Hi Statalist
I have looked at previous posts regarding logit models with IV and it seems that the common solution is to just use -ivprobit- in logit and Probit essentially do the same thing
Thus, I proceeded to carry out an IV Probit regression.
Due to the coefficients of Probit not being interpretable in the traditional sense (of marginal effects) after -ivprobit- I use the -margins, dydx(*)- command
However, I have just read that these marginal effects are uninterpretable for interaction terms in the Probit model (I generated a variable x1x2=x1*x2) and used it in the model and thus if I understand correctly, my marginal effects for x1x2 are unusable.
I was wondering if there is a way to get the marginal effects of the interaction term using the Probit model (I assume not). And if not would someone please enlighten me on how to carry out an IV logistic regression with a single endogenous variable
Thank you
0 Response to Binary Dependent Variable Analysis with an Instrumental Variable
Post a Comment