Election forecasts cloudy with a chance of being dead wrong

By Dana Milbank
Monday, November 1, 2010; 6:08 PM

Election results won't come until Tuesday night at the earliest. But luckily, you don't have to wait. This is because you have prognosticators.

The Economic Club of Washington hosted three of this species Monday for a luncheon and panel discussion titled "The Mid-Term Election Results a Day Early."

Pundit No. 1, Time magazine's Mark Halperin, informed the assembled lawyers and business people how many seats Republicans will gain in the House: "at least 55, and I think it could be as many as 85." While admitting his predictive science is imperfect, Halperin added: "If you want an exact number, 75.2."

"I'm going with 58," offered pundit No. 2, ABC News's Claire Shipman.

"It's possible it could be lower than 50," submitted pundit No. 3, Politico's Jim VandeHei. Or, he added, "Maybe 85. I don't think it's inconceivable it could be much bigger both in the House and Senate than everyone's anticipating."

So, to recap: Republicans will probably gain 75 seats, or perhaps 58, but their possible pickup range is from 55 to 85 -- except if it's lower than 50, or higher than 85.

It's time again to haul out that hoary convention in journalism and punditry: the biennial election prediction. Participants must state with conviction that which they cannot possibly know.

The first Monday in November is, of course, the busiest prognostication day of them all. In this case, it began with a memo from Politico's Mike Allen, author of the "Playbook."

"SIREN," he wrote. "Nine GOP Senate pickups are now possible. . ."

Or are they? A few paragraphs later, Allen reported the latest forecast from handicapper Charlie Cook, who is changing his "outlook to reflect a net gain for Republicans of six to eight seats, down from seven to nine."

Allen forecast gains for the GOP in the House "in the mid-60s or (maybe much) higher."

Dissenting from that forecast were Allen's bosses, VandeHei and John Harris, who in the Outlook section of the Post on Sunday forecast a 46 seat pickup for the GOP.

Cook, the most influential of the handicappers, delivered his forecast in the manner of a weather man warning of bigger snow accumulations at high altitude: "A Democratic net loss of 50 to 60 seats, with higher losses possible."

Stu Rothenberg, Pepsi to Cook's Coca-Cola, went with 55 to 65, while Larry Sabato went with a straight 55 and the Web site Real Clear Politics called for 67. New York Times blogger Nate Silver is going with 53 -- with a "95 percent confidence interval" ranging from 23 seats to 81.

With Republicans only in need of 39 seats to take charge of the House, the only one predicting with any confidence that Democrats will keep the House is Chris Van Hollen -- but that's part of his job as chairman of the Democratic Congressional Campaign Committee. "The one thing the American people don't like," he told Fox News's Chris Wallace on Sunday, "is Washington pundits telling them in advance what they're going to do."

Van Hollen's optimism may be suspect, but he's got a point about the tendency for Washington to declare an election over before the votes have been cast. This is particularly true because the basis for the forecasts ¿ the polls -- are of dubious value. Is the prognosticator basing the prediction on the Gallup poll, which has a 15-point advantage for Republicans, suggesting "the largest Republican margin in House voting in several generations"? Or is the prognosticator giving weight to the Newsweek poll, showing DEMOCRATS with a three-point advantage?

It's a bit like filling out a March Madness bracket, but with unreliable seeding. At the Economic Club luncheon, in fact, they handed out "Election Contest" cards to attendees; the lucky winners get iPads or four seats in the Trustees Box at the Kennedy Center.

Looking out over tables labeled Bank of America, Akin Gump, HSBC and the like, Carlyle Group managing director David Rubenstein, the moderator, said his panel of pundits would "give you the results a day in advance."

Shipman forecast that Nancy Pelosi would retire in six months. VandeHei forecast that Defense Secretary Bob Gates would step down early next year. Halperin forecast that with Republican gains in the House of more than 68 seats, the party would also seize control of the Senate.

Naturally, the three pundits pulled all such predictions straight out of their imaginations. Of the three, Halperin was the most prolific. He predicted: that Hillary Clinton may switch jobs with Vice President Biden; that write-in candidate Lisa Murkowski will win the Alaska Senate race handily; and that Carly Fiorina could beat Sen. Barbara Boxer in California as part of a GOP wave.

But when asked to predict "the biggest surprise" of the election, VandeHei offered a real shocker: Democrats keep control of the house. This, he said, "would actually prove what we all know: That conventional wisdom is always wrong."

Seems you'll have to stay up late Tuesday night after all.


© 2010 The Washington Post Company