[R-sig-ME] lmer vs SAS results

[Darren Norris]

Hi Beth,
I am sure other members will add their thoughts as to why the 
differences are occurring (I know how much they love such comparisons).
However, I just wanted to add that between 5 May 2008 and 20 September 
2010 there have been around 115 messages with "SAS" on this list.
I know it is hard but it could be worth looking through some of these 
previous posts.

Secondly your AIC values are not that different between any of the 3 
models. Are you expecting particularly strong support for one over any 
other?
You might be beating yourself up needlessly?

Members of this list have made and maintain several amazingly accessible 
packages and contributed time to the wiki ( http://glmm.wikidot.com/ ) 
etc....
The fact you are using this list and the packages to me suggests that 
you do not necessarily need the "help" of the statistician -
at least when it comes to mixed models....... oh did you think about 
using BIC (seems to "agree" pretty well.....).
Best wishes,
Darren