I estimate a risk of cardiovascular disease by categories of weight (1->8) throug a Cox model.
And it seems that the risk is high for low and high weight, and low for middle weight. Thus I would like to modelize this by a "U shape" model.
When I trace the plot "Incidence rate" By "Weight categories" , I get something which looks like a U curve. I would be able to prove it statistically.
Thanks for your help