Hi,
I am trying to do a mediation analysis with a dependent count variable. The predictor variable is binary and the mediator is continuous. I have done separate analyses for the path a (linear regression) and for path b (negative binomial regression with an offset). The problem is that I haven't managed to compute the combined indirect effect for a x b. I haven't managed to find a package or code that can analyze the indirect effect for linear regression in the path a and NBR with an offset in path b. Does anyone of you have an idea?