st: How to calculate margins after estimating a SEM that includes dependent variables measured as a proportion?

[Michael Simon <michel-simon@gmx.net>]

Title: How to calculate margins after estimating a SEM that includes dependent variables measured as a proportion?

Dear statalist-members,

I’m estimating a Simultaneous Equation Model (5 equations) with 3SLS. Due to the fact, that some of the dependent variables are given as a proportion, I have done a Logit Transformation with these variables prior to the 3SLS estimation.
Now:

1. Is it possible to directly interprete the coefficients of the 3SLS results?
2. If not, I have to compute margins. The problem is, that the dependent variables now have negative values after the logit transformation. Thus, STATA doesn’t calculate margins for non-integer values. What can I do?
   

Thank you very much for support.

Michael