topic Panelized likelihood estimation for PROC GLIMMIX in Statistical Procedures

Panelized likelihood estimation for PROC GLIMMIX

Kim2009 — Mon, 26 Apr 2021 17:30:21 GMT

Hello -

I would like to know if there is a way to replicate the Firth's regression (PROC LOGISTIC) in PROC GLIMMIX? I have clustered data and a rare binary outcome.

My outcome variable: Visits within 30 days (viz30day)

Covariates: group (1=treatment, 2 = control), Age at baseline (agebase ; continuous), race (1=white, 2=black, 3=other race/ethnicity), insurance status (0=no, 1 =yes).

Any resources or code you are able to provide is greatly appreciated.

Thank you, Kim

Re: Panelized likelihood estimation for PROC GLIMMIX

SteveDenham — Tue, 27 Apr 2021 12:38:22 GMT

All of the factors you list seem (at least to me) to be fixed effects. I know you said your data was "clustered" and I can see that driving to GLIMMIX, but if you are in need of penalized estimation that won't help. What about considering the clusters as a fixed effect or as strata, and getting narrow-inference space results.from LOGISTIC? I suppose this is a relatively gentle way of saying GLIMMIX really doesn't have an analog to the Firth algorithm that I am aware of, and that LOGISTIC is likely the best approach (or PHREG, as it looks like there are some examples in the documentation that might be of interest)..

SteveDenham

Re: Panelized likelihood estimation for PROC GLIMMIX

Kim2009 — Tue, 27 Apr 2021 12:47:19 GMT

Thank you, Steve. Your suggestions are very helpful.