Odds ratio logit result on psmatch2

[Imam Agus Faisal]

Hi, so I am writing a paper for my final year on college. I am using propensity score matching on doing my research/analysis, particularly the psmatch2 on stata.
I see it more digestible and easier to interprete when using odds ratio than log odds, especially to describe the relationship between the increase of the x variable to the possibility/likelihood of getting treatment.
I have used "odds" command, but it only matches the sample using odds ratio, but not showing the logistic regression on odds ratio.
Running logistic regression separately oddly gives a different result, is there any way to do this?

Thanks in advance.

[Carlo Lazzaro]

Imam;
welcome to this forum.
Please follow the FAQ on how to poste more effectively.
Unfortunately, as per your description even really intrerested listers cannot reply positively without seeing, at least, what you typed and what Stata gave you back (as per FAQ again). Thanks.

[Imam Agus Faisal]

Sorry Carlo;
I would try to rearrange my post accordingly,
So I ran this command,

psmatch treatment xlist, out(outcome) neighbor(1) caliper(0.2) logit common trim(5)

and I got this as a result,

Logistic regression                                    Number of obs =  38,777
                                                       LR chi2(10)   = 4342.93
                                                       Prob > chi2   =  0.0000
Log likelihood = -15468.182                            Pseudo R2     =  0.1231

------------------------------------------------------------------------------
internetdesa | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
        umur |  -.0256812   .0018333   -14.01   0.000    -.0292743    -.022088
      gender |  -.0999081   .0317676    -3.14   0.002    -.1621714   -.0376448
  pendidikan |   1.152372   .0355897    32.38   0.000     1.082617    1.222126
 disabilitas |  -.3201624   .0784494    -4.08   0.000    -.4739203   -.1664045
  jumlah_art |  -.0526627   .0092556    -5.69   0.000    -.0708032   -.0345221
   sts_nikah |   .1491105   .0389806     3.83   0.000     .0727099    .2255112
lama_bekerja |   .0270969   .0022315    12.14   0.000     .0227233    .0314705
     migrasi |   .0347966   .0360546     0.97   0.334    -.0358692    .1054624
      kursus |   .7346739    .035532    20.68   0.000     .6650324    .8043153
   pertanian |   -1.12768   .0436472   -25.84   0.000    -1.213227   -1.042133
       _cons |   1.501147   .0809691    18.54   0.000      1.34245    1.659843
------------------------------------------------------------------------------
----------------------------------------------------------------------------------------
        Variable     Sample |    Treated     Controls   Difference         S.E.   T-stat
----------------------------+-----------------------------------------------------------
             inc  Unmatched |  2892202.7   2385128.44   507074.261   32489.1987    15.61
                        ATT | 2891134.24    2546959.8   344174.443    46262.091     7.44
----------------------------+-----------------------------------------------------------
Note: S.E. does not take into account that the propensity score is estimated.

 psmatch2: |   psmatch2: Common
 Treatment |        support
assignment | Off suppo  On suppor |     Total
-----------+----------------------+----------
 Untreated |         0      6,568 |     6,568
   Treated |     1,610     30,599 |    32,209
-----------+----------------------+----------
     Total |     1,610     37,167 |    38,777

I try to change the result of logit regression from log odds to odds ratio, by adding "odds", so i run the command:

psmatch treatment xlist, out(outcome) odds neighbor(1) caliper(0.2) logit common trim(5)

But, it still shows me the same result on logit regression table:

 Logistic regression                                    Number of obs =  38,777
                                                       LR chi2(10)   = 4342.93
                                                       Prob > chi2   =  0.0000
Log likelihood = -15468.182                            Pseudo R2     =  0.1231

------------------------------------------------------------------------------
internetdesa | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
        umur |  -.0256812   .0018333   -14.01   0.000    -.0292743    -.022088
      gender |  -.0999081   .0317676    -3.14   0.002    -.1621714   -.0376448
  pendidikan |   1.152372   .0355897    32.38   0.000     1.082617    1.222126
 disabilitas |  -.3201624   .0784494    -4.08   0.000    -.4739203   -.1664045
  jumlah_art |  -.0526627   .0092556    -5.69   0.000    -.0708032   -.0345221
   sts_nikah |   .1491105   .0389806     3.83   0.000     .0727099    .2255112
lama_bekerja |   .0270969   .0022315    12.14   0.000     .0227233    .0314705
     migrasi |   .0347966   .0360546     0.97   0.334    -.0358692    .1054624
      kursus |   .7346739    .035532    20.68   0.000     .6650324    .8043153
   pertanian |   -1.12768   .0436472   -25.84   0.000    -1.213227   -1.042133
       _cons |   1.501147   .0809691    18.54   0.000      1.34245    1.659843
------------------------------------------------------------------------------

Why the table showing the same result? Why is it still showing "coefficient" or log odds.
I need odds ratio because it is easier to interpret than log odds.

Thanks.

[Carlo Lazzaro]

Imam:
as per the community-contributed module -psmatch2- -helpfile-:

odds match on the logarithm of the odds ratio of the propensity score (stored in _pscore).

Therefore, I'm under the impression that the module is not expected to give back odds ratio in the outcome table.

[Imam Agus Faisal]

Is there any alternative to psmatch2? the one i could modify the logit regression to show/give back odds ratio in the outcome table

[Carlo Lazzaro]

Imam:
not that I know.