two-way anova with mi-data

[Joerg Eulenberger]

Dear Statalisters,
i want to run a two-way anova (like: anova weight rep7##foreign) with imputed Data.
However, mi does not support anova command. I read in the forum, the command regress are a alternative to do two-way anova with mi-data.
But the regress output does not include the F-Value for each variable like in the anova-output. How can I calulate the F and the Prob>F-value for the two variables an the intercationterm after regress like in the anova-output?
Thanks a lot,
Jörg

[Bruce Weaver]

Joerg, have you tried contrast after regress? I have not tried it with mi, but it certainly gives what you want under ordinary circumstances. E.g.,

clear *
webuse systolic
anova systolic drug##disease
regress systolic drug##disease
contrast drug##disease

Please let us know if this also works with mi.

Cheers,
Bruce

[Joerg Eulenberger]

Thanks Bruce,
contrast is what I want, but it does not work with mi:

code:
webuse mheart5
mi set mlong
mi register imputed age bmi
set seed 29390
mi impute mvn age bmi = attack smokes hsgrad female, add(10)
mi estimate: regress bmi hsgrad##female
mi estimate: contrast hsgrad##female

mi estimate: command not supported
contrast is not officially supported by mi estimate

Any other ideas?
Cheers , Joerg

[Marcos Almeida]

I believe this text may be helpful to you, in spite of the approach of somewhat different problem.

[Joerg Eulenberger]

Thanks a lot for the considerations and links,
Jörg