Nicholas Christakis (at the Harvard School of Public Health) (together with James Fowler at U. Cal. San Diego) has done very interesting research with the landmark Framingham Heart Study looking at the likelihood of obesity. They find, after tracking some 12,000 individuals through the 32-year long study (1971-2003), that even controlling for all the genetic markers for obesity (like parents’ obesity) and weight at the study’s baseline, having fat friends increases the chance that you will be fat. [The Framingham study participants could list only up to 3 friends, so the friends being discussed here are closer, stronger friends, rather than weaker ties.]
Having an obese friend increased one’s likelihood of obesity by 57%, with a smaller effect for siblings and spouses. This risk of developing obesity rose to 171% for the closest mutual friendships. Having very large neighbors had no effect on obesity if those neighbors weren’t friends of the individual in question.
The study found that the network impact of friends on obesity could be seen in as small as 2-4 year increments.
Christakis discounted other likely factors such as environmental (since it didn’t matter how close geographically ones’ friends were, making it less likely that they are both responding to something in the neighborhood, for example). They did code the data for density of fast food restaurants and it did not make this effect go away. [The fact that long-distance close friends influence obesity as much as nearby close friends is very surprising; for sure, if respondents are still listing this long-distance friend as one of their three closest friends, they must have stayed in regular contact, but I would have thought that obese friends made respondents feel more comforable being obese partly by physically seeing an obese friend, and this is presumably less common with geographically distant friends.
And Christakis thinks the mechanism is “induction”; having an overweight person list YOU as their friend doesn’t increase your likelihood of being obese, but your listing an overweight person as YOUR friend does. The impact of these friendships decreases with social distance (in other words, your friends affect your weight more than your friends’ friends, which is more consequential than the weight of your friends’ friends’ friends, etc.) but they still have an impact out to three degrees of separation. And same sex friends influence your weight gain more than opposite sex friends.
Christakis thinks that people with heavier friends either come to think of themselves as less fat or else it validates their obesity in a way that wouldn’t be the case if their friends were thinner. (But the study showed that it wasn’t the simple story of a change in physical exercise or eating habits from these friends.) And there is homophily in obesity — in other words, fat people are more likely to choose overweight friends and thin people are more likely to pick thin friends. And if overweight people have thin friends or vice-versa, these relationships tend to be less stable over time (the thin people are more likely to drop their overweight friends or become more obese).
The New York Times notes that: “Science has shown that individuals have genetically determined ranges of weights, spanning perhaps 30 or so pounds for each person. But that leaves a large role for the environment in determining whether a person’s weight is near the top of his or her range or near the bottom. As people have gotten fatter, it appears that many are edging toward the top of their ranges. The question has been why. ” (“Find Yourself Packing It On? Blame Friends, NYT, 7/26/07). The study suggests that social contagion of obesity through networks may be the explanation.
Christakis thinks that the role of social networks may be one of the explanations for the increasing obesity in America (along with other factors like exercise, change in eating habits like supersizing of food and more fast food and soft drinks, etc.). The social networks may have changed norms and made weight gains more acceptable, even outside of any change in behavior. Moreover Christakis thinks that we might be able to use the structure of social networks to fight obesity, by for example fighting obesity in groups rather than with individuals. If there can be a social contagion of obesity, how might we start a social contagion of weighing less? Co-author James Fowler noted that having a friend that was able to lower his or her weight down made it easier for one to lose weight; that’s why weight loss programs often function using groups, to reinforce the attempted change in behavior. But this question of spurring a social contagion for good is exactly the kind of question that Malcolm Gladwell in The Tipping Point considers.
Christakis and colleagues have also found similar network properties in this study to the spread of smoking, but the cessation of smoking was not a factor in individuals gaining weight or not.
Finally Christakis thinks that we may undervalue health interventions since we look only at the impact of the intervention on one individual rather than examining the multipliers on this investment through his/her social network.
The authors note that there are many social and health benefits of friendship so their study is not a reason not to develop friends with anyone.
The Christakis et al article is published today in the New England Journal of Medicine and a summary of some of the findings is available in this Harvard Gazette story.
There’s a neat visual representation of the obesity spreading through the social networks over time in this video.
See also, Clive Thompson, “Is Happiness Catching?” (NYT Sunday Magazine, 9/13/09)