I am running a negative binomial regression in Stata and would like to know if I need to include the vce(robust) option in the model. I know the negative binomial regression is the best for the data I have. I've combed through so many sources and it seems like negative binomials already seem to take care of some heteroskedasticity, but I would really like a test that would let me know if there is still some in the model that needs to be taken care of via the vce(robust) option. I know there is the hettest but I can only use that in a regress model, not a nbreg (negative binomial regression) model. I also saw this advice:

"You could try plotting the absolute value of Pearson residuals from your NB regression against covariates (at least, covariates you think might affect the overdispersion) or against the fitted means. If there's a discernible trend, this suggests non-constant variance."

But what constitutes 'a discernible trend'? How do I know if that discernible trend would be significantly better taken care of by the vce(robust) option or it wouldn't matter?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.