Public opinion polls are central to the political process, but now there is a new contender in predicting the winners of many races — Twitter.
An Indiana University study found that Twitter could be a key means of determining the results of political contests. A candidate’s “tweet share,” the percentage of total tweets mentioning a specific candidate compared to that of their opponent, is an accurate prediction of their chances of winning.
In examining the 2010 mid-terms and the 2012 elections, researchers found tweets correctly forecasted the outcome in 404 of 406 House races.
Surprisingly, it was not the number of total tweets that mattered, or if the messages were positive or negative, but the percentage of mentions to that of the competition.
“Think of this as a measurement of buzz,” said Fabio Rojas, the IU Bloomington associate professor of sociology in the College of Arts and Sciences in a statement. “We call this the ‘all publicity is good publicity’ finding. Even if you don’t like somebody, you would only talk about them if they’re important.”
Rojas, along with colleagues in the Department of Sociology and School of Informatics and Computing, analyzed nearly 537 million tweets to see if online social media behavior could determine real-world political behavior.
The study, called “More Tweets, More Votes: Social Media as a Quantitative Indicator of Political Behavior,” used a giant Twitter database collected by the Center for Complex Networks and Systems Research at Indiana University’s School of Informatics and Computing. The Indiana site gives academic researchers access to the largest sample of tweets in the world.
Just mentioning a candidate is not enough to correlate with the winner of the race, since some high-profile candidates — like U.S. Rep. Michelle Bachman of Minnesota — inherently get more attention. It is only when the raw number of tweets a candidate earns is set in the context with others in the race that the numbers become significant.
This new way of examining popularity could revolutionize the way politicians direct political campaigns, especially when other information is scarce. Without spending millions on polls and surveys, candidates could simply follow Twitter statuses. It could provide an advantage to campaigns with limited budgets.
“With the right planning, someone could monitor races in 2014 on their personal computer,” said Karissa McKelvey, a doctoral student at IU and co-author of the study.
“The point is it’s cheap,” Rojas told the Alex Roarty of the National Journal. “Once you start up software for collecting tweets, it’s very cheap. It took one of my Ph.D. students a couple of weeks to set it up.”