Test of random effect in lme4

There is quite a bit on the message board. For example this string:
https://stat.ethz.ch/pipermail/r-sig-mixed-models/2008q2/000743.html

Dr Anna R. Renwick
Research Ecologist
British Trust for Ornithology, 
The Nunnery, 
Thetford, 
Norfolk, 
IP24 2PU, 
UK
Tel: +44 (0)1842 750050; Fax: +44 (0)1842 750030 

-----Original Message-----
From: Stefano Leonardi [mailto:stefano.leonardi at unipr.it] 
Sent: 11 March 2010 16:03
To: Anna Renwick
Cc: r-sig-mixed-models at r-project.org
Subject: Re: [R-sig-ME] Test of random effect in lme4

There has been a lot of discussion previously whether we should remove
random effects based on LRT. The reason is that you added the random

effect

based on your study design and whether it is significant or not it should
remain in there. I am not sure there is any definite rule and maybe it
depends on your study and personal view point.

Dr Anna R. Renwick
Research Ecologist
British Trust for Ornithology,
The Nunnery,
Thetford,
Norfolk,
IP24 2PU,
UK

Can someone please give some literature references about this 
discussion. I am very interested.

According to my experience in Ecology it often happens  that someone 
needs to understand if a variable or factor is 
important/relevant/significant in determining some other variable.

I would like to understand up to which point the arbitrariness of the
experimenter is considered tolerable.

Thanks
Stefano

======================================================================
  Stefano Leonardi
  Dipartimento di Scienze Ambientali
  Universita` di Parma               E-mail:stefano.leonardi(at)unipr.it
  Viale Usberti 11a                             Phone : +39-0521-905659
  43100 PARMA  (Italy)                          Fax   : +39-0521-905402