Print Friendly and PDF

Tomorrow’s Expectations

This article appears in the 2015 Mackinac Policy Conference edition of the Detroiter Magazine. Click here for more information on the Conference.

Page 58

By Dawson Bell

Nate Silver urges caution for precise predictions.

Nate Silver has a way of driving people crazy. He does it by telling them things they don’t want to hear – and by being right.

The 37-year-old East Lansing native and proprietor of the FiveThirtyEight blog has been doing it since 2007, when he transitioned from projecting the potential of professional baseball prospects to predicting the outcome of elections. He correctly foresaw the 2008 wave for Barack Obama and nailed all 35 U.S. Senate races.

Not much has changed in the interim, as the University of Chicago grad who left a nascent career as an economic analyst with KPMG to play professional poker, has trafficked his brand of data analysis for Baseball Prospectus, The New York Times and ESPN.

Silver’s consistent forecast that Barack Obama would be re-elected in 2012, for instance – even as some individual polls showed GOP nominee Mitt Romney either slightly ahead or tied with the president – made conservative heads explode right up until election day. In the end, the FiveThirtyEight blog correctly predicted the outcome of the election in all 50 states.

Two years later, he panicked Democrats by suggesting eight months before the election that Republicans were more likely than not to regain a majority in the U.S. Senate in 2014. In the end, they did – with room to spare. Of course, as Silver is the first to admit, data and modeling have limits.

At FiveThirtyEight’s current home at ESPN, this year’s NCAA men’s basketball tourney provided some humbling results, as the site’s bloggers pegged Kentucky as an overwhelming favorite, then, when the Wildcats were knocked off, predicted a Wisconsin victory over Duke in the title game. You can be sure Mitt Romney wishes FiveThirtyEight had been equally infallible in 2012.

Despite his reputation, Silver is remarkably frank about how far from precise the science of prognostication is. In his 2012 best-selling book, The Signal and the Noise: Why So Many Predictions Fail – But Some Don’t, Silver calls humility one of the most important attributes for those who seek to define the future. In it, he discusses the “prediction paradox,” that understanding the limits of knowledge – i.e., exercising humility – actually improves accuracy of forecasts.

Silver has said he was drawn to political prognostication, in part, by the glaring absence of humility, or even self-awareness, on the part of so many pollsters and pundits practicing the trade in public at the time. Too many lay people, and even experts, seemed to mistake confident predictions for accurate ones, he said.

Mark Grebner, a veteran political analyst who has run the firm Practical Political Consultants in East Lansing for nearly four decades, said Silver has an astonishing ability to “understand the contours of a problem” behind faulty analysis and fashion a way to address it. “He doesn’t just average out all the poll results available. He thinks through what causes polls to vary,” Grebner said. “He’s simply the most brilliant statistical modeler I’ve ever read.”

More recently, Silver was asked by the Wall Street Journal for his views on potential uses of data analysis in other fields. He suggested, while urging caution about limited or misleading data, that evaluating educational or health care outcomes are both promising. Both, he said, involve issues “with a lot on the line,” and the amount of hard information about results is accumulating fast. But, he said, “You can’t predict things very well if you don’t have definite data on the status quo.” And, as Silver also unfailingly points out in characteristically straightforward prose, things change.

“He doesn’t just average out all the poll results available. He thinks through what causes polls to vary. He is simply the most brilliant statistical modeler I’ve ever read.” – Mark Grebner, Political Analyst, Practical Political Consultants.

In a recent post on the 2016 presidential race called “Is (Jeb) Bush Doomed In The General?” Silver noted the former Florida governor’s exceedingly poor favorability ratings might suggest just that.

“But there’s a particular reason for Democrats to not get giddy,” he wrote. Much of the reason for Bush’s disapproval is because he’s not all that popular with Republicans. That well could change if he nevertheless survives the GOP primary and winds up running against Hillary Clinton.

Silver’s answer to his own Bush question is: not if he gets there. It’s the kind of answer that tends to drive partisans crazy, and the kind Nate Silver has become famous for.