st: From: Ankit Sakhuja <a.sakhuja@gmail.com>

[owner-statalist@hsphsun2.harvard.edu]

Dear Members,

I am trying to understand the effect of interaction between two
categorical variables in a model:

Logistic DIED i.agegroup i.sex i.race i.procedure i.procedure#i.sex

I can visualize the interaction effect using margins and margins plot as below:

margins procedure#sex
------------------------------------------------------------------------------
             |            Delta-method
             |     Margin   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
    procedure#sex |
        0 0  |   .0960571   .0013119    73.22   0.000     .0934859    .0986284
        0 1  |   .1568595   .0082394    19.04   0.000     .1407106    .1730084
        1 0  |   .0462026   .0008488    54.43   0.000      .044539    .0478663
        1 1  |   .1160865   .0086644    13.40   0.000     .0991046    .1330684
------------------------------------------------------------------------------
marginsplot

However, I am not sure how to interpret the above table accurately
(from my understanding it shows the % probability of mortality for
each scenario) and how to find the odds ratios for mortality for the 2
scenarios as below (I can find these ORs by stratifying the population
into males and females & then running 2 separate regression but I am
not sure if there is a way to get those using the interaction term in
the full model):
Males undergoing procedure in comparison to Males not undergoing procedure
Females undergoing procedure in comparison to Females not undergoing procedure

I would really appreciate your help.
Thanks
Ankit
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/