Campaign Crawlers

Is Obama’s Victory a Statistical Certainty?

Polls are not predictions, and elections aren't baseball.

By 8.6.08

Send to Kindle

Analogies from sports are often employed to describe the competitive clash of politics and I've used a baseball metaphor myself to describe the current presidential campaign.

However, my eyebrows were raised when I learned that Internet poll guru Nate Silver of FiveThirtyEight (538 = total number of Electoral College votes) comes to political punditry from a background of baseball statistics. (See the Newsweek article here.)

The real problem with this is that, unlike the cold reality of baseball statistics, campaign polls are a flawed instrument that yield only an approximation of public opinion at any given moment. Pollsters like to say they provide a "snapshot" of current opinion, and the snapshots are often quite blurry.

Also, unlike a baseball game, the outcome of elections is dependent upon the reactions of the spectators.

The results on Nov. 4 may very well match Silver's projections (currently 295 EVs for Obama, 243 for McCain), but this is merely the extrapolation of current numbers over an imaginary 90-day event-free path to Election Day, which cannot be assumed.

INDEED, IF WE COULD could extrapolate the current trend, John McCain would be a shoo-in for the White House. Silver's poll "super tracker" shows that, since peaking near plus-5 on June 19, Obama has declined to plus-2 in less than six weeks. If the "super tracker" were a popular-vote measurement, and this six-week trend were projected over the next 13 weeks, then Obama would be minus-4 on Election Day -- with McCain getting 52 percent of the two-candidate total to Obama's 48 percent. Such a popular-vote margin for McCain would be as large as Bush's 2004 win over Kerry, and would almost certainly mean McCain topping the 270 EVs needed for victory.

That's only an extrapolation, however, and cannot account for whatever real-world campaign events will transpire between now and Nov. 4. The most likely scenario is for a seesaw struggle all the way to Election Day, with Obama winning by a narrow margin. Yet a complete electoral meltdown is possible for either candidate. (Possible, I said, which isn't the same thing as likely.)

What team Obama is counting on is that the situation will remain at least competitive for their candidate going into Nov. 4, so that their campaign's acknowledged superiority in get-out-the-vote operations will make the decisive difference. Yet there are any number of scenarios that might render that advantage moot, should Obama suffer some scandal or blunder that causes his support to collapse in the interim. If Halloween arrives with Obama so unpopular that he's on the losing end of a double-digit poll disadvantage, all the canvassing and phone-banking in the world won't elect him.

That's exactly why the fallibility of poll data puts an asterisk beside any summer projection of November results. Early polls have traditionally favored Democrats by margins that fail to predict the final result. Michael Dukakis led by 17 points in July 1988 and got trounced on Election Day. The July polls could not predict events like the Democrat's idiotic tank ride or his politically tone-deaf response to Bernie Shaw's rape question.

For more than a week after Obama's return from his nine-day foreign trip, Nate Silver resisted the evidence that the momentum of the election was swinging against the Democrat. Yet events were intervening to disrupt the statistical clarity.

THE McCAIN CAMPAIGN had rolled out ads that hit hard on the energy issue, Obama's canceled visit to wounded troops, and the grandiosity of his Berlin speech. An admitted Obama supporter, Silver evidently shares the campaign's view that these were unfair "attacks," and last week declared that the "decline in Obama's numbers...has halted -- and has possibly begun to reverse itself."

Instead, the decline continued. The day after Silver's declaration, Gallup's Gallup's daily tracking poll showed the race deadlocked at 44 percent for each candidate, and by Monday, the Rasmussen daily tracking poll showed McCain inching ahead -- "the first time McCain has enjoyed even a statistically insignificant advantage of any sort since Obama clinched the Democratic nomination on June 3."

Silver was compelled to admit the latest polls "indicate some tightening in the race" -- quite an understatement, considering that Obama had seen a one-time 9-point Gallup lead evaporate in the span of a week.

Silver's concept can be statistically predicted based on past performance. However, the notion of Obama's inevitability is based on the narrative of his primary campaign as a glorious triumph. That narrative, however, fails to account for contradictory evidence: The overconfidence that led Obama to vacation in the Virgin Islands while Hillary was stumping all over Pennsylvania, his weak performance in the April 15 debate, and the late-primary fade that left his pledged delegate count 300 short of a nominating majority

Does the past predict the future? As Republicans this week brandish tire gauges as a symbol of Obama's clueless claim that tire pressure is as important to the gas crunch as increased oil production, the question Silver should be asking is, "Which past are we talking about?"

Obama may yet win the election, but only if he can avoid the historic bumbles that have hurt so many of his Democratic predecessors. So far, the rookie's prospects aren't promising.

Like this Article

Print this Article

Print Article
About the Author

Robert Stacy McCain is co-author (with Lynn Vincent) of Donkey Cons: Sex, Crime, and Corruption in the Democratic Party (Nelson Current). He blogs at The Other McCain.