Skip to contents

Extract the marginal likelihood bridge sampling object from a brma model. The marginal likelihood must first be computed using add_marglik.

Usage

# S3 method for class 'brma'
bridge_sampler(samples, ...)

Arguments

samples

a brma model object.

...

additional arguments (currently not used).

Value

An object of class "bridge", or "bridge_list" for repeated bridge estimates, as returned by bridge_sampler.

Details

This function extracts the raw bridgesampling object underlying the marginal-likelihood result stored by add_marglik. If the marginal likelihood has not been computed, an error is thrown. Product-space model-averaging objects (BMA.norm, BMA.glmm, BMA.mv, RoBMA, and RoBMA.mv) do not expose bridge-sampling marginal likelihoods.

Fully fixed models can have a zero-dimensional marginal likelihood evaluated exactly. Such fits have no bridge-sampling object, so this method raises a RoBMA_exact_marglik_no_bridge error; use logml to retrieve their exact log marginal likelihood. For Bayesian model comparison, pass fitted brma objects directly to bf or post_prob.

When a stored marginal-likelihood result has no raw bridge object, extraction raises RoBMA_bridge_unavailable. Its reason is "exact_zero_dimensional" for an exact result, or "missing_upstream" when a stored result lacks its raw bridge object. For the latter, recompute the result with add_marglik.

Examples

if (FALSE) { # \dontrun{
if (requireNamespace("metadat", quietly = TRUE)) {
  data(dat.lehmann2018, package = "metadat")
  fit <- brma(yi = yi, vi = vi, data = dat.lehmann2018, measure = "SMD")

  fit <- add_marglik(fit)

  bridge <- bridge_sampler(fit)
  print(bridge)
}
} # }