R help - Jul 2006 - Questions about extract-lme.cov

Hi, all:

I am using extract.lme.cov to extract the covariance matrix of lme. But
the results are not expected. 

For example, 

 b <- lme(travel~1,Rail,~1|Rail)

The default correlation for lme is no correlation within groups.
>extract.lme.cov(b,Rail)
The part of covariance matrix looks like:

         1        2        3        4        5        6
1 631.4778 615.3111 615.3111   0.0000   0.0000   0.0000
2 615.3111 631.4778 615.3111   0.0000   0.0000   0.0000
3 615.3111 615.3111 631.4778   0.0000   0.0000   0.0000
4   0.0000   0.0000   0.0000 631.4778 615.3111 615.3111
5   0.0000   0.0000   0.0000 615.3111 631.4778 615.3111
6   0.0000   0.0000   0.0000 615.3111 615.3111 631.4778

Where does the covariance come from? Maybe I'm missing sth here?

Thanks,

Hua Li 

	[[alternative HTML version deleted]]

Li, Hua <HUL <at> stowers-institute.org> writes:
> I am using extract.lme.cov to extract the covariance matrix of lme. But
> the results are not expected. 
> 
>  b <- lme(travel~1,Rail,~1|Rail)
> 
> The default correlation for lme is no correlation within groups.
> 
> >extract.lme.cov(b,Rail)
> 
> The part of covariance matrix looks like:
> 
>          1        2        3        4        5        6
> 1 631.4778 615.3111 615.3111   0.0000   0.0000   0.0000
> 2 615.3111 631.4778 615.3111   0.0000   0.0000   0.0000
> 3 615.3111 615.3111 631.4778   0.0000   0.0000   0.0000
> 4   0.0000   0.0000   0.0000 631.4778 615.3111 615.3111
> 5   0.0000   0.0000   0.0000 615.3111 631.4778 615.3111
> 6   0.0000   0.0000   0.0000 615.3111 615.3111 631.4778
> 
> Where does the covariance come from? Maybe I'm missing sth here?
Maybe 

VarCorr(b) 

gives all you need, because

extract.lme.cov2(b,Rail,start.level=2)

or even 
extract.lme.cov(b,Rail,start.level=2)

are a bit redundant.

Dieter