Nate Silver separates signal from noise at the Athenaeum

Nate Silver separates signal from noise at the Athenaeum

Statistician-extraordinaire Nate Silver spoke to a packed audience at the Athenaeum on Thursday, Sept. 19, about how we look at data and information, sharing insights from his famously accurate predictions of the state-by-state outcomes of the 2012 Presidential election, and conveying some best practices for accurately analyzing and using data and information, separating the “signals” from the “noise.”

Silver began with a history of key technologies such as the printing press, that spurred historical shifts, transforming how humanity dealt with information. He noted how history and prediction are two sciences closely linked, since historical data is often key to predictions. He discussed the changes that are occurring from the emergence of “big data” but warned that large volumes of data do not supplant the scientific method and theory, structure for organizing and understanding of data and information.

In discussing failures of prediction, Silver said not enough robust analysis of data can be one culprit. Another problem in our current age is the rapid speed of data transmission, which can spread false information quickly. These problems can be compounded by the bias of pundits and news sources, also by news organizations that are more apt to report false information in their rush to be the first to report a story.

In discussing the lessons of 2012 and how his organization,, got the Presidential election right, when so many got it wrong, Silver cited three simple steps that led to that accurate prediction:

1)    Averaging all the poll data available—

“Reduce the noise,” he said. “Instead of being distracted by all the outliers here and there, the question is ‘what’s the central tendency.’” According to Silver, news organizations tend to focus on the outliers. “If you can count polls, you are ahead of most of the pundits in America,” he said.

2)    Looking at the Electoral College path to victory for each candidate, counting up to 270– the number of Electoral votes it takes to win the election

3)    “Thinking in terms of probability”

He continued: “A lot of models … are really just putting a lot of simple and intuitive things together in a coherent way.”

His suggestions for accurate analysis:

1)    Think probabilistically

2)    Survey the data landscape

3)    Know where you are coming from—know your own positions, biases, and weak spots when looking at data, since they affect your perception of the information

4)    Try and err

What makes data rich, according to Silver?

5)    Quantity (“large volume of data”)

6)    Quality (“reliable inputs”)

7)    Variety (“observations collected under a wide variety of conditions”)

Nate Silver is one of the nation’s most influential political forecasters and popular statisticians. In 2008, Silver founded his blog,, which provides a data-driven analysis of everything from politics to technology to sports. The site drew acclaim for providing in-depth and accurate political forecasts of the 2008 and 2012 national elections. According to one statistic, on the eve of the national election in 2012, one in five people going to The New York Times website were visiting FiveThirtyEight.

Silver’s influential role in political forecasting was by no means a given for an economics major from the University of Chicago interested in baseball. In 2002, Nate Silver was working as a consultant for the accounting firm KPMG when he developed a new method for predicting the performance of baseball players, which was quickly acquired by the Web site Baseball Prospectus. Using his knowledge of statistics, Silver engaged himself for a time playing professional poker before becoming interested in political predictions. After the 2008 presidential election when he correctly predicted the results of the primaries and the presidential winner in 49 states, his blog became an overnight sensation.

In September 2012, Silver published, The Signal and the Noise: Why So Many Predictions Fail—but Some Don’t.