Model to estimate over-coverage on population

The `oc_model()` function is able to get an estimate of over-coverage in a population based on register data and log-linear models.

Usage

oc_model(
  model_formula,
  freq_table,
  censored,
  nsample = 2000,
  null.move.prob = 1,
  n.burnin = 1000,
  thin = 1,
  prob_level = 0.95,
  ...
)

Arguments

model_formula: Model formula to be used in the log-linear model.
freq_table: Frequency table with all observational data.
censored: Indexes of all individuals who are not observed in any of the registers.
nsample: Number of posterior draws in the MCMC estimation process.
null.move.prob: Parameter to control model selection algorithm. See `conting::bict()` for more information.
n.burnin: Number of burnin samples to be discarded from the MCMC algorithm.
thin: Thinning parameter in the MCMC algorithm.
prob_level: Probability level to be used when calculating the credible interval for the overcoverage level.
...: Additional arguments to be passed to `conting::bict()` function, which is slightly changed here to function properly in the latest versions of R.

Value

A list with overcoverage estimates and summaries of the number of false positives based on estimates of the model. The object model itself is also returned.

Examples

if (FALSE) model_oc <- oc_model(
 qty ~ bin1 * list1 + cat1 * list2,
 freq_table,
 cens_ind)