Finding a Silver lining
The 2012 presidential election was groundbreaking and unique in a lot of ways. Ignore the super political action committees , the voter scamming and the ill-timed hurricane — the biggest game-changer this year was a New York Times affiliate named Nate Silver.
Prior to election day, Silver predicted a close but secure win for Obama on his blog, fivethirtyeight, with a close popular vote and a larger gap in the electoral college. This isn’t a remarkable prediction — a lot of pundits were predicting a close Obama victory. What is remarkable is his accuracy. His predictions on the morning of the election were correct for every state and for 31 of the 33 Senate races.
Fine, he did well. Predicting elections is what pundits and poll analyzers do. Let’s get into the underlying mathematics a little.
Silver’s specific methods are a close-kept secret. However, by weighing the results of several polling agencies and averaging them, he comes up with statistical odds for the results in each state. As part of his analysis, he comes up with margins of error and a confidence interval — effectively, the odds he’s wrong.
What this means is Silver was, in this case, 95 percent confident that the results of the election would fall within the margins of error of his predictions. Accommodating for the 8 percent chance he gave to a Democrat win in North Dakota and the 34 percent chance to a Democrat win in Montana, he was 96 percent correct with his predictions. Silver predicted the 2012 elections and the accuracy of his predictions.
Silver is not well-connected within politics. His previous notoriety came from applying his statistical aptitude to baseball. Critics such as Geoffry Dunn of the Huffington Post have made their disdain for his lack of experience within politics very clear. Yet all of the election predictions from the well-connected were less accurate than Silver’s.
This is why Silver represents what is hopefully a new direction in election reporting. Instead of looking at macroeconomic trends, approval ratings, the word from party heads or his own gut, he looks at the polls and does his math with emotional distance from the results. Silver saw what was coming with nothing but cold mathematics.
Compare him to Karl Rove. After Fox News gave Ohio to Obama with around 73 percent of the vote in, he insisted several traditionally-Republican counties weren’t in yet and it was too soon to call.
Fox newscaster Megyn Kelly asked him if his response was “just math you do as a Republican to make yourself feel better or … real?” Kelly then went downstairs to the station’s results analysts and asked them to explain why Ohio was going Democrat.
That few minutes of television should act as a tombstone for belief, instinct and gut feeling-based election analysis. Calling polls biased across the board without a solid explanation reeks of willful ignorance.
For a person trying to predict the results of a contest of some sort, willful ignorance is a terrible thing. Unpleasant news must be investigated and either substantiated or debunked if an accurate prediction is to be made.
The point of all this is Nate Silver was diligent and it paid off. Wouldn’t election season be nicer without all the wild, flailing gesticulation and party-line predictions? When pundits disagree, they might explain why one position or mathematical model is more valid than the other. And maybe, just maybe, they might admit to missing something.
For those like Rove and Dunn, Silver’s statistically-grounded success is a black cloud over the sun. Maybe for those of us who’d like a little less from their “guts” every two years, it’s a silver lining.