re: st: -xtmixed- and random terms nested within fixed treatment terms?

[David Airey <david.airey@vanderbilt.edu>]

.

Have you seen the very helpful pages on xtmixed at UCLA? They are:

for classical experimental designs:

http://www.ats.ucla.edu/stat/stata/faq/xtmixed.htm

For the examples in Kreft and de Leeuw's textbook on multilevel  
modeling:

http://www.ats.ucla.edu/stat/stata/examples/mlm_imm/default.htm

For examples from Singer's paper using SAS but with other mixed model  
packages, including xtmixed:

http://www.ats.ucla.edu/stat/paperexamples/singer/default.htm

I don't know these answer your question.

-Dave



> I have a uestion on the new xtmixed. I am trying to match a model  
> run in S-Plux lme, and I am confused by the (lack of) nesting in
> -xtmixed-
>
> Suppose I have subjects assigned to treatment with outcomes  
> measured at repeated occasions, and a covariate. It seems that the  
> syntax for -xtmixed- should be something like this, where id 
> (treatment) indicates nesting of id within treatment:
>
> xtmixed outcome covariate treatment occasion || subject(treatment):
>
> I might expect an adaptation of the example for xtmixed on page 184  
> [XT] to look like this:
>
> xtmixed weight preweight treatment week || id(treatment):
>
> It seems to me that id must be nested in treatment, but treatment  
> is a fixed effect, not random, so the following syntax would be  
> incorrect:
>
> xtmixed weight preweight treatment week || id : treatment
>
> It does not seem possible to nest subject (as a random term) within  
> treatment in -xtmixed-; which seems the appropriate specification.  
> Am I incorrect in thinking that this nesting is necessary?
>
>
> Thanks
> Buzz Burhans
*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/