Re: [R] Correct nested design for GLM

I guess your model specification is incoherent. It's seems to me that you incorporated some kind of lmer() style with "(0 + Micro/Trt/Year)" term. Consider the toy data and code below. Nesting model is m2. Models m1 and m3 are the same so showing that terms in brackets don't are used. da <- expan

[R] Correct nested design for GLM

Hi, I am currently running the following negative binomial GLM: glm89.nb <- glm.nb(AvGUD ~ Year*Trt*Micro + (0 + Micro/Trt/Year)) where Year has 3 levels, Trt has 2 levels, and Micro has 3 levels. >From what I have read the above model has a 3 way interaction (Year*Trt*Micro), and Micro is nest