Re: units as a model term
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: units as a model term

Dear Nick,

I do not follow your comments.

You cannot add 'units' without 'changing' the model.

How important the change is will depend on the data.

Adding 'units' typically increases the autocorrelation.

For example, if the nugget variance was 3, the correlated variance was 2
and the AR correlation was 0.5, the implied Lag 1 covariance is 0.5 x 2 = 1

If the nugget variance is omitted from the model, the correlated variance
has to accommodate all the variance (i.e. becomes 5 in this example).

The lag 1 covariance is still 1 leading to a correlation of 0.2.

In ASREML, the model without units would give  S2=5, phi=0.2
           the model with units would give     S2=2, units=3, phi=0.5
           This reckoning is not exact in that it is based only on the lag 1
           correlation.  Obviously ASREML is also taking some notice of
           higher order lag correlations.
   Also, note the LogL may or may not change much, and the fit of the
   variance structure is not changing much even though the
   actual parameters printed appear quite different.  Any other variance
   component may not change much in absolute terms but its gamma value
   would change becasue it is expressed relative to S2.       

Does this help?

> X-Authentication-Warning: petidomo set sender 
to using -f
> Date: Mon, 14 Dec 1998 11:21:48 +0800
> Mime-Version: 1.0
> To:
> From: "N.W. Galwey" <>
> Subject: units as a model term
> Inclusion of units as a random term in an ASREML linear model causes small
> differences in the maximum likelihood, the parameter estimates and the shape
> of the variogram.   My purpose in inculding units was to separate the
> spatial and nugget components of the residual, not to change the model.   
> What's going on?
> _____________________________________________________________________
> N.W. Galwey,
> Faculty of Agriculture,
> University of Western Australia,
> Nedlands, WA 6709, Australia.
> Tel.: +61 9 380 1959 (direct line)
>       +61 9 380 2554 (switchboard)
> Fax:  +61 9 380 1108
> --
> Asreml mailinglist archive:

Arthur Gilmour PhD                    email:
Senior Research Scientist (Biometrics)                 fax: <61> 2 6391 3899
NSW Agriculture                                             <61> 2 6391 3922
Orange Agricultural Institute               telephone work: <61> 2 6391 3815
Forest Rd, ORANGE, 2800, AUSTRALIA                    home: <61> 2 6362 0046

ASREML is still free by anonymous ftp from pub/aar on
    or point your web browser at 

To join the asreml discussion list, send the message  

The address for messages to the list is

Asreml mailinglist archive:

                        <> <> <> <> <> <> <>
"For the wages of sin is death,
  but the gift of God is eternal life in Christ Jesus our Lord." Romans 6:23

Asreml mailinglist archive: