I wonder if there is a way to calculate an "adjusted percentage" in
SAS when we run a logistic regression adjusting for standard variables
like age and gender; something similar to what we do with the LSMEANS
statement when we run a GLM.
I don't understand what you're referring to. Logistic regression computes odds ratios as its parameters. You can use it to calculate probabilities, but I'm having trouble seeing how you'd calculate percentages. Care to elaborate?
I'll try to explain better. when you run a GLM in the way
proc GLM; model continuous outcome= categorical IndepVar + confounders ;
you can either compute means of your DV over your IV strata with the MEANS option, or calculate in the same way standardized means with the option LSMEANS adjusting for other variables included in you model.
Now I have a dichotomous outcome and I would like to do the same, to have adjusted percentages over my independent variable adjusted for other variables. Proc freq will be fine for actual percentages, but I don't know if its possible to calculate "adjusted percentages". So that's my question.
Hmmm...I've never heard of anyone wanting to calculate percentages inside each strata of proc logistic before, so I can't really help you on that path.
However, what you're asking for could easily be done with nested loops in a macro. You'd just divide each of the independent variables up into strata and calculate percentages. This might be good practice to get better at macros. Think you're up for it?