Why Not Choose Obesity?

After reading the paper by Cohen-Cole and Fletcher and discussing it in class I could not help but to question their choice of choosing to study headaches, acne, and height. The authors chose these specific characteristics because they felt that they were not heavily determined by behavior or ones environment but I disagree with that claim. They say that these are unlikely to be transmitted socially but headaches, acne, and height can all be affected by an individual’s lifestyle and social network. For example high levels of stress and fatigue can cause headaches. Stress and fatigue can be linked to social networks and someone can be more at risk for stress according to the people that they are connected to. Similar social network effects can also be seen when examining acne and height. In light of this I wonder why Cohen-Cole and Fletcher decided to use these health measures. I agree with the authors that there could be environmental confounders that affect the Christakis and Fowler data and applaud them for challenging the idea but I think they should have focused on obesity and BMI when challenging Christakis and Fowler. If they devised methods to adjust for environmental factors with headaches, acne, and height why could they have not used those same methods to adjust the data for obesity? When trying to disprove a theory about obesity, proving that obesity shows no network effects makes more sense than showing that acne has no network effects.

While I was questioning the choices of Cohen-Cole and Fletcher I began to think that maybe they could not gain access to the Framingham Heart Study data like Christakis and Fowler were able to do. Regardless of the reason behind the authors not studying the Framingham data though, they still could have studied obesity data using the Add Health data. The Add Health data includes height and weight, the two measures for BMI, so they could have performed the same adjustment measures on obesity data. If they were able to prove that obesity showed no network effects when adjusting the Add Health data, then Christakis and Fowler’s argument would be significantly weakened. Instead they chose to examine headaches, acne, and height, which leads one to believe that perhaps they could not disprove the network effects on obesity and instead settled for data that they could use to argue the points made by Christakis and Fowler.