Skip to content

A Silver Lining in the Endless Analysis of White House Polls

Nate Silver knows he won’t be as popular after the presidential election as he is now.

The Chicago-based Silver runs the blog FiveThirtyEight.com, where he collects and analyzes political polls. Thanks to the 2008 presidential election, he is rapidly gaining prominence, averaging more than 250,000 page views a day since he launched the site only eight months ago. The site’s name refers to the number of electoral votes at stake in the White House contest.

Silver’s timing couldn’t have been better, with a prolonged fight for the Democratic nomination, a hard-fought general election and a baker’s dozen of competitive Senate races.

An average FiveThirtyEight post will list the most recent polls of the day from both well-known polling companies and more obscure ones. And while there are plenty of polls out there, it’s hard to discern their accuracy, bias and trends. Silver attempts to fix that by weighing more heavily the polls that have been more accurate historically, as well as the polls’ sample sizes and recentness.

The math behind the polls is complicated, but Silver has a good grasp of the mechanics and can explain clearly why, for instance, one poll drastically favors one candidate over others.

Silver, 30, originally started the site to vent his political musings.

“It just developed into something organically, which is, frankly, a lot of time right now,” Silver said. “But it wasn’t planned from the outset. It was supposed to be kind of, ‘Hey, I’ll maintain this blog and get 500 hits a day,’ and it turned into something a lot bigger.”

This is Silver’s second successful blog. He made his name on the Internet by inventing PECOTA, an algorithm that predicts baseball players’ performance that is used by Silver’s employer, Baseball Prospectus, a think tank. Though Baseball Prospectus brought Silver, who graduated from the University of Chicago with honors in economics, prominence within the baseball world, most of his fame can be attributed to FiveThirtyEight.

Today, Silver has a fairly full schedule. He runs FiveThirtyEight, which makes him a little money, continues to work for Baseball Prospectus, blogs for the New Republic, and has been a guest on a number of radio and television programs, including “The Colbert Report” on Tuesday.

That’s a lot for one man’s plate, so Silver employs fellow blogger Sean Quinn, whom he met online at a major liberal blog months ago. Although Quinn is not a statistics expert like Silver, he has spent much of the past few months on the road reporting on political campaigns for FiveThirtyEight. With only a month left until Election Day, Quinn says his and Silver’s goal for the site is to extend it beyond the elections and make it the premier venue for the political numbers game.

“Basically, the two of us are trying to make this thing last,” Quinn said. “We know it’s going to last for at least another four years, until 2012 for sure.” The goal is not only to grow the site by bringing in readers, but also to teach them something new, he said.

“Teach” is a good word to explain the style of Silver’s and Quinn’s posts, because they read almost like a college textbook.

“Just because a poll is an outlier doesn’t necessarily mean that it’s doing something wrong,” Silver writes in a post. “Pollsters may have legitimate reasons for having a different perspective on the election, and they may also occasionally produce odd results due to chance alone.”

It’s a good tone for Silver’s audience. He aims FiveThirtyEight at people who aren’t political insiders, which has been relatively easy so far with the extremely high interest in politics thanks to the presidential election.

Silver will focus more on House and Senate races in the next election cycle. Analyzing House race polls has been a challenge so far, Silver said, because “the polling data is much more sparse and you get a lot of internal polls from the campaigns who have an incentive to say, ‘Hey, we’re doing good so you should donate money to us instead of somewhere else,’ or vice versa.”

Even now, Silver regularly posts Senate polling updates, which list polls on each Senate race. These polls offer less analysis than Silver’s presidential posts, where he occasionally injects his own opinion as a Democrat supporting Sen. Barack Obama (D-Ill.).

It’s hard to say which is more appealing for the average Internet surfer, the gathered polls or the analysis. But the site is attracting the attention of professional pollsters.

Jay Leve, chief executive officer of the automated polling firm SurveyUSA, which conducts some polls for Roll Call, visits FiveThirtyEight to compare his company’s polls to others. He compares the site to a newspaper because it offers not only news and analysis, but also opinion.

Silver “doesn’t bring any prejudice or bias. He just puts it in a database,” Leve said. “It just goes into the sauce, he runs the numbers, he spits out the pollster rankings.”

Bloggers hold Silver’s analysis in high regard. Reihan Salam, an associate editor at the Atlantic magazine and a blogger at political and cultural review Web site TheAmericanScene.com, wrote in an e-mail that Silver is careful to keep his personal opinion and analysis separate.

“I think Nate Silver is a dazzlingly intelligent guy,” Salam said. “He is always very careful about numbers, his political allegiances never overwhelm his analytical powers and he’s an engaging read. The blogosphere is overstuffed with opinionated junkies; Silver knows how to present data in attractive and compelling ways, which is a rare strength that I hope we’ll see more of.”

Still, Silver’s reputation hinges on this election in many ways. As of Wednesday, FiveThirtyEight gave Obama a 90.5 percent chance of beating Republican nominee Sen. John McCain (Ariz.).

A week before McCain withdrew his presidential campaign from Michigan, Quinn and Silver predicted that Michigan had lost its status as a swing state.

Still, even if they accurately predict all the presidential and lower-ticket races, Silver expects the site traffic to decrease after Election Day.

“I’m expecting a substantial reduction, whether it’s 80 percent or 90 percent, for sure,” Silver said. “I will say though that I think there will be more interest in politics — I mean, there’ll be more going on with the tumultuous nature of the economy and the other kind of trouble our country has gotten itself into now, but also because both these presidents will get more done than a president with a 30 percent approval rating.”

Recent Stories

Five races to watch in Pennsylvania primaries on Tuesday

‘You talk too much’— Congressional Hits and Misses

Senators seek changes to spy program reauthorization bill

Editor’s Note: Congress and the coalition-curious

Photos of the week ending April 19, 2024

Rule for emergency aid bill adopted with Democratic support