May 15, 2014 | Research Feature

Friendly Influence

A new study sheds light on how the structures of social networks help or hinder the spread ideas, fashion, and even social movements.

Almost 60 percent of Americans now use Facebook, according to the most recent Pew Research Center study of the social network. What brought all those users on board? Sociologists have shown that dense ties are more influential than sparse ties: people are more likely to be influenced by friends who already know their other friends than by friends who don’t share the same social circle. So the conventional wisdom is that most people joined Facebook because others from their social circle invited them.

But the shift to ever-increasing online connectivity may be carving out some exceptions to this sociological rule, says Professor Dan Wang, who studies how peer influence and social ties shape tastes, consumer habits, and beliefs. He worked with Jure Leskovec of Stanford University to analyze 40 million members of an online social network spread over 40,000 online social network communities that, similar to Facebook, allows users to create groups based on interests, from book clubs to hot rod fan clubs. The online communities often have subgroups: the Harry Potter Fan Club Community might spawn the Harry Potter Fan Club of New York, and so on. Users can also create offline events and invite other network members to attend.

Wang and Leskovec looked at invitations to join online groups and attend offline events to learn how users responded to invitations for both kinds of activities. “It turns out there’s no clean answer. Whether you do something based on your friends’ recommendations depends on whether they share the same friends as you and on the kind of action they are asking you to take.”

Consider someone facing three different invitations to join Facebook. Past research has found that this person is more likely to join if the friends who extended the invitation are not also friends with each other — it’s the disconnectedness among friends that prompts joining, an action that doesn’t require much effort. “But if a bunch of different people you know were inviting you to a protest, or even a party? You’re much less likely to make that significantly bigger commitment when the invitation comes from a disconnected set of friends than if it comes from a connected set of friends,” Wang says, summarizing one of the researchers' key findings

Wang and Leskovec explain this by noting two main reasons we are influenced by our peers to begin with. The first is that people adopt behaviors that are perceived as socially legitimate and require verification by many independent sources, as in the Facebook example. The disconnectedness of those issuing the invitations suggests a widespread acceptance of the action, fashion, or idea that’s in question. In other instances, people may not particularly care one way or another but adopt fashions or ideas in order to solidify social relationships and feel as if they are part of a group. “You might or might not actually believe in the cause of a protest, but you do it because three of your friends who are all very close to each other are doing it, and you are all doing it in service to that relationship,” Wang explains.

Specific group interests don’t seem to matter: people from wine-tasting groups and babysitting groups and car fan clubs all behaved the same. But group size mattered a lot: People were more likely to join groups of about 55 or smaller when their friend networks were dense; once groups grew larger, new members were likely to join when their sparsely networked friends invited them. “It reflects how social movements and protests form,” Wang explains. “A small, new group grows because people join out of a commitment to friends they share. But once there is a critical mass, the group gains legitimacy by appealing to a broader audience.”

The findings have implications for everything from marketing to public health. “The firms that are successful at viral word-of-mouth campaigns choose their targets wisely. They understand that they have to appeal to potential consumers who are clustered with one another — they are very similar and highly networked close friends. And the only reason those consumers are adopting the product is because their small, tightly bound group is doing it.” In the case of public health, understanding why some individuals respond to social influence for one set of health behaviors but not others can inform the design of policy. Some types of social networks may be effective at promoting the adoption of healthy behaviors such as getting regular exercise. “It depends on the cost of the behavior,” Wang says. “Hitting the gym early every morning before you hit the office could be relatively high-cost compared to taking a walk during your lunch break. And, each behavior could spread differently depending on the connectedness of your social network.”

Dan Wang is Assistant Professor of Business in the Management Division at Columbia Business School.

Read the Research

Dan Wang, Jure Leskovec

"The structural roots of collective influence in social networks"

View abstract/citation 

Have You Read Columbia Business?

Columbia Business School’s alumni magazine connects alumni with each other and the School; celebrates alumni milestones and accomplishments; and chronicles the impact of Columbia Business School alumni, faculty members, and students on the global business landscape.

Read Columbia Business>