# Error in lmer - how to fix

I am asking another question (as each one is a new issue and I am trying to get these models run). However I don't want to become a nuisance either. However I ran the following and wonder whether anyone could see what I have done wrong. I know I am comparing a lot of factors but really want to find out which ones really contribute and would like to be able to figure out what is wrong so that I will know what to do in the future. The code I used is as follows: Note I was going to use MuMNi to run all the model combinations:

ABmodel <- read.csv(file.choose(), header=T)


I typed out as.factor() for Brood, Site, Age.class, Tide, tide.h.l, tide.inc.out and MF.vs.OF then attached lme4:

local({pkg <- select.list(sort(.packages(all.available = TRUE)),graphics=TRUE)
if(nchar(pkg)) library(pkg, character.only=TRUE)})


Then I attempted run the following model (I know some factors seem to be related but they are different):

fullABmodel1 <- lmer(Feeding ~ Site + Age.class + tide.h.l + tide.inc.out +
Tide + HDp + NHDp + Foraging.Eff + MF.vs.OF + Age +
HDr + NHDr + AllDr + (1|Brood), data=ABmodel, REML=FALSE)


But I got the following Error

Error in mer_finalize(ans) : Downdated X'X is not positive definite, 7.


I wish that I could attach my CSV file so that you could see it (although it contains additional columns for other analyses).

Not sure what this means. Please note my understanding of the intricacies of models is limited. Does it help to know when I ran it as an lme in nlme I got a message saying

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

-
Hi there, could you please add your sample size and an explanation of your variables to this question. I've not seen this error in lmer but in another context it tends to mean you have run out of degrees of freedom - i.e. sample size too small for number of variables in model. Not sure if that applies here. –  Michelle Mar 24 '12 at 6:22
Isn't your dataset already attached to a cross-post on R-help? mail-archive.com/r-help@r-project.org/msg164547.html –  chl Mar 24 '12 at 10:37
Yes, it is. I hope it is okay that I asked the question in two unrelated places. I got up at 3am with an idea. I ran the model by taking out one factor at a time and found that 'Tide' was the issue. I wonder whether it is because I have three different tide effects and technically the numbers in tide.inc.out and tide.h.l are always the same for a tide level, so maybe it was realising this. I am therefore going to run it without this third factor. I hope I don't get this message again. I apologise for all of my questions but am in great need of getting this done. –  Dragonwalker Mar 24 '12 at 14:23