Even as the presidential campaign ended with a triumph for President Bush on Wednesday, armchair strategists and capital insiders were still scratching their heads over exit poll results on Tuesday that strongly, and erroneously, suggested Democratic Sen. John F. Kerry was going to the White House.
"The 7-Hour Presidency of JFK2" was the ironic day-after headline on Slate's Web log called "kausfiles." The headline referred to the period of time on Tuesday when raw exit poll numbers favoring Kerry were flying through newsrooms and around the Internet.
Such data caused Sen. Susan Collins (R-Maine) to become so despondent at one point Tuesday afternoon that she e-mailed her mother: "All is lost."
Similarly, respected election watchers John Zogby and Frank Luntz declared Bush defeated before the sun had set on Washington. "I thought we captured a trend, but apparently that result didn't materialize," Zogby said in a statement posted Wednesday on his website.
Convulsions over exit polls, which sample the opinions of voters as they emerge from polling places, rippled up to the highest levels of both parties. President Bush was briefed on the data by advisor Karl Rove, according to White House Press Secretary Scott McClellan, and there was concern in the Bush camp in the late afternoon.
By contrast, the Massachusetts senator and his top aides were buoyed by raw -- and entirely ephemeral -- numbers that suggested he would carry such key states as Florida and Ohio, both of which ultimately went for Bush.
Pollsters and other analysts interviewed Wednesday said the exit polls -- commissioned by a consortium of broadcast and cable television networks -- had actually served their true function. They are not designed to predict winners and losers, but rather to help news analysts spot demographic and other trends.
The problem Tuesday arose when the raw exit poll data were treated by some who received it as the equivalent of a full-scale poll, without considering its limitations. Often exit polls, which are conducted quickly with a relatively small sampling of voters, fail to capture the true overall shape of the electorate.
In addition, the tight time frames can magnify distortions, especially in samplings taken early in the day, before a full spectrum of voters has been measured. This is especially true in a close, volatile election.
Though the early exit poll data proved misleading, experts said, the election results generally tracked closely with the findings of full-scale preelection polling.
For example, an average of the final week's nonpartisan polls showed Bush with a 2-percentage-point lead over Kerry in the head-to-head national horse race, according to the website RealClearPolitics.com.
That was close to the 3-percentage-point victory margin Bush ultimately claimed, and it was within the margin of error.
Final battleground polls also forecast with relative accuracy the winners of most key states. Only in Wisconsin, where polls generally showed Bush with a tiny edge, did the outcome -- a slim Kerry victory -- belie the prediction.
"The preelection polls did a pretty good job -- they mostly showed it either even or a small Bush lead," said Andrew Kohut, director of the nonpartisan Pew Research Center, whose final poll nailed the outcome with a prediction of a three-point spread for Bush.
The final Los Angeles Times poll found a 49%-48% Bush lead nationally among likely voters -- near enough to the final result to be within the margin of error.
Frank Newport, editor in chief of the Gallup Poll, acknowledged that Tuesday's outcome did contradict one major preelection assumption of many pollsters: that undecided voters would break in Kerry's direction.
Despite the exit polls' limitations, they were eagerly inhaled by impatient amateur analysts -- and plenty of political pros -- as soon as raw numbers began to flow in starting at about 2 p.m. EST Tuesday. They were rapidly leaked to websites such as right-leaning Drudge Report and left-leaning dailykos.com.
Drudge posted a headline -- "Kerry Finds Comfort in First Batch of Exit Polls" -- that alarmed many Republicans.
By the evening, dailykos.com posted a batch of exit poll results that showed Kerry leading Bush 51% to 49% in Ohio and Florida and running better than expected in some other states.
The Times, which purchased portions of the survey data, was told Kerry had a 51% to 49% lead in Ohio and that the Democrat and Bush were locked in a dead heat in Florida. But Times Poll Director Susan Pinkus said the exit pollsters warned the newspaper that the states were too close to call.
The exit polls were conducted by Mitofsky International and Edison Media Research. Edison's Joe Lenski, who helped oversee the surveys, said Wednesday he was happy with the results. The networks, which made no erroneous projections, were also pleased.
"I'm not designing polls for some blogger who doesn't even understand how to read the data," Lenski said. "It's like if you were graded by your readers on the first draft of your article."
But on election day, everyone in politics craves real-time data. Republican pollster Whit Ayres scanned the early numbers Tuesday and heard rumblings of fear from within his party.
"There were a lot of folks on my side who thought it was over," Ayres said. "I worried, but once I looked carefully at the data, I realized it was ridiculously off."
Times staff writer Esther Schrader contributed to this report.