[Limdep Nlogit List] LR test in ordered probits with weights

[Andreas Drichoutis]

Dear all,

 

Is it valid to use a likelihood ratio test after estimation of nested
ordered probits when using probability weights? I was warned that the
likelihood after a weighted estimation is not a true likelihood. If this is
true, then are goodness of fit measures (like pseudo r-squared and AIC)
valid for comparing between non-nested models?

Can anyone advice on other alternatives?

 

Best,

Andreas Drichoutis