Upending predictions, young voters made a strong showing at the polls. Did they go because "everybody's doing it," as they saw on Facebook?
You didn't know it at the time, but when you logged into Facebook on Election Day, you became a subject in a mass social experiment. You went about your day, clicked around Facebook, and you may have voted. Now your behavior is data that social scientists will scrutinize in the months ahead, asking one of the core questions of democracy: What makes people vote? If patterns from earlier research hold true, the experiment's designer James Fowler says that it is "absolutely plausible that Facebook drove a lot of the increase" in young-voter participation (thought to be up one percentage point from 2008 as a share of the electorate). It is, he continued, "even possible that Facebook is completely responsible."
Assuming you are over the age of 18 and were using a computer in the United States, you probably saw at the top of your Facebook page advising you that, surprise, it was Election Day. There was a link where you could find your polling place, a button that said either "I'm voting" or I'm a voter," and pictures of the faces of friends who had already declared they had voted, which also appeared in your News Feed. If you saw something like that, you were in good company: 94 percent of 18-and-older U.S. Facebook users got that treatment, assigned randomly, of course.* Though it's not yet known how many people that is, in a similar experiment performed in 2010, the number was *60 million*. Presumably it was even more on Tuesday, as Facebook has grown substantially in the past two years.
But here's the catch: six percent of people didn't get the intervention. Two percent saw nothing -- no message, no button, no news stories. Another two percent saw the message but no stories of friends' voting behavior populated their feeds, and a final two percent saw only the social content but no message at the top. By splitting up the population into these experimental and control groups, researchers will be able to see if the messages had any effect on voting behavior when they begin matching the Facebook users to the voter rolls (whom a person voted for is private information, but whether they voted is public). If those who got the experimental treatment voted in greater numbers, as is expected, Fowler and his team will be able to have a pretty good sense of just how many votes in the 2012 election came directly as a result of Facebook.
In a country where elections can turn on just a couple hundred votes, it's not far-fetched to say that Facebook's efforts to improve voter participation could swing an election, if they haven't already.
They've done a very similar experiment before, and the results were significant. In a paper published earlier this year in Nature, Fowler and his colleagues announced that a Facebook message and behavior-sharing communication increased the probability that a person votes by slightly more than 2 percent. That may not seem like a huge effect, but when you have a huge population, as Facebook does, a small uptick in probability means substantial changes in voting behavior.
"Our results suggest," the team wrote, "that the Facebook social message increased turnout directly by about 60,000 voters and indirectly through social contagion by another 280,000 voters, for a total of 340,000 additional votes." This finding -- remarkable and novel as it may be -- is in concert with earlier research that has shown that voting is strongly influenced by social pressure, such as in this 2008 study which found that people were significantly more likely to vote if they received mailings promising to later report neighborhood-wide who had voted and who had stayed at home.
Although months of door knocking, phone calls, and other traditional campaign tactics surely bring more people to the polls, those measures are expensive and labor-intensive. Nothing seems to come even close to a Facebook message's efficacy in increasing voter turnout. "When we were trying to get published," Fowler, a professor at UC San Diego, told me, "We had reviewers who said, 'These results are so small that they're meaningless,' and other reviewers who said, 'These results are implausibly large. There's no way this can be true.' " In a country where elections can turn on just a couple hundred votes, it's not far-fetched to say that, down the road, Facebook's efforts to improve voter participation could swing an election, if they haven't already.
Now it must be said that of course Facebook is not trying to elect Democrats. Facebook has an admirable civic virtue and has long tried to increase democratic participation in a strictly nonpartisan way. "Facebook," Fowler said to me, "wants everyone to be more likely to vote. Facebook wants everyone to participate in the fact of democracy."
But that doesn't mean the effects of Facebook's efforts are not lopsided. Outside of Facebook's demographic particularities, there are reasons to believe that improved voter turnout in general helps Democrats, though there is a debate about this within political science.
