City Research Online

Abstract

This study examined the performance of six monetary-unit sampling methods in substantive auditing under various population conditions. The investigation involved both a theoretical point estimator analysis and an empirical study of upper bound estimates of the total error amount using the Stringer, Cell and Moment bounds with two methods of error assignment, three nominal confidence levels and three sample sizes.

For a range of thirty audit populations simulated from real accounting populations of debtors from commercial entities in the Public Sector in Ireland, it was found that the differential effects of simple random, systematic, cell and sieve sampling were independent of the bound used, the error assignment method and the nominal confidence level. The reliability and tightness of the bounds were found to be similar for all selection methods but differences in the precision of the bounds did exist. In terms of the precision of the estimates, systematic and cell sampling favoured populations with large line items and sieve sampling favoured populations with large line items and low error rates.

Lahiri sampling, a selection method not used in auditing previously and proposed in this study as a practical alternative to simple random sampling was not found to be significantly different from simple random sampling with respect to any of the performance measures.

Stabilised sieve sampling, a new monetary-unit sampling method developed in this study as an alternative to simple random sampling and sieve sampling, was found to be reliable for the range of audit populations on which it was tested. In populations with small line items, stabilised sieve sampling tended to have a tightness similar to that of simple random sampling and sieve sampling for any given error rate, taint size, sample size and bound and in populations with large line items, stabilised sieve sampling was more conservative than simple random sampling and sieve sampling but the differences were not significant in any case. It was more precise than simple random sampling and its precision was similar to that of sieve sampling in most cases. As stabilised sieve sampling overcomes the primary disadvantage of sieve sampling by returning a constant sample size of monetary units, it was concluded that it may be a useful alternative to simple random sampling and sieve sampling in real substantive auditing environments.

Publication Type:	Thesis (Doctoral)
Subjects:	H Social Sciences > HF Commerce > HF5601 Accounting
Departments:	Bayes Business School Bayes Business School > Bayes Business School Doctoral Theses Doctoral Theses

Export

Downloads