[R-sig-ME] corr structure AR(1) and lmer function

[Ben Bolker]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

DUYME Florent wrote:
> Dear R users
> 
>  
> 
> I use the following design : 3 complete blocks, in each of them several
> varieties are randomly planted.
> 
> I'd like to model a gradient along each block
> 
>  
> 
> To do that, lme function is quite useful, with cor structure :
> correlation=corAR1()
> 
> The model is a crossed random effect spatial model.
> 
>  
> 
> To avoid using lme with such a model, I'd like to use lmer function. 
> 
>  
> 
> My question is : how do you model the corr structure (ie AR(1)) in the
> lmer function ?
> 
>  

  This is not currently possible with lmer, and Doug Bates has said
that it's not near the top of his "to do" list ...

  Ben Bolker
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFIySHkc5UpGjwzenMRAtSyAJ9/EemZufGn9Jw1KGoalY5NQfaN5wCcDWjI
H1qA4ypeEg2piDFQWT67EtI=
=0ux3
-----END PGP SIGNATURE-----