Skip to content
Prev 2312 / 7420 Next

Re : guidance required (GLM?)

On Wed, 2011-08-24 at 11:16 +0100, momadou sow wrote:
In general, no you don't - a GLM is far more than a logistic regression.
Otherwise, why would glm() allow for more families than just binomial?

As biomass is positive (or at least non-negative) and likely to exhibit
a non-constant mean-variance relationship, lm(log(biomass) ~ ....) or
glm(biomass ~ ...., family = Gamma) might be more appropriate. For the
latter you might need to investigate the various link function options.

HTH

G