I’m currently working on some statistical analysis. In my study I have 12 forested sites with 4 sites in one of 3 treatment groups. At each site, I had 3 plots, and within each plot I recorded whether each individual tree was alive or dead, along with other some variables.

I am interested in seeing if there is a difference in the proportion of dead trees between my 3 treatments. Would it be statistically valid to run a mixed effect binomial model using each tree as an observation (1 or 0) nested within Site and Plot to understand if there are differences in the proportions of dead trees?