In practice, though, there is no such thing as pure a get-out-the-vote, one whose tide raises all votes, and Facebook is no exception. It skews toward both women and younger voters, two groups which tended to prefer Democrats on Tuesday. Eighteen-to-29-year-olds voted 60 percent for Obama, compared with 37 percent for Romney. The next-older age group, 30-to-44-year-olds, gave Obama 52 percent of their support. Among Americans older than 45, Romney won. The implication is clear: If Facebook provides a cheap and effective way to get more people to the polls, and it seems that it does, that is good news for Democrats. For Republicans, well, it's an uncomfortable situation when increasing voter participation is a losing strategy.
* * *
Facebook's effect -- however big it is -- is at the margins, and in a country where elections are so close, the margins can matter a lot. But there are long-term trends underfoot that, for Republicans, mean their troubles go beyond Facebook. This year was the third presidential election in a row where young-voter participation hovered around 50 percent (meaning that half of eligible young people actually voted), Peter Levine of the Center for Information and Research on Civic Learning and Engagement at Tufts University told me, and this is up from a low of just 37 percent in 1996. Obviously, that sort of shift is not the result of a Facebook message. "This seems to be a significant generational change," he said.
The reasons for this generational change are complex and not totally understood. One certain factor is simply that recent campaigns have tried harder to reach out to young people. "In the 1990s the conventional wisdom was that young people don't vote," Levine said. "So they would literally look through a contact list of potential voters to reach out to, and just delete the young people -- not to discriminate but because they were trying to be efficient."
"In the 1990s the conventional wisdom was that young people don't vote," Levine said. "So they would literally look through a contact list of potential voters to reach out to, and just delete the young people."
But that ended with what Levine calls the "50/50 nation situation" -- elections so close that campaigns could no longer afford to write off huge parts of the population. "It had already begun to change, but Obama won on the strength of young votes in '08 -- not only in the general election but especially in the primaries. They gave him Iowa. Without Iowa, no President Obama."
For years it was thought that one reason young people voted in such low numbers is because they are so mobile: If you don't live in a place for years, voting is, as political scientists term it, "costly." You have to re-register every time you move, find your polling place, and, if you plan to vote in local races, spend time getting up-to-speed on complicated municipal politics. But this calculus of inaction may be changing: With the Internet, it's much easier to find where your polling place is. And our online media environment privileges national politics over local politics so much, that national politics alone may entice more people to the polls.
Additionally, Constance Flanagan of the University of Wisconsin argues, there's been a backlash on college campuses to voter-suppression efforts. "The voter-suppression thing did make people more aware," she said. "Our university newspaper had a front-page story about what are your rights, do you have to produce an ID. ... It was a conversation topic among young people and something they passed on to one another." Particularly, she said, that minority groups who felt targeted really responded by organizing themselves and making sure people voted. (Ta-Nehisi Coates and Andrew Cohen have both written about this backlash here at The Atlantic.)
Even with these recent improvements, Levine reminds us to bear in mind that we're still only at 50 percent, and the other 50 percent, those who are not voting, is not a random sample of the population -- typically lower income, lower education levels, and not into politics. With social media becoming an increasingly important part of political communications, campaigns and activists should ask whether there's any way to use social media to get to them. "I think we do incrementally, because there's some serendipity where your old high school friend gets into politics and draws you in, but does it happen at a large scale? I don't think we know," he said.
All of this adds up to a shifting electoral environment, as young people come out to the polls in greater numbers, and are more easily reached online. "In terms of good news for the Democrats," Donald Green of Columbia University told me, "the fact that you have this age cohort that has been socialized into very strong presidential support for Democrats, is in some ways the countervailing force to the age cohort that was socialized into strong Republican support under Reagan." Voting once is known to be habit forming. Those who were brought to the polls by the wave of enthusiasm for Obama in 2008 will likely vote in greater numbers for the rest of their lives, even in a more humdrum election, as this one felt.
And that's perhaps the worst news of all for the Republicans. A wave of enthusiasm is one thing, an anomaly in an otherwise 50/50 nation. But now the tide has gone out, and the seashore looks changed.
* November 9, 2012: The original version of this piece said that the intervention appeared for 96 percent of people, with the four remaining percent getting other conditions. Fowler has today updated these numbers to me by email, and the piece has been edited to reflect this.