Statistical methods in auditing
Date
1996
Authors
Swinamer, Kathryn Joy
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
In performing an audit, an auditor typically employs statistical sampling methods. Information from the sample is used to form an estimate of the total error in an audit population. This estimate is often expressed as a one-sided or two-sided confidence bound. By computing a one-sided upper bound for total audit error, an auditor has a certain level of assurance that the total error does not exceed the upper confidence bound. Most research has centered on finding suitable upper bounds for the error in an audit. Several such methods will be reviewed in this report.
A suitable bound is one which is both reliable and efficient. For example, a 95% upper bound is reliable if, when used repeatedly, the bound exceeds the true audit error 95% of the time. Efficiency measures the size of the bound; the smaller
the bound is, the more efficient it is said to be. A method which yields reliable, efficient bounds is most suitable for use by auditors in order to avoid costly errors or overauditing.
The results of an extensive simulation study comparing various methods are preĀsented. Both real and simulated data are used for this purpose. In each case measures of reliability and efficiency are provided. The performance of the various methods depend on the distribution of the population. No one method was found to be superior.
The multinomial-Dirichlet method of Tsui, Matsamura and Tsui [50] demonstrated the best reliability for a variety of populations. Other Bayesian methods, such as the Bayesian normal bound [31] and the Cox and Snell bound [12] are reliable and more efficient than the multinomial-Dirichlet bound for particular populations.