Check if Prior is Informative — check

Performs a simple test to check whether the prior is informative to the posterior. This idea, and the accompanying heuristics, were discussed in Gelman et al. 2017.

Usage

check_prior(model, method = "gelman", simulate_priors = TRUE, ...)

Arguments

model: A stanreg, stanfit, brmsfit, blavaan, or MCMCglmm object.
method: Can be "gelman" or "lakeland". For the "gelman" method, if the SD of the posterior is more than 0.1 times the SD of the prior, then the prior is considered as informative. For the "lakeland" method, the prior is considered as informative if the posterior falls within the 95% HDI of the prior.
simulate_priors: Should prior distributions be simulated using simulate_prior() (default; faster) or sampled via unupdate() (slower, more accurate).
...: Currently not used.

Value

A data frame with two columns: The parameter names and the quality of the prior (which might be "informative", "uninformative") or "not determinable" if the prior distribution could not be determined).

References

Gelman, A., Simpson, D., and Betancourt, M. (2017). The Prior Can Often Only Be Understood in the Context of the Likelihood. Entropy, 19(10), 555. doi:10.3390/e19100555

Examples

# \donttest{
library(bayestestR)
model <- rstanarm::stan_glm(mpg ~ wt + am, data = mtcars, chains = 1, refresh = 0)
check_prior(model, method = "gelman")
#>     Parameter Prior_Quality
#> 1 (Intercept)   informative
#> 2          wt uninformative
#> 3          am uninformative
check_prior(model, method = "lakeland")
#>     Parameter Prior_Quality
#> 1 (Intercept)   informative
#> 2          wt   informative
#> 3          am   informative

# An extreme example where both methods diverge:
model <- rstanarm::stan_glm(mpg ~ wt,
  data = mtcars[1:3, ],
  prior = normal(-3.3, 1, FALSE),
  prior_intercept = normal(0, 1000, FALSE),
  refresh = 0
)
check_prior(model, method = "gelman")
#>     Parameter Prior_Quality
#> 1 (Intercept) uninformative
#> 2          wt   informative
check_prior(model, method = "lakeland")
#>     Parameter  Prior_Quality
#> 1 (Intercept)    informative
#> 2          wt misinformative
# can provide visual confirmation to the Lakeland method
plot(si(model, verbose = FALSE))

# }