On Mon, Jun 21, 2010 at 10:01 PM, Bert Gunter gunter.ber...@gene.com wrote:
1. This discussion probably belongs on the sig-mixed-models list.
Correct, but I'll keep it here now for archive purposes.
2. Your claim is incorrect, I think. The structure of the random errors =
model covariance
Hi Joris (CC Simon),
Thanks for your kind replies and for being so responsive.
I think this post boils down to two main questions (which I feel are very
important for gams modelling):
1- Is it appropriate to use LR tests in gamm to test model reduction?
2- If yes, which degrees of freedom
Hi Carlo,
You should get the book of Simon Wood and read it thoroughly. It's all
explained in there, but it will lead me too far to copy it all in a
mail. In short : random effects are part of the error structure of the
model, not of the model itself. They're added to correct the error on
the
, testing gamm vs lme,which degrees of
freedom?
Hi Carlo,
You should get the book of Simon Wood and read it thoroughly. It's all
explained in there, but it will lead me too far to copy it all in a
mail. In short : random effects are part of the error structure of the
model, not of the model itself
4 matches
Mail list logo
