Hi Statalist,
I used hausman test to my panel data to see whether a fe/re fits better. The result is significant so a fixed effect logit (xtlogit,fe) is suitable. However, since I also want to observe the coefficients of the time-invariant variables (as a beginner, not sure whether it's correct to say so), I choose the mixed-effect logit (melogit). But how to justify this choice? Can I use hausman test between melogit and fe/re and how to interpret its significance? Thanks!

Regards,
Melody