Displaying 1 result from an estimated 1 matches for "cens1".
Did you mean:
cens
2006 May 05
1
trouble with step() and stepAIC() selecting the best model
...plant
species using a negative binomial glm.
Sometimes, summary(step(model)) or summary(stepAIC(model) does not
select the best model (lowest AIC) but just stops before.
For some species, step() works and stepAIC don't and in others, it's the
opposite.
The model is
model=glm.nb(sdg.cens1 ~ seed.cens1 + log(in.lit+1) +
log(trans.dir+1)+ log(trans.dif+1) + slope +log(pH+1) + log(CN+1) + site,
data=data)
For example, for species A using stepAIC() I get:
Call:
glm.nb(formula = tot.rec ~ slope + log(pH + 1) + log(CN + 1),
data = data, init.theta = 0.0575099635396228, l...