Discussant's response to "Unresolved issues in classical audit sample evaluations";

              Discussant's Response to "Unresolved Issues in Classical Audit Sample Evaluations"
Abraham D. Akresh
Laventhol & Horwath
I am happy to be here at Kansas to discuss the paper by Nichols, Srivastava and Ward. Our firm uses an assertion based audit approach; we use classical variables sampling when we expect to find many errors or when we perform accounting applications. In general, I have little question that the authors understand the mathematics of classical variables sampling and the various approaches. While their mathematics are generally right, I am not sure they have considered all of the practical aspects. I will discuss some of the practical problems that are based on the many telephone calls received in our National Accounting and Auditing Department from our practice offices.
Discussion Points
1. The paper places equal emphasis on the risks of incorrect rejection and acceptance when evaluating sample results. For accounting applications, these risks might be equally important. For auditing applications, however, auditors are much more concerned about the risk of incorrect acceptance. Incorrect acceptance leads to audit failures. Incorrect rejection leads to audit inefficiencies. In today's environment, with insurance difficult to obtain, incorrect acceptance and audit failures are "unacceptable." Incorrect rejection is a cost of doing business that is directly or indirectly passed on to clients. In the short run, we may even realize revenue when there is incorrect rejection.
Audit efficiency can be controlled by means other than sample size; for example, proper planning and supervision, analysis of error risks and determination of materiality levels, selection of nonsampling procedures when justified, and use of modern technology to reduce clerical time. Thus, auditors do not rely solely on risk of incorrect rejection levels to control audit efficiency.
We recognize that the rejection method is conservative and may yield higher than necessary sample sizes. We also recognize that teaching people the more efficient method will be expensive, more so than the sampling cost to be saved.
As a result, auditors determine the risk of incorrect acceptance for sampling applications, then select a somewhat higher risk of incorrect rejection. This selection is somewhat arbitrary and is based primarily on the difficulty of extending procedures if unacceptable results are obtained. For example, an auditor might select a 10 percent risk of
120