John F. Kerry leads President Bush, 47 percent to 46 percent, in the latest AP/Ipsos Public Affairs poll.
Yet Bush leads Democrat Kerry by 8 percentage points in a new TIPP survey.
And Republican Bush leads Kerry, 47 percent to 45 percent, in the new Reuters/Zogby Survey.
These see-saw results in national polls point to new and unsolved challenges in polling because of new telecommunications products.
The daunting advances are cell phones, caller ID and Do Not Call directories.
Polling industry officials insist that in spite of consumers' growing use of high technology, this year's voter surveys are as reliable as they were in the close presidential election of 2000.
But some independent students of the business, as well as some pollsters themselves, strongly disagree.
New consumer technology "is completely screwing things up," said Darrell West, director of the Taubman Center for Public Policy at Brown University.
"It is getting harder and harder to get reliable data (from voters over the telephone) because of cell phones, answering machines, caller ID devices and the resulting low response rates," West said.
In addition, a record number of newly registered voters, including young voters and blocs of African-Americans, may not yet show up on the computerizing call lists for polling companies to check. This is further complicating polling this year.
"Nobody's coping very well with these changes," said Canisius College analyst Michael Haselswerdt, who does his own polling. "We are in the last stages of sampling the voters the way we've always done it."
More than 161 million cell phones -- which are not found in the phone directories that polling companies use -- are now in operation across the country. And cell phone use is growing at a rate of about 13 percent a year, according to estimates by the Federal Communications Commission.
Nationally, 67 percent of households now have someone who uses a cell phone, according to Scarborough Research. In the Buffalo Niagara market, the figure is 56 percent.
Wireless phone use is concentrated among younger, more affluent Americans. Critics of the national election polls say they are missing out on some of this volatile and growing sample of the voting public because pollsters do not call cell phones.
Another problem is implementation of the Do Not Call list barring cold calls to homes by telemarketers. That list of 62 million land line subscribers has an indirect negative effect on pollsters' ability to reach voters.
Congress exempted pollsters and political candidates from the Do Not Call prohibitions. But few voters know of these exemptions, and the list increases the likelihood of hang-ups when the political surveyors call.
Nancy Belden, president of the American Association for Public Opinion Research, contends the polling business has successfully adjusted for these changes.
To begin, she says, cell phone-only households still comprise only 5 percent of all homes -- a figure not large enough to affect poll reliability. "And we are still finding plenty of respondents," Belden said.
In addition, major polls have coped by "weighting" polling results.
Few, if any political polls, actually reach a voter sample that is truly balanced in terms of gender, age and ideology.
So pollsters "weight" their results. That means pollsters need to puff up real samples to make up for groups that the pollsters can't actually reach on the telephone. If a survey can't contact enough real minority households, for example, the company increases the minority share to agree with census or other estimates.
But Cornell University's expert on polling, government professor Walter E. Mebane, said "weighting" may not really work this year.
"Most of these guys are going to use guesswork," Mebane said of the practice.
"And their guesswork may be no better than yours," he added.
John Zogby, president of Zogby International, bristled at Mebane's suggestion that "weighting" poll samples involves guesswork.
"There are pathways of party identification that can be tracked with people who turn out to vote," said Zogby, who was one of only two major pollsters to predict that Al Gore would win the popular vote in 2000. "My 'weighting' for political IDs is based on sound historical trending."
He said all the changes have been accounted for in his surveys by weighting and careful tracking of party trends.
"Let's face it, though, 20 percent of this profession is art," Zogby said about weighting techniques.
One big fly in the ointment this year, Mebane said, is the surge in new registrations around the nation, particularly in battleground states.
As a result of the unrest over the Iraq war and cultural issues, independent organizations such as America Coming Together on the left and evangelicals and lay Catholics on the right are spending millions on registration, advertising and get-out-the-vote drives.
Independent Republican and Democratic groups so far have spent $439 million on such efforts, according to Political Moneyline, an independent bipartisan tracking group.
The edge on independent spending so far goes to the Democratic side, with the liberal patron of Moveon.org, George Soros, leading the pack of givers with $26.5 million.
As a result of these drives, Utah has increased registration by 69,000 over 2002; Georgia is up nearly 400,000; Louisiana is up 133,000; and Philadelphia alone has registered 225,000 new voters this fall.
Georgia is seeing late surges in African-American enrollments, and New Mexico registered 17,000 new Hispanics in October.
But few pollsters are mechanically set up to chart changes in enrollment across the nation, Mebane said. One reason, he said, is the wide variety of registration deadlines among the states.
If these new registrants are young voters, the massive increase in cell phone use since 2000 makes their political leanings even more elusive for pollsters to track.
"So no one has a clue as to who will be voting . . . ," Mebane said.
Industry spokeswoman Belden says voting patterns of the younger people are unpredictable. "Young voters have always been difficult to reach," she added.
Barry Zeplowitz, head of Buffalo-based Zeplowitz & Associates, also wonders whether the major public surveys are getting an accurate reading on young voters.
"The numbers are all over the place," he said.
Zeplowitz, who is doing private polls on state legislative and congressional races, noted that at the end of last week there was "a nine-point spread" among three national polls -- ranging from Bush trailing Kerry by three points to beating the Massachusetts senator by six points.
Zogby, whose polling indicated an 8 percent margin of victory for Bill Clinton in 1996, just 0.5 percent off the actual vote, voiced confidence in his own surveys this year. But he still has worries.
"The one thing that gives me a queasy feeling are the number of voters who make up their mind only on Election Day -- a day we don't poll," he said.
News Washington Bureau assistant Anna L. Miller contributed to this article.