tobit post-estimation

[Anya hewertson]

I am trying to do a avplot after my tobit estimation to check for outliers/ extreme values

here are my results

tobit percent_expshare_wine_on log_price_wine_on log_price_wine_off logincome i.socio_group i.gor_recode i.sexhrp , ll(0)

Refining starting values:

Grid node 0:   log likelihood = -47747.393

Fitting full model:

Iteration 0:   log likelihood = -47747.393  
Iteration 1:   log likelihood = -32372.522  
Iteration 2:   log likelihood =  -27312.95  
Iteration 3:   log likelihood = -25237.826  
Iteration 4:   log likelihood = -24929.648  
Iteration 5:   log likelihood = -24924.116  
Iteration 6:   log likelihood = -24924.114  

Tobit regression                                Number of obs     =     34,301
                                                   Uncensored     =      6,509
Limits: lower = 0                                  Left-censored  =     27,792
        upper = +inf                               Right-censored =          0

                                                LR chi2(15)       =    2514.40
                                                Prob > chi2       =     0.0000
Log likelihood = -24924.114                     Pseudo R2         =     0.0480

------------------------------------------------------------------------------
percent~e_on |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
log_pri~e_on |   .5162378   1.183649     0.44   0.663    -1.803753    2.836228
log_pr~e_off |  -1.085473   1.175537    -0.92   0.356    -3.389565    1.218619
   logincome |   1.373729   .0415701    33.05   0.000      1.29225    1.455208
             |
 socio_group |
          2  |   .0873766   .0682073     1.28   0.200     -.046312    .2210652
          3  |  -.7677631   .0790945    -9.71   0.000     -.922791   -.6127353
          4  |  -1.453442   .2787138    -5.21   0.000    -1.999731   -.9071538
          5  |   .2039505   .1875383     1.09   0.277    -.1636307    .5715317
          6  |  -.0079945   .0619298    -0.13   0.897     -.129379      .11339
             |
  gor_recode |
          2  |  -.3123735   .0904925    -3.45   0.001    -.4897419   -.1350051
          3  |  -.3112034   .0886092    -3.51   0.000    -.4848804   -.1375264
          4  |  -.0465098   .0879178    -0.53   0.597    -.2188316    .1258121
          5  |  -.3931184   .1339016    -2.94   0.003      -.65557   -.1306667
          6  |  -.3035262   .1086274    -2.79   0.005    -.5164394   -.0906129
          7  |  -.4784591   .1237286    -3.87   0.000    -.7209713    -.235947
             |
      sexhrp |
     Female  |   .1663446   .0487767     3.41   0.001     .0707407    .2619485
       _cons |  -12.30692   1.824931    -6.74   0.000    -15.88385   -8.729995
-------------+----------------------------------------------------------------
var(e.perc..)|   8.360073   .1699498                      8.033515    8.699906
------------------------------------------------------------------------------

and then the code i am trying to use to do an avplot is

avplot

but it says last estimates not found,
this is also happens when i try and do a residual plot, can anyone explain why

[William Lisowski]

The output of help tobit postestimation does not suggest that avplot is available following a tobit regression. Since help tobit postestimation also does not suggest residuals can be obtained using the predict command following a tobit regression, that likely explains the failure of a residual plot to work.

[Phil Bromiley]

Many of the tools available after regress are not available after other estimators. However, since their use is largely heuristic, using regress and avplot (and the other tools) might help you identify potential outliers and non-linearities for tobit as well.