Hello,
This question is about mimrgns, Daniel Klein's great command available on the SSC. I am running a GEE on multiply imputed data in Stata 14. The model includes an interaction term and I would like to obtain the marginal effects of the variables in the interaction. I have read the helpful discussion between Klein and Richard Williams on mimrgns and interaction effects here. However, I am receiving an error when using the mimrgns command after the xtgee model. Here is an example of what I am trying to do:
The error says "imputations(): invalid numlist".
When I run the model as a regular OLS, the post-estimation works great. But, because there is a good deal of clustering in the data, I am trying to use xtgee, which is also what has used for the same data in previous research. Is there a way to do this? Should I abandon the MI data so I can do the estimation?
Many thanks for your help,
Robbie Dembo
This question is about mimrgns, Daniel Klein's great command available on the SSC. I am running a GEE on multiply imputed data in Stata 14. The model includes an interaction term and I would like to obtain the marginal effects of the variables in the interaction. I have read the helpful discussion between Klein and Richard Williams on mimrgns and interaction effects here. However, I am receiving an error when using the mimrgns command after the xtgee model. Here is an example of what I am trying to do:
Code:
mi estimate: xtgee y i.x1##c.x2 x3 x4, i(id) corr(exch) robust mimrgns, dydx(x2) at(x1= (0 1))
When I run the model as a regular OLS, the post-estimation works great. But, because there is a good deal of clustering in the data, I am trying to use xtgee, which is also what has used for the same data in previous research. Is there a way to do this? Should I abandon the MI data so I can do the estimation?
Many thanks for your help,
Robbie Dembo
Comment