Frankel: Electoral forecast models prove accurate, expose failed punditry sensationalism

During the weeks leading up to the presidential election, a rather unassuming figure proved to be one of the most divisive names in politics. Political analyst and electoral forecaster Nate Silver was not a candidate, but a statistician blogging for The New York Times.

Critics accused his model of being overhyped and liberally biased, despite Silver ultimately predicting the electoral outcome of every state correctly. Silver eschewed the standard model of election predictions in favor of a sophisticated, mathematical approach, which should become the new standard for electoral journalism.

Silver operates in a field populated by pundits who rely on gut instincts to call elections. Political strategist Dick Morris, for example, predicted former Gov. Mitt Romney would carry 325 electoral votes en route to victory. He based this claim on the assumption that the 2008 election’s turnout was an “excrescence,” rather than on polls and figures.

Pollster Frank Luntz called the election for Romney, writing, “Mitt Romney can win this election. No doubt about it. And it will be because of the first presidential debate on October 3.” Luntz’s prediction severely overestimated the impact of debates on elections.

According to political scientist John Sides, “When it comes to shifting enough votes to decide the outcome of the election, presidential debates have rarely, if ever, mattered.”

On the other side of the coin, commentator Jim Cramer said he predicted that President Barack Obama would receive 440 electoral votes. He offered little in the way of explanation for this dramatic prognostication, other than a tweet saying, “No one is going to recall the guy who picks Obama by 10 electorals if it turns out to be 150 margin. Believe me.”

The president ended up receiving 332 electoral votes to Romney’s 206. Silver went 50 out of 50 in his predictions. Unsurprisingly, pollsters such as Simon Jackman, Markos Moulitsas and Josh Putnam, who also called the election with 100 percent accuracy, relied on aggregated polling data rather than intuition to predict the results.

The method Silver and others used to accurately call the election is built on a systemic approach to poll reading. Silver’s model takes a combination of economic data and a weighted average of polls, accounting for factors that tend to skew standalone surveys.

His findings heavily favored Obama in the weeks leading up to Election Day, prompting critics to accuse him of liberal bias. Such attacks, however, are not merely an indictment of a single Times blogger: they are an indictment of mathematics.

Silver’s model is not perfect. Any prediction, even a correct one, does not prove the infallibility of its model. That said, both Silver’s success – this year and in 2008 when he predicted 49 of 50 states correctly – and the abject failure of traditional pundits to call the election signal the need for a paradigm shift in election coverage. There is no longer any need to listen to political strategist Karl Rove talk about how his “base predictions” indicate a Romney victory.

There are plenty of political issues that are ripe for partisan bickering. The very nature of our political discourse invites it. Leave those issues to the talking heads on cable news and Capitol Hill. As for election predictions, it’s high time we hand those over to the statisticians. You know – the people who know what they’re talking about.