Alternatives for -tab- on imputed data

[Sigrid Vikjord]

Dear Statalist,

When I do Cox regression on regular data using a categorical dependent variable, I run the command

tab depvar if e(sample)

after -stcox- to assess the number of individuals within each category. -tab- is not possible to use on multiple imputed data, as far as I know. Does anyone know of a code or syntax that could do the same on mi data?

Best regards,
Sigrid

[Marcos Almeida]

You may try something like:

. webuse mheart5
(Fictional heart attack data)

.  mi set mlong

. mi register imputed age bmi
(28 m=0 obs. now marked as incomplete)

. set seed 1234

. mi impute mvn age bmi = attack smokes hsgrad female, add(5)

Performing EM optimization:
note: 12 observations omitted from EM estimation because of all imputation variables missing
  observed log likelihood = -651.75868 at iteration 7

Performing MCMC data augmentation ... 

Multivariate imputation                     Imputations =        5
Multivariate normal regression                    added =        5
Imputed: m=1 through m=5                        updated =        0

Prior: uniform                               Iterations =      500
                                                burn-in =      100
                                                between =      100

------------------------------------------------------------------
                   |               Observations per m             
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
               age |        142           12        12 |       154
               bmi |        126           28        28 |       154
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. mi xeq: tabulate attack

m=0 data:
-> tabulate attack

    Outcome |
     (heart |
    attack) |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |         85       55.19       55.19
          1 |         69       44.81      100.00
------------+-----------------------------------
      Total |        154      100.00

m=1 data:
-> tabulate attack

    Outcome |
     (heart |
    attack) |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |         85       55.19       55.19
          1 |         69       44.81      100.00
------------+-----------------------------------
      Total |        154      100.00

m=2 data:
-> tabulate attack

    Outcome |
     (heart |
    attack) |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |         85       55.19       55.19
          1 |         69       44.81      100.00
------------+-----------------------------------
      Total |        154      100.00

m=3 data:
-> tabulate attack

    Outcome |
     (heart |
    attack) |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |         85       55.19       55.19
          1 |         69       44.81      100.00
------------+-----------------------------------
      Total |        154      100.00

m=4 data:
-> tabulate attack

    Outcome |
     (heart |
    attack) |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |         85       55.19       55.19
          1 |         69       44.81      100.00
------------+-----------------------------------
      Total |        154      100.00

m=5 data:
-> tabulate attack

    Outcome |
     (heart |
    attack) |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |         85       55.19       55.19
          1 |         69       44.81      100.00
------------+-----------------------------------
      Total |        154      100.00

. */ only the fifth dataset

. mi xeq 5: tabulate attack

m=5 data:
-> tabulate attack

    Outcome |
     (heart |
    attack) |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |         85       55.19       55.19
          1 |         69       44.81      100.00
------------+-----------------------------------
      Total |        154      100.00

Hopefully that helps.

[Marcos Almeida]

Below, a more realistic example, after generating missing values for the binary variable "attack":

. webuse mheart5
(Fictional heart attack data)

. */ create missing data for the binary variable attack

. replace attack = . in 1/10
(10 real changes made, 10 to missing)

. */ check the tabulation

. tab attack, missing

    Outcome |
     (heart |
    attack) |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |         78       50.65       50.65
          1 |         66       42.86       93.51
          . |         10        6.49      100.00
------------+-----------------------------------
      Total |        154      100.00

. */ now, the multiple imputation

.  mi set mlong

. mi register imputed age bmi attack
(37 m=0 obs. now marked as incomplete)

. mi impute chained (regress) bmi age (logit) attack = smokes hsgrad female, add(5)

Conditional models:
            attack: logit attack age bmi smokes hsgrad female
               age: regress age i.attack bmi smokes hsgrad female
               bmi: regress bmi i.attack age smokes hsgrad female

Performing chained iterations ...

Multivariate imputation                     Imputations =        5
Chained equations                                 added =        5
Imputed: m=1 through m=5                        updated =        0

Initialization: monotone                     Iterations =       50
                                                burn-in =       10

               bmi: linear regression
               age: linear regression
            attack: logistic regression

------------------------------------------------------------------
                   |               Observations per m            
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
               bmi |        126           28        28 |       154
               age |        142           12        12 |       154
            attack |        144           10        10 |       154
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. */ now, tabulations for the variable attack

. mi xeq: tabulate attack

m=0 data:
-> tabulate attack

    Outcome |
     (heart |
    attack) |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |         78       54.17       54.17
          1 |         66       45.83      100.00
------------+-----------------------------------
      Total |        144      100.00

m=1 data:
-> tabulate attack

    Outcome |
     (heart |
    attack) |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |         85       55.19       55.19
          1 |         69       44.81      100.00
------------+-----------------------------------
      Total |        154      100.00

m=2 data:
-> tabulate attack

    Outcome |
     (heart |
    attack) |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |         85       55.19       55.19
          1 |         69       44.81      100.00
------------+-----------------------------------
      Total |        154      100.00

m=3 data:
-> tabulate attack

    Outcome |
     (heart |
    attack) |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |         81       52.60       52.60
          1 |         73       47.40      100.00
------------+-----------------------------------
      Total |        154      100.00

m=4 data:
-> tabulate attack

    Outcome |
     (heart |
    attack) |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |         84       54.55       54.55
          1 |         70       45.45      100.00
------------+-----------------------------------
      Total |        154      100.00

m=5 data:
-> tabulate attack

    Outcome |
     (heart |
    attack) |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |         82       53.25       53.25
          1 |         72       46.75      100.00
------------+-----------------------------------
      Total |        154      100.00

You can see that only the "original" data set shall present missing values for "atttack".

Hopefully that helps.

[Sigrid Vikjord]

Thank you Marcos, this was really helpful!