Conflicting campaign polls put confusion in the lead

By The Dallas Morning News and Chicago Tribune

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

Related stories Campaign Notebook: Nader on Florida's ballot; denied spot in New MexicoKerry having a ball

Unless it's a dead heat.

It all depends on which poll you believe.

With Election Day a little more than six weeks away, Americans are being bombarded with often-conflicting polls, adding another layer of confusion to a contest muddled by Vietnam-era war records and different views on the economy and Iraq.

Surveys released this week range from a 13-point Bush edge to a one-point Kerry comeback, reflecting the fact that polls can be as volatile as the political races they try to reflect.

"Opinions are shifting as people think about something one day and think about something different the next," said Andrew Kohut, director of the Pew Research Center. "The convention period created instability in the electorate that probably won't stabilize until after the debates."

Republicans love the new Gallup poll, which gives Bush a lead of 55 percent to 42 percent among likely voters. The lead shrinks among registered voters, 52 percent to 44 percent. Democrats prefer a Harris poll giving Kerry a one-point lead, 48-47.

Even political sages such as Kohut were bewildered when they saw Bush's wide lead in the latest Gallup poll, which partners with USA Today and CNN.

"It certainly must be confusing to people who aren't in the business," Kohut said, "and even confusing to those of us who are."

Sometimes the variances are within one survey. One wave of questions in the Pew poll, posed Sept. 11-14, showed a 46-46 tie among registered voters. But questions taken the three previous days gave Bush a 12-point lead, 52-40. Both surveys had error margins of 3 percent.

Some analysts attributed the differences to a fast-paced race: violence in Iraq, uncertainty over the economy, Republican claims of Kerry flip-flopping, Kerry accusations that Bush is misleading the nation, questions about Bush's National Guard service, and disputes over whether documents used to report on that service were real.
 
The ways polls are conducted may be a bigger factor, analysts said. In an already volatile race, the difference between the Gallup and Pew results touched off an active debate yesterday in an e-mail discussion group for the nation's pollsters.

"I don't know the answer to it," said J. Ann Selzer, a pollster with The Des Moines Register's Iowa Poll who was following the online conversation. "That just seems crazy to me."

Many polls have shown a sizable difference in the preferences of registered voters vs. likely voters. Bush typically has done better among likely voters, a more-limited group formulated by asking poll participants questions about their voting history and future voting plans.

Several surveys also weight the responses of poll participants based on how they have answered questions about the likelihood of their voting. The answers of someone already registered to vote and with a history of voting may be given greater weight than those who have not voted in the past.

Conventional wisdom has suggested that polls using likely voters are more reliable than those using registered voters after Labor Day. This year could be an exception, though, since Republicans staged their convention much later than usual, possibly resulting in later voter swings.

Another possible reason for the wide variations: Pollsters long have debated whether polling samples should be weighted to reflect the historical ratio of Republicans, Democrats and independents.

Some firms say polling data should allow for demographic shifts, but Democrats had a 3 to 4 percent advantage over Republicans in party identification in the past three presidential elections.

The sample for a Newsweek poll that showed Bush with an 11-point lead immediately after the Republican convention was 38 percent Republican, 31 percent Democratic and 31 percent independent. Because the magazine does not weight its polls, some critics have said the numbers were skewed.

In Pew's case, the poll that gave Bush a double-digit lead used data that included more Republicans than Democrats, and the survey that called the race even used data that included more Democrats than Republicans.

With Bush enjoying a lead in most surveys, some Democrats have suggested polls are not picking up new voters being signed up through intensive registration drives in urban areas and on college campuses.

"We are sensitive to that and are carefully monitoring for it," said Frank Newport, editor in chief of the Gallup poll.

Despite the confusion, the polls are difficult to dismiss because results drive perceptions, shape strategy and influence the campaign story line.

"A lot has been said about the Bush lead over the last few weeks, but we are looking at a race that is tight nationally and in battleground states," said Joe Lockhart, a Kerry adviser.

Scott Stanzel, a Bush campaign spokesman, agreed. "We have always indicated from Day 1 this election would be close, perhaps as close as 2000," he said. "That's what we're preparing for."

Seattle Times desk editor David Birdwell contributed to this report.