A list-sequential sampling scheme with applications in financial auditing

External auditors have the responsibility of estimating the level of error in accounts presented to them by their clients, and establishing whether or not this exceeds materiality. Usually the large volume of accounts necessitates the use of sample information to estimate the error amount, which is obtained by randomly choosing a subset of the line items for auditing. Since the account amounts may vary considerably, it is desirable to select them with probability proportional to the book value size, and while numerous such procedures exist, most have implementation problems which prevent their widespread application to practical situations: some select items with replacement, others become too complex when the sample size is large, and yet others return a variable sample size. This paper presents a selection method that mitigates these problems. It returns a fixed-size sample of distinct line items, and is easy to implement no matter how large the sample. The results of a series of simulation experiments with a variety of audit conditions indicate that the new method provides reliable bound estimates of the total error amount, which are more precise than methods currently in use.