Re: Change in total Vp when adding variance components

From: Arthur <asremlforum_at_VSNI.CO.UK>
Date: Mon, 2 Mar 2009 10:25:51 +0000

Dear Celine,

First, I note that the model mu !r ID
is an inapproriate model for this data.

mu !r ID ide(ID)
is a more appropriate model.

When you fit an inappropriate model, REML tries to choose parameters that fit best but can't do a very good job. It is possible that there a several local maxima
and ASReml has converged to one of them. A round ped in a square hole does not fit very well.

Otherwise, it is trying to accomodate covariances which means inflating the variances.

The residual variance (between repeats for each individual) has not changed much (72 v 74).
I do not have a very rational explanation of the link between the other values

Arthur Gilmour

Principal Research Scientist (Biometrics)
NSW Dept Primary Industries

-------------------- m2f --------------------

Sent using Mail2Forum (

Read this topic online here:

-------------------- m2f --------------------
Received on Tue Mar 02 2009 - 10:25:51 EST

This webpage is part of the ASReml-l discussion list archives 2004-2010. More information on ASReml can be found at the VSN website. This discussion list is now deprecated - please use the VSN forum for discussion on ASReml. (These online archives were generated using the hypermail package.)