I happened to be talking to some students today about their social network research, and we’re following up on that. Our penumbra paper—And then came the controversial 2000s study by Nicholas Christakis and James Fowler on the epidemic nature of obesity.
We covered this topic in this space in 2010 and 2011.
Controversy over social contagion
Controversy over Christakis-Fowler study findings on obesity epidemic
Christakis-Fowler Update
There we discussed critiques of Christakis and Fowler’s work, the work of economists Jason Fletcher and Ethan Cohen-Cole, mathematicians Russ Lyons, political scientists Hans Noel and Brendan Nyhan, and statisticians Cosma Shalizi and Andrew Thomas, and a response from the original authors, who wrote:
We do not claim that this study is definitive, but we believe it provides novel, informative, and provocative evidence on social contagion in longitudinally tracked networks. Along with other scholars, we are working to develop new methods for identifying causal relationships using social network data, and we believe this field is well-suited for statistical development, as current methods have known and often unavoidable limitations.
Here’s a quick summary:
1. Christakis and Fowler were doing interesting and innovative social science. They just went overboard with interpreting the data. You know the saying, “high risk, high reward”? That’s exactly what happened here. Despite the potential high reward, the study was ultimately a failure. Except that failure can help us avoid certain dead ends in the future.
2. The claim of a social epidemic of obesity is not supported by data from the Framingham Health Study. The critics (Fletcher, Cohen-Cole, Lyons, Noel, Nyhan, Shalizi, Thomas) were right.
3. Attitudes and behaviors have social effects, and they are difficult to study. As Christakis and Fowler write, this field lends itself to statistical development and is also suitable for experimental design and data collection.
I was curious how this work is cited now, 15 years later. There are 7,000 citations on Google Scholar. I searched for citations of the year, and here are the first few.
The first link above is to the book, and here’s what it says:
The following is an excerpt from a review article that cites the Christakis-Fowler paper as reference 31.
The following is a review of peer effects on “weight-related behaviors in adolescents,” with an incorrect summary.
I’m worried about the rest of the literature review!
The following is a paper on the community impact of microfinance loan defaults in a crisis, citing Christakis and Fowler for their methods rather than their substantive arguments.
In the following paper, reference number is 40.
And here it is cited too readily as reference 15.
And there’s this too:
They quoted Brian “Pizzagate” Wansink! Not good. If this blog was a drinking game, everyone should take a sip right now.
summation
There are false claims about an obesity epidemic out there, and hardly anyone seems to justify them with a comment about the critics. Sorry, Fletcher, Cohen-Cole, Lyons, Noel, Nihan, Shaliji, Thomas. Your efforts have been (almost) in vain.
P.S. In the comments section, Lyons shares a horrific example of a persistently refuted claim that is apparently being taught to medical students. Future doctors! This ongoing episode speaks ill of science, media, and academia.
