Date:         Fri, 30 Jun 2006 11:25:10 -0400
Reply-To:     Peter Flom <Flom@NDRI.ORG>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Peter Flom <Flom@NDRI.ORG>
Subject:      Re: help on proc mixed
In-Reply-To:  <6250203B042D8349A920AA61038309360172F735@UTHEVS3.mail.uthouston.edu>
Content-Type: text/plain; charset=US-ASCII

>>> "Swank, Paul R" <Paul.R.Swank@UTH.TMC.EDU> 6/30/2006 11:05 am >>>
wrote
<<<
It seems to me you should have two random statements, rather than 1
random and 1 repeated, if you are trying to do a growth curve. However,
with only two assessemtns, I don't think you can get a slope at the
individual level with only two assessments. I suggest:

  proc mixed data=treatment ;
  class fid id genotype sex ;
  model outcome= genotype time sex age time*genotype/solution;
  random intercept time/subject=id(fid) type=un;
  random intercept time/subject=fid type=un;
  run;

But like I said, you may have to remove time form the random statement
at the id level.
>>>

Thoughts:
1) I think you are right about the slopes with two time points
If you put them in, I think slopes will be artifactaully related to
intercept,  due to regression to the mean.  Also saturation problems

2) I've not seen the two random statements on one proc mixed before. A
quick perusal of the documentation didn't yield insight into what it
does.....can you elucidate?

thanks

Peter



Peter L. Flom, PhD
Assistant Director, Statistics and Data Analysis Core
Center for Drug Use and HIV Research
National Development and Research Institutes
71 W. 23rd St
http://cduhr.ndri.org
www.peterflom.com
New York, NY 10010
(212) 845-4485 (voice)
(917) 438-0894 (fax)