Virtually all social science research related to obesity uses body mass index (BMI), usually calculated using self-reported values of weight and height, or clinical weight classifications based on BMI. Yet there is wide agreement in the medical literature that such measures are seriously flawed...

As an example of the value of fatness in predicting social science outcomes, we show that while BMI is positively correlated with the probability of employment disability in the PSID, when body mass is divided into its components, fatness is positively correlated with disability while fat-free...