Thanks for the replies from both Arthur and Bruce. Arthur got me
fixed up by finding a little syntax error that has been giving me
grief for way too long now. Bruce made me realize that I should
provide a little more information in my next email to the list.
Arthur said
> First, the R structure is defined the wrong way around.
> You say the data was sorted and presumably the variance
> for the first third should be less than for the middle third etc.
>
> The structure as you have it implies data is order as 2547 lots
> of 3. You actually wanted 3 lots of 2547.
>
Yes I had wanted 3 lots of 2547. If I had 2547 lots of three then why
in the output (.asr file) did I have only three residual variance
estimates? I had thought I was fitting three lots of 2547 and when I
got three residual variance estimates, I had no indication that I was
fitting 2547 lots of three.
Arthur said do this
> 1 2 3 !STEP 0.1
> 3 0 DIAG 9804.57 9761.96 9214.61 !GUUU !S2==1
> 2547
when I had done this
1 2 3 !STEP 0.1
2547
3 0 DIAG 9804.57 9761.96 9214.61 !GUUU
again close does not count in programming
To clarify my model, I have cattle beasts, lets call them "animals"
,measured three times after purchase for fat depth. They were
purchased at weaning from many different "Herds" and these animals
were from a number of different "sires". With a nice design to use
sires across herds.
rib ~ mu wt wt.wt cg cg.wt cg.wt.wt n2 n2.wt n2.wt.wt,
br br.wt br.wt.wt !r he he.wt s s.wt an an.wt
My model, includes the random regression components of herd of origin,
sire and animal. Animal in this case should not be confused with the
"animal" in "animal model" where we fit a relationship matrix. Sire
is the genetic component, we have only one generation here so we just
fit sire as a random effect with no account of relationships among
sires (old fashioned sire model). Herd of origin was found to be
important from the initial investigations, especially at low
weights... makes sense, above and beyond the genetic effect (sire) the
herd the calf was raised in will affect the fat depth at weaning.
Then we fit "animal" this is the effect unique to that animal on fat
depth that is not attributable to the average effect of the herd of
origin or the additive genetic effect (sire). We can fit this because we have more than one obs per animal. Things that might be
included in the "animal" effect estimated here would be any genetic
effects (additive and non-additive) not accounted for from the sire
effect, basically a permanent environment effect.
Fixing up the error variance seems to have helped. Thanks for the
input.
Steve Miller
--
Asreml mailinglist archive: http://www.chiswick.anprod.csiro.au/lists/asreml