Problem with mi estimate, post and margins, post

[Jonathan Horowitz]

I figure someone has to have answered this already on Statalist, but I can't seem to find it. If so, please point me towards it. If not, help?

I'm using mi estimate, post to run a fixed effects model, like the following:

xtreg y i.x1 i.x2 i.x3 [pweight=wt], fe vce(robust)

Then I want to use margins, so I use the following command:

margins i.x1 i.x2 i.x3, nose post

And I get the following error:
e(sample) does not identify the estimation sample

I'm guessing this has something to do with the multiple imputation datsets (since it worked just fine before imputation). But I can't seem to figure out how to solve it. Advice is appreciated.

Best,
Jonathan

[daniel klein]

Try mimrgns (SSC) but omit the post options when not absolutely necessary. mimrgns will also not allow you to suppress standard errors.

Best
Daniel

[Jonathan Horowitz]

Thanks! I will give that a shot and report back if I can't figure it out from there.

[Jeph Herrin]

You may also find this helpful: https://stats.idre.ucla.edu/stata/fa...-imputed-data/

cheers,
Jeph