Selected Upper Error Limit Methods Applied to an Accounting Population

摘要

The purpose of this thesis was to study the performance of the Cox and Snell Upper Error Limit Model. Twelve Cox and Snell bounds were constructed by changing the prior error probabilities incorporated in the model. The performance of the twelve bounds was compared to the performance of the Stringer bound. The study constructed twenty-four study populations from an actual accounting population, sampled from these study populations, and compared the error estimates produced by the Cox and Snell model with the true dollar error in the study populations. The objective was to find the best upper error limit estimator from the Cox and Snell bounds tested. The bounds were examined in terms of robustness, mean relative tightness, effect of nominal confidence level, and statistical power. In order to analyze the results of the simulation, the study populations were divided into four categories by error amount intensity.