[S] aov and NAs

Christian Hoffmann (christian.hoffmann@wsl.ch)
Thu, 02 Jul 1998 10:05:14 +0200


Hi there,

For my application of manova, Y = X*B + U, I do not want to lose all the
rows of X where there are one or more NAs in them. As I can see from the
description of %*% (which, I suspect, is used to form X'X), NAs are
allowed in the factors, but the resulting sum will be NA.

Question:
Can aov be altered so that in the computation of X'X terms of NA*something
will be treated as zero?
Are there theoretical reasons against it?

In the worst case I would have to program aov myself to reflect the desired
bahaviour.

Any help?

PS: This problem has to do with my earlier post:

>> (beta-hat-ij,beta-hat-kl) = sigma-jl*g-ik, with g the elements of G <-
>> (X'X)^(-1), where X is the design matrix. I have the feeling that G is
>> constructed inside method "aov", but somehow I cannot find it. I want to
>> make it appear as a result parameter $G.

Thank you

Christian W. Hoffmann
Swiss Federal Institute for Forest, Snow and Landscape Research
CH-8903 Birmensdorf, Switzerland
phone: ++41-1-739 22 77 fax : ++41-1-739 22 15 e-mail: hoffmann@wsl.ch
WWW: http://www.wsl.ch/wsidb/leute/cho/choHPENG.html
I prefer WWW pages which can be viewed best with *any* browser, contain
*no* cookies, *no* Java, *no* frames, and *no* required images !
-----------------------------------------------------------------------
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