topic Re: How to retain variables in backward elimination in SAS Programming

How to retain variables in backward elimination

angouldi — Sun, 20 Nov 2022 20:04:00 GMT

Hello,

I am trying to complete a backward elimination analysis to select covariates for a logistic regression model. I would like to retain my key exposure variable (quartile_SDI) in the model, and only consider removing potential confounding variables, but I am struggling to figure out the code. Below is the code that I am using:

proc logistic data=pas.ranked_quart_3;
class RBC_4 (ref='0') ante_dx (ref='1') anemic (ref='0') anyDM (ref='0')/ param=ref;
model RBC_4 = quartile_SDI age BMI_delivery prior_CS ante_dx anyDM anemic
/ selection=backward fast slstay=0.1
ctable pprob=(0 to 1 by 0.05);
run;

Any guidance would be appreciated. Thank you for your help!

Alison

Re: How to retain variables in backward elimination

FreelanceReinh — Sun, 20 Nov 2022 22:58:58 GMT

Hello @angouldi,

Use the INCLUDE= option of the MODEL statement: Insert include=1 after the slash (e.g., between your existing SLSTAY= and CTABLE options) and SAS will use the first explanatory effect listed in the MODEL statement, i.e., quartile_SDI, in every model during the variable selection process.

Re: How to retain variables in backward elimination

angouldi — Mon, 21 Nov 2022 04:40:22 GMT

Thank you! This is exactly what I was looking for.