Goodness of fit -xtreg, re vce(cluster var)- and -xtlogit-

[Melina Selia]

Hello everyone,

I am running a random effects panel regression with -xtreg, re- and clustered standard errors. Since there is no adjusted R^2 available for the -re- option I was wondering how I can judge the goodness of fit of my model?

In addition to the above mentioned regression, I am also running a panel logit regression with random effects (-xtlogit, re) and again clustered standard errors. Also there, Stata is neither reporting residual deviance nor AIC and there seems to be no option to postestimate AIC. Same question as above, how can I judge the goodness of fit?

Thank you very much in advance for your help!!

Melina

[Andrew Musau]

There are some proposals for goodness of fit measures in linear mixed models, but there is no consensus on the appropriateness of these proposals. If you are in Economics, I would advise you not to bother with this and just report results of the robust Hausman test and the RE results. If you are in another field, look at published papers there and see what is the norm.