Hello,
My data are numbers of trees in plots sampled in a number of forest stands.
Some stands were subjected to a treatment, others not. Several plots were
sampled per stand to get a better idea of what the stand means were, but
replication is really at the stand level. Therefore I think this is a
split-plot design.
I would like to know whether the treatment affected the number of trees, so:
results<-aov(trees ~ as.factor(treatment) + Error(stand))
However, the frequency distribution of trees is highly skewed, with lots of
zeros. I was therefore considering using a generalized linear model, perhaps
with Poisson error. However, the glm function does not seem to support
adding an Error() term to the model.
My question is: is there any way of modelling the experimental design in
glm, or should I transform the data as best as I can and stick with aov?
Many thanks,
Dan Bebber
____________________________
Dr. Daniel P. Bebber
Department of Plant Sciences
University of Oxford
South Parks Road
Oxford
OX1 3RB
Tel. 01865 275060
[[alternative HTML version deleted]]
______________________________________________
[EMAIL PROTECTED] mailing list
https://www.stat.math.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
