Skip to content
Prev 3300 / 7420 Next

Zero-inflated model inquiry

Hi Peter,
With a simple model you may get satisfaction with glht() in multcomp, 
but once your models get more complicated, success with glht tends to 
wane - at least in my experience. [If anyone can provide code to use 
glht for multiple factors or whether addressing each factor one at a 
time works in this situation, please post.]

One approach to getting those 'missing' comparisons is to change the 
base in the contrast treatments. For example:

contrasts(data$impact_period)=contr.treatment(levels(data$impact_period),base="BEFORE")

Someone can correct me if I am wrong, but because these are treatment 
contrasts (mean effect), moving the reference level has no effect on 
coefficients.

Often I put the line above in a loop and change the base accordingly.

cheers,
trevor
On 9/27/2012 7:00 AM, r-sig-ecology-request at r-project.org wrote: