s-news
[Top] [All Lists]

testing of significance of factor variable when using gee

To: s-news@lists.biostat.wustl.edu
Subject: testing of significance of factor variable when using gee
From: Karin <twentestat@yahoo.com>
Date: Tue, 16 Nov 2004 02:57:03 -0800 (PST)
Comment: DomainKeys? See http://antispam.yahoo.com/domainkeys
Domainkey-signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com; b=QfqkOT/GQ0C5TBHHFoAwN8tL7OF9lWHXjztl/aUp5Zc6Zr10jMOO3LJzpvelEpgICJQ3ZnGO+HM3eDwo3R12UBuyXAtKxBKHKto7ggiklwJIC4ErKb2TsHTahTo2GnOAHQDmFCeO2rYF7jHYA1W/HnCGuxO84QkJyBvSUv3B+cE= ;
Hello,
 
I am using the function gee of CorrelatedData library to do logistic regression analysis with repeated measures data. I have a factor with three levels (representing the different patientgroups). Now I want to test whether there is a difference between the three groups. Is it possible to extract somehow the p-value of this overall test? Usually one can use anova(fitobject) but this does not work for a gee object.
 
Thanks,
 
Karin


Do you Yahoo!?
Meet the all-new My Yahoo! ? Try it today!
<Prev in Thread] Current Thread [Next in Thread>
  • testing of significance of factor variable when using gee, Karin <=