Simulated audit-sampling populations for a MUS precision-bound comparison study

These three datasets support a Monte Carlo simulation study comparing precision-bound estimators used in Monetary Unit Sampling (MUS) for audit sampling: the Horvitz-Thompson-type bound (HH), the Poisson-Stringer and Binomial-Stringer bounds, and the Moment bound. They are synthetic (simulated) populations, not real financial records. They were constructed to reproduce the statistical characteristics of a real accounting population — a small number of large book-value items alongside a large majority of much smaller items, and a book-value-to-error ("tainting") structure calibrated against a real audited population's dispersion — while keeping full control over ground truth, since the true total error and its distribution must be known exactly to evaluate an estimator's coverage and precision, which is not possible with real audit data. The three files form a pipeline, each built from the one before it: 1. first_book_value_population.xlsx — the base book-value population. 2. book_value_populations.xlsx — three high-value (HV) concentration scenarios derived from the base population. 3. simulated_error_populations.xlsx — simulated errors injected into each scenario, across a grid of error-frequency, error/book-value correlation, and error-rate targets. These are the populations the simulation itself draws audit samples from.