[R] lme function

[Spencer Graves]

      RSiteSearch("lme spatial correlation", "functions") produced 10 
hits for me just now.  The sixth title on that list was "spatial 
correlation structure" 
(http://finzi.psych.upenn.edu/R/library/nlme/html/corSpher.html).  This 
is the help page for the "corSpher" function.  The Examples section 
there includes references to selected pages in Pinheiro and Bates (2000) 
Mixed-Effects Models in S and S-Plus (Springer), which  for me is 
essential documentation for 'lme' and is the best book I know on 
mixed-effects models generally.  The value of that book is greatly 
enhanced by the availability of script files "ch01.R", "ch02.R", ..., 
"ch06.R", "ch08.R" (in the "~R\library\nlme\scripts" subdirectory of 
your R installation directory).  These contain R code to reproduce all 
the data analyses in the book.  There are a very few cases where the 
syntax is different between R and that documented in the book [e.g., x^2 
must be I(x^2)].  Before I found the script files, I couldn't understand 
why I got substantially different results from the book when just typing 
the commands into R. 

       Hope this helps. 
      Spencer Graves

Mark Wilson wrote:
> Hello.
>
> As advised by Mick Crawley in his book on S+, I'm trying to use the lme
> function to examine a linear relationship between two variables measured at
> 60 locations in 12 sites, while taking account of any spatial
> autocorrelation (i.e. similarity in variation between the two variables that
> is due to site). I am using the function as follows:
>
> model<-lme(yvariable~xvariable,random=~xvariable|site)
>
> If you know your way around this function, I would be very grateful if you
> could confirm that this approach is a valid one, or point out why it isn't.
> I'd also be very keen to hear any suggestions regarding alternative ways to
> address the spatial autocorrelation in my data (I'm hoping to arrive at a
> slightly more elegant solution than simply taking site averages for each of
> the two variables and running a correlation using these mean values).
>
> Thanks,
>
> Mark
>