Discussant's response to an examination of the status of probability sampling in the courts;

              Discussant's Response to
An Examination of the Status of
Probability Sampling in the Courts
Kenneth P. Johnson
Coopers & Lybrand
It may be clear to some that the courts are increasingly dealing with issues of probability, or statistical, sampling and how it should be used. It may also be clear to some that auditors who continue to use judgment sampling in the per­formance
of their audits should be prepared to defend their logic for doing so. However, the discussion by Professors Randall and Frishkoff will not support these conclusions. Further, as the authors indicate, it is certainly not clear how the courts will appraise the results of statistical sampling as part of the evidence used to reach audit conclusions.
On reviewing this paper, it occurs to me that there may be distinctions and observations that the authors have not made, which may assist in anticipating what attitude the courts may adopt in ruling on sampling questions that arise from audits of financial statements.
Considerations by the Courts
Only some of the cases cited turn on a court decision concerning the appro­priateness
of a sampling plan or particular statistical technique used. For example, in Johnson vs. White, the court reviewed the sampling plan and determined that it was appropriately devised and applied based on a review of the techniques employed. Two of the cited cases do not seem to deal directly with probability sampling. In Commissioner vs. Indiana Broadcasting Corp., the issue seemed to turn on whether a Poisson distribution was either applicable or appropriately applied. This entailed consideration of whether proper judgments had been made in the planning stage, including whether the defined population had the appropriate attributes for the application of this statistical concept. Another cited case, Super Foods, appears to have been decided on a legal issue rather than on the issue of statistical techniques employed.
Putting aside these cases, I believe there is another characteristic implicit in the remaining cases that should be discussed. The paper notes that "references to sampling usually arose where the technique was being used to gather evidence rather than where it was used prior to the litigation, such as in an auditor's test of transactions." I believe it is significant that the cases cited typically seem to deal with specific attributes of a more or less well-defined population. However, the court's acceptance of probability sampling plans in these circumstances, in my opinion, does not indicate what its attitude would be in the much more com­plicated
auditing environment. The expertise required of an auditor in devising
103