WASHINGTON, Sept. 20 (UPI) -- Privately, the campaigns of President George W. Bush and Sen. John F. Kerry both say that internal polling shows Bush leading in the 2004 race for president by anywhere between 3 and 5 percentage points.
Bush is ahead largely as a result of his strong performance at the Republican National Convention. It is the first significant lead either candidate has shown, if not all year then certainly since Kerry became the Democrats' de facto nominee.
The internal numbers, however, are wildly different from what national political surveys show.
The name-brand independent and media polls suggest everything from the race still being too close to call to there being an emerging Bush blowout in the offing.
In the latest national Gallup poll, Bush had a 13-point lead over Kerry among likely voters polled. The CBS/New York Times poll had Bush ahead by 8 points, 50 percent to 42 percent, among registered voters asked.
The Zogby America poll from Sept. 19 has Bush leading by 3 points among likely voters nationwide, just inside the 3.1-point error margin, while the Pew poll from Sept. 14 has Bush and Kerry separated by a 6-point gap among registered voters and by 1 point among respondents likely to vote in November.
Adding to the confusion is the Sept. 13 Harris poll, which has Kerry at 48 percent, Bush at 47 percent and liberal independent Ralph Nader -- who is increasingly less likely to be a factor in the 2004 outcome because of his inability to gain access to the presidential ballot in several key states -- at 2 percent, with a margin of error of plus or minus 4 points.
Some of the variation can be explained, pollster David Winston said, by the difference between the registered-voter number, which represents all potential voters, and the likely-voter sample, which is determined in a non-standardized fashion.
"The problem with the likely-voter screen," Winston said, "is that you are using a matrix that is based on assumptions rather than fact.
"It's an artificial device that lets you get at a smaller group of voters; therefore," he said, explaining that most of the screens used to measure voting behavior and establish whether a survey participant is likely to vote in the next election "measure intensity at the moment rather than identify the people who are actually going to vote on Election Day."
Bloggers and Internet bulletin boards have, in the days since the conflicting polling numbers were made public, worked to undermine public confidence in their authenticity.
A dissection of the Gallup survey posted on the blog The Daily KOS argued Republicans, at 40 percent, were over-represented in Gallup's likely-voter sample -- while the Democrats, at 33 percent, were somewhat more in line with recent election results. Media Matters for America, a leftwing media-criticism Web site run by ex-journalist David Brock, echoed the charge and applied it to the CBS/New York Times poll as well.
Moore Information, a GOP polling firm, likewise has worked to deconstruct the Pew survey, which showed Bush's double-digit lead post-convention has collapsed in a matter of days.
"The difference in the results of these two surveys is almost entirely a factor of the partisan makeup of Pew's samples," the Moore Information analysis said, pointing out that the sample used in the later poll had a 9-point shift in partisan identification among the respondents, "large enough to explain (the) 11-point drop off" in support for Bush.
"There is this concept of what a poll is supposed to do," Winston, who has polled for a number of prominent Republicans, said, emphasizing that what many consumers of survey information want is for the snapshot taken today to predict the outcome tomorrow -- but that's only part of the story.
"Why do the number move, what happened to generate the shifts," Winston said, is often more important than the "who is in front, who is behind" horse race.
The issue of intensity is particularly important, other pollsters suggest, because so much of U.S. election results are based on the last two weeks and last two days before the majority of voters go the polls.
One example is voting by inner-city blacks, who do not tend to show up in a statistically significant way in pre-election polls of likely voters. These same voters, however, tend to turn out in larger numbers than the polls indicate thanks to late-in-the-campaign mobilization efforts -- particularly but not exclusively from the pulpit.
The same can be said for Latinos, union members, white evangelical Christians, gun owners and other groups for whom intensity of feeling is an important factor in motivating a trip to the polls on Election Day.
"The likely-voter screen," Winston said, measures intensity of the moment. "It's not surprising that Bush's numbers spiked right after the GOP convention," he said, because the Republicans and Republican-leaning voters were left with a good or better opinion of Bush while Kerry voters were further disheartened.
What that misses, Winston said, "is the fact that, among the registered voters, there was a clear shift toward Bush."
"There is a general consensus the electorate has shifted significantly toward Bush," Winston said, "but the morass of polls has made it difficult to put a finger on a number."
The size of the number matters, say strategists in both parties, because it determines the kind of campaign that will be waged in closing weeks. A campaign that believes it is in a close race can run a state-specific campaign, meaning it can target messages to specific states -- like Ohio or Michigan -- where a shift of a few thousand voters could make the difference in which ticket's votes are cast in the Electoral College.
A campaign that is not close, one that is behind by 5 points or more, does not have that luxury and must first begin writing off states in which it hoped to be competitive going into the home stretch, as Kerry has done with North Carolina and Missouri.
There are 200 million registered voters in the United States, meaning each 1-point shift in the polls can indicate 2 million voters have moved toward or away from a particular candidate; 5 points, therefore, is 10 million people, a number far too large to be swayed on a state-to-state basis. In that environment, campaign experts say, the campaign running behind has to work to shift national attitudes -- something they say is a difficult proposition this late in the game.
--
(Please send comments to [email protected].)
