[R] LME without convergence

[Douglas Bates]

On 2/28/07, Ralf Sommerfeld <sommerfeld at mpil-ploen.mpg.de> wrote:
> Dear R-help list readers,
>
> I am fitting a mixed model using the lme function (R V 2.3.1 for
> Windows). This is an example:
>
> dep<-c(25,40,33.33,60,70.83,72,71.43,50,40,53.33,64,54.17,60,53.57)
> yes<-c(0,1,2,3,4,5,6,0,1,2,3,4,5,6)
> treat<-c(1,1,1,1,1,1,1,0,0,0,0,0,0,0) #factor
>
> If I now fit a model with random slopes as well as intercepts:
>
> model1<-lme(dep~yes,random=yes|treat)
>
> R encounters the following problem:
>
> Fehler in lme.formula(dep ~ yes, random = ~yes | treat) :
>          iteration limit reached without convergence (9)
>
> Following other instructions on that problem, I tried to adjust the
> iteration limits and tolerances using lmeControl(), however this did not
> work.
>
> Can someone help me?

Not really.  You're trying to estimate three variance-covariance
parameters from two groups and just about any way you try to do that
will encounter difficulties.