Code:
* Example generated by -dataex-. To install: ssc install dataex
clear
input byte(owndecision treat gender)
1 1 1
1 1 1
1 1 1
1 1 1
0 1 1
1 1 1
1 1 0
1 1 0
1 1 0
1 1 1
0 1 1
1 1 0
1 1 1
1 1 1
0 1 0
1 1 1
owndecision = 1 if defect, 0 otherwise
treat=0,1,2; (3 treatments: 0=common, 1=asymmetric, 2=private)
gender=1 if female, 0 otherwise

I would like the average marginal effects of defection (owndecision=1) by gender for asymmetric and private and to produce a graph that looks like this:
Array

Code:
logit owndecision i.gender#i.treat

------------------------------------------------------------------------------
 owndecision |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
gender#treat |
        0 1  |   1.466337   .7372854     1.99   0.047     .0212843     2.91139
        0 2  |   2.590267   1.179689     2.20   0.028     .2781187    4.902415
        1 0  |   1.041454   .6522961     1.60   0.110     -.237023    2.319931
        1 1  |   1.977163   .6868733     2.88   0.004     .6309158     3.32341
        1 2  |          0  (empty)

 margins, dydx(treat) over(gender)
------------------------------------------------------------------------------
             |            Delta-method
             |      dy/dx   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
1.treat      |
      gender |
          0  |   .3472222   .1606046     2.16   0.031      .032443    .6620015
          1  |   .1828704   .1157482     1.58   0.114    -.0439919    .4097326
-------------+----------------------------------------------------------------
2.treat      |
      gender |
          0  |   .5138889   .1600699     3.21   0.001     .2001576    .8276201
          1  |          .  (not estimable)
------------------------------------------------------------------------------
Note: dy/dx for factor levels is the discrete change from the base level.

marginsplot
Since the coefficient on gender=1 when treat=2 is empty then the marginal effect is not estimable, and the graph produced by marginsplot is missing that effect:

Array

Any help would be appreciated. Thank you.