Do friends wear the same clothes or see the same movies because they have similar tastes, part of the reason they became friends or, once a friendship is established, do individuals influence each other to adopt like behaviors? 

Social scientists don't know for sure and are still trying to understand the role social influence plays in the spreading of trends because the real world doesn't keep track of how people acquire new items or preferences. 

But the virtual world of "Second Life" does. Researchers from the University of Michigan have tried to use this information to study how "gestures" make their way through this online community. Gestures are code snippets that Second Life avatars must acquire in order to make motions such as dancing, waving or chanting.

Roughly half of the gestures the researchers studied made their way through the virtual world friend by friend. 

"We could have found that most everyone goes to the store to buy gestures, but it turns out about 50 percent of gesture transfers are between people who have declared themselves friends. The social networks played a major role in the distribution of these assets," said Lada Adamic, an assistant professor in the School of Information and the Department of Electrical Engineering and Computer Science. 

Adamic is an author of a paper on the research that graduate student Eytan Bakshy will present on July 7 at the Association for Computer Machinery's Conference on Electronic Conference in Stanford, Calif. Bakshy is a doctoral student in the School of Information.

"There's been a high correspondence between the real world and virtual worlds," Adamic said. "We're not saying this is exactly how people share in the real world, but we believe it does have some relevance." 

This study is one of the first to model social influence in a virtual world because of the rarity of having access to information about how information, assets or ideas propagate. In Second Life, the previous owner of a gesture is listed. 

The researchers also found that the gestures that spread from friend to friend were not distributed as broadly as ones that were distributed outside of the social network, such as those acquired in stores or as give-aways.

And they discovered that the early adopters of gestures who are among the first 5-10 percent to acquire new assets are not the same as the influencers, who tend to distribute them most broadly. This aligns with what social scientists have found. 

"In our study, we sought to develop a more rigorous understanding of social processes that underlies many cultural and economic phenomena," Bakshy said. "While some of our findings may seem quite intuitive, what I find most exciting is that we were actually able to test some rather controversial and competing hypotheses about the role of social networks in influence."

The researchers examined 130 days worth of gesture transfers in late 2008 and early 2009. They looked at 100,229 users and 106,499 gestures. They obtained the data from Linden Lab, the maker of Second Life. Personally-identifying information had been removed. 

Paper: "Social Influence and the Diffusion of User-Created Content."

The research is funded by the National Science Foundation. Physics graduate student Brian Karrer is also a co-author.