A single Facebook message on a congressional election day in 2010 prompted about a third of a million more Americans to cast their vote, scientists said on Wednesday.
The biggest-ever experiment into social networking confirms that peer pressure in cyberspace helps get out the vote in the real world, they said.
On November 2, 2010, 60 million Americans who logged onto Facebook saw a non-partisan prompt at the top of their newsfeeds.
It comprised a message that said "Today is Election Day"; a clickable "I Voted" button; a link to local polling places; a counter on how many Facebook users had already reported voting; and up to six pictures of close Facebook "friends" who said they had already voted.
A further 600,000 people were assigned to see a modified message. It was identical in all respects to the first, except that it had no pictures of friends who had voted.
A third group, also comprising 600,000 people, were a comparison, or "control." They did not receive any "Today is Election Day" prompt at all.
The researchers then examined how the Facebook users behaved.
By looking at public records from polling stations -- but without revealing the identity of the voter or how they voted -- the scientists estimate that those who received the informational message or no prompt voted in the same prevalence.
Where there was a difference was among the group that got the prompt along with the pictures of friends who had voted.
Those who received this "social" message were 0.39 percent likelier to cast their vote than the other groups.
Translated a different way, this amounts to an extra 60,000 votes cast by the "social message" recipients, say the scientists.
But they say the figure can be multiplied roughly fourfold nationwide because of social contagion -- when friends tell others that they have voted, this can prompt peers to follow suit.
So by this count, some 340,000 people voted because of the Facebook "poke."
"Social influence made all the difference in political mobilisation," said James Fowler, a professor of political science at the University of California at San Diego, who led the experiment.
"It's not the 'I voted' button or the lapel sticker we've all seen that gets out the vote. It's the person attached to it."
The result also shows how the virtual world can have an impact on the real world, especially in electoral contests resting on a knife edge, he said.
In the 2000 US presidential election, George W. Bush beat Al Gore in Florida by 537 votes, or less than 0.01 percent of the vote -- a result that gave Bush the White House thanks to the US electoral college system.
In a commentary, also appearing in the science journal Nature, New York University academic Sinan Aral said the test gave scientific backing to anecdotal evidence about the effectiveness of "viral marketing."
"But such interventions also have the potential to promote positive social changes, such as increasing the rate of HIV testing, reducing violence, improving adherence to exercise, or increasing political mobilisation and awareness," said Aral.
"In this way, the science of social influence may have dramatic implications for products, politics and public health."