Play Live Radio
Next Up:
Available On Air Stations
KPBS Midday Edition

UC San Diego Examines Influence Of Social Network On Voting

Illustrative map of part of the social network of “close friends” from Abilene, Texas, who logged in on Election Day in 2010.
James Fowler, UC San Diego
Illustrative map of part of the social network of “close friends” from Abilene, Texas, who logged in on Election Day in 2010.
Facebook Voting Button Increased 2010 Turnout
UC San Diego Examines Influence Of Social Network On Voting
GUESTJames Fowler, professor of political science and medical genetics at UC San Diego and author of the book "Connected: How Your Friends' Friends' Friends Affect Everything You Feel, Think, and Do."

CAVANAUGH: This is KPBS Midday Edition. Those "I voted" stickers you get at the polls are quite popular. They allow people to display the fact they voted and serve as reminders to others who haven't gotten to the polls yet. What works in real life works exponentially in social media. According to a new study in the magazine Nature, researchers found that a little "I voted" button on Facebook may have inspired 300,000 voters to get to the polls in 2010. James Fowler is a professor of political science and molecular genetics at UCSD, author of the book, connected, how your friends' friends' friends affect everything you feel, think and do. He led the study on the button. Welcome to the show. FOWLER: Thank you very much. CAVANAUGH: Now, the sheer size of this study, 61 million people, that's astounding. Could you explain how this experiment was conducted? Facebook had to be in on the plan, right? 92 this is a collaboration with the head of the science team at Facebook. And they put a message at the top of the news feed saying today is election day. And when I started collaborating with them, they said how do you know it works? They said well, we're not sure. I said nonot shoe an experiment? So we showed the message to some people but also didn't show the message to some people, and we could compare not only how they behaved online, but we matched 6 million of the 61 million people to voter records and we were able to show that not only was it the case that people who saw the message voted, but also the friends of the people who saw the message vote, and that got 280,000 people to the polls. CAVANAUGH: What did the I voted message for people who received it look like? FOWLER: Just a little button that said I voted, you could click on it. If you did that, it showed in your newsfeed, a news story. James Fowler has voted today. Some of this happened offline as well. We looked at the quality of the friendships that spread the behavior. And we found that those relationships that were most likely to exist in real life were the ones that were doing all the work. CAVANAUGH: Not everybody got the same message. Tell us how that was broken down >> We gave two messages. One was today is election today, it had the I voted button. There was also a link you could clink on to your polling place, and a counter for the number of people who had clicked on it that day. Another version showed all those things as well and pictures of your friends who had voted earlier in the day. When we looked at the information-only message, without the pictures of friends, there was no difference between that and receiving no message at all. What this means is that what was really driving the behavior was seeing pictures of your friends. CAVANAUGH: That's fascinating. And you did have some of these 61 million, you broke it down into three type, and some of those people didn't get any messages at all right? They were the control group? FOWLER: That's right. CAVANAUGH: So how can you tell that the people that motivated those people to go out and vote after they saw pictures of their friends that they were close friends? FOWLER: We actually did know additional study in relationship to this study where we asked about 1,000 people to name for us their close friends in real life. Then we used the Facebook information to see what kinds of communication is there between real-world friends? So we developed this mechanism where we can use the information to predict with about 90% accuracy who among your Facebook friends is most likely to be your best friend in real life CAVANAUGH: You count close friends as those people that you actually know, that you see on a regular basis. Can't you be close friends with someone on the Internet who you don't see with any regularity? FOWLER: This is why it's only 85 or 90%. Will I say your close friend are the people you spend the most time with on Facebook. But they say, I know this one guy, I read all the stuff he posts, and we're not close. We have those examples. But most of the time, your close friend, you communicate them via e-mail, telephone, face too face. Through a variety of media. CAVANAUGH: How is this different from one friend basically bumping into another one, he's got an I voted sticker on his lapel, and you notice it and say, okay, well, he voted. I should go vote too. Or you come home in the evening, and all your family has voted or everybody in the workplace has voted. How is this different? FOWLER: One thing that's different here is the scale. This may be the largest experiment that's ever been conducted with 61 million people. But other than that, I think it's working very much the same way as real-world social networks always worked. There was a lot of research before in that suggested friends and family were important for mobilizing political motivation. Although the average person on Facebook has 150 friends, it was just their 10 closest friends that made a difference. Is this just a new way for us to reach those real-world social networks. CAVANAUGH: And how did you target out the exponential effect of this from one friend to another friend to another friend? FOWLER: We were able to track the online behavior, people clicking the I voted button not only to the friends but to the friends of friends as well, up to 2 degrees of separation. So if your friend's friend refers this message on Facebook, you were more likely to click on the I voted button yourself. CAVANAUGH: How many people clicked on it who didn't actually vote? FOWLER: Oh, that's really interesting. In this study, we found that it was about 4%, and that's lower than for other studies. Other studies, we find that it's about a 20-30% rate of people saying they voted when they didn't actually vote. And this is another sign that voting is a socially desirable outcome. You don't want to admit to the fact that you didn't vote if you didn't have time. And a lot of ways we measure this to get people to tell us the truth, we'll say people don't really have time sometimes, but by the way, did you vote? And we have to prompt them to make it okay for them to say no. Even though, we find this difference. CAVANAUGH: And how did you actually find out who voted and who did not? FOWLER: We matched about 6 million of the Facebook records to publicly available voter