Appendix E: Statistical Sampling Plans

Abstract

Appendix E Statistical Sampling Plans 306 There are many different statistical sampling plans, but we shall discuss just two, whose characteristics lend themselves to the partial audits taken from dis­ covery. The first, fixed-size attribute sampling, can accommodate developing systems where system deviation may be relatively high and easily measured. The other is stop-or-go sampling, which can be used in stable systems with low deviation rates and hard-to-find errors. E.1 Fixed-Size Attribute Sampling Plan Sampling is efficient and practical when there is a trail of documentary evi­ dence, or when the auditor can take the time to observe a process in action. Apostolou & Alleman (1991a) propose a seven-step strategy for a fixed-size attribute sampling plan: 1) Determine the objectives of the test 2) Define the attribute and deviation conditions 3) Define the population 4) Determine the method of sample selection method 5) Determine the sample size 6) Perform the procedures of the sampling plan 7) Evaluate the sample and express conclusions E.1.1 Determine the Objectives When we speak of testing operations, invariably we are speaking of testing its controls. The objective of the tests of controls is to assess the effectiveness of control procedures. One control might be a required sign-off signature or revi­ sion date on a document. Others might be a receipt inspection, feed rate on a machine, or completeness of a purchase order. The basic job of the analyst is to determine if the error rate of a system of controls is less than the acceptable level. As no control is perfect, there will always be output variation in any sys­ tem. Therefore, given that a system of controls is not perfect, how good is it? Specifically, how does the measured system deviation rate (SDR) compare to the acceptable deviation rate (ADR)? If the measured deviation rate is lower than the acceptable rate, then the control procedures are deemed effective. Otherwise, they are deemed ineffective.