Egger's Regression Test for Small-Study Effects (Bayesian)

Usage

egger(
  data,
  studyvar,
  n_ctrl,
  n_int,
  event_ctrl = NULL,
  event_int = NULL,
  mean_ctrl = NULL,
  mean_int = NULL,
  sd_ctrl = NULL,
  sd_int = NULL,
  likelihood = c("binomial", "gaussian", "poisson"),
  heterogeneity = c("multiplicative", "additive"),
  alpha_prior = NULL,
  beta_prior = NULL,
  kappa_prior = NULL,
  gamma_prior = NULL,
  d_prior = NULL,
  tau_prior = NULL,
  credible_level = 0.9,
  chains = 4,
  iter_warmup = 2000,
  iter_sampling = 4000,
  adapt_delta = 0.95,
  seed = 1234,
  custom_model = NULL,
  custom_data = NULL,
  ...
)

Arguments

data: A data frame with one row per study.
studyvar: Character. Column name of the study identifier.
n_ctrl, n_int: Character. Column names of control and intervention sample sizes.
event_ctrl, event_int: Character. Column names of event counts (binomial / Poisson likelihoods).
mean_ctrl, mean_int, sd_ctrl, sd_int: Character. Column names of arm means and SDs (Gaussian likelihood).
likelihood: Character. One of "binomial", "gaussian", "poisson".
heterogeneity: Character. "multiplicative" (default) or "additive".
alpha_prior: Prior on the intercept.
beta_prior: Prior on the slope (the Egger coefficient).
kappa_prior: Prior on the multiplicative heterogeneity coefficient.
gamma_prior: Prior on the dispersion parameter.
d_prior: Prior on the overdispersion parameter.
tau_prior: Prior on the between-study SD for additive heterogeneity.
credible_level: Numeric in (0, 1). Credible-interval level. Default 0.90.
chains, iter_warmup, iter_sampling, adapt_delta, seed: MCMC settings.
custom_model: Optional character scalar of Stan code overriding the generated program.
custom_data: Optional named list merged into the Stan data list.
...: Passed to cmdstanr::sample().

Value

An object of class c("bayesma_egger", "bayesma").