Mastering Audit Sampling – From Principles to Practice

A. Defining the Audit Objective 

The objective of the auditor when using audit sampling is to obtain a reasonable basis for drawing conclusions about the entire population from which the sample is selected. Accordingly, the auditor needs to clearly define whether the sampling is intended for testing control effectiveness, verifying transaction accuracy, or validating account balances—ensuring that the sampling approach aligns with these audit objectives.

B. Determining Sample Size

The auditor must determine a sample size sufficient to reduce sampling risk to an acceptably low level. This involves applying professional judgement based on the nature of the population and the assessed risks.  

Stratified sampling, which involves dividing the population into subgroups with similar characteristics, can enhance sampling efficiency, particularly when high-value or high-risk items require focused attention.

For example, if the population is divided into three groups: Group A (50 items), Group B (100 items), and Group C (200 items), the auditor may select 10, 15, and 20 items, respectively, from the respective groups based on factors like risk, value, or significance. This method enables focused testing of critical areas while ensuring broad coverage of the population. 

C. Selecting the Sampling Method

Materiality guides how detailed and extensive the audit checks should be.

1. Simple Random Sampling:
   Also known as the Method of Chance Selection, it involves selecting items in such a way that each item in the population has an equal chance of being chosen. It is primarily used for general-purpose testing.
2. Systematic Sampling:
   Unlike Random Sampling, here audit samples are chosen after setting up a fixed sample interval, such as every nth item after a random start would be chosen for sampling. This method is used primarily with respect to large and homogeneous populations. It provides a structured approach to sampling, but care must be taken to ensure that the interval does not coincide with any hidden periodicity in the data. 
3. Monetary Unit Sampling (MUS):
   MUS treats each monetary unit as a separate sampling unit, thereby giving higher-value items a greater chance of selection. This method is particularly useful when the auditor wants the sample to reflect the monetary significance of the population.
4. Non-statistical Sampling:
   This approach relies on the auditor’s judgement rather than statistical methods to determine sample size and selection. It is often used when the assessed risk of material misstatement is low or when a more flexible, judgement-based selection is appropriate.
   

   Common techniques under this method include: 

   • Haphazard sampling, where items are chosen without any structured technique but with care to avoid bias,  
   • Block sampling, where a continuous group of items (e.g., transactions from a particular day or period) is selected.

   While the non-statistical sampling method provides practical flexibility, especially in lower-risk areas, it does not offer the same level of measurable assurance as statistical sampling. 

D. Performing Audit Procedures

Once the sample is selected, the auditor performs appropriate audit procedures to obtain evidence. These may include: 

• Inspection – examining supporting documents or records related to the sampled items. 
• External Confirmation – verifying balances or transactions directly with third parties. 
• Recalculation – checking the mathematical accuracy of amounts.
• Re-performance – independently executing procedures or controls to assess whether they were applied correctly. 

If a selected item cannot be tested, for example due to missing documentation, the auditor should either select a replacement item or consider it a deviation or misstatement, depending on the circumstances. 

E. Evaluate Results & Project Misstatements

The final step would be to assess the nature and cause of any errors or deviations. If no anomalies are found, conclusions can be extrapolated to the population. As per SA 530, any identified misstatements should be projected to the entire population unless clearly proven to be an anomaly.