Hello Statalist,

I am using Stata 16 and I am trying to obtain the exponentiated form of my estimates with multiple imputation (mi estimate) using xtgee with the eform option, however, the estimates I obtain are the same as when there is no eform option.

Before using mi estimate, I used mi impute chained.

A simplified version of my code without eform is:

Code:
. mi estimate: xtgee vf gender age, family(binomial) link(logit) corr(exch) vce(robust)

Multiple-imputation estimates                   Imputations       =         30
GEE population-averaged model                   Number of obs     =        496

Group variable:               clinic_bsl~m      Number of groups  =         16
Link:                                logit      Obs per group:
Family:                           binomial                    min =         10
Correlation:                  exchangeable                    avg =       31.0
Scale parameter:                         1                    max =         64

                                                Average RVI       =     0.0000
                                                Largest FMI       =     0.0000
DF adjustment:                Large sample      DF:     min       =          .
                                                        avg       =          .
                                                        max       =          .
Model F test:                    Equal FMI      F(   2,      .)   =       5.32
Within VCE type:                    Robust      Prob > F          =     0.0049

                        (Within VCE adjusted for clustering on clinic_bsl_num)
------------------------------------------------------------------------------
          vf |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
      gender |   .3896755   .1532997     2.54   0.011     .0892136    .6901374
         age |  -.0589721   .0390038    -1.51   0.131    -.1354182     .017474
       _cons |    .601567   .6309789     0.95   0.340    -.6351288    1.838263
------------------------------------------------------------------------------


Then with eform, I get the same estimates (among which a negative coefficient indicating that they have not been exponentiated):

Code:
. mi estimate: xtgee vf gender age, family(binomial) link(logit) corr(exch) vce(robust) eform

Multiple-imputation estimates                   Imputations       =         30
GEE population-averaged model                   Number of obs     =        496

Group variable:               clinic_bsl~m      Number of groups  =         16
Link:                                logit      Obs per group:
Family:                           binomial                    min =         10
Correlation:                  exchangeable                    avg =       31.0
Scale parameter:                         1                    max =         64

                                                Average RVI       =     0.0000
                                                Largest FMI       =     0.0000
DF adjustment:                Large sample      DF:     min       =          .
                                                        avg       =          .
                                                        max       =          .
Model F test:                    Equal FMI      F(   2,      .)   =       5.32
Within VCE type:                    Robust      Prob > F          =     0.0049

                        (Within VCE adjusted for clustering on clinic_bsl_num)
------------------------------------------------------------------------------
          vf |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
      gender |   .3896755   .1532997     2.54   0.011     .0892136    .6901374
         age |  -.0589721   .0390038    -1.51   0.131    -.1354182     .017474
       _cons |    .601567   .6309789     0.95   0.340    -.6351288    1.838263
------------------------------------------------------------------------------

I have tried using the or option instead as odds ratio would be relevant as well, but that option is not allowed.

Does anyone know why it gives the same output? Is there a way to obtain exponentiated coefficient?

Thank you in advance for your help!