On Tue, 02 May 2000 03:42:52 GMT, Mike and Michele Hewitt
<[EMAIL PROTECTED]> wrote:
> Can anyone tell me the conditions for using Roy's Largest Root for
> multivariate repeated measures rather than the Pillai's, Wilks, or
> Hotelling's which may be "more conservative and perhaps less powerful".
I know about multivariate; I am less sure about your exact context of
"multivariate repeated measures" but I think this applies.
The different tests have been written as weighted combinations of the
eigenvalues. Wilks test spreads the test across all of the roots of
the eigen problem.
If the "largest root" is what is interesting, then you think the
important effects are in the first eigenvector.
I usually think the effect will be an obvious, first-eigenvector
effect; but I also think that I should be able to define the contrast
in advance: So I can do a t-test (say) on an obvious "Summary score"
instead of doing an obscure test on an newly defined vector. Since it
does not have to reckon on capitalizing on chance, the test on the
summary will be more powerful -- unless I have been badly mistaken in
defining it.
You use Roy's if you think there is a simple effect and (for some
reason) you can't describe that in advance.
--
Rich Ulrich
http://www.pitt.edu/~wpilib/index.html
===========================================================================
This list is open to everyone. Occasionally, less thoughtful
people send inappropriate messages. Please DO NOT COMPLAIN TO
THE POSTMASTER about these messages because the postmaster has no
way of controlling them, and excessive complaints will result in
termination of the list.
For information about this list, including information about the
problem of inappropriate messages and information about how to
unsubscribe, please see the web page at
http://jse.stat.ncsu.edu/
===========================================================================
