Identifying Endogenous Peer Effects in the Spread of Obesity

Recent research in the New England Journal of Medicine (NEJM) purports to show the existence of peer effects in the spread of obesity. Using a dataset of 5124 residents from Framingham, Massachusetts spanning the years 1971 to 2003, the authors show correlations between own weight gain and friends’ and relatives’ weight gain over this period. They find, furthermore, that these results are strongest for males and weaker for females. We use the Adolescent Health Survey, a nationally representative dataset of seventh through twelfth graders in 1994 and 1996 to examine the effect of peers on weight gain. Despite the differences in the samples, we are able to replicate the pattern of results in the NEJM study. However the results are not robust to alternative definitions of the outcome variable. Furthermore, due to the various identification issues that are unresolved in both this and the NEJM paper, we conclude that the evidence for contagion effects in the spread of obesity is only suggestive at best.