Audit Sampling: How Auditors Select and Test Samples

What Is Audit Sampling?

Audit sampling involves applying audit procedures to less than 100% of items within a population of audit relevance, so that all sampling units have a chance of selection, and the auditor can draw conclusions about the entire population.

Under ISA 530, when designing a sample, the auditor must consider the purpose of the audit procedure and the characteristics of the population from which the sample will be drawn.

Statistical vs Non-Statistical Sampling

Statistical sampling: Uses random selection and probability theory to evaluate results. This allows the auditor to measure sampling risk and project the likely error in the population with a quantifiable confidence level. Requires more rigour in sample selection and evaluation.

Non-statistical sampling: The auditor uses professional judgement to select items and evaluate results without using probability theory. Valid under ISA 530, but the auditor cannot quantify sampling risk statistically.

Sample Selection Methods

Random selection: Every item in the population has an equal chance of selection. Typically requires a random number generator. Eliminates selection bias.

Systematic selection: A starting point is chosen randomly, then every nth item is selected (e.g. every 50th transaction). Efficient for large populations.

Monetary unit sampling (MUS): Each monetary unit (e.g. each £1) has an equal chance of selection, so larger-value items are more likely to be selected. Particularly effective for testing the overstatement of assets or income.

Haphazard selection: Items selected without following a structured technique but avoiding any conscious bias. Less rigorous — only acceptable for non-statistical sampling.

Block selection: Selecting a block of consecutive items (e.g. all invoices in March). Generally not representative of the full population — rarely used alone.

Determining Sample Size

Sample size is influenced by: the acceptable level of sampling risk (lower risk = larger sample), tolerable error or misstatement (lower tolerable error = larger sample), expected error in the population (higher expected error = larger sample), and the size of the population (has a smaller effect than commonly assumed).

For tests of controls, tolerable deviation rates and expected deviation rates drive sample size. For substantive tests, tolerable misstatement and expected misstatement are the key drivers.

Evaluating Sample Results

When errors are found in a sample, the auditor must project the error to the whole population (for monetary sampling) or calculate the deviation rate (for tests of controls).

If the projected error exceeds tolerable misstatement, or the deviation rate exceeds the tolerable deviation rate, the auditor must take further action — additional procedures, adjustments, or a qualification of the audit opinion.

Sampling vs 100% Testing

100% testing is appropriate when the population is small, automated controls make full testing efficient, or the risk is very high. For most high-volume transaction populations, sampling is more practical and equally effective when properly designed.