — and that led the TV networks that commissioned the polls in the year 2000 to rash predictions, followed by hasty backtracking. We also noted that VNS has since been replaced by the National Election Pool, a new consortium of CBS, Fox, ABC, NBC, CNN, and the Associated Press.

“Whether NEP will produce more accurate results based on exit polls than VNS did, however, remains an open question,” we wrote.

Well, it doesn’t remain an open question anymore. Yesterday afternoon and evening, half a dozen blogs, along with Slate’s Jack Shafer, set both liberals and conservatives all a-twitter by posting woefully flawed NEP exit poll results that uniformly overestimated John Kerry’s support. A few posters cautioned that exit polls aren’t gospel, but they didn’t bother to explore such niceties as margin of error, sampling bias, size of sample, geography of sample or bias in the questions asked — or inform readers whether or not they were looking at raw data or weighted data.

Consequently, as late as 7:30 p.m. EST — before most polls closed in the east, and four hours and a half before most polls closed in the west — Shafer posted the following numbers on critical states:

Florida: Kerry 51, Bush 49
Ohio: Kerry 51, Bush 49
Michigan: Kerry 52, Bush 46
Pennsylvania: Kerry 53, Bush 46
Iowa: Kerry 51, Bush 49
Wisconsin: Kerry 51, Bush 48
Minnesota: Kerry, 52, Bush 46
New Hampshire: Kerry 54, Bush 44
New Mexico: Kerry 50, Bush 48

What’s wrong with this picture? And what does this tell us? Well, among other things, it tells us the head-to-head results gathered by NEP, like the ones gathered by its predecessor, are about as valuable as a bucket of warm spit. The only difference is that, while gullible bloggers more or less swallowed the numbers whole, this time the networks that actually paid for the exit polls didn’t let those polls lead them into premature calls.

(We have to wonder for how long the consortium is going to be willing to commission exit polls that are useless to it members, but that get splattered all over the Internet, to amuse, dismay or stun the electorate.)

As to what effect the distorted exit poll data had on voters, who knows? Maybe none. But you could certainly make a hypothetical case that some would-be Kerry voters stayed home because they read Slate, or The Corner, or Wonkette, and thus it appeared to them that Kerry had it sewed up and didn’t need their help — or that some Bush voters were alarmed enough by the early exit polls to get off their butts and rush to the polls. Or vice versa.

That is the possibility that some Internet journalists, like Shafer of Slate, are loathe to admit. Shafer’s boss, Jacob Weisberg, on Monday dismissed the idea of withholding exit poll data (flawed or not) until a few networks had called the election, or better yet, until voters had actually finished voting, as “paternalistic,” and he voiced confidence that Slate readers are not the sort to be swayed by such data, flawed or otherwise.

In a way, it’s touching that Shafer and Weisberg think that the readers of Slate are a sort of steely-minded super race, immune to the normal human emotions — elation, despondency or uncertainty — that influence behavior. But somehow we doubt it.

Steve Lovelady

Has America ever needed a media watchdog more than now? Help us by joining CJR today.

Steve Lovelady was editor of CJR Daily.