Behavior of Statistical Estimators in Multilocation Audit Sampling.

Key Points

• To evaluate the performance of various statistical estimators in two-stage sampling for dispersed accounting populations.
• Conducted a simulation study involving 64 accounting populations across multiple locations.
• Examined global Stringer bound with cell sampling of dollar units and mean-per-unit bound with stratified sampling of line items.
• Compared the accuracy of different sampling strategies in achieving nominal confidence levels.
• The global Stringer bound achieved nominal confidence levels in all scenarios tested.
• The mean-per-unit bound with stratified sampling only achieved nominal confidence levels for half of the populations.
• The global Stringer bound was often stricter but could lead to rejecting satisfactory populations.