Dear All,
I am working with a loglineal model implying two variables
y=x+error we would like to estimate the 95-quartiles or the p-quartiles of x
and not the expected value as the Poissonian model gives to us. Does any one
knows a method that doesn't implies the using of regression. And if it's
impossible is this method implemented in SAS?
Thank you so much, Pep Puigvert.
_________________________________________________________________
Chat with friends online, try MSN Messenger: http://messenger.msn.com
|