Mixed-effect with interactions of categorical variables

[Yaliu He]

I am running a mixed command to test a random coefficient model: patients were nested within therapists.

It includes a three-way interaction of measure x time x treatment modality. Both measure and treatment modality are categorical variables with more than two categories. Measure was coded as having 5 categories (1-5) and treatment modality had 3 categories (1-3).

STATA command:

mixed clinicalscore i.measure##c.time##i.modality||therapist: ||clientid: time,ml

STATA output:

I want to get only one omnibus test to have ONE p-value for this three-way interaction. In other words, I want to have one p-value to tell me if this three-way interaction is significant or not. Can anyone help me to get the command to do so? Thank you very much!

[Clyde Schechter]

testparm i.measure#i.modality_atsession#c.rtime

[Yaliu He]

Thanks for the response. It was very helpful. One follow-up question: in addition to the p value, what code can I use to get a beta and standard error (SE) for this three-way interaction? Thanks a lot!