108



Thanks to lessons learned four years ago when big media made some wrong calls, the average American watching television Tuesday night got a pretty accurate picture of how the election was going.

But for about seven hours in the afternoon and early evening, several million "insiders" with access to exit-poll data -- blog readers, print journalists, TV executives, politicos and their e-mail buddies -- had a different impression.

Aided by wishful thinking in some cases, many were convinced that John Kerry was benefitting from a powerful voter surge. The news largely wasn't reported on TV or in newspapers, but for much of the day, the nation's tech-savvy intelligentsia was zigging when everyone else was zagging.

What happened? The early exit polls were wrong in some cases. And some insiders disregarded warnings from the exit pollsters that theirs is an inexact science. By comparison, most pre-election polls were quite accurate: They gave President Bush a narrow lead nationally, and correctly identified the battleground states.

The first wave of exit-poll data, compiled by a media-owned consortium called the National Election Pool, went out to subscribers at about 2 p.m. The consortium consists of the five big purveyors of TV news plus the Associated Press. Other media, such as some newspapers and magazines, pay thousands of dollars to see the results. (The NEP was assembled to replace Voter News Service, an earlier consortium whose faulty exit polls caused TV networks to call Florida for Al Gore in 2000.) All had entered into agreements not to leak or make the information public prematurely.

But the 2 p.m. poll results were leaked almost immediately to the Web-log operators called bloggers. The "blogosphere," as this community likes to call itself, was split on whether to release the confidential data. Several blogs did -- bolstered by the respected Microsoft Corp.-owned magazine Slate -- often with caveats attached.

At 1:58 p.m. Eastern time, , a political blog, posted exit-poll results from 12 states, with the caution that they were "early numbers." They showed Kerry with a four-point lead in Ohio and a three-point lead in Florida. At 4:27, the site added another set of numbers, commenting: "Kerry continues to lead Florida overall as well. Again, these are exit poll numbers, so doubt them, but it looks great!"

Slate posted its first results at 3:15 p.m.; earlier, it had posted a long note explaining its decision to publish the exit polls, including a disclaimer about their potential inaccuracy.

"There's this election-day charade that goes on every four years where journalists and friends of journalists are all trading this exit-poll information which historically has had pretty high predictive value, and they've kept that information from readers and viewers while hinting at it," said Slate's editor, Jacob Weisberg. "My view is we can't be in the position of holding back information that's accurate."

But it turned out to be inaccurate, Mr. Weisberg said yesterday. As late as 7:20 p.m. Slate displayed polls showing Kerry leading by two points in Florida and Ohio. "My layman's analysis is they were pretty significantly off," he said of the exit polls.

Exit polls from later in the day were generally within a few points of the final voting tallies. But significantly, exit polls reported at 7 p.m. predicted that Sen. Kerry would do slightly better than he actually did in 15 of 16 states, suggesting some systemic pattern in the methodology.

Warren Mitofsky, a longtime CBS election analyst who co-heads the NEP, said the consortium spotted problems with its early data in Florida, South Carolina, Virginia and six other states. "We had called the networks earlier in the day and told them about a handful of states where we didn't think the results were accurate. But we didn't think we were supposed to share that with the leaker and the leakee."

Indeed, Mr. Mitofsky said: "All the people getting leaked stuff were getting [it] from people who weren't always accurate . . . and were premature, and now they are complaining about it. We didn't mislead the people we were working with. We made any number of projections. All of them were correct."

Complaints about the NEP data stem from a fundamental misunderstanding about such polling, Mr. Mitofsky added: "In a football game, after you get the score at halftime, do you get mad at the announcer if it doesn't turn out that way?"

"I hope I've had some role in killing exit polls," Editor Ana Marie Cox, one of the bloggers who reported exit polls, said in an interview yesterday. "To the extent that blogs provide people with bad or misleading information, I hope that teaches people not to trust media in general."

It wasn't only bloggers that reported exit polls. , the Web site of this newspaper, posted an article between 3 p.m. and 3:30 p.m. saying the early exit-poll data "purported to give Mr. Kerry an early lead in several key states" but raised questions about the validity of the numbers. The article linked to , which posted the figures. "To have a story about how the election is playing out on the Web and not mention the exit polls would be a disservice," said Bill Grueskin, the managing editor of .

Reuters news service ran a story at 6:17 p.m., citing political Web sites and their exit-poll data indicating a strong Kerry lead. Reuters also quoted an article from the conservative National Review casting doubt on the validity of the polls.

Adding to the confusion, pollster John Zogby late Tuesday afternoon predicted Mr. Kerry would capture 311 electoral votes to President Bush's 213, with two states too close to call. "Even though I think our pre-election polling was, by and large, good, we've got some new developments that we need to somehow take a look at," Mr. Zogby said in an interview yesterday. "One is the growing number of voters who make up their mind on election day." He said he used data that went up to 2 p.m. to make his prediction "but some didn't vote until five or six or seven o'clock."

Larry Sabato, a University of Virginia political scientist and expert on polling, contended the exit polls skewed news coverage toward Mr. Kerry and may even have had the perverse result of delaying the reporting of actual results in some states because they didn't match projections. He blamed, in part, what he described as too-high numbers of women and minority voters in the exit-poll samples.

But other experts in public-opinion research say they think the exit-polling system worked quite well overall. Michael Delli Carpini, dean of the Annenberg School for Communication at the University of Pennsylvania, said the media suffers from "a bit of ignorance about what you can take away from an exit poll, especially the early waves. They are still polls, and polls have a certain amount of error associated with them. And ... a two or three percent difference is within the margin of error."

Even though the TV networks were largely scrupulous about not revealing the poll information, observant viewers could see a certain implicit wink-and-nod drift to Kerry, at least early on. Fox News business reporter Neil Cavuto reported in his late-afternoon program that Mr. Kerry was doing well, while CNN's "Crossfire" had pundits analyzing how Mr. Kerry had managed to turn the race around in his favor, and raising questions about Mr. Bush's ability to capture Ohio.

Early in the evening, both the broadcast and cable networks were very reluctant to call several Southern states, including Virginia and South Carolina, both of which traditionally go Republican but seemed to be leaning towards Sen. Kerry.

"I know the White House expressed concern that we not take the early exit polls as a true guide," said CNN Political Director Tom Hannon, adding that he shared those doubts.

"The public should beware," said Linda Mason, vice president of public affairs at CBS News, said: "In some cases, people are criticizing the media for making these forecasts. We didn't make these forecasts, we use them as a tool, and if someone else gets it and misuses the tool, we can't be blamed for that."

All in all, Mr. Fair makes a dazzling case for the methods of the social sciences, whose practitioners have long nursed an inferiority complex vis-à-vis natural scientists. I once saw a study claiming that social scientists had a lower mean IQ than natural scientists, but that economists (like Mr. Fair) pulled the social-science average up, whereas chemists pulled the natural-science average down. I don't remember whether economists were found to be smarter than chemists, but perhaps that is not the point.

-----------------------

Wall Street Journal, November 4, 2004

How Insiders Were Fooled

BLOGGERS LEAKED SECRET DATA GIVING KERRY AN EARLY LEAD, BUT NETWORKS HONORED RULES

By Jesse Drucker and Glenn R. Simpson

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download