Side-by-side comparison
| Dimension | Monetary unit sampling | Classical variables sampling |
|---|---|---|
| Sampling unit | Individual currency units (e.g., each €1 in the population) | Physical items (e.g., each invoice or balance, regardless of monetary value) |
| Selection bias | Built-in bias toward high-value items; every euro has equal selection probability | No built-in value bias; every physical item has equal probability regardless of size |
| Best population type | Right-skewed populations with few expected errors (e.g., trade receivables) | Populations with expected errors distributed across all value ranges, or where understatement is the risk |
| Evaluation model | Projects using tainting factors applied to the sampling interval | Estimates total population value or misstatement using statistical estimation (mean-per-unit, difference, ratio) |
| Understatement testing | Weak: low-value items are less likely to be selected, so understated items may be missed | Effective: selection probability is independent of recorded value |
| Zero-error assumption | Produces the smallest sample when zero errors are expected; sample size increases sharply with expected errors | Sample size is less sensitive to the expected number of errors |
Key Points
- MUS is weighted toward high-value items by design, making it efficient when you expect few misstatements in a right-skewed population.
- CVS treats every physical unit equally and works better when misstatements are expected across all value ranges.
- MUS struggles with understatement testing because smaller items have lower selection probability.
- Choosing the wrong method for the population shape inflates sample size or produces weak evidence for the conclusion.
When the distinction matters on an engagement
MUS dominates audit practice because most populations auditors test are right-skewed: a few large items carry most of the monetary value, and the expectation is that misstatements are rare. For those populations, MUS is efficient. But when the risk relates to understatement (completeness of liabilities, completeness of provisions), MUS is structurally weak. An understated liability of €500K that is recorded at €50K has only 10% of the selection probability it should have. ISA 530.A5 requires the auditor to consider the characteristics of the population when designing the sample.
CVS handles understatement testing without this structural disadvantage because it selects physical items regardless of their recorded value. The trade-off is that CVS requires the auditor to estimate population variability, which means the auditor needs to know (or estimate) the standard deviation of the population before calculating sample size. For populations where that variability is high or unknown, CVS sample sizes can become impractically large. The choice between MUS and CVS is a design decision that ISA 530.7 requires the auditor to document before drawing the sample.
Worked example: Berger Einzelhandel AG
Client: Austrian retail chain, FY2024, revenue €95M, IFRS reporter.
The engagement team needs to test trade receivables (€11.6M across 840 line items, right-skewed distribution) for overstatement, and trade payables (€7.8M across 1,200 relatively homogeneous line items) for understatement.
Trade receivables (overstatement risk): MUS applied
Population: €11.6M. Performance materiality: €340K. Expected misstatement: €10K (low error expectation based on prior year). Confidence: 95%.
Calculated sample size: 38 sampling units. The team applied systematic selection with a random start across the €11.6M population, producing a sampling interval of approximately €305K.
Documentation note: "MUS applied to trade receivables overstatement. Population: €11.6M, 840 items. Parameters: tolerable misstatement €340K, expected misstatement €10K, confidence 95%. Sampling interval: €305K. 38 sampling units selected with systematic selection. MUS chosen because population is right-skewed and error expectation is low. Ref: ISA 530.7, ISA 530.A5."
Result: zero misstatements found. Upper misstatement limit at 95% confidence: €305K (one sampling interval), below tolerable misstatement of €340K. Conclusion: sufficient evidence for the accuracy assertion.
Trade payables (understatement risk): CVS applied
Population: €7.8M, 1,200 items. The risk is understatement (completeness). MUS would not work here because understated or unrecorded items have reduced or zero selection probability. The team applied a difference estimation approach.
Estimated standard deviation of differences (based on prior year): €120 per item. Required precision: €150K. Calculated sample size: 52 physical items, selected using simple random sampling from the payables listing.
Documentation note: "CVS difference estimation applied to trade payables completeness. Population: €7.8M, 1,200 items. Standard deviation estimate: €120 per item (prior year basis). Required precision: €150K. 52 items selected by simple random selection. CVS chosen because the risk is understatement and MUS would under-sample low-value and unrecorded items. Ref: ISA 530.7, ISA 530.A6."
Result: average difference per item in sample was €18 (understatement). Projected total understatement: €21,600. Precision at 95% confidence: ±€145K. Projected misstatement plus precision fell within tolerable misstatement.
If the team had used MUS for the payables completeness test, the sample would have been biased toward the largest recorded payables — precisely the items least likely to be understated. The weakest point in the population would have received the least coverage.
What reviewers get wrong
Teams frequently apply MUS to completeness assertions without recognising the structural limitation. ISA 530.A5 requires the auditor to consider whether the sampling technique is appropriate for the assertion being tested. A MUS sample over a completeness assertion produces a conclusion that is statistically valid for the recorded amounts but says little about amounts that should have been recorded and were not.
When MUS finds errors, teams sometimes evaluate the results using simple projection (misstatement divided by item multiplied by population) instead of the tainting factor method. The tainting factor approach (misstatement as a percentage of the item's recorded value, applied to the sampling interval) is the correct MUS evaluation method. Using direct projection understates the upper misstatement limit and can produce a false pass.
Key standard references
- ISA 530.5–8: Requirements for sample design, including the choice between MUS and CVS.
- ISA 530.A5: Requires the auditor to consider population characteristics when selecting the sampling technique.
- ISA 530.A6: Addresses stratification and other design considerations relevant to CVS.
- ISA 530.14: Requires projection of misstatements found in the sample to the population.
Related terms
Related tools
Related reading
Frequently asked questions
Why is MUS unsuitable for testing completeness of liabilities?
MUS selects sampling units proportional to their recorded monetary value. An understated liability recorded at a low amount has a correspondingly low probability of selection. An unrecorded liability has zero probability. ISA 530.A5 requires the auditor to consider population characteristics when designing the sample, and for completeness assertions, classical variables sampling avoids this structural limitation because it selects physical items regardless of recorded value.
How does the evaluation method differ between MUS and CVS?
MUS uses tainting factors: each misstatement is expressed as a percentage of the item's recorded value, then projected across the sampling interval. CVS uses statistical estimation methods such as mean-per-unit, difference estimation, or ratio estimation to project the total population misstatement. Using direct projection instead of the tainting factor method on a MUS sample understates the upper misstatement limit and can produce a false pass.