R help - Aug 2006 - lmer(): specifying i.i.d random slopes for multiple covariates

Dear readers,

Is it possible to specify a model

y=X %*% beta + Z %*% b ; b=(b_1,..,b_k) and b_i~N(0,v^2) for i=1,..,k

that is, a model where the random slopes for different covariates are i.i.d., in
lmer() and how?

In lme() one needs a constant grouping factor (e.g.: all=rep(1,n)) and would
then specify:
lme(fixed= y~X, random= list(all=pdIdent(~Z-1)) ) ,
that?s how it's done in the lmeSplines- documentation.

Any hints would be greatly appreciated- I'm trying to write a suite of
functions that will transform additive models into their mixed-effects
representation like lmeSplines but using lmer() instead of lme().

Thank you for your time,
Fabian Scheipl
-- 


Echte DSL-Flatrate dauerhaft f?r 0,- Euro*. Nur noch kurze Zeit!

The first example in the 'lmer' help page is the following: 

     (fm1 <- lmer(Reaction ~ Days + (Days|Subject), sleepstudy))

      If this does not answer your question, please provide commented, 
minimal, self-contained, reproducible code with your explanation of why 
and how it does not. 

      Hope this helps. 
      Spencer Graves

Fabian Scheipl wrote:
> Dear readers,
>
> Is it possible to specify a model
>
> y=X %*% beta + Z %*% b ; b=(b_1,..,b_k) and b_i~N(0,v^2) for i=1,..,k
>
> that is, a model where the random slopes for different covariates are
i.i.d., in lmer() and how?
>
> In lme() one needs a constant grouping factor (e.g.: all=rep(1,n)) and
would then specify:
> lme(fixed= y~X, random= list(all=pdIdent(~Z-1)) ) ,
> that?s how it's done in the lmeSplines- documentation.
>
> Any hints would be greatly appreciated- I'm trying to write a suite of
functions that will transform additive models into their mixed-effects
representation like lmeSplines but using lmer() instead of lme().
>
> Thank you for your time,
> Fabian Scheipl
>