Mixlogit- "Some variables are collinear, check your model specification"

[Rosa Mak]

Hello,

I am doing a mixlogit regression for a DCE.

This is the command that I am using- mixlogit choice smartfeature, id(id) rand(highenergyefficency mediumenergyefficiency lowenergyefficiency) group( iid)

I collected this data regarding the preferences of consumers. When I try to run this without medium e.e, the p-value comes out to be very high but the regression does run. whereas, In all other cases, the regression does not run and shows the following message- "Some variables are collinear - check your model specification".

I will really appreciate all the help that I can get.

Thank you.

[Hong Il Yoo]

Am I right in thinking that "highenergyefficency mediumenergyefficiency lowenergyefficiency" are three zero-one indicators of mutually exclusive categories that add up to 1? If yes, you can look up "dummy variable trap" and "perfect (multi-)collinearity" on Google or in almost any of popular econometrics textbooks.

[Rosa Mak]

Yeah exactly, I figured out and will be using one of the levels as a base variable.