Lecture by UC San Diego professor James Fowler, part of the Hard Problems in Social Science symposium.
Frequent collaborators Fowler and Christakis are co-authors (along with Jaime Settle) on the recent paper, Correlated genotypes in friendship networks:
Maps of the friendship networks show clustering of genotypes and, after we apply strict controls for population stratification, the results show that one genotype is positively correlated (homophily) and one genotype is negatively correlated (heterophily).
But not everyone is convinced that friends connect on a genetic level:
The findings are based on patterns of variation in two out of six genes sampled among friends and strangers. But the claim is a hard sell for some geneticists, who say that the researchers have not analysed enough genes to rule out alternative explanations.
. . .
“If this was a study looking for shared genes in patients with diabetes, it would not be up to the standards of the field,” says David Altshuler, a geneticist at the Broad Institute in Cambridge. “We set these standards after 10 years of seeing so many irreproducible results in gene-association studies.”
Genetic Future is also skeptical:
I suspect most regular readers are already thinking suspicious thoughts, and with some justification: Genetic association studies on behavioral traits are notoriously littered with false positives, and the more headline-worthy the findings, the more caution we should apply. That’s not to say that all behavioral genetic studies should be discarded outright, but certainly we should be careful to set our threshold high for a study like this one.
Finally, a paper by Russell Lyons (PDF) criticizes the methods used for social network analysis:
We present cautionary examples of what can go wrong when assumptions behind statistical procedures are insufficiently examined, even when the analysis is performed by highly reputed and otherwise careful practitioners. Our examples come from a series of recent papers by Christakis and Fowler that claim to have demonstrated the existence of transmission via social networks of various personal characteristics, including obesity, smoking cessation, happiness, and loneliness. Those papers also assert that such influence extends to three degrees of separation in social networks.
More from Christakis and Fowler.