Hi all:

My dependent variable, v6mouses, is drug use; 60.15% of the result is 0 and 26.92% of the result is 1, which is overdispersion. Hence, I used "xtnbreg v6mouses IV, fe" to run my analysis. However, I have a concern xtnbreg is not a true fixed effects model.

I referred to the previous post, and I am using xtpoisson with robust option. Here is my code:

xtpoisson v6mouses l.exptots l.cynics l.mordiss l.parmnts l.impulss l.peerprs l.hoods i.wave, robust i(caseid) fe

I also use "l." option for creating lagged one variables for all IV to establish temporal sequences. I am not sure whether this would be suitable.

Thank you very much!!

Steven