Audit sampling applies procedures to selected items from a defined population so their results can support a conclusion about that population. The application of sampling techniques requires a suitable selection design, appropriate procedures, and evaluation of uncertainty arising from examining fewer than all items.
These principles apply to nonissuer audits under the auditing standards for nonissuers and issuer audits under Public Company Accounting Oversight Board (PCAOB) standards. Neither regime requires control sampling in every financial statement audit; understanding controls and testing their operating effectiveness serve different purposes.
Three examination approaches produce different conclusions:
Automated tools can generate random selections, build cumulative monetary totals, identify duplicates, and isolate unusual transactions for investigation. Automation improves processing capacity, but the auditor still determines whether the selected records support the planned audit objective.
- A duplicate-payment search identifies potential repeated disbursements, which require investigation because identical amounts can represent valid separate purchases.
- An audit data analytic procedure can flag unusual period-end entries or transactions involving unfamiliar counterparties that warrant additional evidence.
Common mistakes
- Convenient records. Selecting 100 readily available payments can bias a sample when missing records are associated with failed controls.
- Wrong inference boundary. Testing every invoice above $20,000 does not establish the characteristics of smaller invoices without evidence addressing that remainder.
- Interval periodicity. Selecting every tenth transaction can systematically capture the same transaction type when the register repeats a ten-item pattern.
Bottom line
- A population inference requires appropriate sampling; specific-item testing supports conclusions about the selected items.
- The objective determines the population, sampling unit, period, and exception definition.
- Attribute sampling evaluates rates; monetary estimation evaluates amounts.
- Statistical sampling requires both random selection and probability-based evaluation of sampling risk.
Exam shortcut
Start with the requested conclusion. Circle the control characteristic or monetary assertion, then identify a source population that can reveal the relevant error. Separate selection mechanics from evaluation. A random-number generator answers how items were chosen; it does not establish how sampling risk was measured. In an evaluation exhibit, locate the upper limit and its matching tolerance before focusing on the observed rate or raw dollar errors.
The full lesson (about 5,778 words, 39 min read) adds 2 worked examples, all 6 common mistakes, a self-check, free in the app.
Learning objectives
- III.C1
Browse all free CPA AUD lessons or jump into free CPA AUD practice questions.