Skip to contents

Prior predictive checks

Source: R/check_priors.R
check_priors.Rd

Simulates from the prior marginal distribution of the data to assess the consistency of the chosen priors with domain knowledge (Gabry et al. 2019) and creates a visualization from the prior predictive checks.

Usage

check_priors(model = NULL, ...)

# S3 method for class 'stanreg'
check_priors(model = NULL, predictors = NULL, ...)

Arguments

model

A Bayesian model of class stanreg or brmsfit. Note that this model should include draws from the prior predictive distribution. This can be achieved, e.g., by using bayestestR::unupdate() on the fitted model, or setting prior_PD = TRUE (when using rstanarm) or sample_prior = "only" (when using brms).

...

Currently not used.

predictors

Character vector with names of one or more model predictors for which prior predictive checks should be visualized.

Value

A ggplot2 object visualizing the prior predictive distribution, overlaid with boxplots indicating where the probability mass of the prior distribution is located.

Details

Prior predictive checks allow researchers to verify whether the mathematical definitions of their priors accurately reflect the assumed underlying reality. Data are sampled from the model using only the prior distributions (the prior predictive distribution), before any observed data are considered. By visualizing the prior predictive distribution, researchers can assess whether their chosen priors generate plausible data. This provides a crucial sanity check on the model's assumptions and their consistency with domain knowledge.

References

Gabry, J., Simpson, D., Vehtari, A., Betancourt, M., & Gelman, A. (2019). Visualization in Bayesian Workflow. Journal of the Royal Statistical Society Series A: Statistics in Society, 182(2), 389–402. doi:10.1111/rssa.12378

See also

See documentation of see::plot.see_check_priors() for available arguments to change the plot appearance.

Examples

# \dontrun{
# model with correctly defined priors. outcome is binary, prior
# predictive checks indicate the predicted probability mass based
# on the prior distributions - the resulting pattern aligns with
# our real-world assumptions
model <- insight::download_model("stan_prior_checks_1")
check_priors(model, "mmse")


# model with default (weakly informative) priors, which is poorly
# calibrated. It pushes probability mass almost exclusively to the
# extremes of 0 and 1, leaving the plausible middle range largely
# unsupported
model <- insight::download_model("stan_prior_checks_2")
check_priors(model, "mmse")

# }