Who needs Nate Silver? A study says Twitter can predict elections

Last year, statistician (and recent ESPN hire) Nate Silver correctly picked the winner of every state in the 2012 presidential election, earning the enmity of Politico writers and conservatives alike.

But those looking elsewhere for political prognostications can take heart: It appears that Twitter could hold the key to predicting election winners. A new study by researchers at Indiana University found that a candidate was more likely to win if the candidate was mentioned more often on Twitter than his or her opponent.

It was irrelevant if the tweets were negative or positive. It didn't matter if a candidate was the subject of a flood of tweets or only a trickle.

Subscribe to The Week

Escape your echo chamber. Get the facts behind the news, plus analysis from multiple perspectives.

SUBSCRIBE & SAVE

Sign up for The Week's Free Newsletters

From our morning news briefing to a weekly Good News Newsletter, get the best of The Week delivered directly to your inbox.

From our morning news briefing to a weekly Good News Newsletter, get the best of The Week delivered directly to your inbox.

Sign up

The only factor that was important was a candidate's "tweet share" — the percentage of tweets in a congressional district that are about each candidate.

Fabio Rojas, one of the authors of the study, explained why in The Washington Post:

How effective is Twitter at picking winners? The researchers' data correctly predicted 404 out of 406 House races.

Rojas insisted that his Twitter model for predicting elections isn't just a novelty.

He noted that it could be a boon for poorly funded candidates who can't afford to conduct extensive polling, claiming that an analysis of the data can be done by "anyone with programming skills" who "can write a program that will harvest tweets, sort them for content and analyze the results."

This hardly means the end of traditional polling. As Alex Roarty of the National Journal argued, "Professional polling isn't likely to disappear from politics any time soon," mainly because "it's used for more than just the horse race — campaigns test a variety of things with polls, including their message."

For the rest of us, however, Twitter could provide insight into races that the national media doesn't find flashy enough to cover, or in remote districts that never get polled. Rojas theorized that one day an app could automatically sort tweets by congressional district and give you real-time election predictions on your phone.

"Once you start up software for collecting tweets, it's very cheap," Rojas told the National Journal. "It took one of my Ph.D. students a couple of weeks to set it up."