WASHINGTON (AP) — Here's something most politicians can "like": Facebook friends played a big role in getting hundreds of thousands of people to vote in 2010, a new scientific study claims.
Facebook researchers and scientists at the University of California, San Diego conducted a massive online experiment in the mid-term congressional election to test and measure the political power of online peer pressure.
They found that people who got Facebook messages that their friends had voted were a bit more likely to go to the polls than those who didn't get the same reminder. And from there the effect multiplied in the social network, they reported in Thursday's journal Nature.
The friend-prodding likely increased voter turnout by as much as 340,000 in the non-presidential election that voted in a new Republican congress, the scientists calculated. They said that it could potentially change the outcome of close elections.
"Our study is the first large-scale scientific test of the idea that online social networks affect real world political behavior," said study lead author James Fowler, a professor of medical genetics and political science at the University of California, San Diego.
He has studied friend and social media influences on public health and politics over the past decade. While pundits have pointed to social media-inspired revolutions in the Arab world, this is more verifiable scientifically because it is a controlled study comparing groups that had different inputs. It's the voting equivalent of testing real drugs versus sugar pills.
Outside experts say the new study makes sense and fits with other research about how effective get-out-the-vote drives are, but say Fowler's numbers may be a bit high. That's because they factor in a large indirect effect, calculations which some didn't find as convincing.
Nearly every American of voting age who logged into Facebook on Election Day 2010 was part of the experiment, even though they didn't know it.
Most of them — more than 60 million — saw an announcement on top of their Facebook news feed: Today is Election Day. It showed how many Facebook users as well as their friends had clicked an "I voted" button and showed up to six pictures of those friends. It also linked to a list of polling places.
Researchers compared voter turnout with two groups that didn't receive that same message. One group of 611,000 people simply got a generic announcement encouraging voting, but no pictures or count of friends. Another 613,000 users didn't receive any message.
Those who got the peer pressure message were less than half a percent (0.39 percent) more likely to vote than those who got no message or the generic one. While that seems like a very small increase, it is statistically significant and it adds up, Fowler said.
There was no difference in voting found between the generic and no-message groups. Nor was there any difference seen in friend-prompted turnout between self-identified conservatives and liberals, Democrats and Republicans.
Fowler and colleagues didn't just take the word of people who clicked the "I voted" button. They checked public voting records in 13 states for that election, and found about 4 percent of those who said they voted hadn't really cast ballots.
Of those who saw the peer pressure posting, Fowler calculated that 60,000 voted who wouldn't have. On top of that, he said, another 280,000 people voted who wouldn't have because their friends saw the online message and spread the get-out-the-vote word.
It's a form of social contagion with people noticing that the original message recipient voted, so the message spread in person, by word of mouth and online, he said.
"The network is key," he said.
Columbia University political scientist Donald Greene said the 60,000 direct voter number makes sense and fits with other research done, which shows that the more personalized the appeal the better the result. But he said he had difficulty buying the calculations used to come up with the 280,000 indirect votes.
George Mason University political science professor Michael McDonald, an expert in voter turnout, said the study seemed reasonable to him, adding "anything we can do to increase turnout is a good thing."
In the 2010 election, about 38 percent of the voting age population cast ballots, up from about 37 percent in 2006 and 36 percent in 2002. Voting is usually much higher in presidential elections, so the overall effect of a social media get-out-the-vote push might be lower in 2012 because people were already more likely to vote, Fowler and others said.
Fowler and Facebook scientist Cameron Marlow said no decision has been made about doing a similar study or voting drive on Facebook this November.
The study was initiated by Fowler, who got Facebook involved, and was funded by the University of Notre Dame and two private foundations.
Seth Borenstein can be followed at http://twitter.com/borenbears