Probabilistic (Monte Carlo) sensitivity analysis — probabilistic_bias

Draws bias parameters from prior distributions and returns the distribution of bias-adjusted estimates. Cross-references episensr (episensr::probsens) for the canonical multi-bias version with separate selection-bias and misclassification-bias models; use episensr directly when you need those.

Usage

probabilistic_bias_analysis(
  estimate,
  se,
  n_simulations = 10000L,
  bias_parms = NULL,
  seed = 42L
)

Arguments

estimate: Observed estimate.
se: Standard error.
n_simulations: Number of MC draws. Default 10000.
bias_parms: Named list with (mean, sd) pairs for rr_ud, rr_eu, prevalence. Defaults supplied.
seed: RNG seed. Default 42.

Value

Named list with bias-adjusted distribution summaries.

Examples

set.seed(1)
str(probabilistic_bias_analysis(0.5, 0.15, n_simulations = 2000L),
    max.level = 1)
#> List of 8
#>  $ original_estimate: num 0.5
#>  $ median_adjusted  : num 0.662
#>  $ mean_adjusted    : num 0.697
#>  $ ci_2.5           : num 0.317
#>  $ ci_97.5          : num 1.33
#>  $ pct_null_included: num 0
#>  $ pct_same_sign    : num 100
#>  $ n_simulations    : int 2000