Nate Silver's Election Predictions a Win for Big Data, The New York Times
Blog Shows Power of Predictive DataBy: John McDermott, Published: Nov 07, 2012
Despite many claiming it was a horse race, the networks began calling yesterday's presidential election for President Barack Obama shortly before 11:30 p.m., only several minutes later than the media outlets did in 2008.
Outside of politics, the biggest winner may have been Nate Silver, the economist turned New York Times blogger -- who was also on our 2009 Creativity 50 list -- whose algorithms correctly projected the presidential vote in at least 49 states before a single ballot was cast. (The tally could include all 50 if Obama wins Florida, where the ballots were still being counted at press time.) Just before the election, Mr. Silver put the probability of an Obama re-election at 90.9% .
As of Monday, one in every five visits to the Times website included a visit to the FiveThirtyEight blog, a Times spokeswoman said. Expect that ration to rise once the data is crunched for Election Day.
For many, the accuracy of Mr. Silver's projections, which appear on the Times blog FiveThirtyEight, signal a shift in how the media covers elections. Rather than reporting on nebulous aspects of a political race such as "feel" and "momentum," Mr. Silver's analysis relies solely on statistics.
"It's not just about Nate Silver, but the use of statistics in general and to express prediction in terms of probabilities," said Andrew Lipsman, VP-marketing and industry analyst at digital analytics company ComScore. "It's interesting because so much of how the election moves is driven by the media narrative and in some instances it can be a self-fulfilling prophecy. The Romney momentum narrative didn't exist."
Mr. Silver first gained fame as a prognosticator in 2008 with his self-run blog FiveThirtyEight.com. The site's traffic skyrocketed during that election period from 34,000 unique visitors in August to 169,000 in September, 426,000 in October and 436,000 in November, according to ComScore.
While most pundits were focusing their coverage on the results of a single poll, Mr. Silver developed his own methodology that aggregated a host of poll results and weighted them based on their historical accuracy. Those numbers were then used to determine a candidate's probability of winning in each state. Mr. Silver's method correctly predicted the presidential voting outcome in 49 states in 2008.
Mr. Silver's site is now hosted by the Times where his traffic (and notoriety) have continued to climb.
Mr. Lipsman said Mr. Silver's move to the Times has been a mutually beneficial arrangement.
This September, the Times' FiveThirtyEight blog attracted 1.2 million unique visitors, more than seven times the traffic his site received in September 2008, according to ComScore. Current traffic numbers were not available, but visitors likely increased as the election neared and Mr. Silver was criticized by others in the media.
Joe Scarborough, host of MSNBC's "Morning Joe," called Mr. Silver an ideologue in October for predicting that Mr. Obama had a 73% chance of winning the election. Dylan Byers, media reporter for Politico, also questioned the accuracy of Mr. Silver's predictive models, saying he "could be a one-term celebrity."
But it turns out he won't. If anything, Mr. Silver's star continues to rise while the importance of qualitative political analysts has come into question.
For Mr. Lipsman, Mr. Silver and other statistics-driven political analysts will become increasingly popular in subsequent election cycles.
"Now that people have seen this proven over a couple of cycles, people will be more grounded in the numbers," Mr. Lipsman said. "I think the prognosticators that ended up being the most [off the mark] had a partisan agenda."