Re: [R] lmer for 3-way random anova

From: Douglas Bates <dmbates_at_gmail.com>
Date: Sun 11 Dec 2005 - 01:41:42 EST

If you are running on a Linux system you can try an experimental version of the Matrix package that is available from the SVN archive. The URL is

https://svn.r-project.org/R-packages/branches/Matrix-mer2

The code in that branch uses what is called a supernodal sparse Cholesky factorization which works very well for these types of models. Once I get the Laplace and AGQ methods for generalized linear mixed models working in this formulation I will merge the branch back in to the trunk and release a version of the Matrix package using this.

I haven't created a Windows binary for this experimental branch because I still can't get the compilation system to work on my Windows computer. If someone wants to try compiling on Windows from the SVN sources I would appreciate it.

On 12/9/05, Dick Beyer <dbeyer@u.washington.edu> wrote:
> I have been using lme from nlme to do a 3-way anova with all the effects treated as random. I was wondering if someone could direct me to an example of how to do this using lmer from lme4.
>
> I have 3 main effects, tim, trt, ctr, and all the interaction effects tim*trt*ctr. The response variable is ge.
>
> Here is my lme code:
>
>
> dat <- data.frame(ge=ge,trt=factor(trt),tim=factor(tim),ctr=factor(ctr))
> dat$grp = as.factor(rep(1, nrow(dat)))
>
> # dim(dat) = 216x5
>
> w <- lme(ge ~ 1,data=dat,
> random = list(
> grp = pdBlocked(list( pdIdent(~ trt - 1)
> ,pdIdent(~ tim - 1)
> ,pdIdent(~ ctr - 1)
> ,pdIdent(~ trt:tim - 1)
> ,pdIdent(~ trt:ctr - 1)
> ,pdIdent(~ tim:ctr - 1)
> ,pdIdent(~ trt:tim:ctr - 1)
> )
> )
> )
> )
>
> I was trying the following as a starting place:
>
> lmer(ge~1+(1|tim)+(1|trt)+(1|ctr), data=dat)
>
> but this causes my R session to terminate.
>
> > sessionInfo()
> R version 2.2.0, 2005-11-15, i386-pc-mingw32
>
> attached base packages:
> [1] "methods" "stats" "graphics" "grDevices" "utils" "datasets" "base"
>
> other attached packages:
> lme4 lattice Matrix
> "0.98-1" "0.12-11" "0.99-2"
>
> Thanks very much for any help or pointers,
> Dick
> *******************************************************************************
> Richard P. Beyer, Ph.D. University of Washington
> Tel.:(206) 616 7378 Env. & Occ. Health Sci. , Box 354695
> Fax: (206) 685 4696 4225 Roosevelt Way NE, # 100
> Seattle, WA 98105-6099
> http://depts.washington.edu/ceeh/ServiceCores/FC5/FC5.html
> http://staff.washington.edu/~dbeyer
>
> ______________________________________________
> R-help@stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
>



R-help@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html Received on Sun Dec 11 01:49:37 2005

This archive was generated by hypermail 2.1.8 : Fri 03 Mar 2006 - 03:41:35 EST