Date: Wed, 22 Mar 2006 16:30:54 -0800
Reply-To: David L Cassell <davidlcassell@MSN.COM>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: David L Cassell <davidlcassell@MSN.COM>
Subject: Re: Specifying a residual correlation matrix
Content-Type: text/plain; format=flowed
>I wonder if it is possible to specify a tailor-made residual correlation
>matrix (R) to be used in a mixed model analysis (PROC MIXED or PROC
>The reason for the question is that we have observations on a number of
>individuals (one obs per individual) and we are interested in estimating
>the effects of some (fixed) explanatory variables on the outcome. However,
>we know that these individuals are relatives and would like to account for
>this similarity. The idea was then to force SAS to use an R-matrix that we
>have defined and that show the relationship between them, since we believe
>that e is not ND(0,I*sigma_e), but rather ND(0,R*sigma_e). I believe this
>is pretty straight forward MME, but I am not sure that SAS can handle it.
I see that you have already received some expert advice. Follow it.
And also, have you considered that you have a more natural PROC MIXED
problem if you think of the family as the subject at the first level and the
person as the subject at the second level? That would allow you to use
standard PROC MIXED covariance structures with SUBJECT=FAMILY and
the individual people nested within family.
Finally, it's time once again for one of my pet peeves, upon which I am
irrationally intent. :-) Are these data from a sample survey, or are they
from an observational study? If they come from a sample survey, then
you're using the wrong proc and you should be focusing on a multi-stage
sample and PROC SURVEYREG. That would address the covariance
structure without your having to define it first.
David L. Cassell
3115 NW Norwood Pl.
Corvallis OR 97330
On the road to retirement? Check out MSN Life Events for advice on how to
get there! http://lifeevents.msn.com/category.aspx?cid=Retirement