Error with marginal effects after oprobit

[Oscar Ll]

Dear users,

I'm estimating an ordered probit model. My dependent variable has four categories, ranging from 0 to 3.
I want to test if one of my independent variables (continuous) has an inverted u-shaped effect over the highest level of my dependent variable (3).
To do so, I want to compute and plot marginal effects. But I get an error.

Here's the oprobit regression:

oprobit Variety_1_Output2 ///
c.EI_7categs##c.EI_7categs ///
c.tertius_avg_c ///
$controls, robust cluster(tipo_org_2)

Then, I want to run the marginal effects for the full range of my variable:

qui sum c.EI_7categs, d
margins, at(c.EI_7categs =(`r(min)'(0.1)`r(max)')) predict(outcome(3))

But I get the following error:

0 invalid name
(error in option xb())

I've been testing different alternatives but I haven't been able to figure out where's the problem.

Many thanks in advance.

[Richard Williams]

I just semi-cloned your example without problems. Can you post an extract of your data using dataex that reproduces the problem?

[Oscar Ll]

Yes, here it is:

* Example generated by -dataex-. To install: ssc install dataex
clear
input float(Variety_1_Output2 EI_7categs tertius_avg_c)
1        0   .0984943
0        .  -.7348387
3 .7142857  1.9318278
0        1   .5984943
1      -.6  1.2651613
1        1  .26516128
0        1   .7651613
1        0   .9318278
0       -1  -.5681722
1 .6666667 -1.0681722
end

I don't show all the control variables that I employ in the regression. But when I drop controls from the oprobit, I still get the same problem.

I use Stata 14.0

[Richard Williams]

This works for me. What version of Stata do you have? Is it up to date?

 * Example generated by -dataex-. To install: ssc install dataex
clear
input float(Variety_1_Output2 EI_7categs tertius_avg_c)
1        0   .0984943
0        .  -.7348387
3 .7142857  1.9318278
0        1   .5984943
1      -.6  1.2651613
1        1  .26516128
0        1   .7651613
1        0   .9318278
0       -1  -.5681722
1 .6666667 -1.0681722
end
oprobit Variety_1_Output2 ///
c.EI_7categs##c.EI_7categs ///
c.tertius_avg_c ///
, robust 
qui sum c.EI_7categs, d
margins, at(c.EI_7categs =(`r(min)'(0.1)`r(max)')) predict(outcome(3))

. margins, at(c.EI_7categs =(`r(min)'(0.1)`r(max)')) predict(outcome(3))

Predictive margins                              Number of obs     =          9
Model VCE    : Robust

Expression   : Pr(Variety_1_Output2==3), predict(outcome(3))

1._at        : EI_7categs      =          -1

2._at        : EI_7categs      =         -.9

3._at        : EI_7categs      =         -.8

4._at        : EI_7categs      =         -.7

5._at        : EI_7categs      =         -.6

6._at        : EI_7categs      =         -.5

7._at        : EI_7categs      =         -.4

8._at        : EI_7categs      =         -.3

9._at        : EI_7categs      =         -.2

10._at       : EI_7categs      =         -.1

11._at       : EI_7categs      =           0

12._at       : EI_7categs      =          .1

13._at       : EI_7categs      =          .2

14._at       : EI_7categs      =          .3

15._at       : EI_7categs      =          .4

16._at       : EI_7categs      =          .5

17._at       : EI_7categs      =          .6

18._at       : EI_7categs      =          .7

19._at       : EI_7categs      =          .8

20._at       : EI_7categs      =          .9

21._at       : EI_7categs      =           1

------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
         _at |
          1  |   3.88e-15   2.14e-13     0.02   0.986    -4.16e-13    4.24e-13
          2  |   7.25e-11   2.64e-09     0.03   0.978    -5.11e-09    5.25e-09
          3  |   1.04e-07   2.36e-06     0.04   0.965    -4.52e-06    4.73e-06
          4  |   .0000185   .0002469     0.08   0.940    -.0004655    .0005025
          5  |   .0006546   .0047667     0.14   0.891     -.008688    .0099972
          6  |   .0069615   .0263364     0.26   0.792    -.0446568    .0585799
          7  |   .0322502   .0643065     0.50   0.616    -.0937883    .1582887
          8  |     .08702     .10055     0.87   0.387    -.1100544    .2840944
          9  |   .1658776   .1295816     1.28   0.201    -.0880978    .4198529
         10  |   .2506954   .1591342     1.58   0.115    -.0612019    .5625927
         11  |    .322385   .1896684     1.70   0.089    -.0493582    .6941283
         12  |   .3688054   .2131961     1.73   0.084    -.0490513    .7866621
         13  |   .3848245   .2249847     1.71   0.087    -.0561375    .8257865
         14  |   .3692251   .2236826     1.65   0.099    -.0691847    .8076348
         15  |   .3231888   .2083838     1.55   0.121     -.085236    .7316136
         16  |   .2517749    .180001     1.40   0.162    -.1010206    .6045704
         17  |    .167017   .1436743     1.16   0.245    -.1145794    .4486135
         18  |   .0879448   .1039123     0.85   0.397    -.1157195    .2916092
         19  |   .0327792   .0615373     0.53   0.594    -.0878317    .1533901
         20  |   .0071391   .0240336     0.30   0.766    -.0399659    .0542441
         21  |   .0006804   .0043282     0.16   0.875    -.0078027    .0091634
------------------------------------------------------------------------------

[Richard Williams]

I dug out Stata 11 and it runs fine there too.

[Oscar Ll]

Dear Richard,

I just found out that the problem occurs when I specify the option "robust cluster(tipo_org_2)" in the oprobit regression.
When I only use "robust" in the regression, it works perfectly. But if I cluster errors, it doesn't work.
In fact, you have run the model without the option "cluster(tipo_org_2)".

Could you check if it works for you when you cluster errors?

* Example generated by -dataex-. To install: ssc install dataex
clear
input float(Variety_1_Output2 EI_7categs tertius_avg_c) double tipo_org_2
1         0   .0984943 3
0         .  -.7348387 1
3  .7142857  1.9318278 3
0         1   .5984943 3
1       -.6  1.2651613 3
1         1  .26516128 3
0         1   .7651613 3
1         0   .9318278 1
0        -1  -.5681722 1
1  .6666667 -1.0681722 1
1        .4   .4318278 1
0        -1          . 1
3 .14285715  .26516128 1
0         0 -1.0681722 1
0         1   .7651613 1
end
label values tipo_org_2 tipo_org_2
label def tipo_org_2 1 "university", modify
label def tipo_org_2 3 "public research org", modify

Many thanks.

[Richard Williams]

I didn't use the cluster option because it wasn't in the earlier file. But with it, it still works fine for me, assuming I have exactly reproduced what you ran. If I haven't, show the exact command you ran with the controls dropped. Also indicate your version of Stata. I am using 15.

* Example generated by -dataex-. To install: ssc install dataex
clear
input float(Variety_1_Output2 EI_7categs tertius_avg_c) double tipo_org_2
1         0   .0984943 3
0         .  -.7348387 1
3  .7142857  1.9318278 3
0         1   .5984943 3
1       -.6  1.2651613 3
1         1  .26516128 3
0         1   .7651613 3
1         0   .9318278 1
0        -1  -.5681722 1
1  .6666667 -1.0681722 1
1        .4   .4318278 1
0        -1          . 1
3 .14285715  .26516128 1
0         0 -1.0681722 1
0         1   .7651613 1
end
label values tipo_org_2 tipo_org_2
label def tipo_org_2 1 "university", modify
label def tipo_org_2 3 "public research org", modify

oprobit Variety_1_Output2 ///
   c.EI_7categs##c.EI_7categs ///
   c.tertius_avg_c ///
   , robust cluster(tipo_org_2)
qui sum c.EI_7categs, d
margins, at(c.EI_7categs =(`r(min)'(0.1)`r(max)')) predict(outcome(3))

. margins, at(c.EI_7categs =(`r(min)'(0.1)`r(max)')) predict(outcome(3))

Predictive margins                              Number of obs     =         13
Model VCE    : Robust

Expression   : Pr(Variety_1_Output2==3), predict(outcome(3))

1._at        : EI_7categs      =          -1

2._at        : EI_7categs      =         -.9

3._at        : EI_7categs      =         -.8

4._at        : EI_7categs      =         -.7

5._at        : EI_7categs      =         -.6

6._at        : EI_7categs      =         -.5

7._at        : EI_7categs      =         -.4

8._at        : EI_7categs      =         -.3

9._at        : EI_7categs      =         -.2

10._at       : EI_7categs      =         -.1

11._at       : EI_7categs      =           0

12._at       : EI_7categs      =          .1

13._at       : EI_7categs      =          .2

14._at       : EI_7categs      =          .3

15._at       : EI_7categs      =          .4

16._at       : EI_7categs      =          .5

17._at       : EI_7categs      =          .6

18._at       : EI_7categs      =          .7

19._at       : EI_7categs      =          .8

20._at       : EI_7categs      =          .9

21._at       : EI_7categs      =           1

------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
         _at |
          1  |   9.47e-07   2.17e-06     0.44   0.662    -3.30e-06    5.20e-06
          2  |   .0000333   .0000627     0.53   0.596    -.0000895     .000156
          3  |   .0004755   .0007309     0.65   0.515     -.000957     .001908
          4  |   .0033791    .004235     0.80   0.425    -.0049214    .0116796
          5  |   .0141956   .0146377     0.97   0.332    -.0144938     .042885
          6  |   .0404866   .0348105     1.16   0.245    -.0277406    .1087139
          7  |    .086686    .062885     1.38   0.168    -.0365663    .2099383
          8  |   .1495932   .0924915     1.62   0.106    -.0316867    .3308732
          9  |   .2192219   .1170521     1.87   0.061    -.0101959    .4486398
         10  |   .2838647   .1335963     2.12   0.034     .0220209    .5457086
         11  |   .3344861   .1428484     2.34   0.019     .0545083    .6144639
         12  |   .3659204   .1470744     2.49   0.013     .0776598     .654181
         13  |   .3759237   .1481844     2.54   0.011     .0854876    .6663598
         14  |   .3639433   .1468418     2.48   0.013     .0761387     .651748
         15  |   .3306505   .1422547     2.32   0.020     .0518364    .6094647
         16  |   .2784964   .1324306     2.10   0.035     .0189371    .5380557
         17  |   .2129741   .1151502     1.85   0.064    -.0127162    .4386645
         18  |   .1434574   .0899461     1.59   0.111    -.0328338    .3197486
         19  |   .0817198   .0601743     1.36   0.174    -.0362196    .1996592
         20  |   .0373136   .0325958     1.14   0.252     -.026573    .1012001
         21  |   .0126924   .0133132     0.95   0.340     -.013401    .0387858
------------------------------------------------------------------------------