Date:         Wed, 16 Oct 1996 13:55:00 -0400
Reply-To:     Sullivan_T <SullivaT@OEUS.PSB.BLS.GOV>
Sender:       "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From:         Sullivan_T <SullivaT@OEUS.PSB.BLS.GOV>
Subject:      Logistic Regression and Monte Carlo

I have run a logistic regression of the proportion of successes to a
various doses of drug.  I have 50 subjects in each of 9 dosage levels. I
now have a model of the form log(p/(1-p)) =  a +b*(dose) and am
concerned with the 50% response dose.  My question is how can I use SAS
to simulate 1,000 datasets using the parameters of the model to come up
with 1,000 new estimates of the 50% response dose.
 
Thanks for your help,
 
T.J. Sullivan