topic Re: Proc Logistic Standard Errors in New SAS User

Proc Logistic Standard Errors

JJ_83 — Thu, 03 Dec 2020 20:26:33 GMT

I am working with a matched case control dataset and I am running the following proc logistic procedure.

proc logistic descending data = data;
	strata newgroup;
	model new_case = exp;
run;

newgroup is the matching variable.

new_case is the binary outcome variable (1/0).

exp is the binary exposure variable (1/0).

I am trying to replicate results that my coworker did in stata.

So my first question is:

How can I run logistic regression in sas with robust or cluster standard errors?
What kind of standard errors does proc logistic use? (I am assuming it is not robust or cluster)

My apologies if these are really stupid questions!

Reeza — Thu, 03 Dec 2020 20:30:46 GMT

I think you want a CLASS statement for your categorical variable as well.

SAS Procedures are very well documented, did you review the documentation and not find what you were looking for?
https://documentation.sas.com/?cdcId=pgmsascdc&cdcVersion=9.4_3.5&docsetId=statug&docsetTarget=statug_logistic_details.htm&locale=en

JJ_83 — Fri, 04 Dec 2020 15:36:18 GMT

Yes, I posted because I have not been able to find any answers through the documentation.