How to write logistic regression equation using proc logistic

htst1003 — Sun, 21 Nov 2021 21:52:06 GMT

Hi, this is data I have.

dependent var: outcome (1=cured 2=not cured)

independent var : treatment (1=A, 2=B, 3=C)

diagnosis (1=contaminated, 2= uncontaminated)

This is code for logistic regression

proc logistic data=test;
class outcome treatment diagnosis/ param=reference;
model outcome= treatment diagnosis;
run;

and I got this (ref variable treatment 3 and diagnosis 2)

If I have to write model, is this correct? I dont have to include reference variable in the full equation?

logit(cured)=1.42 +0.58(Treatment1)+1.56(Treatment2) - 0.96(diagnosis 1)

How can I find full model? Am I doing the right thing?

Re: How to write logistic regression equation using proc logistic

PaigeMiller — Sun, 21 Nov 2021 23:09:09 GMT

Except for the fact that you have rounded the regression coefficients to two decimal places (which I would not do), it appears that your equation is correct.

This is the "full model". SAS parameterizes models with class variables to have the last value alphabetically set to zero, that's why you don't see a treatment 3, because the coefficient is set to zero. You can request other parameterizations, but you still get the same fitted model and the same predictions.

topic How to write logistic regression equation using proc logistic in SAS Studio

How to write logistic regression equation using proc logistic

Re: How to write logistic regression equation using proc logistic