records using first name, last name, and birth date. And one interesting thing, Facebook was concerned about privacy. When we proposed this, they said no. We said what if we Adesign a procedure to allow us to do this at the group level, so Facebook doesn't know if any one individual has voted but we do the statistics? They said okay. CAVANAUGH: We're talking about a study published in the magazine Nature about an experiment with the I voted button on Facebook and what the study found out. You have been telling us, and thank you for it, you've been breaking down how this study was conducted, how it influenced -- seems to have influenced people to go out to the polls. But what does the study actually tell us? FOWLER: I think it tells us that the online world matters. And believe it or not, there are people that don't think that. I've had a lot of people who have looked at some of the research we've done and people say ah, that's just for playing farmville. And we have another group of people that are extreme on the other is. They'll say oh, my God! Online social networks, it's possible to have a deep, close, meaningful relationship with a million people online. Neither of those things is true. What the study shows is that, yes, online social networks matter. Just a single message on Facebook got 300,000 people to engage in a real-world activity. But the reason they matter is not because we have different brains or different ways of making friends, it allows us to reach the real-world friends we've always had. CAVANAUGH: So in light of the study's finding, do you think candidates are using social media correctly? FOWLER: It's interesting that you say that. I've met with some local campaign managers recently. They saw the study and they were interested in chatting. And I thought that they were further along than they were. But it sounds like the thing with these campaigns, if it's not broke, don't fix it. So apparently they're still using technology from before the social media existed. So I think we're going to see big changes. An increasing use of experiments in which candidates are elected to do, because they want to send the message to everybody. And I think we're going to see an increasing reliance on measuring not just the effect of the messages on the people who receive them, but on the friends of the people who receive them. In our study, we found that if you didn't look at the friend, you would have missed the whole story. For every one person whose behavior changed because they saw the message, there were four friends who changed. CAVANAUGH: We just found online kind of an example of how social media was used in an actual campaign to save a library in Troy Michigan. It was a proposed tax increase that would help the library keep its doors open, but it was opposed by the local tea party. All they kept saying was taxes, taxes, taxes. So the library supporters launched a book burning party site on Facebook. It outraged the public, and then supporters revealed finally their real purpose, after all this outrage was expressed on Facebook. NEW SPEAKER: Once it reached a fevered pitch, we vealed the true intent of our campaign. A vote against the library is like a vote to burn books. And people started posting, tweeting, and reporting all over again. In the end, we had changed the conversation completely from taxes, taxes, taxes, to library, library, library. And on August 2nd, the yes voters, voters who don't normally turn out to vote, turned out at levels 342% greater than projected. CAVANAUGH: This is obviously an effective use of social media. Why do you think it worked? FOWLER: It's a combination of broadcast which is what we're doing right now, in traditional media, and exercising this real-world social network. The new thing is that you can tap right into the real-world N. You can see the conversation, participate in the conversation, and energize not just the ties between strangers but ties between real-world friends. In a small community like this, using Facebook with a media savvy campaign that gets people's attention, it's one way to really multiply the effect of whatever small effect a broadcast message might have. CAVANAUGH: I want to go back to the broadcast message. How then does what works on social media differ from the kinds of messages that we usually get in the broadcast media like on TV? What will work on social media that doesn't necessarily work on TV and vice versa? Do we know that yet? FOWLER: Well, this is the thing we're trying to study. And I would say that a big part of the message from our study is that Facebook works very similarly to traditional media in one sense. We need to descend 61,000 messages to get people to change their behavior. What we haven't been able to report is to see what happens to their friend or to make it easy for people to contact their friends in order to get this multiplier effect from the network. CAVANAUGH: Another aspect of this is that instead of -- because you're finding that close friends are the real motivators when it comes to people going out to vote in this study, there's also a flip side of social media in that it doesn't necessarily expand our world the way we think it does, but that one's world view can become smaller rather than larger because you associate only with the people who share your opinion, really. FOWLER: Yeah, there's a natural human tendency to reach out and connect to people who are just like us. When you only have access to 150 people, that's sort of a natural human group size that we evolved in, are the chances are that there's going to be a lot of diversity. People are going to have differing opinions. But if you're searching for that certain someone in a group of a billion people, the chances are very good you're going to find someone who has exactly your same idiosyncratic beliefs. One thing that worries me, these online networks may make us more polarized. We may be able to reach out and find those people who are telling us the same things we already believe. CAVANAUGH: I see. And I was also interested in finding out that political campaigns don't only have viral effects but it's one of the major reasons beam defriend each other. FOWLER: Along with religion and sharing too much, politics can be a real motivator for people to cut ties online. CAVANAUGH: Will you be studying the effectiveness of how political campaigns are using social media this election cycle? FOWLER: We are talking with Facebook about doing additional experiment, but nothing has been worked out. There are a couple things I think are critical. One is what kinds of messages work best. We've already seen that faces mattered and the message didn't. So are there other text messages that could motivate people to go to the polls? And the other is who is most important? Is it older people or younger people who are more influential? Is it people with more friends or fewer friends? Those are the kinds of questions I'd really like to be able to answer. CAVANAUGH: That is fascinating.

