Choosing weakly informative priors for population-level effects in a Poisson GLMM

[torkar]

Yeah, that blog post by @anon75146577 was a joy to read. But this link, where you explain things very nicely, together with @richard_mcelreath’s examples in his book, taught me to always plot priors no matter what:

Use of `sample_prior='only'` in brms

@paul.buerkner already solved the problem with the intercept, but before that I was thinking about this For poisson(link=log) they can be too wide, and compared to your data they seem to be too wide. If I read your model right, the intended model has four additive terms for log intensity with each having a prior N(0,1), and thus the prior for the log intensity would be N(0,4^2), which corresponds to assuming that there is 5% probability that the intensity is larger than 720. This may or may n…

@jocateme, @Solomon’s work on the 2nd edition of Richard’s book also has more examples on prior predictive checks:

Statistical rethinking with brms, ggplot2, and the tidyverse: Second edition