I have run an IV regression with a dichotomous outcome (Y) and my endogenous ("instrumented") regressor (X1) is also dichotomous. I am assessing whether continuous variable (X2) is moderating the effect of X1 on Y.
When X2 is not in the model, there is a negative, significant association between Y and X1.
When X2 is added to the model, X2 exacerbates the negative association between Y and X1. So it is moderating the relationship. In this model, there is a positive, significant association between Y and X2.
Does this mean that as X2 increases, the negative effect of X1 on Y increases, or is that misinterpreting it?