[R] Error in MEEM

[Vaquero-Alba, Iker]

   Hello everyone:

   It's the first time I write to this mailing list. Sorry in advance if my doubt has already been posted before, but I have been checking the archives and I haven't been able to find anything satisfactory.

   I am running a mixed effects model with nested effects (site and pair, referred to barn swallow nests located in different places in different farms). My dependent variable is a measure of plumage coloration. When I include four independent variables in the model everything is Ok, it works, but when I try to write a model with five independent variables like this:

   insmodel<-lme(briventral~sex*inslarge*temp*humid*wspeed,random=~1|site/pair,method="ML")

   I get this error message:

   Error in MEEM(object, conLin, control$niterEM) :
   Singularity in backsolve at level 0, block 1

   I haven't been able to find many people who usually work with mixed effects models, but apparently this problem is due to the fact that the model “runs out” of degrees of freedom when including the fifth variable.

   I’ve tried to remove the 5 and 4-way interactions, but the model only works when at least two of the four 3-way interactions not involving “sex”  are removed, no matter if the ones involving “sex” are kept in or not.

   Do you know what that error message means and how I can fix it, or am I trying to include too many independent variables?

   Thank you very much.

   Iker