A simple categorical model for alligator food choice as a function of alligator size:
Gators <- read.table("http://www.stat.ufl.edu/~aa/cat/data/Alligators.dat", header = TRUE)
require(brms)
mod.cat.brms <- brm(y ~ x, prior = prior(normal(0, 5), class = Intercept) + prior(normal(0,2.5), class = b), family = categorical, data = Gators)
The model fits without incident, but then there’s a warning:
Warning messages:
1: Specifying global priors for regression coefficients in categorical models is deprecated and may not work as expected.
2: Specifying global priors for regression coefficients in categorical models is deprecated and may not work as expected.
However, the parameter estimates look just fine. They are essentially identical to those of a corresponding frequentist model. Clearly nothing has gone awry, so why the warning?
This is a simple example where it’s easy to confirm that nothing is actually wrong. But the categorical models I’m fitting in my own research are hierarchical, complex and take very long to fit, so seeing the same warning in that context is much more unnerving.