Dear Dr. Bates
I replied to your email before but apparently it didn't make it so I am
replying again. I would really appreciate if you could send me an example
on how you suggest to build a cell means model for fixed effects using the
NLME library. I am not sure whether you suggest to create a separate factor
for each unique combination of levels of a factor or whether you suggest to
create a factor for each level of the original factors. Wouldn’t the first
case generate collinearity problems under some situations?
Many thanks again for your help
Francisco
>From: Douglas Bates <bates at stat.wisc.edu>
>To: "Francisco Vergara" <gerifalte28 at hotmail.com>
>Subject: Re: [R] (no subject)
>Date: 08 Oct 2003 19:21:13 -0500
>
>"Francisco Vergara" <gerifalte28 at hotmail.com> writes:
>
> > I want to specify the contrasts to build a cell means model on LME
> > when there are several fixed effect as factors in the model and also
> > interactions between them. Can anybody give me a hint on how to do
> > accomplish this? How can I override the default
"Contr.Treatment" for
> > linear models? I tried removing the intercept but this will only give
> > me the mean values just for the first factor included in the model but
> > then will use "contrast treatment" for all the other
factors.
>
>You can't do what you want to do by defining interactions.
>
>If you really want a cell means model then construct a new factor
>with one level for each unique combination of your fixed factors and
>use this constructed factor and - 1 in the formula.
>
>That's what a cell means model is. There is one parameter for each
>unique combination of factor levels from the original design.
>
>Let me know if this answer is too vague and I will provide an example.
>
>--
>Douglas Bates bates at stat.wisc.edu
>Statistics Department 608/262-2598
>University of Wisconsin - Madison http://www.stat.wisc.edu/~bates/
_________________________________________________________________
See when your friends are online with MSN Messenger 6.0. Download it now