mibeta command and statistical output

[Pat Burns]

Dear Stata Community:

I am trying to ascertain the following statistical outcomes, including R^2, F, etc. after the mibeta regression. Can anyone help with my coding error below.
Appreciate anyone's thoughts-
Pat

. mibeta ZMATH ib3.MOMDEG [pweight=W1P0]
. outreg2 using regression.xls, replace excel dec(2) alpha(0.001, 0.01, 0.05) addstat( ///
> dfModel, e(df_m_mi), dfError, e(df_r_mi), F, e(F_mi), R2, e(r2_mi)

matrix e(b) not found; run/post a regression, or specify varlist for non-regression outputs
r(111);

end of do-file

[Andrew Musau]

You do not specify where mibeta is from and you lack a reproducible example. The command is from Stata Journal by Yulia Marchenko and here is an example of the output.

sysuse auto
mi set mlong
mi register imputed rep78
mi impute mlogit rep78 mpg weight length price turn trunk, add(5) rseed(231729)
*Fit linear regression model to imputed data and obtain standardized coefficients
mibeta mpg weight length i.rep78

I see two tables and statistics at the bottom. What exactly do you need to extract?

. mibeta mpg weight length i.rep78

Multiple-imputation estimates                   Imputations       =          5
Linear regression                               Number of obs     =         74
                                                Average RVI       =     0.0263
                                                Largest FMI       =     0.0970
                                                Complete DF       =         67
DF adjustment:   Small sample                   DF:     min       =      52.34
                                                        avg       =      59.22
                                                        max       =      64.92
Model F test:       Equal FMI                   F(   6,   64.7)   =      23.98
Within VCE type:          OLS                   Prob > F          =     0.0000

------------------------------------------------------------------------------
         mpg |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
      weight |  -.0028379   .0016612    -1.71   0.092    -.0061557    .0004799
      length |   -.097916   .0562784    -1.74   0.087    -.2103143    .0144824
             |
       rep78 |
          2  |  -.0827336    2.60311    -0.03   0.975    -5.295529    5.130062
          3  |  -.6184872   2.388466    -0.26   0.797    -5.402655    4.165681
          4  |  -.5727751   2.473571    -0.23   0.818    -5.535595    4.390045
          5  |   2.234269   2.563839     0.87   0.387    -2.904531    7.373069
             |
       _cons |   48.32177   6.465051     7.47   0.000     35.40862    61.23492
------------------------------------------------------------------------------

Standardized coefficients and R-squared
Summary statistics over 5 imputations

             |       mean       min        p25     median        p75       max
-------------+----------------------------------------------------------------
      weight |  -.3812313     -.394  -.3936288  -.3746896  -.3737662      -.37
      length |  -.3768436     -.387  -.3821648  -.3820722   -.367529     -.365
             |
       rep78 |
          2  |  -.0043114    -.0554  -.0075797  -.0051404   .0114916     .0351
          3  |  -.0531988     -.124  -.0637297  -.0493614  -.0435566     .0149
          4  |  -.0439306     -.117  -.0523171  -.0512322  -.0288462       .03
          5  |   .1442454     .0823   .1406105   .1475846   .1485822      .202
-------------+----------------------------------------------------------------
    R-square |    .689109      .688    .687761   .6898957   .6900493       .69
Adj R-square |    .661268       .66   .6597993   .6621252   .6622926      .662
------------------------------------------------------------------------------

[Pat Burns]

Andrew: Thank you so much for your response. I did see Yulia Marchenko's post together with a number of others in the FAQ section, but none seem to answer my question. I have successfully used the following command in Stata 15 to run a regression analysis with imputed data:

.mi estimate, dots post: reg ZMATH ib3.MOMDEG [pweight=W1P0]
outreg2 using regression.xls, replace excel dec(2) alpha(0.001, 0.01, 0.05) addstat( ///
dfModel, e(df_m_mi), dfError, e(df_r_mi), F, e(F_mi))

I am also separately running the same regression this time using -mibeta for the sole purpose of getting R2 results as follows:

.mibeta ZMATH ib3.MOMDEG [pweight=W1P0]

My question is whether or not there is a way to combine the mi estimate and the mibeta command together to get all of the results in a table using outreg2.
For example, I would like to have:
outreg2 using regression.xls, replace excel dec(2) alpha(0.001, 0.01, 0.05) addstat( ///
dfModel, e(df_m_mi), dfError, e(df_r_mi), F, e(F_mi)) ......PLUS......... (R2, e(r2_mi) in my output

I appreciate your help-
Best,
Pat

[Andrew Musau]

*RUN mibeta TO PICK OUT STATS
mibeta ZMATH ib3.MOMDEG [pweight=W1P0]
*R2
local r2mi= e(r2_mi)
*ADJ. R2
local ar2mi= e(r2_adj_mi)

*RUN YOUR COMMAND
mi estimate, dots post: reg ZMATH ib3.MOMDEG [pweight=W1P0]

*OUTPUT USING OUTREG2
outreg2 using regression.xls, replace excel dec(2) ///
alpha(0.001, 0.01, 0.05) addstat( dfModel, e(df_m_mi), dfError, ///
e(df_r_mi), F, e(F_mi), R-squared, `r2mi', Adjusted R-squared, `ar2mi')

Typically, you will not want to report both the R2 and Adjusted R2 in the same table, so pick one. I present both here to illustrate the method. The estimation results are stored in e following most estimation commands in Stata, so you can run

ereturn list

to access them and store what you want in a local macro as I have done above. Finally, because you are using MI, it is important to mention that the reported R2 statistic is the average from estimations over the imputed data sets.

[Pat Burns]

Andrew:
Yes, I understand. This worked perfectly. Thank you so much for your help!
Best,
Pat