Date:         Fri, 7 Jun 1996 11:10:32 -0700
Reply-To:     Leora Lawton <leora@DOCK.BELLCORE.COM>
Sender:       "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From:         Leora Lawton <leora@DOCK.BELLCORE.COM>
Organization: Bellcore
Subject:      using ordered logistic regression

My colleague and I have been trying to figure out exactly how SAS
calculates the logits in the ordered logit (PROC LOGISTIC; where the
dependent variable is ordered but not linear, e.g. a scale of attitudes
where 1=poor and 4 = excellent).  I'm not statistician, so it's beyond
me.  My colleague is a statistician, but unfamiliar with the method.  The
SAS manual doesn't seem to give a clear enough explanation (no surprise
there...).  If anyone can shed some light so that we can exactly
interpret the coefficients beyond something more than direction and
statisticical significance, that would be wondeful.   What exactly is the
equation SAS uses?
oh, the Agresti book (as referenced in the SAS manual does explain the
method, but not SAS's method to its madness.).
thanks!
leora lawton
leora@dock.bellcore.com