|
Abstract:
|
Performance audit standards require an auditor to obtain sufficient evidence to draw robust conclusions about an audit population. Because it is usually not possible for auditors to collect, examine and assess information for all units in an audit population, a subset (sample) of the population is often selected for review. When an appropriate probability sampling method is implemented, results can be generalized to the audit population and the sufficiency of the evidence can be quantified in terms of sampling error. Probability samples of varying complexity can be used to design samples that are both statistically and operationally efficient. These designs include simple, stratified, cluster, and multi-stage random probability samples. This paper discusses common applications of simple and complex sample designs and presents examples of samples the Government Accountability Office has conducted to generate sufficient performance audit estimates. These will include examples of probability samples designed to estimate multiple population parameters of different types (attributes, ratios, means, and totals) and demonstrate how the sample design contributed to the efficiency of the au
|
