This week we have a special statistical election edition of Masters in Business radio podcast, speaking with Nate Silver of fivethirtyeight.com. Silver received his Bachelors in Economics in 2000 from University of Chicago, and is the author of The Signal and the Noise: Why So Many Predictions Fail–but Some Don’t.
Silver first came to the public attention’s with his shockingly accurate forecasts of political elections, from U.S. Senate to the President. He “ran the table” in 2010, 2012, and 2014 using a statistical methodology adopted in part from baseball. Eschewing the anecdote driven approach the mainstream media embraces, Silver takes a statistical average of all surveys and polls, adjusting for know tendencies and biases. The result has been a stunningly accurate run of predictions in the U.S. (the method has not worked as well in the UK, where election cycles are so much shorter).
He began publishing under a pen name at Daily Kos, before the editors at the New York Times made fivethirtyeight.com part of the Times site. During the 2012 electoral peak, fivethirtyeight was 20% of the NYT’s web traffic.
Be sure to listen to the segment on where we are in the electoral process and what it means for the general election. His insights into The Donald, Bernie Sanders, and Hillary Clinton are profound.
The full podcast is available soon after on iTunes, SoundCloud and on Bloomberg. Earlier podcasts can be found on iTunes and at BloombergView.com.
Next week, we speak with Paul McCulley, former chief economist at PIMCO.
Opinion polls too often skim superficial opinions. To determine a serious level of opinions, maybe polls should ask two questions: How interested are you in the topic of this poll? How well informed? Respondents could assess themselves on a scale of 1 to 10 when appropriate.If 90 percent of respondents aren’t particularly interested in an issue but up to ten percent are highly interested, the opinions of the latter small minority may be more significant than the majority’s.
Those questions would be an improvement on the typical “don’t know”, “no opinion” poll responses,especially with educated respondents who would not be confused by the 1 to 10 scale.