Can -mlogit- accommodate repeated choice tasks

[Anat Tchetchik]

Dear all,
I was wondering whether Stata's Multinomial logistic regression can accommodate the presence of repeated choice observations. I assume that using VCE(cluster id) is a good path but I'm not sure is this satisfactory for correcting for the repeated choice observation problem. All the examples I find for -mlogit- use one choice observation per individual.
any response will be appreciated.
Anat

[Andrew Musau]

If you have panel data, see xtmlogit in Stata 17+.
https://www.stata.com/new-in-stata/p...inomial-logit/

[Anat Tchetchik]

Thank you Andrew, it's not a panel, there is no time dimension. Each individual was facing 6 choice tasks sequentially. I'm not sure whether -mlogit- can handle this..

[Andrew Musau]

With sequential decisions, there is a time element. Does an individual choose among the same set of alternatives across the choice tasks? If that is the case, I do not see an issue using the conditional fixed effects estimator as choices are not independent across choice tasks. You do not need to specify a time variable if you are not using time-series operators. If you look at the example in the link in #2, only the group variable is used in the -xtset- command.

[Anat Tchetchik]

Thanks again Andrew Musau yes, each individual chooses among the same set of alternatives across the choice tasks. I'll use the xtmlogit.