I have a choice experiment survey, but we implemented the survey by assigning each respondents with only one choice card. This means that in the stata commend using mixlogit, my group(gid) and id(pid) are the same. Could this cause a problem using mixlogit? I have many attributes, and it seems like they don't converge. and there are many collinearity problems while assigning dummy variables.
Any suggestions?
Any suggestions?
