I already shared with you my excitement about acclaimed statistician Nate Silver (who I’ll have the chance to hear speak at the Online News Association Annual Meeting this October) making the move to ESPN and ABC. Silver appeared on Morning Edition on NPR yesterday and talked more in depth about the move and the role of statistics in reporting.
Here’s the audio:
And some snippets of the interview.
DAVID GREENE: Do you feel like, in all of these types of fields, whether politics, sports, that the numbers have not played a large enough role?
SILVER: It’s a case-by-case basis. I mean, I think certainly that sports coverage, for example, has come a long way since Michael Lewis wrote Moneyball about 10 years ago now. In baseball in particular, almost every team has a statistical analyst on their payroll. Sometimes, the so-called stat head is running the team.
So baseball and other sports are the one place where you’ve seen, I think, already reaching equilibrium. But it’s way behind, I think, still, in coverage of elections, for example. Part of the problem is people are looking for narratives. It’s always a better narrative when, oh, the candidate who’s behind is coming back. Oh, it’s a really close race, down to the wire. But sometimes that isn’t true. Sometimes you have a close, but clear lead, the kind of rooting-for-the-story component, wanting to sell copy – to put it a bit more cynically – kind of goes against what the statistical evidence, what the history would say, I think. And we try and provide clarity on that for people.
Silver got a lot of flack from the journalism community for this stance, and he admitted in this interview that it got to him a little. For the total cynic, Silver’s ideas might have you believe that he believes he’s able to do some number crunching, and have all the answers you ever need in the world about any and everything. But that’s just not true, and he addressed that.
You know, I think I have less of a critique of kind of traditional shoe-leather reporting. I think that’s very valuable. But, yeah, I felt that somewhat became important to kind of put what I was saying in context. And I’m not a guy who says, oh, we can just press a button and predict everything. I’m saying that we have to be more careful about how we weigh information. We have to be more accountable about how we characterize future events and not just kind of flippantly say, oh, it’s a tossup when one candidate’s ahead in most, but not all of the polls. And that led, I think, to some ideological clash.
Greene ended the interview with a really good question that anyone who deals with sports statistics I’m sure has had to answer thousands of times, but Silver supplies a great answer: Things like leadership and other intangibles are impossible to truly quantify, so how much stock should we really put into stats?
We have to be careful here. I think sometimes staties can make a mistake of assuming that because something is hard to quantify that it doesn’t matter. That’s not quite true. In science, you’d want to set up some kind of a testable hypothesis, right? So take all the guys who are seen as good clubhouse leaders and see what happens when they leave. Does a team play inexplicably worse? And my guess is what you’d find is that if you take all these clubhouse leader guys and look at what happens when they’re released or traded or injured doesn’t make a ton of difference. But I don’t know. I think both sides should be willing to contemplate actually setting up an experiment for that.
He can’t get started soon enough.