Date:         Thu, 18 May 2000 12:52:46 GMT
Reply-To:     Kattamuri.Sarma@RESPONSEINSURANCE.COM
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Kattamuri.Sarma@RESPONSEINSURANCE.COM
Subject:      Re: Need some statistical advice....
Comments: To: Kyeeha Park <kyeehp@APSTAT.KOREA.AC.KR>
Content-type: text/plain; charset=us-ascii

Start using Proc Logistic..  Check to see if the proportional odds
assumption is valid.  In the SAS output you
   will see a test  for this.  This is the case where intercept is same,
and slopes are different.  If this assumption
   is not met you can go to PROC CATMOD.   However you may run into
problems, when using continuous
   explanatory variables.  It is best to convert the continuous explanatory
variables into groups such a way
    that the groups do not have empty cells.  My experience is that CATMOD
worked best when  I converted
   my continuous variables into discrete variables.

   PROC GENMOD is a general method for linear models,  it can do what
Logistic and CATMOD can do.
   The output statistics given may be different.  Version 8 some new
things, and I am still exploring it.

    I suggest you read :   Logic Regression Examples, by SAS Institute.
                                              Paul Allison ( I hope I have
not made a mistake),
                                              wrote two books : one on
Logistic Regression, and
                                               Survival Analysis ( covers
Proportional Hazards Model).
                                             I am not sure of the titles of
Paul Allison's books.. but they are published by
                                             the SAS  Institute  (user
publication).

   PROC GENMOD is in the on-line documentation for Version 8.   I am
currently
   deciphering it .

    Hope this helps.
   Kattamuri Sarma

    Kattamuri Sarma