Wednesday, September 12, 2012

Facebook influences voting behavior, you, your friends

Maybe, maybe not.
By Emily Willingham

This one will have you wondering if you have inadvertently participated in a science experiment by way of Facebook. Actually, you probably did.

In 2010, a team of researchers led by UC San Diego political science professor James Fowler managed to get more than 60 million people to see a "get out the vote" message at the top of their Facebook news feed. The date, in case you want to review your memory, was November 2, 2010, the day of the U.S. Congressional elections. The message, described in a news release from UCSD as "non-partisan," carried a reminder to vote--"Today is Election Day"--and a button Facebook users could click to let everyone know "I voted." Also available to users were a counter showing how many Facebookers had reported voting and a link to a site for finding local polling places. I have a terrible memory, but I know that a button like the "I voted" one would be exactly something I'd've clicked out of pride in participating in our democratic process.

I feel so used. Sort of. But in the name of Science, right?

Of the 61 million people who saw the message, a total of 60,055,176 of them also saw up to six pictures of their Facebook friends--was I one of them?--associated with the message. These friends were subcategorized as close or not to the Facebook user based on their history of interactions. Another 611,044 people got everything but the friend pictures, and another 613,096 or so got no Facebook message about election day at all, unwittingly serving as the "control group" for this grand experiment. 

The Facebook responses indicated that people who saw the message along with pictures of friends were about 2 percent more likely to report having voted and about 0.39% more likely to vote compared to users who got the message without friend images and associations. And the effect was contagious: Even users who didn't see the information message at all--with or without friend images--but had a friend who received the message were 0.22% more likely to vote, and that likelihood increased by that amount for each close friend who'd gotten that message. People were also a little more likely to click on the link to find a polling place if a close friend had gotten the informational message. 

These percentages may seem unimpressive at first look. But in a commentary on the study, both published in Nature, Sinan Aral writes:
Although these estimates may seem small, they translate into significant numbers of  votes. A social message saying that a Facebook friend had voted generated 886,000 additional ‘expressed’ votes (clicks on the ‘I voted’ button), and messages involving a close friend gener­ated an additional 559,000 expressed votes. 
In fact, the authors say that the social message they distributed on Facebook might have directly increased voter turnout that day by 60,000 people--enough to take Florida these days in a presidential election--and that the social contagion of indirect influence from friends might have sent another 280,000 voters to the polls who might otherwise not have gone. In other words, social contagion had the bigger influence. 

That reference to actual votes is based on the researchers' analysis of Facebook users' voting behavior and matching with voter records, which some states make publicly available. These states include Arkansas,California, Connecticut, Florida, Kansas, Kentucky, Missouri, Nevada, New Jersey, New York, Oklahoma, Pennsylvania, and Rhode Island, and the authors report that "about 1 in 3 users" were matched successfully with their voting records. 

If it seems a little creepy that a group of complete strangers might influence your voting behavior via Facebook and then check that against public records ... well, it is, I think. Basically, if you logged into your Facebook account on November 2, 2010, you placed yourself unknowingly into a grand experiment into the viral effects of social networks on political action. The study authors also accessed information you may have provided on Facebook about your political affiliation and ideologies and assessed your interactions with people on Facebook to determine your level of connectedness and, the researchers hypothesize, social influence and contagion. Of course, if you spend a lot of your time on Facebook fighting with the same people, the effect might have been the opposite of what they assumed. I'm not as disturbed by scientists observing human behavior unknown to the humans as I am about Facebook's making this information available. Did we sign off on that somewhere?

What might come as no surprise to anyone deeply engaged in social media, whether Twitter or Facebook or other networks, is that "online messages may influence a variety of offline behaviors." What the study findings seem to show is that we don't have to see each other's faces to pick up and spread behaviors, but we do have to have close relationships, whether virtual or otherwise, for the behavioral "contagion" to have an effect. That said, face-to-face interaction still carries a big punch.

Women are known for their strong social networks in real life, and they dominate some of the virtual platforms. One thing the study doesn't mention is differences in the strength of social networks and their influence based on sex, so I asked the senior author, Dr. Fowler, about that. He said that the team "would be taking a look at differences in effect size between men and women in future work," so I guess we can anticipate more unwitting involvement in science to come. 

Because I also am interested in the influence of these networks--real world or virtual--in the spread of misinformation and pseudoscience, I asked Dr. Fowler if this kind of virality could translate to these other contexts. "Absolutely," he said, noting that a wide range of beliefs and behaviors can spread in networks "up to three degrees of separation." He added, "I don't see why the spread of pseudoscientific beliefs would operate any differently."

My take-home from that? Those Facebook memes that say "Wow, your Facebook post about politics has changed my mind and my vote... said no one, EVER!" might be off base. Want to be influential? Get thee to your social media and begin spreading your messages about voting or evidence-based science or child-rearing or quick dinners in 20 minutes. And if you're skeeved out by involuntary participation in scientific studies like this by way of Facebook (are you? Do you think you participated in this one?), you might want to avoid "public service" voting messages from here on out. 

The views expressed in this post do not necessarily reflect or conflict with those of the DXS editorial team.

No comments:

Post a Comment