Sampling risk vs. nonsampling risk in the auditor's logic process;

              3
Sampling Risk vs. Nonsampling Risk in the Auditor's Logic Process
William L. Felix, Jr.
University of Washington
Most of the larger auditing practice units in this country can be described as either making considerable use of statistical methods in current practice or engaging in research and development that in the near future will encourage the use of statistical methods whenever they are appropriate. While the level and nature of the use of these methods varies considerably across these auditing firms, frequently a pervasive deficiency appears to exist in the documented in­corporation
of the results of such procedures in the audit logic process. In the following paper an examination of this problem is presented, followed by an analysis of the role of sampling evidence in the auditor's logic process. The paper concludes with some recommendations for action.
An Interpretation Problem
The use of statistical sampling methods as a structure for applying auditing procedures results in a confidence interval or an accept/reject decision depending on whether the auditor is using an estimation or a testing approach. (Although only an estimation approach is discussed in this paper, the comments apply with equal force to both approaches.) As an example of the estimation approach, an auditor might specify a 95% confidence interval of $6,934,000 plus or minus $141,700 in sending out positive confirmations of customer accounts receivable and use a mean-per-unit or an auxiliary estimation method to construct the re­sulting
confidence interval. In documenting the results of such an application in the auditor's working papers, a conclusion similar to the following is often found:
"Based on the above tests, I am 95% confident that the accounts receivable balance of $7,037,000 at 6/30/X7 is fairly stated."
While there are a number of issues in this conclusion that could be argued, the major concern of this paper is that the conclusion implies that the risk of a non-representative sample (the risk of sampling error) is the only audit risk of concern in the confirmation of a sample of accounts receivable. This implication is never correct. Of equal or possibly even greater significance in evaluating the results of the auditing procedure are the auditor's perceptions of (1) how well assistants executed the procedures and computations, (2) the ex post appropriate-
47