Dear members
My dependent variable is ordered and the data are clustered.
I would like to run a population-average (also known are a marginal) ordered probit model. However, xtgee (STATA 13) does not have this option (I can run a probit, but not an ordered probit).
A different approach would be to use random effects model. For instance, the command meoprobit runs a multilevel mixed effects ordered probit regression. However, I am reluctant to use a random effects model because I do not think that the unobserved, cluster level variables are uncorrelated with the other dependent variables.
Any ideas? I would really appreciate any help.
Thank you for your time,
Simone Silva
My dependent variable is ordered and the data are clustered.
I would like to run a population-average (also known are a marginal) ordered probit model. However, xtgee (STATA 13) does not have this option (I can run a probit, but not an ordered probit).
A different approach would be to use random effects model. For instance, the command meoprobit runs a multilevel mixed effects ordered probit regression. However, I am reluctant to use a random effects model because I do not think that the unobserved, cluster level variables are uncorrelated with the other dependent variables.
Any ideas? I would really appreciate any help.
Thank you for your time,
Simone Silva