Model Comparison and Specification of Priors

[xprockox]

Hi there,

In the vignette, there's a really nice tutorial on using Bayes Factors for model comparison. The sample provided constructs a few variations of models to compare:

intercept only model

m0 <- brm(Sepal.Length ~ 1, data = iris,
save_pars = save_pars(all = TRUE), refresh = 0, backend = "rstan")

main effects models

m1 <- brm(Sepal.Length ~ Petal.Length, data = iris,
save_pars = save_pars(all = TRUE), refresh = 0, backend = "rstan")

m2 <- brm(Sepal.Length ~ Species, data = iris,
save_pars = save_pars(all = TRUE), refresh = 0, backend = "rstan")

additive main effects model

m3 <- brm(Sepal.Length ~ Species + Petal.Length, data = iris,
save_pars = save_pars(all = TRUE), refresh = 0, backend = "rstan")

full model

m4 <- brm(Sepal.Length ~ Species * Petal.Length, data = iris,
save_pars = save_pars(all = TRUE), refresh = 0, backend = "rstan")

Then, models can be easily compared using bayestestR::bayesfactor_models().

However, from what I've read, computing Bayes Factors requires that models specify informative priors, which doesn't appear to have been done in these models. Is the specification of priors only required for testing model parameters? Or am I missing something here?

Any feedback would be greatly appreciated!

[mattansb]

Oh my god, you are 110% right! 😨😨😨😨😨

I will fix the vignette!!!

[mattansb]

Fixed!!

[xprockox]

Excellent, thank you!

[bwiernik]