[R] var-covar matrices comparison:

Aldi Kraja aldi at wustl.edu
Tue Feb 21 16:19:03 CET 2006


Thank you Patrick for your thoughts.

I was expecting someone had written a function that does maybe a 
hierarchical comparison of var-covar matrices using the flury hierarchy 
as proposed by
Philips and Arnold in Evolution 53(5), 1999;1506-1515. But your email 
has an interesting idea, so I will try to test it.
Thank you,

Aldi

Patrick Burns wrote:

> My first thought is to use a random permutation test.
> In this setting the main question you need to ask is
> what distance measure do you want to use between
> variance matrices -- there are lots of choices.  One
> that I've found useful is the absolute value of the
> maximum eigenvalue of the difference of the matrices.
>
> If you have a hypothesis about how the variances may
> differ, then you should be able to come up with a more
> powerful statistic.
>
> Patrick Burns
> patrick at burns-stat.com
> +44 (0)20 8525 0696
> http://www.burns-stat.com
> (home of S Poetry and "A Guide for the Unwilling S User")
>




More information about the R-help mailing list