With six days to go, the list of battleground states seems to be growing ever smaller. Of the 11 states identified as up for grabs by the New York Times on Sunday, the conventional wisdom has it that Nevada and Colorado are likely to go for President Bush, while John Kerry looks good in Pennsylvania, Michigan, Minnesota, and New Hampshire. That leaves just five “true” battlegrounds: Ohio, Wisconsin, Florida, Iowa, and New Mexico. This year’s Florida, pundits say, is likely to be one of those five.

It’s hard to blame journalists for predicting which states are still up for grabs — especially when the candidates themselves seem to have come to a similar conclusion. But a look at some poll results from this stage of the 2000 race suggests that those who make bold predictions about which states will be close could end up eating crow on November 3.

On November 1, 2000, six days before voters went to the polls, the Pittsburgh Post-Gazette reported on a Zogby tracking poll that looked at 10 battleground states. It had Al Gore winning 284 electoral votes, to Bush’s 254.

That wasn’t too far off the mark — a few hundred more Gore votes in Florida would have yielded a result very close to that. But here’s what’s interesting: the Post-Gazette reported that, “Pollster John Zogby maintained an air of caution in making a prediction based on his poll results because several states — Pennsylvania, Michigan, Missouri, Tennessee and Washington — are so close.”

So six days out, Zogby singled out those five states as particularly close. Zogby was wrong. The five states where the vote was the closest turned out to be Florida, New Mexico, Oregon, Wisconsin, and New Hampshire.

The following day — just five days out — the Post-Gazette reported on a new round of Zogby results, telling readers: “Ground Zero in the race for the White House may be shifting west to Illinois, Wisconsin and Missouri, a new daily tracking poll shows.” Meanwhile, Florida looked to be in the bag for Gore: he led 51 to 39.

Wisconsin did indeed end up being agonizingly close, but Bush won Missouri by four points, Illinois ended up going for Gore by 12 points — a virtual landslide — and Zogby was off by 12 points in Florida, where we all know what happened.

Even polls released just hours before voting began failed to accurately predict the closest states. A Harris Poll released on election eve found Gore leading the electoral college, and showed just one state, Washington, too close to call. As it turned out, the vote in 14 states was closer than it was in Washington, which went for Gore by 50 to 45.

The point here isn’t to castigate pollsters; it’s no secret that polls differ in methodology, and can and do fluctuate wildly. (And to be fair, Zogby’s polling in the final two or three days of the race proved much more accurate than those conducted a week out.) It’s simply to point out we don’t have a clear idea in which states the vote will be closest on Election Day — no matter how much pollsters and pundits may wish we did. And also to note that, at a time when many observers seem confident that only a few states have the potential to become this year’s Florida, if the year 2000 is any gauge, any of ten or twelve states could fit the bill.

Thanksgiving in Honolulu, anyone?

Zachary Roth

Zachary Roth is a contributing editor to The Washington Monthly. He also has written for The Los Angeles Times, The New Republic, Slate, Salon, The Daily Beast, and Talking Points Memo, among other outlets.