Home | Opinion | Columnists | Mark Mellman

Just like 2000, the polls are off

Where does this presidential race stand? Unfortunately the plethora of public polls does more to confuse than to clarify. Contorting and distorting the data with absurd “likely voter” models only makes matters worse.

Cutting through the clutter, a few facts are clear:

• Averaging the recent polls together gives George W. Bush a one-point lead, 47 percent for Bush to 46 percent for John Kerry.

• About 5 percent of the electorate remains undecided.

• Almost every poll that has released battleground-state results shows Kerry ahead there. For instance, the ABC/Washington Post poll that had Kerry four points behind nationally said he leads by 10 in the swing states.

• Historically, between 70 percent and 100 percent of the late undecideds break toward the challenger.

• This group of undecideds should be no exception. They are much more likely to give Bush negative ratings than the electorate at large and much more likely to say that the country is seriously off on the wrong track.

• Historically, winning incumbents have had 60 percent of the vote at this stage of the campaign. No president has been reelected whose vote did not exceed 51 percent of the two-party vote at this point in October.

But what should one make of all the polls and their varying results? A note today from my Kerry polling colleague Diane Feldman sparked an idea. Let history be our guide.

National polls purport to tell us about the distribution of the national popular vote. Al Gore won that by a half-point margin in 2000. Exactly how much insight did we get from the “likely voter” polls two weeks out in 2000? Check out the list:

Actual: Gore +.5

ABC News 10/24: Bush +4

CBS/NYT 10/21: Bush +2

CNN/Time 10/26: Bush +6

Gallup/CNN/USA Today 10/26: Bush +13 (+7 registered voters)

Newsweek 10/27: Bush +8

Pew 10/29: Bush +4

Tarrance/Lake Snell Perry 10/23: Bush +6

Barrels of ink were spilled analyzing every one- or two-point movement in these polls. Each was thought to reveal a meaningful fact about the forthcoming election.

Though analysts noted that they were only “snapshots” in time, each was interpreted as foreshadowing the ultimate result.

And each offered a completely inaccurate and misleading portrait of what was to happen a couple of weeks later.

Just as a parlor trick, look at this year’s results, “correcting” for 2000 error.

CBS/New York Times was closest in 2000, off by three points. Say they are off by the same three points this year. That would give Kerry a one-point lead. Gallup was off by 13 points.

If history repeats itself, precisely (an unlikely event, to be sure), Kerry will win by five. Time magazine’s 2000 error would also give Kerry a five-point win today. ABC would have Kerry ahead by one.

Admittedly, it is a silly game. But no more foolish than expecting today’s polling results to foretell November’s outcome.

Apply today’s polls to November’s results at your own peril.

Mellman is president of The Mellman Group and has worked for Democratic candidates and causes since 1982, including Sen. John Kerry (D-Mass.) this year.