s-news
[Top] [All Lists]

modelling question

To: s-news@wubios.wustl.edu
Subject: modelling question
From: "Richards, Tom" <richards@pci.upmc.edu>
Date: Mon, 18 Dec 2000 05:20:00 -0500
Hello:

I want to fit a model to count data where many counts are zero.  I would
like to posit that the baseline value of an immunologic marker impacts
whether or not a patient experiences any events at all, and the conditional
(on being  > 0) expectation of the remaining counts as a function of the
percent inhibition of the immunologic marker.  I guess I want to use a
logistic model for the probability of experiencing any events, and a Poisson
model for the conditional expectation.  This sounds like it should be a
familiar problem.  Can someone point me in the right direction?  Thanks in
advance for any help.

Tom Richards

<Prev in Thread] Current Thread [Next in Thread>