A controversial 2007 study concluding that traits such as obesity can spread among friends via a "social contagion" processes now faces a wave of criticism, most recently from SFI External Professor Cosma Shalizi and former SFI-affiliated researcher Duncan Watts.
The original study, co-authored by Nicholas Christakis and James Fowler, found that people who declared themselves friends often shared traits and behaviors such as obesity, happiness, depression, and smoking. They surmised that people calibrate their self-images to match those of friends, so when one friend adopts a certain behavior it spreads throughout his social network. The researchers also claimed that their statistical analysis allowed them to separate the effect of this social contagion from other potential explanations.
Now, Shalizi and a fellow statistician from Carnegie Mellon University have published a proof that demonstrates that social contagion cannot be differentiated from other explanations, such as shared environment or homophily -- people’s tendency to seek out friends like themselves.
“I think what the various critiques have driven home to many of us in the research community is the extraordinary difficulty of differentiating contagion from other plausible explanations, in particular homophily,’’ Watts comments in the Boston Globe.
Christakis and Fowler soon plan to publish a paper in response to their critics.
Read the New York Times article (August 8, 2011)
Read the Boston Globe article (August 7, 2011)
Read the Shalizi-Thomas proof (revised November 30, 2010)