Social contagion? Maybe not…

July 28, 2010

(This article was first published on Brokering the Closure » R, and kindly contributed to R-bloggers)

Recently there was a lot of racket about results presented in papers by a pair Nicholas Christakis & James Fowler. Their papers gained a lot of attention both in the academic world as well as in media. In their papers they claim to provide evidence for social contagion (transmission through social networks) of several types of individual characteristics. They wrote articles addressing, among other, social transmission of: obesity (friends of obese people are likely to be obese too), smoking (smokers tend to be friends with smokers), happiness (friends of happy people tend to be happy too), or loneliness (people are more likely to feel lonely if their friends feel lonely too, my personal favourite…). All of these are based on data from Farmingham Heart Study about which I wrote some time ago. For the reference, the relevant papers are listed at the bottom.

Authors claim to find very strong effects which they attribute to social transmission of the studied characteristic (obesity, smoking etc.) dismissing other potential explanations based on various arguments. I am sure for anybody interested these things the results were jaw-dropping. For example, in the case of obesity, the reported result was that if your friend is obese then you risk for becoming obese increases by 57% (!) as compared to the situation if your friend did not become obese.

For an overly-critical person like me the results were not so much jaw-dropping but eyebrow-raising. Among other things, to me, the results were based on somewhat strange analytical techniques where, at the same time, social networks literature suggests different approaches. Christakis & Fowler were not referring to the existing methodology at all. Most notably to SIENA models for network and behavior dynamics or models for social selection and social influence developed by people at MelNet.

I was not the only one not fully convinced, see for example here or here.

Anyway, I’m not going to review all the results here as somebody did that pretty well recently. The paper The Spread of Evidence – Poor Medicine via Flawed Social Network Analysis was uploaded couple of days ago to the Arxiv. The author, Russel Lyons, mathematician from Indiana University, takes a closer look at the papers I mentioned. In general, he finds flaws that range from problems in the arcane details of estimation strategy to undergraduate-level mistakes in interpreting confidence intervals. All the flaws Lyons finds fall into two categories:

  1. Certain aspects of statistical techniques used by Christakis & Fowler are not justified.
  2. The numerical results obtained are misinterpreted.

The bottom line is: substantive claims Christakis & Fowler make are not supported by results they show. Again, I’m not going to copy-paste from Lyons’ paper. Have a look yourself here.

Couple of end-thoughts:

  • It’s great that somebody took a detailed and closer look at this research. It is a contribution to the public good.
  • How come the mentioned papers by Christakis and Fowler passed the reviews in, what seems to be, quite respected journals? Especially that some of the mistakes seem to be very basic.
  • Perhaps we need to move forward from the present journal reviewing system to something like Open-Source Science 2.0? All the scientific publications should be transparent as it goes for data and methods used. I’m thinking about systems in the flavor of “literate statistical analysis” like Sweave in R. Moreover, all scientific publications could be reviewed and commented upon publicly on the Web? Much like Talk pages on Wikipedia…

Some of the papers by Christakis & Fowler I’m referring to:

  • Christakis & Fowler (2007) “The Spread of Obesity in Large Social Network over 32 Years”, N. Engl. J. Med., 357:370-379
  • Christakis & Fowler (2008) “The Collective Dynamics of Smoking in a Large Social Network”,N. Engl. J. Med., 358:2249-2258
  • Fowler &  Christakis (2008) “Dynamic spread of happiness in a large social network: longitudinal analysis over 20 years in the Framingham Heart Study,” Brit. Med. J., vol. 337, p. a2338, 2008. doi:10.1136/bmj.a2338.
  • J. T. Cacioppo, J. H. Fowler, and N. A. Christakis, (2009) “Alone in the crowd: the structure and spread of loneliness in a large social network,” J. Personality Soc. Psych., vol. 97, no. 6, pp. 977–991.



(2010-07-30) Seems like Lyons paper is a bit of an old news. The first version was available on his website at least in April, as featured by this piece at Slate. Also, see this article at NYT from September 2009.

Filed under: networks, R, statistics

To leave a comment for the author, please follow the link and comment on their blog: Brokering the Closure » R. offers daily e-mail updates about R news and tutorials on topics such as: Data science, Big Data, R jobs, visualization (ggplot2, Boxplots, maps, animation), programming (RStudio, Sweave, LaTeX, SQL, Eclipse, git, hadoop, Web Scraping) statistics (regression, PCA, time series, trading) and more...

If you got this far, why not subscribe for updates from the site? Choose your flavor: e-mail, twitter, RSS, or facebook...

Tags: , , , , , , , , ,

Comments are closed.


Mango solutions

RStudio homepage

Zero Inflated Models and Generalized Linear Mixed Models with R

Dommino data lab

Quantide: statistical consulting and training



CRC R books series

Six Sigma Online Training

Contact us if you wish to help support R-bloggers, and place your banner here.

Never miss an update!
Subscribe to R-bloggers to receive
e-mails with the latest R posts.
(You will not see this message again.)

Click here to close (This popup will not appear again)