Calibrate the KL normalized power prior on binomial likelihoods — npp_kl_calibrate_design

The criterion of calibrate_npp_kl() - the same two hypothetical target results, reference distributions, weights, bounds and optimiser - with the posterior of the power parameter computed from the binomial marginal likelihood of each result, npp_kl_binomial_log_marginals(), instead of the normal one.

Usage

npp_kl_calibrate_design_binomial(
  source_counts,
  source,
  target_data,
  theta_0,
  settings,
  n_lattice = 1000L,
  n_nodes = 80L
)

Arguments

source_counts: List of the four source counts.
source: The source data, with treatment_effect_estimate.
target_data: Target study data for the scenario, a binary design.
theta_0: Boundary of the null hypothesis space.
settings: Output of npp_kl_settings().
n_lattice: Number of lattice points.
n_nodes: Quadrature nodes of the Beta prior.

Value

A list with the fields of calibrate_npp_kl()'s result.