Comment by plutomeetsyou

> Adjusted R2 values for each model were 0.085 (linear), 0.118 (quadratic), and 0.125 (cubic)

I get the study design is not causal and all, but this R^2 looks very underfit for a study that claims a stronger predictor?