Politics, polls and PECOTA: the story of Nate Silver
    Nate Silver speaks at South by Southwest in 2009. Photo by JD Lasica on Flickr. Licensed under Creative Commons.
    Nate Silver speaks at South by Southwest in 2009. Photo by JD Lasica on Flickr. Licensed under Creative Commons.

    The Huffington Post knows who really won Tuesday’s election. No, not Barack Obama – the real winner of this year’s presidential contest was Nate Silver: poll aggregator, controversy magnet, potential witch and Lord of the Algorithms. “Nate who?” you may ask. Well, imaginary reader, pull up a chair, because it’s time you got learned on math.

    Who is Nate Silver?

    The story of Nate Silver goes back to the 2008 presidential election. Offline, he ran PECOTA, a baseball program that used a player’s career statistics to compare him to past players of similar age and skill, and then projected his future success based on those numbers (think Moneyball). Online, Silver was an anonymous blogger by the name of Poblano, who applied his knowledge of baseball statistics to another sport dominated by corporate interests and old white men: American politics.

    On his blog, FiveThirtyEight, he aggregated the results of various national and state polls and weighted them according to sample size, when the poll took place and various other factors. He then applied some math to project the results of the presidential race, as well as Senate races nationwide (House races had less polling, and therefore received less analysis). He rocketed to success following the election, which he had correctly called for Obama as early as March.

    Following the election, he got snapped up by the New York Times, which was very eager to acquire a poll aggregator of its own. He performed very well in the 2010 midterm elections, and midway through last year began handicapping the 2012 presidential race.

    What did he have to say about this year’s election?

    Silver’s stance was fairly consistent throughout the primary process and election: Though the conventional wisdom on the economy and unemployment indicated that Obama ought to lose, the polls and actual numbers, especially in swing states (especially the most important of swing states, Ohio), said he would win. Because Silver’s model is built primarily on numbers and not opinions, his election forecast predicted an Obama victory as early as March.

    On Election Day, Silver was projecting a 91 percent chance of an Obama victory (the remaining 9 percent mostly accounted for the fact that the polls could all have been collectively wrong), with the most likely outcome being that Obama would receive 332 electoral votes. These projections basically came to fruition in the actual election, as he picked every state correctly and nailed the electoral vote count.

    So what made him so controversial?

    Silver was consistent in predicting an Obama victory, and the probability of an Obama win in the FiveThirtyEight forecast grew higher and higher as the election neared. Some pundits, who were convinced the race was a tossup, were not enamored of that model. In particular, Joe Scarborough of MSNBC’s Morning Joe took issue with Silver’s model. “Anybody that thinks that this race is anything but a tossup right now is such an ideologue, they should be kept away from typewriters, computers, laptops and microphones for the next 10 days, because they're jokes,” Scarborough said.

    Silver, miffed at having his professional work questioned by people who did not have formal training in mathematics, responded with impulsive bets, hilarious tweets and extended analysis of why his opponents were wrong. Mainstream news sources quickly picked up on the feud, and in the week leading up to the election, it became a fairly popular topic. “Nate Silver vs. the pundits” became a major narrative of the election.

    But Nate Silver won that contest, right?

    According to the internet, yes. The truth of the matter, of course, is probably somewhere in-between.

    The media narrative surrounding Nate Silver focused on qualitative analysis versus quantitative analysis. Think of it like the World Series: Most sports journalists predicted a win for the Detroit Tigers based on their strong fundamentals, especially their pitching (a quantitative approach), while a few holdouts said the San Francisco Giants would win because they are scrappy underdogs who seem to always come from behind (the qualitative approach). The two of you reading this who also watched the World Series will know that in this particular case, the qualitative approach was right.

    Such a model of conflict, however, may not be entirely accurate. Dr. Georgia Kernell, an assistant professor in the Weinberg Political Science department, says that the qualitative versus quantitative narrative ignores the fact that pundits were not claiming that Nate Silver’s use of math was somehow inherently wrong. What they were saying was that the polls Silver relied on were missing key points of data. “It’s not so much qualitative versus quantitative, but ‘we are going to trust the polls’ versus ‘there are outside forces the polls haven’t taken into account yet,'" Kernell says.

    Professor Yanna Krupnikov, also an assistant professor in the Political Science department, agrees. To her, there is no real contest between the math-driven analysis of Silver and those like him and the gut-based predictions made by pundits like Joe Scarborough. The election itself, after all, is not driven by heart, guts and momentum. It is a decision made by a quantifiable metric: votes. “You can’t bypass math. At some point, it has to get to math," Krupnikov says.

    So does any of this matter?

    For the next couple of weeks, you are probably going to keep hearing about Nate Silver. He was a major story of the election, after all, and newspapers will do their follow-up pieces talking about his “victory."

    Krupnikov, though, thinks the media is probably going to drop the story once something more interesting comes along. Silver, in her view, was less important for helping the media decide who “won," and more for the endless reams of copy his story could fill. “In the last 30 days of this election, there wasn’t that much else to talk about other than Nate Silver predicting things," Krupnikov says. "It was Hurricane Sandy, Chris Christie or Nate Silver. It filled a gap in coverage."

    Perhaps the most important takeaway from the controversy surrounding Nate Silver is that numbers and math do matter. Though we can dispute the “momentum” of a race, or who won the news cycle, there are quantifiable ways to examine who is leading, what voters think and what is actually going on.

    “We can argue about whether or not things are true," Kernell says. "But there are ways to test whether or not things are true."


    blog comments powered by Disqus
    Please read our Comment Policy.