I am working with a data set containing nested groups and am wondering how to properly specify the models in my generalized (glmmTMB) and linear mixed effect modellings (lmer) for tree survival and growth rate studies in R. My experiment is about understanding the effects of my treatments on tree survival and relative growth rate.
My experimental design - I have 4 sites, 5 blocks (serving as treatment replicates) nested within each site and 4 treatment plots within nested each block. In each treatment plot, 3 sapling tree replicates are planted closely, hence I collected 3 observations of the survival and height data in each plot. Due to site variation, I will use site as my fixed factor while modelling block (or potentially plot) as my random effects. However, I would also like to show nestedness of plot within block within site, as I am hoping to account for variation within blocks and plots, mainly due to autospatial correlation nature of trees being planted more closely to each other. However I wasn't sure if I could include (1|Site/Block) as my random effects since I have Site as fixed factor. Below shows my ideal structure:
glmmTMB(Survival~Treatment*Site+Height3M_c+(1|Block2),
family=binomial,data=dataM24)
In terms of including plot as random variable, I am aware I only have three sample size in each plot, hence it may be overrepresented, causing larger p values in my data. Any advice on that too? enter image description here
I have tried creating Block2 column with stringed words: Site1Block1Plot to indicate independence between blocks from different site (so that Block 1 from Site 1 is treated differently than Block 1 from Site 2 if that makes sense?). The results/contrasts are quite different from using (1|Block) as random effect. So how can I show nestedness?