I am dealing with over-dispersed count variable in my model and to my best knowledge, negative binomial model is ideal for this type dependent variable. I would like to use FE d