The percentage of votes for Republican and Democratic candidates could be predicted by the percentage of tweets that mentioned those candidates -- and it didn't matter whether the tweets were positive or negative, they said.
Indiana University researchers report analyzing a sample of 537 million tweets to examine whether online social media behavior could be used to assess real world political behavior.
"Think of this as a measurement of buzz," sociology Professor Fabio Rojas said. "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."
While polls and surveys are the traditional ways to gauge public attitudes, social media can yield massive amounts of data, the researchers said.
"Our findings show there is massive, untapped reliable data out there that can give insights into public opinion," doctoral student Joseph DiGrazia, the study's lead author, said.
Twitter can be a valuable tool for analyzing races, particularly when other data is in short supply, the researchers said.
"With the right planning, someone could monitor races in 2014 on their personal computer," through the study of tweets, study co-author Karissa McKelvey said.