Shameless Pollsters Already Making 2004 Predictions

by Arianna Huffington

The poll, conducted by the Marist Institute for Public Opinion, found that "if the 2004 presidential election were held today" 49 percent of registered voters would vote for Bush, 44 percent would pick Gore, while 7 percent were undecided.

That's quite an "if." Why not just phrase the question, "If the world were to stop spinning and all life were placed in a state of suspended animation, who would you like to see in the Oval Office when you thawed out?"

I'm not even sure what I want for dinner tonight, but 49 percent of "registered voters" -- or, at least, the 742 of them who didn't hang up when Marist's telephonic inquisitors called -- know who they'd support for president 1,371 days from now? Talk about decisive.

And what's the deal with the 7 percent who were unsure? Are they still waiting for the final- final- final Florida count, or are they just a bunch of foot-draggers holding off until they see how W. handles his second week in office before making up their minds?

Still agitated by the vacuity of the "Bush leads Gore in 2004" poll, I put in a call to Dr. Lee Miringoff, director of the Marist Institute, making the obvious point that it's four years until the next presidential election. He didn't miss a beat: "But it's only three years until New Hampshire, and only two years until it's one year to New Hampshire." I can almost taste Steve Forbes' free barbecue at the Iowa straw poll.

What's worse is that this fluff is being reported as if it were a story of importance. Why are grown people wasting their time conducting such meaningless polls? And, more importantly, why do grown people in the media insist on disseminating them as newsworthy? Suddenly, all my suppressed anger about the ludicrous dominance of polls during the last campaign came rushing back to the surface.

In the spirit of forgiveness sweeping the land -- Mark Rich has been pardoned and Ted Kennedy is going to be munching popcorn with George W. in the White House screening room -- I had "somewhat or strongly agreed" to let bygones be bygones and allow old wounds to heal. But if the polling industry is going to keep offering up this garbage, then forgiveness be damned. It's time to remind everyone what the pollsters would rather you forgot -- just how far off the mark they were in projecting this year's presidential election.

According to the polling industry's own data, eight of the 10 national polling organizations got it wrong, failing to accurately project Gore's popular vote victory. Only CBS News and Zogby/Reuters showed Gore leading Bush, while the Harris poll had the race tied. The other seven -- including Gallup/CNN/USA Today, ABC News/Washington Post, and NBC News/Wall Street Journal -- had Bush winning the popular vote, with Battleground/Voter.com tallying a 5-percentage point W. rout.

You'd think an 80 percent inaccuracy rate would be a source of great embarrassment, shame and regret. But apparently you've never dealt with pollsters -- they can explain everything away with that deus ex machina, the margin of error. So, instead of shutting up and doing the honorable thing -- like hiding in Switzerland with Mark Rich -- the polling industry actually went on the offensive, trying to spin its abject failure into admirable success.

In fact, the National Council on Public Polls, an industry trade group, released a study that concluded that the pre-election polls of 2000 were among the most accurate of the last 40 years, "surpassed only by the polls of 1976 and 1960." This pretty much sums up the state of polling: 20-percent accuracy puts you in the Hall of Fame. As proof of this year's sterling achievement, the study touts the fact that while seven out of 10 polls "overstated" Bush's vote, "four years ago, all nine polls erred in favor of overstating Democrat Clinton." Color me impressed.

"It was a remarkably good year," boasted pollster Humphrey Taylor, chairman of Harris Interactive. "The accuracy exceeded our expectations." I'm sure it did. Set the bar low enough, and you can't help but clear it.

In a strange sidelight that calls into question the math skills of the study's authors, the report claims "the 2000 election was a tie between Bush and Gore." Really? I always thought that 50,996,116 was "greater than," not "equal to," 50,456,169. Or maybe their definition of "tie" is "anything within 550,000 votes." And Bush called Gore's math fuzzy?

If the polling industry wants to keep itself busy amusing us with parlor game predictions for 2004, so be it. Let's just not pretend it's political news. And I have it on good authority that Elvis somewhat or strongly agrees with me.