Who's Losing The Presidential Race? The Pollsters

Looking through the lens of national public opinion polls, the 2000 Presidential election is as clear as mud. Not only do the leading pollsters disagree on who's ahead and by how much, but they also can't decide who has the momentum. The situation is so bad that seasoned politicos are disregarding the latest numbers. "We have seven different polls with wildly different results," says Democratic National Committee Chairman Joe Andrew. "I don't take any of them to heart."

Nor should voters. Here's a snapshot of the poll-taking chaos: A Sept. 14-15 Newsweek poll gave Democrat Al Gore a 14-point edge in the Presidential race. A week later, the same poll declared that the Veep's lead over Republican George W. Bush had collapsed to just two points. Over the same period, Gallup's daily tracking poll showed Gore ahead narrowly, then detected a Bush surge days later. And the bipartisan Voter.com/Battleground Poll had Gore trailing the entire time--but gaining slightly.

Why the wild discrepancies? One reason: Swing voters keep swinging. Most voters have made up their minds, but 30% of the electorate--predominantly independents--remains up for grabs. The result: a greater potential for rapid shifts in polling results. For example, Bush held a 53% to 26% lead among indies on Aug. 4-5, according to Gallup. By Sept. 21, it showed Gore way ahead, 48% to 34%. A day later, Bush had crept back in front, 41% to 39%.

Another explanation for this volatility: Some polls force uncommitted voters to choose. Those "soft" supporters of a nominee often blow with the latest political wind. So polls listing fewer undecided voters tend to be more volatile. "Voters really aren't terribly committed or passionate this year," says John Gorman, president of Cambridge (Mass.)-based Opinion Dynamics, which polls for Fox News. "If you're pushing people to commit, you get a lot of random responses."

The number of Presidential polls has quadrupled in eight years. While that means lots more information is available to the public, the proliferation of "down-and-dirty" polls makes much of it unreliable. Ideally, pollsters say, surveys should be taken over a period of at least three days and include at least 1,000 voters. But some single-day polls are based on samples as small as 250 people. "One-night polls are totally irresponsible, verging on product liability," complains veteran Democratic pollster Celinda Lake.

"CHEAP AND SLOPPY." What's more, polls taken on Friday night or over the weekend tend to lean toward Democrats, because more Dems and blue-collar voters seem to be at home then, says GOP pollster Ed Goeas. GOP consultant Frank I. Luntz blames media organizations that sponsor quickie polls: "They're getting cheap, and they're getting sloppy. They don't care what the numbers show as long as they get a story. This will discredit the industry in the end."

In 1992 and 1996, most polls slightly overstated Bill Clinton's margin of victory, and Republicans had a right to howl about a Newsweek survey just before the '96 election that had Bob Dole trailing the President by 23 points. Such discrepancies have heightened public skepticism. "I don't think everyone's telling pollsters the truth," says Robert C. Broeker, 55, a truck mechanic from St. Louis. "People are upset with polls and will say anything." If Broeker is correct, it could be a wild ride between now and Nov. 7--when the only poll that counts finally takes place.

    Before it's here, it's on the Bloomberg Terminal.