Facebook has proved that it’s useful for more than posting pictures, stalking exes and playing imaginary agricultural games: It’s a powerful tool for mobilizing people to vote. An experiment involving more than 61 million people reveals that Facebook users who received a message that a friend had voted were more likely to go to the polls themselves.
“Prior work on political mobilization and motivation for voting has been focused on the individual,” says information economist Sinan Aral of New York University. “This paper presents really nice large-scale evidence that political mobilization is a peer-to-peer activity.”
The experiment also suggests that harnessing someone’s digital network of friends might be a tool for influencing other behaviors, such as getting people to quit smoking, says Aral.
During the 2010 congressional elections, researchers led by social scientist James Fowler of the University of California, San Diego worked with Facebook to conduct a massive social influence experiment. On November 2, the day of the election, all Facebook users over 18 years old who logged in experienced one of three treatments: one group got a message at the top of their news feed that urged them to vote, provided a link to find local polling places, encouraged them to click an “I voted” button and showed a counter indicating the number of Facebook users who had voted. A second group received the same message, but they also saw profile pictures of up to six friends who had reported voting and a tally of how many of their Facebook contacts had clicked on “I voted.” A third group received no message. (Because Facebook was interested in getting out the vote, more than 60 million people received the message with pictures of friends; the other two test groups each had about 600,000 people.)
People who saw pictures of their friends along with the message were 2.08 percent more likely to click the “I voted button” than people who saw the message alone, Fowler and his colleagues report in the Sept. 13 Nature. The researchers also validated the voting of about one in three Facebook users in the study by examining publicly available voting records. Those who saw the friend message were 0.39 percent more likely to vote, while turnout among those who saw the generic message or no message was basically the same.
Those percentages aren’t huge and are low compared with other get-out-the-vote tactics, such as door-to-door canvassing, says Fowler. But such numbers can still turn elections. The researchers’ conservative estimate suggests that the message with pictures of friends translated directly into getting about 60,000 people to vote.
Close friends, identified by the frequency and nature of their interactions on Facebook, were much more important than weaker links in terms of driving behavior, the researchers found. Such connections make up only about 7 percent of the friendships on Facebook.
There was also a domino effect: Users were also more likely to vote if friends of their friends saw the message that contained pictures, regardless of whether they received the message themselves. While the researchers can’t discern whether that’s because of offline behavior, such as talking on the phone about the election, the team estimates that this friend-of-friend effect translated into an additional 280,000 voters.
The work nicely demonstrates the influence of digital social signals that society is increasingly exposed to, says Aral. The potential influence of these signals on behavior has big implications, he says, whether for selling products, mobilizing political activity or improving public health.