Hello,

I am doing a ZIP and I model the logit part as the probability of the count being zero. The coefficient on one of the predictor variables is positive for both the logit part and the Poisson part. How can that be? I mean, its saying the odds of an observation being zero increases with an increase in the variable, right? So, higher the value of the predcitor, higher is the probability of the event being zero. But then the Poisson part tells me that the expected number of event changes positively with the value of this variable. I am sooo confused and I think I must have confused all of you out there too. But if anyone understands my problem and is good enought to reply, thanks in advance!

Amy.