Close-up
How did exit polls start avalanche of inaccuracy?

By Newhouse News Service and Knight Ridder Newspapers

MIKE DERER / AP Joe Lenski, co-founder of Edison Media Research, says that because of the sampling error, exit-poll data can help predict elections only if "there is a landslide."

E-mail this article Print this article Search archive Most read articles Most e-mailed articles

The information even was powerful enough to move financial markets.

And, as it turns out, it wasn't true.

John Kerry was not beating President Bush in Ohio and Florida and on the cusp of becoming the new leader of the free world — despite what confidential exit-poll results widely distributed on the Internet on Tuesday seemed to indicate.

In the 2000 election, scoop-hungry TV networks used exit polls to call the key state of Florida for eventual loser Al Gore.

This year, it was the bloggers' turn to be led astray. A few influential Web sites got hold of poll results showing Kerry with a small lead in key battleground states, word spread at hyper-speed, and soon political insiders were convinced that Kerry was going to be the president-elect.

What went wrong with those exit polls?

"I don't know," said pollster John Zogby, who relied partially on exit polls Tuesday to declare Kerry the winner in Ohio. "I'm not blaming everything on the exit polls, but the exit polls were terrible."

The impact of the exit polls — interviews of voters after they cast their ballots — was significant. They set the tone of news coverage on election night, providing the news media insight into which way an election is heading. They also set the agenda for the campaigns, giving them a roadmap for potential last-ditch strategies before the polls close and to gauge whether a candidate should prepare a victory or concession speech.

Tuesday's exit polls were a major shift from several pre-election public polls that showed Bush with a small but definite lead over Kerry of 1 to 3 percentage points. That turned out to be right on the money, a victory for the pollsters' often-criticized art.

Don't blame the exit polls, said Joe Lenski, co-founder of Edison Media Research, the Somerville, N.J., company that conducted the surveys nationwide at the behest of major media companies.

Generally, exit polls are used to get at the demographics behind an election, to quantify where candidates got support and to break the vote down by age, race, gender and other factors.

But determining winners?

Not in these close elections, he said.

Most of the exit poll questions asked Tuesday had an error margin of 3 percent to 4 percent.

In many states, the actual results were much closer than that.

"Because of the sampling error, exit poll data is only useful as a sole source of predicting elections if there is a landslide," he said.

But that didn't stop the bloggers, and by the middle of the day Tuesday, the perception that Kerry was winning had been absorbed into the political mainstream.

Kerry's people saw a stunning victory: The early polls seemed to verify the anticipated surge in the number of newly registered 18- to 29-year-old voters casting ballots and heavy turnout among women, key ingredients in Kerry's recipe for winning the White House.

The Bush people saw major defeat: Preliminary numbers showed the president a staggering 20 points behind in Pennsylvania and losing in every other closely contested state.

Some network correspondents subtly telegraphed the polls to viewers. On the evening news, NBC's David Gregory said Bush "appeared subdued," while ABC's Terry Moran noted the president had expressed a "rare sense of doubt."

"I remember watching CNN ... and seeing Tucker Carlson and Bob Novak sitting there with long faces and James Carville sitting there with a huge grin," said Adam Clymer, political director of the National Annenberg Election Survey.

Carville is a Democratic strategist. Carlson and Novak are conservative commentators.

Late Tuesday afternoon on Republican-friendly Wall Street, the Dow Jones Industrial Average fell by more than 100 points — and analysts blamed the premature news of Kerry's imminent triumph.

Yet Kerry's victory speech never came. As day gave way to evening and vote-counting began, the actual results began to contradict the polls.

Larry Harris, a pollster with Mason-Dixon Polling & Research, said the early exit polls overstated the turnout of young voters — people 18 to 29 — and failed to note that Bush received a 4-percentage-point gain in the people older than 60 who voted for him from the 2000 election.

That said, Harris believes that the media and the campaigns misuse exit polls.

"Exit polls are a wonderful advantage in providing context and texture," he said. "But for trying to make a call, we've learned in two elections in a row that they lead to confusion."

Karyn Barker, a fellow at the American Enterprise Institute, agreed that exit polls are getting a bad rap. It's not the polls, she says, it's how the people use them.

"Anyone who follows exit polls should not use them as a prediction of the turnout of certain groups," she said. "To assume that these numbers were spot-on was foolish. It seems some people, including me, wildly misinterpreted what they meant."

Some say users of exit-poll information never used to have to worry about error margins and statistical sampling.

"If we go back in history to prior presidential elections, those exit polls were dead on," said Dennis Simon, a political scientist at Southern Methodist University in Dallas. "Something has changed to make them less dead on."

Zogby said Tuesday's exit-poll problem probably would send pollsters back to the drawing board to figure out whether they need to apply different formulas, models and weights to their surveys.

Some critics believe the exit polls have become too proprietary. Instead of trying to keep the results secret, the poll results and methodology should be entirely above board, said Jack Shafer, the media critic for Slate, an online magazine that published leaked poll results on Tuesday.

That way, he said, the public can be better informed about how to view poll results — and make a more informed decision about how to interpret them.

"There is very little transparent about what they do," he said. "They are basically accountable to no one."

Background on NBC's Gregory and ABC's Moran was reported by The Associated Press.

Copyright © 2004 The Seattle Times Company

E-mail this article

Print this article

Search archive

More politics headlines...