Logistic model with fixed effects and robust standard error

sdan23 — Fri, 11 Feb 2022 19:18:13 GMT

Dear SAS users,

I have a cross-sectional data but I would like to do state fixed effects with state robust standard errors. My outcome is binary so I was going to use Proc Logistic with the Strata function but it seems like I can't do robust standard errors (I think). Proc Genmod can adjust for robust standard but I couldn't figure out how to do the fixed effects.

So am I only left with the option to use proc glimmix? Did I miss something? Is there a way I can compress the output of the coefficients for state?

proc glimmix data=data;

class X2 state;

model y = x1 x2 x3 state/dist=binary link = logit oddsratio(diff=all) solution;

random intercept / subject=state;

run;

Thank you very much in advance!

Re: Logistic model with fixed effects and robust standard error

sbxkoenk — Fri, 11 Feb 2022 20:30:40 GMT

Hello @sdan23 ,

See the answer from @StatDave on this topic :

Logit with robust standard errors
https://communities.sas.com/t5/Statistical-Procedures/Logit-with-robust-standard-errors/td-p/469174

Cheers,

Koen

topic Re: Logistic model with fixed effects and robust standard error in Statistical Procedures

Logistic model with fixed effects and robust standard error

Re: Logistic model with fixed effects and robust standard error