st: Marginal Means for Imputed Data

[Ada Keding <ada.keding@york.ac.uk>]

Dear all,

After identifying a group effect in a regression, I am having
difficulty extracting adjusted group means to illustrate the effect
when the data is imputed.

In the non-imputed case, I have:

             anova y group y_baseline
             regress
             adjust y_baseline, by(group)

‘Adjust’ gives me the adjusted means for each group, which differ by
the size of the group coefficient in the regression table as I would
expect.

I now use a multiply imputed data set and save the coefficients to
estimate adjusted group means:

             mi estimate, saving(miest): regress y i.group c.y_baseline
             mi predict xb_mi using miest
             mi estimate: mean xb_mi if group == 1
             mi estimate: mean xb_mi if group == 2

This again gives me means for each group, but the difference between
the group means is not that of the group effect coefficient (it is
much bigger).

So I don’t know how to report the group means for the group effect.

Maybe ‘mi predict’ is the wrong way to go about it, but I have not
found any working alternatives.

Any pointers are much appreciated.

Many thanks,

Ada Keding
Statistician
University of York

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/