mixed logit model

[Alan James McMichael]

Hello all,

I am new to the forum and looking forward to talking with all of you, and hopefully at some stage I will be able to help someone but right now I need help. Here is some brief background. I am running a mixed logit model in stata to determine what attributes of a treatment infiluence treatment choice

at present I have treatment choice as the dependent variable and "lung exacerbation, FEV1, headaches, chest tightness" are all my independent variables.

here is my stata code at present

mixlogit treatment_choice, id(id ) group(gid) rand(lung_exacerbations FEV1 headaches chest_tightness) nrep(1000)

this runs fine and I get a coefficient for each of the attributes. However, each attribute has three levels for example "headaches is broken down into " low" for some treatments "medium for others and high for others.

My question is, is there any way I can run the model to get a coefficient, not only for the attribute (which I can do) but also for each level of each attribute? So I would have a coefficient for low headaches, medium headaches and high headaches?

and hopefully I could also do this for chest tightness, FEV1 and lung exacerbations

Any help would be greatly appreciated

best wishes

Alan

[Stephen Jenkins]

Factor variables are your likely friends: help fvvarlist. See also, for post-estimation interpretation, margins

[Nick Cox]

Cross-posted at http://www.talkstats.com/showthread....each-attribute

Please note our cross-posting policy, which is that you should tell us about it.