Use Stan model to estimate bias and estimands

Use Stan model to estimate bias and estimands

get_meta_estimates(
  data,
  sampling_variable = "meta",
  which_estimand = "hidden_size",
  benchmark = list(sample = "pps", estimator = "ht"),
  stan_handler = get_meta_stan3,
  hidden_prior = NULL,
  rel_bias_prior = list(mean = 1, se = 10),
  control_params = list(iter = 8000, chains = 8, thin = 10, seed = 872312, cores = 1)
)

Arguments

data

pass-through meta population or meta sample data frame

sampling_variable

name of variable storing meta analysis sampling information

which_estimand

name of study level estimand for meta analysis

benchmark

named list of length 2 giving benchmark sampling-estimator pair (only accepts one value across studies for now)

stan_handler

function that takes stan_data as input and produces compilable stan model object

hidden_prior

list of two hyperpriors, on means and standard errors of each included sampling-estimator pairs. Names of list objects should be "mean" and "se". If one number provided for a hyperprior it gets expanded to all sampling-estimator pairs

rel_bias_prior

list of two hyperpriors, on means and standard errors of relative bias. Names of list objects should be "mean" and "se".

control_params

list of additional parameters to pass to Stan fit function. These can include number of iterations, chains, thinning, seed and number of cores to use