To boldly go …

August 13, 2013

“With over 500 million active users in 2012, Twitter now represents a new frontier for the study of human behavior.”

When I read this sentence in the summary of a new paper by IU Bloomington sociologists and informaticists, I couldn’t help but hear echoes of the Star Trek mission (“Space: the final frontier … to boldly go where no man has gone before”). Of course, the shape of life on Earth today goes far beyond what the 1960s TV show could have imagined. Twitter is but one example.

Joseph DiGrazia, Karissa McKelvey, Johan Bollen, and Fabio Rojas recently published a study “More Tweets, More Votes: Social Media as a Quantitative Indicator of Political Behavior.” (DiGrazia, the principal author, is in the College of Arts and Sciences’ Department of Sociology. Fabio Rojas is an associate professor of sociology. McKelvey is a doctoral student with the Center for Complex Networks and Systems Research in the School of Informatics and Computing, and Bollen is an associate professor in the SOIC.) DiGrazia and his co-authors write:

Online social networking environments present a tremendous scientific opportunity: they generate large scale data about the communication patterns and preferences of hundreds of millions of individuals, which can be analysed to form sophisticated models of individual and group behavior.

Twitter data is revealing across all realms, from economics and entertainment to personal habits to, of course, politics. The study by DiGrazia et al evaluated a random sample of 537,231,508 tweets posted from August 1 and November 1, 2010. Their findings, they say, demonstrate “a statistically significant relationship between tweets and electoral outcomes.” In other words, as the title of the paper puts it, more tweets=more votes, regardless of the tweet’s positive or negative content.

“We call this the ‘all publicity is good publicity’ finding,” says Rojas. “Even if you don’t like somebody, you would only talk about them if they’re important.”

In short, while Twitter is still rife with celebrity gossip and bad jokes, it also has become a kind of dream database for social scientists. Study co-author Johan Bollen, for example, recently patented a tracking system dubbed “The Twitter Predictor”, which tracks various “mood categories” across millions of tweets to detect public mood and opinion about financial and economic indicators, such as the Dow Jones.

Unlike survey responses or polls, tweets are largely unsolicited, unfiltered, and unvarnished.

“Our findings show there is massive, untapped reliable data out there that can give insights into public opinion,” DiGarzia says. As he and his co-authors conclude in their study, “the models show that social media matters. …[O]nline social networks are not ephemeral, spam-infested sources of information. Rather, social media may very well provide a valid indicator of the American electorate.”

The final frontier in 140 characters? Even Spock might have been surprised.