The papers have recently been full of stories about yet another interesting pattern that researchers have gleaned from the data collected in the Framingham heart study. You can get a taste for the result by watching this annimation. What is shows is a social network (genetic and friends) changing over time. The primary focus of attention is body fat, and over time the member of this community get fatter, a lot fatter.

The sound bite associated with this study is that fat is contagious; and indeed the study showed that friends of fat folk had a higher likelihood of becoming fat than other people. Social networks exhibit a tremendous amount of “birds of feather flock together.” If I buy an conditioner, take up bicycling, start wearing a hat, or take a particular political position it is likely that I am following the lead of some of my friends and that my choice will lead others to come along.

While all behavior is socially viral I am troubled by the way that lots of people have gotten the impression that this study proves that weight gain is viral; i.e. the association with fat people is effectively dangerous because becoming fat is dangerous. I have trouble seeing how that conclusion isn’t exactly like the presumption that association with poor people (think here of the local high school) is likely to make you poor. While both of these may well have a modicum of truth to them the puzzle is exactly how concerned about the risk should a person be?

Reading the paper I’m struck by the apparent absence of references to other work on how ideas, behaviors, and real viruses spread across social networks. I.e. there id no framing of exactly how contagious this effect is relative to other things.

I’d love to see similar animations for other behavioral affectations. For example people who own air conditioners, folks who drive large cars, eating out at chain restaurants. It is inconceivable that the modern marketing industry doesn’t have reams of data like that. The this study is looking at something with a strong health component colors how we think about this. If the same data showed internet usage, or cell phone adoption, then we would presume that the primary driver certainly as much or entirely techo-economic rather than dragging in questions of individual will and infection. Of course ideas like choice, memes, viral spread would remain interesting but the wouldn’t be so highly energized.

Causality is very complex stuff; but I don’t see how this study is a big help in getting at the root causes of obesity. If we had a similar drawing that showed the association with internet users substantially increased the chance of adopting becoming an internet user would we announce that it was contagious? What about gun ownership? What about being a Republican? This animation looks to me like a simple illustration of how a behavior gains market share; i.e. decreasing barriers to adoption enable increased market share. That adoption trickles across the social network seems mind bogglingly obvious and actually quite unexceptional.