[S] Problems with varcomp

Robert Kinley (KINLEY_ROBERT@lilly.com)
Mon, 26 Oct 1998 15:19:46 +0000


According to the documentation, the fitted values returned by varcomp
should be based only on the fixed effects in the model.

However, after fitting a model like :-

y ~ (fixed.factor.1 + fixed.factor.2) / (random.factor.1+random.factor.2)

I noticed that there were too many unique values returned in $fitted .

Checking the code of varcomp I see that the fitted values seem to be formed by
subtracting the residuals from the raw data, which I would have thought was
only appropriate if there are no random factors in the model.

Has anyone else come across this 'feature' ?

Is it a bug, are am I misunderstanding something ?

thanks ................ Bob Kinley .............. rdk@lilly.com

-----------------------------------------------------------------------
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