Date: Wed, 27 Mar 1996 20:07:35 GMT
Reply-To: Jay Weedon <j_weedon@ESCAPE.COM>
Sender: "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From: Jay Weedon <j_weedon@ESCAPE.COM>
Organization: Escape.Com Internet Access and Services
Subject: Re: GEE
Jaya Veeramani <jaya@GCRCA.BIOSTAT.WISC.EDU> wrote:
>Thanks for the reply.The problem with GEE is that it cannot handle
>missing values. We have to remove all the missing values before the
>analysis and this in turn will lead to unequal time points and hence,
> be able to try only identity or an exchangeable covariance matrix for
>the repeated observations.
Yes, that's correct. In many cases, however, you won't be severely
by this - one of the advantages of GEE is that it's fairly robust with respect
to the covariance structure used.
By the way, as far as I'm aware the limitation you mention is specific to the
SAS implementation, rather than to GEE as a statistical tool.