[S] manova()

Rolf Turner (rolf@math.unb.ca)
Sat, 5 Dec 1998 17:07:54 -0400 (AST)


The on-line documentation (Splus version 3.4 for Unix) on
summary.manova indicates that all of the four possible test
statistics (Pillai's trace, Wilks' lambda, the Hotelling-Lawley
trace, and Roy's maximum root) are all transformed to approximate
F statistics in order to assess their significance. The p-values
given for the various statistics indeed appear to be precisely the
p-values for the accompanying ``approx. F'' that summary.manova
provides.

Can anyone tell what the transformations are, and how the
corresponding degrees of freedom are found? Of the references cited
in the documentation (Hand and Taylor, Mardia, Kent and Bibby, and
Seber) Seber is the only book I can immediately lay hands on. And as
far as I can tell, he makes no mention of such a general
transformation to approximate F statistics.

If the transformations are to be found in one of the above mentioned
references, I would appreciate a specific page citation.

Thanks for any help that anyone can provide.

cheers,

Rolf Turner
rolf@math.unb.ca
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news