changing col headers in etable

[Rich Goldstein]

if I estimate two mlogit commands with the same outcome (dependent) variable but different sets of covariates, I will, by default, get from e(table) the name of the outcome variable as the column header (see below); I want to change that to show something else so the columns will be easier to refer to; the following simple (and unrealistic) example shows the result and my attempt to change one of the headers using the eqrecode option:

. sysuse auto
(1978 automobile data)

. mlogit rep78 price

Iteration 0:  Log likelihood = -93.692061  
Iteration 1:  Log likelihood = -93.113707  
Iteration 2:  Log likelihood = -92.974678  
Iteration 3:  Log likelihood = -92.970763  
Iteration 4:  Log likelihood = -92.970763  

Multinomial logistic regression                         Number of obs =     69
                                                        LR chi2(4)    =   1.44
                                                        Prob > chi2   = 0.8368
Log likelihood = -92.970763                             Pseudo R2     = 0.0077

------------------------------------------------------------------------------
       rep78 | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
1            |
       price |   -.000549   .0007284    -0.75   0.451    -.0019766    .0008786
       _cons |   .1169081   3.421563     0.03   0.973    -6.589231    6.823048
-------------+----------------------------------------------------------------
2            |
       price |  -.0000547   .0001431    -0.38   0.703    -.0003352    .0002259
       _cons |  -.9834618    .952914    -1.03   0.302    -2.851139    .8842154
-------------+----------------------------------------------------------------
3            |  (base outcome)
-------------+----------------------------------------------------------------
4            |
       price |   -.000041   .0001027    -0.40   0.690    -.0002424    .0001603
       _cons |  -.2545298   .7025195    -0.36   0.717    -1.631443    1.122383
-------------+----------------------------------------------------------------
5            |
       price |  -.0000622   .0001279    -0.49   0.627    -.0003129    .0001884
       _cons |  -.6201496   .8475732    -0.73   0.464    -2.281363    1.041063
------------------------------------------------------------------------------

. est store m1

. mlogit rep78 mpg

Iteration 0:  Log likelihood = -93.692061  
Iteration 1:  Log likelihood = -86.581485  
Iteration 2:  Log likelihood = -85.767758  
Iteration 3:  Log likelihood = -85.752385  
Iteration 4:  Log likelihood = -85.752375  
Iteration 5:  Log likelihood = -85.752375  

Multinomial logistic regression                         Number of obs =     69
                                                        LR chi2(4)    =  15.88
                                                        Prob > chi2   = 0.0032
Log likelihood = -85.752375                             Pseudo R2     = 0.0847

------------------------------------------------------------------------------
       rep78 | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
1            |
         mpg |   .0708122   .1471461     0.48   0.630    -.2175888    .3592132
       _cons |  -4.137144    3.15707    -1.31   0.190    -10.32489      2.0506
-------------+----------------------------------------------------------------
2            |
         mpg |  -.0164251   .0926724    -0.18   0.859    -.1980597    .1652096
       _cons |  -1.005118   1.822129    -0.55   0.581    -4.576426     2.56619
-------------+----------------------------------------------------------------
3            |  (base outcome)
-------------+----------------------------------------------------------------
4            |
         mpg |   .0958626   .0633329     1.51   0.130    -.0282676    .2199927
       _cons |  -2.474187   1.341131    -1.84   0.065    -5.102756    .1543813
-------------+----------------------------------------------------------------
5            |
         mpg |   .2477469   .0764076     3.24   0.001     .0979908     .397503
       _cons |  -6.653164   1.841794    -3.61   0.000    -10.26301   -3.043314
------------------------------------------------------------------------------
r; t=0.03 8:49:53

. est store m2

. etable, estimates(m1 m2) showeq eqrecode(m1 = mi) cstat(_r_b) cstat(_r_p)

------------------------------------
                        rep78  rep78
------------------------------------
1                                  
  Price                -0.001      
                         0.45      
  Mileage (mpg)                0.071
                                0.63
  Intercept             0.117 -4.137
                         0.97   0.19
2                                  
  Price                -0.000      
                         0.70      
  Mileage (mpg)               -0.016
                                0.86
  Intercept            -0.983 -1.005
                         0.30   0.58
4                                  
  Price                -0.000      
                         0.69      
  Mileage (mpg)                0.096
                                0.13
  Intercept            -0.255 -2.474
                         0.72   0.07
5                                  
  Price                -0.000      
                         0.63      
  Mileage (mpg)                0.248
                                0.00
  Intercept            -0.620 -6.653
                         0.46   0.00
Number of observations     69     69
------------------------------------

here, I attempted to change the first column header to "mi" but no change was made

I think that my problem is that I don't understand what the equation name is for mlogit, but it could be something else???

In case it makes a difference, in my actual data, I will have one column for the data with missing values and one column that comes after multiple imputation and using "mi estimate"

[Andrew Musau]

You can name the estimates using the desired column name and add the option -column(estimates)- to etable. However, this restricts you to a single-word column name, or to using underscores in place of spaces. A more general solution is to use collect to modify the column titles.

sysuse auto, clear
mlogit rep78 price
est store m1
mlogit rep78 mpg
est store m2
etable, estimates(m1 m2) cstat(_r_b) cstat(_r_p)
*LOOK AT ETABLE'S LAYOUT
collect layout
collect levelsof etable_depvar
*CHANGE LABELS
collect label levels etable_depvar 1 "mi" 2 "rep78", modify
collect style header etable_depvar, title(hide) level(label)
collect preview

Res.:

. collect preview

------------------------------------
                         mi    rep78
------------------------------------
1                                  
  Price                -0.001      
                         0.45      
  Mileage (mpg)                0.071
                                0.63
  Intercept             0.117 -4.137
                         0.97   0.19
2                                  
  Price                -0.000      
                         0.70      
  Mileage (mpg)               -0.016
                                0.86
  Intercept            -0.983 -1.005
                         0.30   0.58
4                                  
  Price                -0.000      
                         0.69      
  Mileage (mpg)                0.096
                                0.13
  Intercept            -0.255 -2.474
                         0.72   0.07
5                                  
  Price                -0.000      
                         0.63      
  Mileage (mpg)                0.248
                                0.00
  Intercept            -0.620 -6.653
                         0.46   0.00
Number of observations     69     69
------------------------------------

[Rich Goldstein]

thank you