Statistical Auditing: Attribute Sampling Methods with a Case Study
Statistical Sampling
12:00 pm - 1:00 pm

Statistical sampling is a generally accepted audit technique widely used in tax, internal audit, financial statement auditing, and many other regulatory and assurance functions. Even with this broad use and deep acceptance of sampling, there can be a wide range of practices and a lack of standardization for audit sampling.

We will observationally explore the impact of sample sizes, in attribute samples (i.e., error vs non error), ranging from the unconventional small to the unnecessarily large determining what statistical attributes the samples require to provide assurance on audit and tax populations.  In addition, we will explore combining samples for a “project” error rate; this technique follows stratifcation principles and is often used for variable (i.e., monetary amount) sampling but can also be applied to attribute sampling.    A few of the topics that will be covered follow:

  • What do sample sizes look like as population sizes approach the infinitely large?
  • When does a population become large enough to statistically be considered infinitely large?
  • How do population error rates contribute to the required sample sizes?
  • Standard audit sample sizes for process risk level risk rating.
  • Stratification, combining samples for an “overall” error rate and precision.

In discussing sample size impact, this will be an observational review (i.e., simulations) and does not seek a theoretical approach, but actual data driven analysis, and with more populations than is routinely seen in applicable research.  We should expect to roughly but clearly observe patterns and qualities that authenticate the theories of sampling while providing practical examples that validate industry best practices.