collinearity diagnostics for ordered logistic regression

[Ibraheem Khalid]

Dear researchers,

I am trying to get the VIF after running the xtologit but it doesn't work with me?
Any help would be very appreciated.

Many thanks

[Clyde Schechter]

If you really have nothing better to do with your time than run VIF, then just redo the regression with plain old -regress-, and follow that with -estat vif-. The variance inflation factors depend only on the covariance matrix of the predictor variables, and that is the same in an regression command.

But really, don't bother. See my response at #2 in https://www.statalist.org/forums/for...ring-variables. While the focal question there was about centering, it was in the context of exploring multicolinearity. There really is no need to do VIF as a diagnostic. Just assess the adequacy of your standard errors on the variables of key interest. (By the way, I agree with Jeff Wooldridge's comment later in that thread about centering variables before calculating quadratic and higher power terms--but that is a separate issue.)

[Ibraheem Khalid]

It's really great information. I sincerely appreciate your interest and cooperation.

Million thanks for the help.