Storing the results of suest on MI data

[Kjell Weyde]

Dear Statalist users,
I am doing an analysis where I want to compare whether some associations differ between people who eat much Fish and People who eat less Fish. I have multiple imputed datasets, and use the suest command. The code looks as follows:

cap program drop mysuest
program mysuest, eclass properties(mi)
        version 15.0
 args model1 model2 

        qui `model1'
        estimates store est1
        qui `model2'
        estimates store est2
        suest est1 est2
        estimates drop est1 est2
       
        ereturn local title "Seemingly unrelated estimation"
end


eststo: mi estimate (diffAs: [est1_mean]logAs - [est2_mean]logAs) (diffHg: [est1_mean]logHg - [est2_mean]logHg), post nocoef: ///
mysuest "regress PCT_5mC logAs logHg i.sex i.smoking if TOTAL_Fish_Intake<23.72" "regress PCT_5mC logAs logHg i.sex i.smoking if TOTAL_Fish_Intake>=23.72&TOTAL_Fish_Intake!=."

 est store PCT_5mC_suest_fish

I then use -esttab- to make a table of the results in Word:

esttab PCT_5mC_suest_fish using PCT_5mC_suest.rtf, ci compress nostar nogaps nonumber noconstant nobaselevels brackets replace

¨The table, however, only shows the estimation results of the two linear regressions, not the output from the suest comparison (shown below), which is the one I want to have in the table. How can I achieve this?

Multiple-imputation estimates                   Imputations       =        100
Seemingly unrelated estimation                  Number of obs     =        630
                                                Average RVI       =     0.0108
                                                Largest FMI       =     0.0138
DF adjustment:   Large sample                   DF:     min       = 523,849.92
                                                        avg       =   1.01e+07
Within VCE type:       Robust                           max       =   1.97e+07

       diffAs: [est1_mean]logAs - [est2_mean]logAs
       diffHg: [est1_mean]logHg - [est2_mean]logHg

------------------------------------------------------------------------------
             |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
      diffAs |   .0027628   .0092464     0.30   0.765    -.0153597    .0208854
      diffHg |   -.010537   .0154482    -0.68   0.495    -.0408149     .019741
------------------------------------------------------------------------------

Best regards,
Kjell

[daniel klein]

I have not thought about your general approach here, so I cannot comment on that. Also, this is a bit of a guess since I have not tried, but I believe that esttab (you should have told us where it is from) picks results from e(b) and e(V); the results of your diff* are stored in e(b_Q_mi) and e(V_Q_mi), respectively.

Best
Daniel