Column: The polls slant too

Charlatans. That's what they are, the men and women who produce the daily presidential tracking polls.

Some border on junk science, such as the daily tracking poll the Gallup organization produces for CNN and USA Today. Others are just plain biased, such as the daily poll John Zogby produces for left-leaning Reuters and NBC. Either way, they are, with a few notable exceptions (including the Rassmussen and Battleground polls), not to be trusted.

Indeed, consider the wild swing in CNN/USA Today/Gallup Poll this week. On Monday, the poll indicated that George W. Bush was leading Al Gore by 9 percentage points, 50 to 41. By Wednesday, Gore was up, supposedly, by 1 percent, 46 to 45. And by Thursday, Bush had regained the lead 48 to 43.

Given such volatility, one would expect that something major happened in either the Bush or Gore campaign. But neither candidate has had an exceptional week, good or bad.

It is clear to all, save for maybe CNN and USA Today, that the Gallup Poll is fundamentally flawed. In fact, while Gallup is considered the granddaddy of polls, it is a novice when it comes to daily tracking.

The most obvious defect with Gallup is its polling sample. The fact is, the electorate is split fairly evenly between Democrats and Republicans, with independents, et al. accounting for roughly 20 percent to 30 percent of voters.

Yet, Gallup's tracking poll swings wildly day to day in terms of the percentage of Democrats and Republicans included in its samples.

For instance, Gallup's three-day rolling sample, Oct. 2-Oct. 4, was broken down into 37 percent Democrat, 33 percent independent and only 30 percent Republican. The result: Gore up by 11 percent.

Conversely, the pollster's three-day sample, Oct. 6-Oct. 8, was broken down into 37 percent Republican, 32 percent Democrat, 30 percent independent. The result: Bush up by 8 percent.

So in the space of three days, Gallup showed a 19-point swing in the polls. And it was all attributable to its faulty methodology.

The only Gallup tracking polls that have any credibility are the few that happen to sample a comparable number of Democrats and Republicans. Those balanced polls have shown Bush with a steady 3- to 5-point edge over Gore over the past month.

The Reuters/MSNBC poll has not swung nearly as wildly as Gallup. But it is no more credible, because the poll's director, John Zogby, incorporates an anti-Republican bias into his results.

Unlike Gallup, Zogby controls his daily sample to keep the breakdown of Democrats and Republicans constant. The problem is he builds in a 2-point to 3-point advantage for Democrats.

He claims that his breakdown is based on results of presidential elections past, which show Democrats with such an edge on Republicans.

However, if Democrats truly enjoy a 2-point to 3-point advantage over Republicans nationwide, how does Zogby explain Republican majorities in both houses of Congress and in the governors' mansions?

Then there's the matter of how Zogby weights his raw polling figures.

Thursday on the MSNBC Web site Zogby Thursday claimed that Gore is leading Bush by double digits in the West. Yet, an electoral map accompanying the Zogby's three-day tracking poll shows Bush leading in all but one Western state - California -- with Washington and Oregon too close to call.

So for Gore to offset the lead that Bush enjoys throughout much of the West and to have the double-digit lead in the region that Zogby attributes to him, Gore would have to have a 25-point lead in the Golden State.

And the latest polls in California show Gore with a lead of 5 points to 7 points. So obviously Zogby is cooking his raw polling numbers to make Gore appear stronger out West than he really is.

But why would the pollster do that?

Because, like most of the pollsters tracking this year's presidential election on behalf of the major news organizations (Reuters/NBC News, CNN/USA Today, Washington Post/ABC News, New York Times/CBS News), Zogby is partial to the Democratic Party.

Of course, Zogby and his fellow pollsters insist that, no matter their party registration, no matter their personal political philosophy, they are objective pollsters.

But the fact is, their political biases creep into their polls in much the same way that the political biases of print reporters and network correspondents creep into their coverage of the presidential campaign.

The American public recognized a long time ago that the national media tend to slant toward the Democratic party (consider the presidential endorsements of the New York Times, the Washington Post and the Los Angeles Times). But not until this election year have they learned that the nation's pollsters are no less slanted.

Joseph Perkins is a columnist for The San Diego Union-Tribune and can be reached at Joseph.Perkins@UnionTrib.com.