Friday, February 1, 2013

What to Expect When You’re Electing

Stephen AbbottMiddlebury College 

When the national election finally came to a merciful end in November, there was one universally recognized winner whose name did not appear on any ballot. In a stunning denouement, political blogger Nate Silver may have permanently altered the way elections are reported—and run for that matter—and he did so by staking his claim to the veracity of Bayesian statistics.

Like everything else in an election year, Silver’s story is nearly impossible to separate from its heated political overtones, but in this case it is worth a try. Not only was mathematics well served, but its objectivity emerged as a potential means for making headway into the political storms that lie ahead.

Nate Silver’s first statistical love was analyzing baseball, which he did successfully for a sports media company after college, but in the run-up to the 2008 presidential election Silver began applying his mathematical tools to political forecasting. In March of that year he started a blog called FiveThirtyEight and made a name for himself by correctly predicting the outcome of every state except for Indiana in the Obama-McCain race. With its star on the rise, FiveThirtyEight was picked up by The New York Times, just before the 2010 midterm elections. In anticipation of 2012, the Times signed Silver to a multiyear contract.

And this is where the plot thickens. In addition to being a first-rate statistician, Silver is also a self-professed progressive with ties to the Obama campaign. Thus, when Silver’s blog showed Obama with a comfortable polling edge going into the final weeks of the election, attacks from conservative pundits began to fly. Denigrating the messenger is standard procedure in elections, but Silver’s methods—i.e., his mathematics—also became fair game. An L.A. Times editorial characterized the FiveThirtyEight model as a “numbers racket.”

Referring to Silver, MSNBC’s Joe Scarborough proclaimed that “anybody that thinks that this race is anything but a toss-up right now is such an ideologue [that] they should be kept away from typewriters, computers, laptops, and microphones for the next ten days, because they’re jokes.”

Silver’s series of responses make for some pedagogically compelling reading. “There were twenty-two poles of swing states published Friday,” he wrote in a November 2, 2012, post. “Of these, Mr. Obama led in nineteen polls, and two showed a tie. Mitt Romney led in just one . . . a ‘toss-up’ race isn’t likely to produce [these results] any more than a fair coin is likely to come up heads nineteen times and tails just once in twenty tosses. Instead, Mr. Romney will have to hope that the coin isn’t fair.” Silver then goes on to give a razor-sharp explanation of the difference between statistical bias and sampling error and how one accounts for each in assessing uncertainty.

The FiveThirtyEight author’s mathematical rejoinders only agitated his antagonists, who vowed to make him a “one-term political blogger.” But on Election Day Silver’s model was correct for all 49 state results that were announced that evening. And what about Florida, which was too close to call for several days? Silver had rated it a virtual tie.

Predictably, this “victory for arithmetic” was quickly employed as weaponry in the red versus blue debate. This is as unfortunate as it is counterproductive, and here is why. If we can agree on anything in today’s political climate, it is the need for a more productive means of public discourse. If we ignore Silver’s political orientation for a moment, what we have is an illustration of how mathematics, in the proper hands, can provide an objective foothold when the partisan winds start to blow.

What could mathematics, and a mathematical approach that prioritized proof over punditry, contribute to our ongoing debates about climate change? The national debt? The relationship of gun laws to violent crime? What are the chances that some disciplined mathematical analysis might provide an objective first step in bridging at least some of our philosophical differences?

I’d rate it a toss-up.

Stephen Abbott is a professor of mathematics at Middlebury College and coeditor of Math HorizonsThis article was published in the February 2013 issue of Math Horizons.

Image by Randall Munroe (