Run the asymptotic calibration suite for the elicitation algorithm — run_asymptotic_performance

Replaces the human respondent with a simulated robot agent of known risk parameter and measures mean absolute error (MAE) between the true and inferred parameter across a grid of true values. Used in vignette 5 to validate the elicitation algorithm.

Usage

run_asymptotic_performance_suite(
  param_values = seq(-1.5, 3.5, by = 0.25),
  trials_per_param = 15,
  model = "CRRA",
  val_min = 0,
  val_max = 120
)

Arguments

param_values: Numeric vector of true gamma values to test.
trials_per_param: Number of independent robot sessions per value.
model: Utility model; currently only "CRRA" is supported.
val_min, val_max: Outcome range passed to elicit_risk_preferences().

Value

A data frame with columns True_Param, Inferred_Mean, Abs_Error.