[R-sig-ME] differing variances within different random effects levels

[David Afshartous]

Thanks Reinhold.  I had forgotten about that one also.   Actually, I 
just found the part of the thread the has the summary I was thinking about:
https://stat.ethz.ch/pipermail/r-sig-mixed-models/2007q3/000248.html

On 02/15/2011 12:27 AM, Reinhold Kliegl wrote:
> I think it was this very useful thread David refers to
> :
> Subject:	[R-sig-ME] getVarCov for lmer()	
> From:	Douglas Bates (bat... at stat.wisc.edu)
> Date:	Feb 2, 2008 3:47:12 pm
> List:	org.r-project.r-sig-mixed-models
>
> Reinhold Kliegl
>
> On Mon, Feb 14, 2011 at 10:47 PM, David Afshartous
> <david.r.afshartous at vanderbilt.edu>  wrote:
>> Ben,
>> There are a few old threads that you might find useful, search the
>> following:
>>
>> random effect variance per treatment group in lmer
>> heteroscedastic model in lme4
>> Stratifying level-1 variance with lmer()
>> Different random effects variances for outcomes and groups
>> Behavior of coef() for lmer() model w/ level-2 variance stratified
>>
>> I think in one of them I provided a summary of some of the issues but I
>> can't seem to find it at the moment.
>>
>> Cheers,
>> David
>>
>>
>>
>>
>>
>> On 02/14/2011 02:53 PM, Ben Bolker wrote:
>>>     This is a question that's been bothering me for a while.  The answer
>>> might be "read sections 4.2.2 and 4.2.3 of Pinheiro and Bates 2000
>>> again, more carefully", but if anyone has done this before I wouldn't
>>> mind some hints ...
>>>
>>>     Consider a simplified version of the previous poster's example.
>>> Suppose I have a randomized block design where T treatments are done for
>>> each of N individuals within each of S species. My basic model for this
>>> would be
>>>
>>>     response~treatment+(treatment|species/indiv)
>>>
>>> (or fixed=response~treatment, random=~treatment|species/indiv in lme)
>>>
>>> which would give me random effects
>>>
>>> 1|species
>>> 1|indiv:species
>>> treatment|species
>>> treatment|indiv:species
>>>
>>> (and correlations between intercept and treatment effects at species and
>>> indiv:species level).
>>>
>>>     Suppose now that I think the *variance* among individuals varies among
>>> species.  As far as I can tell, lme/lmer assume that this variance is
>>> constant across species.  If I wanted variance across *observations*
>>> (i.e. residual variance) to vary across species I could do it via
>>> something like weights=varIdent(~species), but this only works for
>>> residual variance.
>>>
>>>     I suspect I could work out some way to do this with MCMCglmm, and I
>>> don't have an immediate need for it, but I thought I would ask ...
>>>
>>>    thanks
>>>       Ben Bolker
>>>
>>> _______________________________________________
>>> R-sig-mixed-models at r-project.org mailing list
>>> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>>
>> --
>> David Afshartous, Ph.D.
>> Research Associate Professor
>> School of Medicine
>> Department of Biostatistics
>> Vanderbilt University
>>
>>
>>         [[alternative HTML version deleted]]
>>
>> _______________________________________________
>> R-sig-mixed-models at r-project.org mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>>


-- 
David Afshartous, Ph.D.
Research Associate Professor
School of Medicine
Department of Biostatistics
Vanderbilt University