Re: positive LogL

My understanding of GLMMs in ASReml is they are based on Schall's (1991) method
which is simply based on a linearisation and hence the first two moments. There
is no log-likelihood and ASReml uses an approach simply to fit in with the LMM
software available. Thus Bruce's quote from Arthur confirms that the LogL
printed is simply a mechanism to monitor convergence (of a pseudo-likelihood).
The question of model selection is therefore completely an open one (in the
Schall context).

I believe the long-term aim is to provide an alternaive GLMM approach.

Ari

Quoting Bruce Southey <southey_at_uiuc.edu>:

> Hi,
> Positive LogLs happen and are expected!
>
> The use of the LogL in GLM's has been previously addressed on this list. One
> of Arthur's explanations was:
>
> "The LogL value reported by ASReml is for the analysis of the working
> variable. As this working variable changes between iterations and especially
> between models, the LogL values are not comparable - even to confirm
> convergence of a particular model."
>
> Regards
> Bruce
>
>
>
> ---- Original message ----
> >Date: Fri, 24 Nov 2006 15:09:09 -0500
> >From: Agus Susanto <asusanto_at_UOGUELPH.CA>
> >Subject: positive LogL
> >To: ASREML-L_at_AGRIC.NSW.GOV.AU
> >
> >Dear ASReml-ers,/
> >/I run the following model for categorical data:
> >lsw !POISSON ~ mu grp parity !r animal ide(animal)
> >On the output I had positive LogL and a Warning: The LogL value is
> >unsuitable for comparing GLM models. Can somebody please help me
> >explain the case.
> >Thanks.
> >//
> >
> >--
> >Agus Susanto
> >Department of Animal and Poultry Science
> >Centre for Genetic Improvement of Livestock
> >University of Guelph, Guelph, Ontario, Canada N1G 2W1
> >Rm.128, Phone (519) 824-4120 X-58353
>

--
Professor Ari Verbyla
School of Agriculture, Food and Wine, The University of Adelaide
and
Statistical Bioinformatics - Agribusiness, CMIS, CSIRO
Phone: +61 8 8303 6760 or 8769
Fax:   +61 8 8303 6761 or 8763
Ari.Verbyla_at_adelaide.edu.au
Ari.Verbyla_at_csiro.au
CRICOS Provider Number 00123M
-----------------------------------------------------------
This email message is intended only for the addressee(s)
and contains information that may be confidential and/or
copyright.  If you are not the intended recipient please
notify the sender by reply email and immediately delete
this email. Use, disclosure or reproduction of this email
by anyone other than the intended recipient(s) is strictly
prohibited. No representation is made that this email or
any attachments are free of viruses. Virus scanning is
recommended and is the responsibility of the recipient.