Arranging predicted probabilities into a table (presentation question)

[Ivan Privalko]

Hi all,

I have some predicted probabilities from the prtab command. These are shown below.

 .  prtab quit skill,novarlbl

ologit: Predicted probabilities for jbsat2

Predicted probability of outcome 1 (not_satis_at_all)

----------------------------
          |      skill      
     quit |       0  skilled
----------+-----------------
        0 |  0.0113   0.0102
        1 |  0.0058   0.0053
----------------------------

Predicted probability of outcome 2

----------------------------
          |      skill      
     quit |       0  skilled
----------+-----------------
        0 |  0.0260   0.0235
        1 |  0.0137   0.0123
----------------------------

Predicted probability of outcome 3

----------------------------
          |      skill      
     quit |       0  skilled
----------+-----------------
        0 |  0.1251   0.1147
        1 |  0.0710   0.0645
----------------------------

Predicted probability of outcome 4 (not_satis/dissat)

----------------------------
          |      skill      
     quit |       0  skilled
----------+-----------------
        0 |  0.0929   0.0872
        1 |  0.0592   0.0545
----------------------------

Predicted probability of outcome 5

----------------------------
          |      skill      
     quit |       0  skilled
----------+-----------------
        0 |  0.3570   0.3511
        1 |  0.2981   0.2849
----------------------------

Predicted probability of outcome 6

----------------------------
          |      skill      
     quit |       0  skilled
----------+-----------------
        0 |  0.3456   0.3667
        1 |  0.4734   0.4915
----------------------------

Predicted probability of outcome 7 (completely_satis)

----------------------------
          |      skill      
     quit |       0  skilled
----------+-----------------
        0 |  0.0421   0.0466
        1 |  0.0789   0.0870
----------------------------

I'm trying to present them in a table similar to the ones made using "outreg". I can't find any code for this, but I suspect there's a way to do it. Long and Freese show similar tables in their chapter on ordinal logistic regression, but don't list any code.


Thanks in advance.

[Richard Williams]

prtab is from Long & Freese's old spost9 routines. Unless you have an ancient version of Stata, I suggest you do

findit spost13_ado

and use the mtable command. Is this more like what you have in mind?

. webuse nhanes2f, clear

. ologit health i.female i.race, nolog

Ordered logistic regression                     Number of obs     =     10,335
                                                LR chi2(3)        =     182.45
                                                Prob > chi2       =     0.0000
Log likelihood = -15673.171                     Pseudo R2         =     0.0058

------------------------------------------------------------------------------
      health |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
    1.female |   -.131102   .0352117    -3.72   0.000    -.2001156   -.0620884
             |
        race |
      Black  |  -.7446752    .057536   -12.94   0.000    -.8574437   -.6319067
      Other  |  -.0856606   .1237325    -0.69   0.489    -.3281717    .1568506
-------------+----------------------------------------------------------------
       /cut1 |  -2.754836   .0441146                     -2.841299   -2.668373
       /cut2 |  -1.361154   .0313824                     -1.422663   -1.299646
       /cut3 |  -.0816749   .0280282                     -.1366092   -.0267406
       /cut4 |   1.057518   .0301285                      .9984673    1.116569
------------------------------------------------------------------------------

. mtable, at(female = (0 1) race = (1 2 3))

Expression: Pr(health), predict(outcome())

           |   female      race      poor      fair   average      good  excellent
 ----------+----------------------------------------------------------------------
         1 |        0         1     0.060     0.144     0.276     0.263      0.258
         2 |        0         2     0.118     0.232     0.309     0.198      0.142
         3 |        0         3     0.065     0.154     0.283     0.257      0.242
         4 |        1         1     0.068     0.159     0.286     0.254      0.234
         5 |        1         2     0.132     0.248     0.308     0.185      0.126
         6 |        1         3     0.073     0.168     0.292     0.248      0.219

Specified values where .n indicates no values specified with at()

           |  No at()
 ----------+---------
   Current |       .n

.

mtable has various options for prettying up the output.

Even if this isn't what you want I would get spost13 regardless as it is much better than spost9. But before installing see

http://www.indiana.edu/~jslsoc/spost13.htm

[Ivan Privalko]

Thanks Richard,

I'm using an older copy of the book, so thanks for this. I have spost13 also, I just haven't made the switch to mtable. Any tips on getting it to run faster? Both mtable and mchange seem to stall once I run them.

[Richard Williams]

They generally work fine for me. If you can't get them to run then I would suggest posting your code and output in a new thread. Or write Scott Long and see what he can tell you. Follow Long's directions from my earlier link to make sure spost9 and spost13 co-exist peacefully. Also you may just need to be patient, especially if you have a big data set or a complicated model.

[Richard Williams]

See appendix A of

http://www3.nd.edu/~rwilliam/xsoc73994/Ologit01.pdf

for examples of how to use margins and mtable after running ologit.

[Ivan Privalko]

This looks helpful, thanks again.

[Antonia Diaz-Valdes]

Hello everyone,

I am having a similar problem with prtab commang after ologit. The error message says: cut1 not found. I updated the Spost13 version, forcing to replace the prior package, and I have Stata 15. But, still having the same problem. Just to be sure, I tried to uninstall spost13_ado but it returned an error message:

. ado uninstall spost13_ado
criterion matches more than one package

Any thoughts about what I could be doing wrong?

Best

[Richard Williams]

Hi Antonia. prtab is not included in spost13. mtable has superseded it. For suggestions on how to use margins and mtable with commands like ologit, see

https://www3.nd.edu/~rwilliam/xsoc73994/Margins05.pdf

You may also want to see

https://www3.nd.edu/~rwilliam/xsoc73994/Margins04.pdf

which includes information on the old spost9 commands if you still want to use them (although I can't imagine why you would).

[Antonia Diaz-Valdes]

Thank you Richard