Facebook Voting Button Increased 2010 Turnout, UCSD Study Finds
As the November elections draw closer, campaigns everywhere are focused on how to get more voters to turn out at the polls. A new study from UC San Diego suggests peer pressure through Facebook could do just that.

Facebook users who saw a message on their account that close friends had voted in the 2010 general election were more likely to turn out to a polling place themselves, according to a study released today by UC San Diego.

The report, published in the journal Nature, randomly posted a message in the news feeds of more than 60 million Facebook users on election day.

The message showed people pictures of their friends who had voted. It also included a running tally of the number of people on Facebook who said they voted, information on where to find your polling place and a clickable "I Voted'' button.

About 600,000 users received the same message but without pictures of their friends, and another estimated 600,000 users received no message at all.

James Fowler, a professor of political science in the UC San Diego Division of Social Sciences and of medical genetics in the School of Medicine, led the study. He said it shows use of social media convinced about 340,000 people to vote who otherwise wouldn't have.


Fowler breaks that 340,000 down into direct and indirect social influence.

An estimated 60,000 people who saw the voting message from their friends then decided to vote, when they otherwise would not have, he said.

"More importantly, we found that the friends of people who received these messages were more likely to vote than the friends of people who saw no messages at all," he said.

Fowler estimates that indirect friend influence caused an additional 280,000 people to vote.

"This is just extraordinary because what this means is the indirect effect was actually much larger than the direct effect," he said. "For every one person the message got to the polls, it got four of their friends to the polls.”

Fowler began the study after meeting Cameron Marlow, a data scientist at Facebook, during a conference. The two men decided to form a collaboration to conduct Fowler's study.

Using public voting records to see whether Facebook users who clicked the "I Voted" button actually voted, Fowler found that many who said they voted on Facebook did not actually appear at the polls.

"We developed this technique where we could use names and birthdays in the voting records and names and birthdays on Facebook to do a matching process that allowed us to see whether or not people who were voting on Facebook were actually voting in real life," he said.

Fowler stressed that the method kept Facebook from knowing which people actually voted.

This new report follows Fowler's previous studies on how social behavior can spread like an epidemic through networks of friends—he’s also the author of a study that found your friends can make you fat.

In this study, people who saw an “I Voted” message without pictures of their friends were not as likely to vote, he said, suggesting the importance of social influence. He added that your close friends have a far bigger influence on your behavior.

“The 10 closest friends on Facebook, the ones you’re most likely to have a face-to-face relationship with, drove the whole effect," he said. "And what this means is that yes it’s true that online networks affect real world behavior. But the way they do it is by stimulating the real world networks we’ve always had before Facebook existed.”

The researchers said they did not find any evidence of differences in effects among self-described liberals and conservatives.