[R-sig-ME] lme error message

[Luca Borger]

Hello,

>  Singularity in backsolve at level 0, block 1

in fact my experience is that in this case there is some problem with the fixed effects (e.g. not enough data for all factor combinations etc.), otherwise it would say level 1 or higher. By fitting a simpler model usually it disappears.

Disclaimer: just limited empirical experience, I have never checked the code to understand better this error message, so take this with caution ;-)


HTH


Cheers,

Luca


----- Original Message -----
From: "Stuart Luppescu" <slu at ccsr.uchicago.edu>
To: "Ken Beath" <ken at kjbeath.com.au>
Cc: r-sig-mixed-models at r-project.org
Sent: Wednesday, July 22, 2009 12:22:56 AM GMT +01:00 Amsterdam / Berlin / Bern / Rome / Stockholm / Vienna
Subject: Re: [R-sig-ME] lme error message

On 水, 2009-07-22 at 08:15 +1000, Ken Beath wrote:
> On 22/07/2009, at 7:54 AM, Stuart Luppescu wrote:
> 
> > Running a nested model with lme, and getting this message:
> >
> > Error in MEEM(object, conLin, control$niterEM) :
> >  Singularity in backsolve at level 0, block 1
> >
> > Can anyone explain this?
> >
> 
> Usually means that some of the random effects are highly correlated or
> possibly zero.

Random effects? Only the intercept is random, and in a more restricted
model there's tons of variance.
-- 
Stuart Luppescu -=- slu .at. ccsr.uchicago.edu        
University of Chicago -=- CCSR 
才文と智奈美の父 -=-    Kernel 2.6.28-gentoo-r5                
Oz: Sometimes when I'm sitting in class...you know,
 I'm not  thinking about class 'cause that would
 never happen... I think  about kissing you. And
 it's like everything stops, it's like,  freeze
 frame: Willow kissage. 
 
 
 
 

_______________________________________________
R-sig-mixed-models at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models