Bruce Southey wrote:
> Sent: Monday, August 07, 2000 9:23 PM
> Subject: Re: R-square statistic
> Note that the residual variance is a bad criterion as the log(y) produces
> a smaller residual than the untransformed value - it does not mean that
> the model fits any better.
I think this is a bit of a red herring. Provided that one sticks to the
same vector of y-values, surely MS(Resid) is quite a good indicator of the
relative goodness-of-fit of any two models? With the caveat that it only
leads to a FORMAL comparison (test of significance) in certain circumstances
(e.g. both models linear, one a sub-model of the other, i.e. the ANOVA
situation).
Nick Galwey
_____________________________________________________________________
N.W. Galwey,
Faculty of Agriculture,
University of Western Australia,
Nedlands, WA 6907, Australia.
Tel.: +61 89 380 1959 (direct line)
+61 89 380 2554 (switchboard)
Fax: +61 89 380 1108
--
Asreml mailinglist archive: http://www.chiswick.anprod.csiro.au/lists/asreml