Dear Allstaters:
I have data for numbers of moose counted in clearcuts (i.e., survey unit)
of various sizes. I would like to fit a Poisson or negative binomial
model describing variation in counts per survey unit relative to
environmental covariates. My question is simple:
Will I violate assumptions of the Poisson distribution if I use moose
density (number counted/area of sample unit = density) which
occur as fractions?
Thanks in advance for any and all suggestions.
Chris
|