Hi Everyone,

I am trying to analyse a panel data using interrupted time series. I have used mi for imputations. I will be using mi estimate now and don't now which models suits my data, GLM family (Poisson) or xtpoisson? I am looking at the difference on hospital admissions following the administration of an intervention. The data is on the individual level, although; the the intervention was implemented as a population based model. Which model suits better in this cause, and i case of xtpoission how I am going to take overdispersion, autocorrelation and seasonality into account? Please remember that I will be using mi estimate. Thank you