I have a binary dependent variable (infection + or -) and both continuous and categorical independent variables. One continuous independent variable (BMI) seems to be nonlinear in that the lowest and highest BMI are more likely to be + and the BMI values in the middle are less likely to be +. So when I plot BMI against the proportion that are +, I get U shaped line. How do I incorporate that into the logistic regression?
Asked
Active
Viewed 16 times
1
-
2If there's a scientific reason for this, the approach I describe at https://stats.stackexchange.com/a/64039/919 would be appropriate. Otherwise the solution pointed to by @jbowman looks good. – whuber Nov 06 '19 at 21:14