If you live outside of the media bubble, there’s some news you don’t know that everybody living inside it—reporters, editors, television producers, media executives, news interns, and their friends and acquaintances—are sharing right now: The exit poll numbers from today’s New Hampshire presidential primary.
Slate made news in during the 2000 presidential primary season when it ended the media conspiracy and started sharing those numbers with readers as we received them. Voter News Service—the media consortium that produced the 2000 numbers for CBS, NBC, ABC, CNN, Fox, and the Associated Press—officially prohibited its members and subscribers from releasing the exit-poll numbers until the majority of New Hampshire polls closed. Even so, the network anchors telegraphed, as they almost always do, the identity of the winners with carefully sculpted language hours before the polls closed.
Slate decided to publish the early VNS numbers coming across our desk on Election Day afternoon to protest the hypocrisy of the networks and make the point that if the networks possessed news—such as the leader of their exit polls—they shouldn’t suppress it. For our labors we earned a cease-and-desist letter from VNS in February 2000, and having made our editorial point about the stupidity of the networks’ policy, folded our tent for the rest of the campaign. (For the back story of Slate’s tussle with VNS, see this piece I wrote for the Wall Street Journal op-ed page.)
VNS is no longer with us. For the 2004 election season, Slate intends to share the numbers cascading into our inboxes. The following is an average of midday exit polls performed by six different news organizations, which—with one exception—were highly consistent with one another. (We averaged the polls in the hope of smoothing out blips or anomalies particular to any one of them).
The outlying poll, i.e., the one not consistent with the other five–but which is included in the above average–showed Dean running just ahead of Kerry.
Midday exit-poll numbers don’t necessarily predict the winner, but the networks do use them as a confidence cushion upon which they seat the early evening coverage. It will be interesting to see how close these numbers parallel the final count. While exit polls tend to be more accurate than other kinds of polling, they’re only one of the tools used in predicting the results of a race before the full official count is in. Election statisticians often need to get their hands on actual vote counts from test precincts to call a race. And as we saw in Florida 2000, the prognosticators aren’t infallible.
The major news organizations collecting this information think the public can’t be trusted with it. Slate, and I, continue to disagree.