More on Poll Panic: A Little History

A quick tour of liberal blogs seems to indicate that gleeful Democrats are ready to pop their champagne corks and start celebrating an Obama victory, considering the current movement in head-to-head polls. Meanwhile, some Republicans seem caught somewhere between fatalistic resignation and despair. But not so fast. It’s a bit early yet to start moving assets offshore and training our tongues to say “President Obama.”

The current situation: today’s Gallup Tracking Poll shows Barack Obama moving out to an 8 point lead on John McCain, while the Rasmussen poll shows a 6 point lead and the Hotline/FD poll shows a 5% Obama lead. According to the current Real Clear Politics average, Obama leads by 4.3%–although that includes a GW/Battleground Tracking poll that’s a couple days old now and shows McCain up by two.

Discounting the GW/Battleground poll, I think it safe to assume that McCain is currently trailing by something like 5-7 points. As I pointed out in a previous diary, I believe that the last two weeks have very nearly constituted a “perfect storm” for Obama, and the current lay of the political landscape (as a result of Wall Street panic) is very unlikely to remain the front page status quo for the next five weeks.

While it’s obviously better to be ahead than behind, a little historical context about presidential polling is in order.

First, a handful of examples related to the instability of polling in the final weeks of presidential elections.

  • At around this time in September of 2000, a Newsweek poll showed Al Gore leading George Bush by a 52-38 margin. We’ve yet to see ANY poll showing anything close to a 14 point lead for Obama, and we know how the 2000 election turned out.

  • At approximately this point in the race in 2004, polls showed George Bush with a lead in most polls that looked slightly greater than Obama’s current lead. (8 in the CNN/USAT/Gallup, 8 in Pew, 7 in the Battleground, 6 in ABC/WJ). By the time the election rolled around, the race was back within virtually every poll’s margin of error, with a RCP average of +1.5% for Bush.

  • In 1996, the CBS/New York Times final pre-election poll predicted an 18% Clinton victory. This was 9.5% greater than the actual result.

I’d also like to point out that when Rasmussen (my personal favorite pollster) put out his final poll for the New Hampshire Democratic primary in January, he showed Obama at 37% and Hillary Clinton at 30%. Clinton did nine points better than that, winning 39-36. The RCP average for this contest was also wrong–predicting an 8.3 Obama victory. Reuters, for instance, predicted that Obama would win by 13 points. They were fully 10 points off. This raises real questions about Obama’s ability to deliver in actual votes what he’s predicted to receive by the polls.

Of course, past performance is never a predictor of future results. All that I’m sure of is that a 5-8 deficit in September polls is not quite as terrible as some might initially think, considering how they’re appearing in the midst of a very bad economic situation.

Had I only been following the news–without ever peeking at the polls–I would have guessed that Obama was currently ahead by 15%. At least. Considering that he’s not, and that something like 20% of likely voters have still not made up their minds, I think it far too soon to predict with any degree of confidence how this election will actually turn out.