An empirical note on imitative obesity and a puzzling result

Using data from the Behavioral Risk Factor Surveillance System for 2006, I test recent theoretical predictions on social comparisons influencing individual Body Mass Index (BMI). I find that in particular the average BMI of individuals in the same county-age-gender-cell as the respective individual influences BMI. Evidence from quantile regressions points towards significant heterogeneity of effects along the distribution. However, there is no evidence of some individuals becoming slimmer as a result of preferences for deviant behaviour. Life satisfaction regressions show a positive effect for BMI relative to the county average. Paradoxically, BMI relative to the cell average seems to have no effect on life satisfaction. These two results contradict most theories of social comparisons.