Longtime sports statistics guru Nate Silver made a name for himself as a political fortune teller when he foresaw the results of the 2008 election that swept then-Sen. Barack Obama into the White House. He has since published a book about the power of data and founded the popular polling and statistics blog FiveThirtyEight. Silver was also one of the few pundits to acknowledge Donald Trump had a fighting chance at winning the presidency in the 2016 election. While the FiveThirtyEight crew was in Boise to record a podcast during Hackfort, Boise Weekly sat down with the data whiz to get his take on pronouncing "Boise," data in journalism and the Trump upset.
You put up a poll measuring how people pronounce 'Boise.' What were the results?
I think only about a third of people got it right. I think some people thought we were making fun of Boise, but we weren't. I was in a playing mood.
You come from a baseball stats background. What are the lessons learned from transitioning from sports to political stats?
The one thing is, people are much bigger idiots about politics than they are about baseball. I think even sports fans who don't think they're data savvy look up statistics all the time. Journalists covering [politics], everything is filtered through this partisan lens. That's quite different.
Do you think there might be more resourceful reporting done if politics were treated more like a sport?
I think a lot of the bad habits that some political journalism has are efforts to cater to partisans or being afraid of being labeled a certain way. That's a dimension that is something I've noticed.
How do you use the data analysis you provide to tell compelling stories?
We want to tell true stories, right? That's the first priority. We also hope they're compelling. We spend a lot of time on our editing process, trying to make sure the truth is as interesting as fake news or alternative facts, potentially. There's all this concern now about being truth-driven. Well, we think the way to be more truthful is to be more empirical, which means experience and being evidence-driven. It means being transparent and not relying so much on inside information where you can't show the reader where you're coming to your conclusions from.
Are you a journalist?
Sure, why not? We do a lot of traditional reporting. Half our stories involve some sort of reporting, which means talking to and interviewing people, but I don't understand why that type of reporting is held up on a pedestal—and particularly that type of reporting where you're just talking to insiders. The thing is that if you have a data set you're looking at, first of all, it could be a different way to have more people represented. Polling is just a more scientific way to interview people than going out into the field and talking to people at a rally or something. There's value to doing both, but we've seen lots of times where a reporter will have a prejudicial view of a story and interview people in a way that confirms those prejudices, whereas polling is a way to interview thousands of people in a way that's more objective and is a check against whatever the reporter's bringing to the table.
What about the fracas over polling and the 2016 elections? How and why did so many people get that so wrong?
I don't think the pollsters got it that wrong. We'd flowed from periods where [Hillary] Clinton had leads of six or seven points and periods where it was quite competitive in part because, at least according to our math, Clinton had an Electoral College disadvantage: Because her vote was concentrated in coastal, urban areas, she could do fairly well in California and Massachusetts and places like that, and really struggle in the middle part of the country. We thought that was clear even though it went against conventional wisdom. The fact that she was struggling so much in Iowa and Ohio should have been a tipoff that Michigan, Wisconsin and Pennsylvania were a problem for her. Her coalition was, like, perfectly engineered to be inefficient in the Electoral College. And also the fact—and I'm in run-on mode here—voters could dislike Trump and vote for him anyway because it was a change election and people were fed up with having the same party have the presidency for eight years.