Compute true marginal quantiles of the predictive distribution

Uses Monte Carlo integration over the study-level location hierarchy to compute the true population-level quantiles of the predictive distribution.

Usage

compute_true_marginal_quantile(
  dist_type,
  mu0,
  tau,
  phi,
  kappa = 1,
  probs = c(0.5, 0.95),
  n_mc = 5000L
)

Arguments

dist_type

Distribution type: one of "lognormal", "gamma", "weibull", "burr12", "gengamma".

mu0

True population mean (location hyperparameter).

tau

True between-study standard deviation.

phi

True distribution-specific shape/scale parameter.

kappa

True third distribution parameter (Burr XII k or GG Q). Ignored for 2-parameter distributions.

probs

Numeric vector of probabilities for which quantiles are computed. Default c(0.5, 0.95).

n_mc

Number of Monte Carlo draws. Default 5000.

Value

A named numeric vector of quantiles (names are e.g. "50%", "95%").