Statistical sampling;

              Statistical Sampling
by LEONARD M. BROOKE Principal, New Orleans Office
Presented before the Society of Louisiana Certified Public Accountants, New Orleans Chapter—October 1965
FOR SEVERAL YEARS our firm has been quite active in promoting the use of statistical sampling, and in several instances I have applied the sampling plan we have developed. I have also worked some with statistical sampling procedures used in selecting revolving credit ac­counts
that are eligible for deferred income treatment as instalment accounts for income tax purposes. The subject is very broad and some aspects of it are quite technical.
Most accounting audiences represent a fairly wide range of under­standing
about statistical sampling. A few within the group may have devoted attention to the subject and would like to discuss it in depth, but the majority probably have not given it much attention and are interested in it only in a general way. On this assumption I shall direct my effort to the general level.
This discussion will cover three areas of statistical sampling:
1) Its general nature
2) The procedures for establishing its use in auditing
3) How it can be used by clients
THE GENERAL NATURE OF STATISTICAL SAMPLING
A logical starting place to explain the nature of statistical sampling is to compare it with something we are all familiar with—that is, the traditional or conventional audit testing. The principal feature these two activities have in common is the purpose. The purpose of testing or sampling is to obtain information, so that the auditor is able to form an intelligent opinion about the whole of something by examining only a part of it. In the traditional case the auditor is concerned about selecting certain transactions to represent a period of time or certain detail items to represent the balance in a control account. This purpose is the same one we have in mind in statistical sampling.
With either of these methods some degree of uncertainty is in­evitable.
Any time anything less than the whole is examined, some degree of uncertainty obviously remains about the part left unexamined. When dealing with something so important that no uncertainty what­ever
can be tolerated, sampling is simply not employed—the whole
195