[R-sig-ME] Post-test on MCMCglmm posteriors distribution

[Jarrod Hadfield]

Dear Stephane,

Do you mean posterior predictions or posterior distributions? If the  
latter, and you want to compare wether Va1 > Va2, for example, you can  
just use HPDinterval on the difference.

Cheers,

Jarrod





Quoting Stephane Chantepie <chantepie at mnhn.fr> on Wed, 11 Jul 2012  
16:01:53 +0200:

> Dear all,
>
> I have a quite simple question but I do not find a clear answer on the
> internet nor in books.
>
> The context : I am currently working on different aspects of bird aging. I
> have done animal models with MCMCglmm to find if there is an additive genetic
> variance (Va) in my sexual life traits. One animal model has been  
> done by age.
> Now, I am interested in testing the differences in Va (posterior  
> mode) between
> ages.
>
> My question : As posterior predictions are supposed to be normally  
> distributed
> (if the model converge), is it possible to use multiple Student test
> (corrected for unequal variance and pairwise multiple comparison) to test
> statiscal difference on posterior prediction? It seems to me that all
> assumptions are meet to use Student test but maybe there some problems that I
> am not thinking about.
>
> If you have a more elegant way to test differences of posterior  
> distributions,
> let me know!
>
> Thank in advance for reply and sorry for this beginner question.
>
> Stephane
>
> _______________________________________________
> R-sig-mixed-models at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>
>


-- 
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.