Using eststo with mi estimate - Help needed

[alex badalyan]

Hi,

I am trying to create a set of estimation tables, similar to eststo and esttab using .rtf to be included in my work. However, with mi estimate: this does not seem to work.

Previously I used something like this:

quietly xtreg HHS1 TFP1 i.year i.code2
eststo
quietly xtreg HMS1 TFP1 i.year i.code2
eststo
quietly xtreg HLS TFP1 i.year i.code2
eststo

esttab using table20.rtf, se b(%10.3f) scalars(N r2 rmse) indicate("sector fe = *code2" "time fe = *year") starlevels( * 0.10 ** 0.05 *** 0.010)

Is there a similar solution for multiply imputed data?

Best,

Alex

[Weiwen Ng]

Hi,

I am trying to create a set of estimation tables, similar to eststo and esttab using .rtf to be included in my work. However, with mi estimate: this does not seem to work.

Previously I used something like this:

quietly xtreg HHS1 TFP1 i.year i.code2
eststo
quietly xtreg HMS1 TFP1 i.year i.code2
eststo
quietly xtreg HLS TFP1 i.year i.code2
eststo

esttab using table20.rtf, se b(%10.3f) scalars(N r2 rmse) indicate("sector fe = *code2" "time fe = *year") starlevels( * 0.10 ** 0.05 *** 0.010)

Is there a similar solution for multiply imputed data?

Best,

Alex

Use -mi estimate, post-

[alex badalyan]

Hi Weiwen, thanks : could you please provide an example? I am struggling to find clues in https://www.stata.com/manuals13/mimi...estimation.pdf

[Weiwen Ng]

Hi Weiwen, thanks : could you please provide an example? I am struggling to find clues in https://www.stata.com/manuals13/mimi...estimation.pdf

Try:

mi estimate, post: xtreg HHS1 TFP1 i.year i.code2
eststo

That should work. The -post- option will post the e(b) and e(V) matrices, and -eststo- uses those. Without the -post- option, -mi estimate- doesn't return those matrices for reasons outlined in the document you linked. However, as far as I know, it is totally OK to use those matrices to tabulate results! The document merely cautions against using other post-estimation commands.

[alex badalyan]

Thanks a lot Weiwen!

[Felipe Alvarez]

Thank you very much Weiwen!