Shining a Light on the Power of Statistics

by Jeremy Loudenback

Discussions of margin of error don’t typically generate full-throated cheers, even at USC.

But most lectures don’t usually feature Nate Silver, the country’s most famous statistician, author, and founder of the popular political blog FiveThirtyEight.com.

Nate Silver - Big Data ... Big Bias?On Friday at Bovard Auditorium, Silver regaled the audience with his usual diverse set of interests in a talk sponsored by the Bedrosian Center as part of the Dennis F. and Brooks Holt Distinguished Lecture series. Drawing together topics such as weather forecasting, poker players, the housing bubble, data-driven electoral redistricting, and the legendary chess match between Garry Kasparov and IBM’s Big Blue computer, Silver demonstrated the ability of statistics to help us understand the world, an idea he also explored in his best-selling book, The Signal and the Noise: Why Most Predictions Fail but Some Don’t.

“We want to see relationships and connections that allow us to make better decisions and predictions going forward,” Silver said. “I’m far from the best communicator in the world, and I’m far from the best statistician in the world, but that overlap of skills is what characterizes FiveThirtyEight. It’s hard to find people who can both communicate and think in a credible way about numbers and data.”

Nate Silver - displaying some Trojan spiritThe theme of communication was an important part of the day. The Holt Lecture series is designed to investigate the connections and interactions between communications and policy, and there is perhaps no more galvanizing figure in the world of political communications today than Silver. Though his accuracy in calling elections has earned him a certain celebrity and notoriety in the Beltway, his work in communicating policy issues through the straightforward use of data and statistics has made him much more valuable to the average citizen. By emphasizing the importance of clear communication of evidence-based results, Silver has had a huge impact on how the country thinks about elections, the value of accurate information, and the role of gatekeepers in the political process. During his run as a writer and editor at FiveThirtyEight.com, he’s made it his mission to separate facts from the bias and spin of talking heads and political insiders that dominate most political discussion.

The Holt Lecture series is designed to investigate the connections and interactions between communications and policy.

“My only bias is toward truth,” Silver said. “When you look at election polling, you can actually be accountable. You verify and say, ‘How objective am I being? How much does my point of view match up to reality?’ People on all sides of politics, including reporters, tend to be very detached from reality.”
Silver’s work with statistics has brought previously arcane topics like Bayes’ theorem to a new audience of media-savvy readers, and he suggested that thinking about statistics requires policy-makers and voters alike to pay closer attention to political data and reports.

“It’s not just about numbers or technique,” Silver said. “You have to think about what you’re really doing in a bigger picture here and also about how do you communicate that to a wider audience.

“Thousands of polls are released in a year. A lot of time if you’re looking at media coverage, it’s the outliers that get the attention. Our goal instead is to find the average, what the central tendency is. The average, as simple as it is, is still the most useful statistical tool ever invented.”

But with the current rash of excitement about the possibilities of Big Data, Silver counseled restraint. With more information now available a lot faster, the need for critical thinking is more important than ever.

“This is the problem with the idea that volume conquers everything,” Silver said. “You have to apply a filter, have some theory, apply some scrutiny. There are terrific insights in the data but not without some human intelligence at the console to tell the signal from the noise.”

Gap between what we really know and what we think we knowWith his work on the FiveThirtyEight blog, Silver has pioneered a new appreciation for data-driven analysis. As the blog moves from the New York Times to ESPN, Silver will continue to provide political analysis about elections and policy issues, but he will also set aside some time to explore the sports analytics that first inspired his love of statistics. And with recent records of data innovation, baseball and politics still offer many opportunities for comparison, though it may be much more difficult to change perceptions about data-driven forecasting in legislative offices than in dugouts.

“Baseball is unique in some ways because there’s a lot of accountability in baseball because there are tangible ways to measure wins and losses,” he said. “You might have a great theory in baseball, but if your team is winning 62 games every year, then you’re not going to keep your job for long. In politics, where you have only one election every four years, it takes much longer to actually see who is skilled at making predictions, and so it’s a much harder history to change.”