<
>

Presidential Fantasy League: An interview with Nate Silver

Man cannot live by bread alone, but bread and politics? Maybe. Getty Images

Where the avenues of sports and politics meet, there is a mailbox where we deposit a weekly letter; a mailbox called the Presidential Fantasy League (see every PFL ever right here).

If one person can be called an "oracle" at this exact moment in our country's history, it might be Nate Silver. The Chicago resident who made a name for himself writing for Baseball Prospectus recently turned his attention to politics and launched his own blog, fivethirtyeight.com, where he makes election predictions. More than once during the Democratic primaries he shocked pundits by out-predicting the national polls. Since then he's become a hot commodity, appearing on CNN and MSNBC and having profiles written about him in The Wall Street Journal, Newsweek, and New York Magazine. We caught up with him by phone while he was waiting in a TV studio to do a spot on the Charlie Rose show. Like we said, guy's in demand.

MAG.COM: So what's the similarity between sports and politics that makes the same prediction models possible for both and why did you start predicting politics and not weather or something?

SILVER: Any one game in baseball doesn't tell you that much, just as any one poll doesn't tell you that much. In that sense they have similar pacing and both require you to look at the big picture. One of our goals at Baseball Prospectus was to build a better kind of mouse trap and be better able to critique the media narrative. For instance, we like on-base percentage better than batting average; it tells you more about how a team creates runs and wins ball games. The same thing is true in politics where a lot of the analysis is so superficial and there is no one to hold the media accountable—that's certainly part of the goal.

One difference, though, is that in politics you have polls and the polls aren't very good, so it's more about cleaning up the data. In baseball you have terrific data and you can be a lot more creative with it.

How effective are polls in general at predicting elections?

They usually get the winner right. But in the Democratic primaries the average poll was off by seven points, and they claimed to only be off by three points. That's a pretty big difference—a touchdown versus a field goal. The problem is that when polls are wrong, they tend to be wrong in the same direction. If they miss in New Hampshire, for instance, they all miss on the same mistake.

What are some "tell signs" people can look for when reading a poll to tell how accurate it is? What's the first thing you look at?

If you track polls for a long enough time you'll notice some polls follow the national trend better than others. They just behave intuitively. Others that kind of jump around every which way you should be more skeptical about. You don't want an Obama poll that rises 10 points when the Jeremiah Wright news comes out. Just last week there was a poll that came out that had John McCain winning young voters 75 to 25. That just doesn't make any sense. They probably weren't asking enough young people. Those kind of things can be a tip off.

What about sample size and polling methods?

All else being equal, the bigger the sample size the better. But if you have poor methodology you could sample all the people in the world and it would still be bad. For instance, these Internet polls conducted by Zogby, they'll often have 20,000 people in their sample but they're terrible. They get people volunteering to to take the poll instead of a random scientific sample.

What's the single most shocking trend you've noticed since you started doing politics?

We'll know more after Election Day, but I'm surprised at how Obama has apparently been able to turn some red states purple and possibly blue—states like North Carolina and Indiana. But it's a reminder also that we get so locked into the 2000 and the 2004 map that we forget that this thing does change now and then. The 2000 map looks different from the 92' and '96 maps. And those maps look different than '88 and '76. I guess I'm surprised that North Carolina is as competitive as it is, but if we look at history it shouldn't be such a surprise.

This year we're seeing athlete endorsements becoming more Democratic than in the past. We've noticed that race, more than anything else, seems like the biggest determinant of which athletes are going to vote for who in this election. Is that trend reflected in the general polling population? Is income still a better determinant of how people will vote, or has race overcome it?

Race is still the No. 1 determinant in every election. The income demographic has really changed though. Income used to help the Republicans, but now it appears that higher income people vote Democratic by and large. It's the exact reverse of what it used to be. And that's why Obama is performing well in states like Virginia and Colorado, which have high income and are highly educated, and not as well in states like West Virginia or Arkansas, that were good for Democrats in past years as recently as the Clinton era. I think to be younger and hipper is part of it too. Race might be part of the reason why Obama's getting some of these athlete endorsements, but it's also because he's not a stodgy, stuffy politician. It's easier for people under the age of 50 to relate to in certain respects.

Have you gotten a sense that endorsements, in particular those from athletes and celebrities, effect public sentiment for a candidate? Are there any reflections in the polls? Can't we give LeBron at least a little credit for Obama's edge in Ohio?

In general, endorsements don't matter. If they do matter, it's more a Colin Powell endorsement more than LeBron James. We had those two primaries in Indiana and North Carolina, which are the two craziest basketball states in the country, where Obama played the basketball card, which I thought was kind of smart. But in general I don't think these endorsements make a heck of a lot of difference. Colin Powell might be worth a few votes at the margin, and he's a bigger name for most political watchers than almost any athlete.

What World Series match up, historically, would you compare this election to in its current state?

Right now it's looking like kind of a sweep. Maybe last year's Series between the Red Sox and Rockies, where you had a great regular season, a great finish and then a sort of anti-climactic World Series. It almost feels like Obama has pulled away enough that we might not have that fight to the finish. He might just kind of walk away with it.

In general, McCain's supporters in the sports world have been extremely wealthy, older, white upper management types. Obama's have been younger, wealthy—but not as much, non-white, and often basketball players. Which would you say is closer to representing the current voting demographic of this country? Are we more Obama or McCain camp?

I think the traditional demographics are changing. It's become a generational thing. There are a lot more Hispanic people now and they tend to be younger, likewise with African-Americans. I think it'll be like 2035 or something where the majority-minority of whites is going to flip. In California that's already the case. A campaign doesn't want to have what I call the Readers Digest problem, where all the subscribers are like 65 years old and are dying off. In the Republican Party support is concentrated among people age 50 and over, and they're not going to be around as long as the 18 to 29 group that came of age during the Bush era and have a negative impression of the Republican brand. This is a problem for them going forward.

Do you think the 30-minute ad Obama ran before the World Series helped him or hurt him?

A lot of people watched it, which was surprising to me. Something like 35 million people watched it. At first, when I heard it was going to come before the World Series I thought it might be a dumb idea. But it didn't have the feel of an ad, it seemed more like a short documentary. If they had done something cheaply and it had been more of an attack ad, I think it would have hurt them.

What should we look out for before Tuesday?

Three states: Pennsylvania, Colorado, and Virginia. If McCain can get the numbers to a tie in two out of those three states, or get at least within the margin of error in the polls, then he might have a chance to eek out a really narrow electoral victory. If he can't do that then I don't think he has any path to a victory. If he can do that he'll have gotten himself to the point where he can try and recover the onside kick.