Assessing the level of effectiveness of random sampling techniques in financial audits

Abstract

In financial audits, to reach a reasonable level of assurance, the auditor will have to express their opinion on the behavior of a population of transactions. However, most of the time, he/she will be able to analyze only a portion of this population using sampling techniques. We analyzed the effectiveness of sampling techniques in financial audit work, considering the statistical techniques mentioned in ISA 530. Using the Monte Carlo method, series containing the accounted values and the actual values were simulated for a total of 1,000 transactions. From the generated series, statistical samples were extracted using systematic sampling, simple random sampling, and stratified random sampling techniques. Considering the extrapolations made from the distortions found in each sample, it was identified that the stratified random sampling technique, using the groups formed from the quartiles of the population, was the technique with the highest level of effectiveness, meaning 60% of total estimated distortions within a 95% confidence interval (79% for overvaluations and 77% for undervaluations), with 100% of the samples are representative from population.