What is expected misstatement?
On about half the sampling files we review, the expected misstatement input is set to zero even though PY testing found errors. That single default produces an undersized sample, and the rest of the methodology (however sound) cannot compensate for it. The concept exists precisely to prevent this: expected misstatement forces the auditor to feed real evidence about the population's error rate into the sample size calculation before any items are selected.
Expected misstatement is the auditor's best estimate of the monetary error likely to exist in a population before sampling begins. ISA 530 .A11–A12 requires this estimate as a direct input to sample size determination. Together with tolerable misstatement, it controls how large the sample needs to be.
The estimate is based on evidence, not assumption. ISA 530 .A12 identifies four sources: misstatement levels found in previous audits, changes in the entity's procedures since the prior period, evidence from the evaluation of internal controls, and results of other substantive procedures already performed in the current engagement.
Expected misstatement works alongside tolerable misstatement (the maximum error the auditor will accept in the population). The relationship between the two drives sample size: when expected misstatement is low relative to tolerable misstatement, smaller samples suffice. When expected misstatement approaches tolerable misstatement (above 40–50%), sample sizes increase sharply because the auditor needs more precision to distinguish between an acceptable and unacceptable level of error.
Key Points
- ISA 530 .A12 requires the estimate to rest on evidence (PY results, control evaluations, entity changes, current-period procedures), not on a default assumption of zero.
- Higher expected misstatement relative to tolerable misstatement means larger samples. Setting it at zero when prior years found errors produces an undersized sample.
- Expected misstatement estimates what the population contains; tolerable misstatement sets the maximum the auditor will accept. Both feed into sample size, but they answer different questions.
- If projected misstatement from the sample exceeds the expected amount, the sample may lack sufficient precision and additional procedures may be needed.
Why it matters in practice
The most common error is setting expected misstatement at zero when PY testing found errors. If last year's testing identified misstatements in a population, the default assumption should be that similar errors exist this year unless there is specific evidence that the entity has remediated the underlying cause. Setting expected misstatement at zero here produces a sample that is too small to detect the error rate that actually exists. This is the input that generates the most RNs on sampling files.
A related mistake is the instinct to just roll it forward from last year's WPs without reconsidering whether the estimate still fits. If the entity changed its revenue recognition process or implemented a new ERP module, last year's error rate may no longer be relevant. The estimate needs to reflect current conditions, not just PY convenience.
A second frequent mistake is confusing expected misstatement with tolerable misstatement. They are both inputs to sample size formulas, but they measure different things. Expected misstatement is a prediction about reality. Tolerable misstatement is a decision about what level of error the auditor will accept. Mixing them up (or using one where the other belongs) produces mathematically valid but logically wrong sample sizes. Internally, some teams call this kind of error a PIOOMA number: Pulled It Out Of Mid Air. The figure looks precise but has no evidential basis.
The impact on sample sizes is non-linear. Moving expected misstatement from 10% to 20% of tolerable misstatement produces a modest increase in sample size. Moving it from 40% to 60% of tolerable misstatement can double or triple the required sample. This is why the estimate matters most when it is already high. Small changes in the assumption produce large changes in the work effort.
Key standard references
- ISA 530 .A11 sets out the factors affecting sample size, including the expected rate or amount of misstatement in the population.
- ISA 530 .A12 identifies the sources of evidence for estimating expected misstatement (PY results, entity changes, control evaluations, current-period substantive procedures).
- ISA 530 .A22 requires the auditor to consider whether the sample provides a reasonable basis for conclusions when projected misstatement exceeds expected misstatement.
- ISA 530.7 requires the sample to be sufficient to reduce sampling risk to an acceptably low level.
Related terms
Related tools
Related tools
Related reading
Frequently asked questions
How do you estimate expected misstatement?
ISA 530.A12 identifies four sources: misstatement levels from previous audits, changes in the entity's procedures, evidence from internal control evaluation, and results of other substantive procedures already performed in the current period. The estimate should be documented with the specific evidence supporting it.
What happens when actual misstatement exceeds the expected amount?
If projected misstatement from the sample exceeds the expected misstatement used to size it, ISA 530.A22 requires the auditor to consider whether the sample was sufficient. The sample was calibrated for a lower error rate, so finding more error means the sample may lack the precision to reliably conclude the population is not materially misstated. Additional procedures may be needed.
What is the difference between expected misstatement and tolerable misstatement?
Expected misstatement is the auditor's best estimate of what the population actually contains, based on evidence. Tolerable misstatement is the maximum error the auditor will accept while still concluding the balance is not materially misstated. Both are inputs to sample size, but they answer different questions. When expected misstatement is a high proportion of tolerable misstatement (above 40-50%), sample sizes increase sharply.