Dear lmer users,
The experiment includes 15 groups of (3 males and 1 female). The female is
characterized by its quality Q1 and Q2. Each male of a group is characterized
by the number of MatingAttempts (with Poisson distribution). I want to examine
if male mating attempts depend on female quality. I can see from graphic
exploration that the within-group heterogeneity of male attempts increases
with female quality Q1.
When including the method weights in the function lmer, I get the message that
variables' length varies and the model does not run.
lmer(MatingAttempts~Q1+Q2+(1|Group),data=file,family="poisson",na.action=na.omit,
REML=FALSE, weights=varExp(form=~Q1))
If I run the same model (fixed effects and random effects) with lme, it
works properly, which shows that there is no problem with data structure.
lme(MatingAttempts~Q1+Q2,random=~1|Group,data=file,na.action=na.omit,
method="ML", weights=varExp(form=~Q1))
I saw on the forum that lmer had problems in taking into account variance
heterogeneity. Yet, the messages were old and there are maybe new solutions.
How can I correct the analyses for this problem of heteroscedasticity?
Should I normalise the within group variance before implementing the model? And
deal with the variance (as a new variable to explain) in another model?
Is there another way to solve this problem?
Thank you in advance for your help
Doris Gomez
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
