03-14-2017 01:32 AM - edited 03-14-2017 01:32 AM
I have an outcome frequency over a variable number of years per patient. I have converted this to a rate per year and would like to investigate differences bet a certain category. The rates are overdispersed and zero inflated. How can I code a zero inflated negative binomial model in SAS for this rate outcome?
03-14-2017 02:11 AM
what about countreg?
proc countreg data=data;
model rate = category covariate / dist=zinb method=qn;
zeromodel rate ~ category covariate;
I was not sure how to model the zeromodel..