Audit Sampling Techniques

Why This Matters

Sampling is the backbone of every audit—you can't test every transaction, so you need methods that let you draw reliable conclusions from a subset of data. The CPA Exam and professional auditing standards expect you to understand not just what each technique is, but when and why you'd choose one over another. You're being tested on your ability to match sampling methods to audit objectives: tests of controls vs. substantive testing, detecting overstatement vs. understatement, efficiency vs. precision.

The key insight is that sampling techniques fall into two broad categories—statistical (where probability theory gives you measurable confidence) and non-statistical (where professional judgment drives selection). Within statistical sampling, you'll encounter methods designed for different purposes: some test for the presence of errors, others estimate dollar amounts, and still others optimize for efficiency. Don't just memorize definitions—know what audit objective each technique serves and when it's the right tool for the job.

---

Statistical vs. Non-Statistical Foundations

Before diving into specific techniques, understand the fundamental distinction: statistical sampling uses probability theory to quantify risk, while non-statistical sampling relies on auditor judgment without measurable confidence levels.

Statistical Sampling

• Probability-based selection—every item has a known, non-zero chance of selection, enabling mathematically defensible conclusions
• Quantifiable sampling risk—allows auditors to calculate confidence levels and precision intervals that satisfy professional standards
• Enhanced objectivity—removes selection bias and provides documentation that withstands regulatory scrutiny

Non-Statistical Sampling

• Judgment-driven selection—auditor expertise guides which items to examine, without random selection mechanisms
• No measurable confidence level—results cannot be extrapolated to the population with statistical precision
• Practical applications—useful when populations are small, items are unique, or statistical methods would be cost-prohibitive

Judgmental Sampling

• Experience-based approach—auditor selects items based on knowledge of the client, industry patterns, and risk factors
• Targeted flexibility—allows focus on unusual transactions, high-value items, or areas of known concern
• Bias risk—must be documented carefully since it lacks the defensibility of random selection methods

---

Probability-Based Selection Methods

These techniques determine how you select items from a population. The goal is ensuring your sample represents the whole population without systematic bias. Each method trades off simplicity, precision, and practical constraints.

Random Sampling

• Equal selection probability—every item in the population has the same chance of being chosen, eliminating selection bias
• Implementation tools—use random number generators, random number tables, or audit software to ensure true randomness
• Foundation for other methods—serves as the baseline technique that other statistical approaches build upon

Systematic Sampling

• Fixed interval selection—choose every nth item after a random starting point (e.g., every 50th invoice starting from invoice #23)
• Efficiency advantage—faster to implement than pure random sampling, especially with physical documents
• Pattern risk—watch for cyclical patterns in the population that could align with your interval and bias results

Stratified Sampling

• Population subdivision—divide items into strata based on shared characteristics (dollar amount, location, transaction type)
• Targeted precision—sample more heavily from high-risk or high-value strata while still covering the entire population
• Reduced variability—produces more precise estimates than simple random sampling when strata are internally homogeneous

Cluster Sampling

• Group-based selection—divide population into clusters (branches, time periods, product lines) and randomly select entire clusters to test
• Cost efficiency—reduces travel and access costs when populations are geographically dispersed
• Homogeneity requirement—works best when clusters are internally diverse but similar to each other; otherwise, variability increases

---

Tests of Controls: Attribute-Based Techniques

When testing whether controls are operating effectively, you're asking a yes/no question: Did the control work or not? These techniques measure the rate of deviations rather than dollar amounts.

Attribute Sampling

• Binary measurement—tests for the presence or absence of a characteristic (signature present? approval documented? segregation maintained?)
• Deviation rate focus—results expressed as a percentage of the population that deviates from the prescribed control
• Tests of controls application—the go-to method for compliance testing under auditing standards

Discovery Sampling

• Critical error detection—designed to find at least one instance of a specific deviation, particularly useful for fraud indicators
• High-risk targeting—applied when even a single occurrence would be material (embezzlement, regulatory violation, intentional override)
• Sample size implications—requires larger samples to achieve reasonable assurance that at least one error will surface if errors exist

---

Substantive Testing: Variable-Based Techniques

When your objective is estimating dollar amounts—account balances, total misstatement, or transaction values—you need techniques that measure magnitude, not just presence.

Variable Sampling

• Magnitude measurement—quantifies the amount of a characteristic rather than whether it exists
• Account balance estimation—used to project total misstatement or verify recorded amounts in substantive procedures
• Greater precision—provides more detailed information than attribute sampling, enabling dollar-value conclusions

Monetary Unit Sampling (MUS)

• Dollar-weighted selection—treats each dollar as a sampling unit, so larger transactions have proportionally higher selection probability
• Overstatement bias—particularly effective for detecting overstatement errors since inflated balances contain more "dollars" to select
• Combined efficiency—blends attribute and variable sampling concepts, making it popular for accounts receivable and inventory testing

---

Efficiency-Focused Techniques

These methods optimize the sampling process itself, reducing sample sizes or combining objectives to save time and resources without sacrificing audit quality.

Stop-or-Go Sampling

• Conditional continuation—begin with a small sample and expand only if initial results don't provide sufficient assurance
• Resource efficiency—avoids testing the full planned sample when early results are favorable
• Predetermined criteria—requires establishing clear stopping rules before testing begins to maintain objectivity

Sequential Sampling

• Staged evaluation—assess results after each item or small batch, making accept/reject/continue decisions throughout the process
• Adaptive sample size—total items tested depends on results obtained, potentially reducing effort when evidence is conclusive early
• Time-sensitive applications—valuable when audit deadlines are tight or access to records is limited

Dual-Purpose Testing

• Combined objectives—simultaneously tests controls and performs substantive procedures on the same sample
• Efficiency gain—reduces total items tested by addressing two audit objectives with one selection
• Design complexity—requires careful planning to ensure sample size and selection method satisfy both objectives; failure of control doesn't automatically mean substantive failure

---

Risk-Driven Approaches

Modern auditing emphasizes risk assessment as the driver of audit effort. These techniques explicitly incorporate risk into the sampling strategy.

Risk-Based Sampling

• Risk allocation—directs more testing toward accounts, assertions, or locations with higher assessed risk of material misstatement
• Scalable intensity—sample sizes and selection methods vary based on inherent risk, control risk, and detection risk assessments
• Standards alignment—reflects the risk-based approach required by AS 2110 and ISA 315, making it essential exam knowledge

---

Quick Reference Table

---

Self-Check Questions

1. Which two sampling techniques both divide populations into groups but differ in whether you sample from groups or sample entire groups? What determines which approach is more appropriate?

2. An auditor wants to test whether purchase orders over $\$10,000$ are properly approved. Which sampling technique measures the rate of control deviations, and how does it differ from a technique designed to find at least one instance of fraud?

3. Compare Monetary Unit Sampling and traditional Variable Sampling: which is better suited for detecting overstatement in accounts receivable, and why does the selection mechanism create this advantage?

4. If an audit team wants to test controls and account balances simultaneously to save time, which technique should they use? What planning consideration is critical to making this approach valid?

5. An FRQ describes a client with 50 branch locations spread across three countries. The auditor has limited travel budget but needs to test cash handling controls. Which sampling method would be most efficient, and what risk must the auditor consider when using this approach?