suest with multiple imputations

[Pat Goldsmith]

Hi all:
This is my first post. I'm an old time SAS user converting to stata. I want to compare coefficients between two models using data with multiple imputations. I'm using areg because I am using state (fips) fixed effects. Here is the code. Achievemath and seshsp are continuous variables


mi estimate: areg achievemath seshsp if newdestination==1, absorb(fips)
est store ndach
mi estimate: areg achievemath seshsp if newdestination==0, absorb(fips)
est store odach
suest ndach odach

I get this error message:

impossible to retrieve e(b) and e(V) in ndach
r(198);

end of do-file


I can't find anything about that message on line. Unfortunately, I cannot produce the results I get from the separate models using regress and lots of interactions.

Any solutions?

[daniel klein]

You could put the b and V matrices where suest expects them with mi estimate's post option. Whether this suffices to get results and whether those results are valid, I do not know. I would probably write a wrapper program; see this FAQ.

Best
Daniel

[Pat Goldsmith]

Thanks for the suggestions. The faq looks very promising.