Can we use -asmixlogit- to run multinomial logistic regression for panel data?

[Linh Nguyen]

Dear everyone,

Stata 15 has the -asmixlogit- command. I read its manual (https://www.stata.com/manuals/rasmixlogit.pdf), but I'm not sure whether -asmixlogit- can be used for multinomial logistic regression for panel data.

Is there anyone having experience of this command?

[Brianne Altmann]

I am sorry that I do not have an answer, but I too am wondering the same thing. I purchased STATA exclusively for the analysis of a discrete choice experiment that has panel data. Did you manage to find the answer to this question? I am missing the old 'group' code from the old 'mixlogit'. Is this the new 'alternatives' syntax?

[Arne Risa Hole]

You can still use mixlogit, which supports panel data (type "ssc install mixlogit" in Stata to install).

asmixlogit cannot be used to fit mixed logit models to panel data, as far as I know.

[Linh Nguyen]

Hi Brianne, I haven't had the answer yet. My solution for this problem is as follows:

- First, I try to re-group 6 categories into 2 categories to run -xtlogit-

- Next, I use -mlogit- with -vce(cluster id)- to partially correct the standard errors (as shown at https://www.stata.com/statalist/arch.../msg00733.html) as an additional finding for -xtlogit-.
and I am looking for a citeable preference for the second step.