topic Re: Logit with robust standard errors in Statistical Procedures

Logit with robust standard errors

MarinaMag — Mon, 11 Jun 2018 11:16:36 GMT

Hello,

I want to run a logit regression.

Is the procedure as follows?

proc qlim data=...;

model=... / discrete(d=logit);

run;

Is there a command to run the logit with robust standard errors (e.g like the "cluster" in "proc surveyreg")?

Thank you.

Re: Logit with robust standard errors

StatDave — Mon, 11 Jun 2018 14:13:54 GMT

See this note for the many procedures that fit various types of logistic (or logit) models. QLIM is generally not the first choice. For randomly sampled data with independent observations, PROC LOGISTIC is usually the best procedure to use. If you have complex sample survey data, then use PROC SURVEYLOGISTIC. If your interest in robust standard errors is due to having data that are correlated in clusters, then you can fit a logistic GEE (Generalized Estimating Equations) model using PROC GENMOD. See the examples in the documentation for those procedures.