Random Effects Probit Model

[Eva Straka]

Dear Statalist community,

I am looking to do a random effects probit model but I am a bit unsure about the process. Do I have to specify certain errors in a certain form or is it simply sufficient to use the
xtprobit y x, re command?

And is there a way to test whether the assumptions for the random effects hold?

Thank you!

[Carlo Lazzaro]

Eva:
welcome this forum.
Your code is correct.
The main requirement of the -re- model is the absence of correlation between the vector of regressors and the panel-wise effect, which follows a normal distribution.
The hypothesis of no correlation is, in fact, assumed, even though you might suspect that it does not hold for your data. If that were the case, there are some (theoretically demanding) solutions, covered in any decent panel data econometrics textbook.

[Eva Straka]

Thank you!!