A staff auditor pulls 60 sales invoices, finds 1 missing approval, computes a 1.7% deviation rate, and signs off as "control reliable." The senior asks two questions the staff cannot answer: was 60 the right size for a 5% tolerable rate, and did the upper deviation limit fall below tolerable after the sampling-risk adjustment? The exam tests both, and AU-C 530 tells you which inputs drive each one.
HIGH-FREQUENCY: AU-C 530 (and PCAOB AS 2315 for issuers) accepts both statistical and nonstatistical sampling. The standard does not prefer one. What it requires of both: define the population so it fits the test objective, give every sampling unit a chance of selection, perform procedures appropriate to the assertion, project misstatements or deviations, and evaluate qualitative as well as quantitative results.
The mechanical difference is measurability. Statistical sampling uses random selection plus probability theory, so the auditor can compute a numeric upper bound on sampling risk.
TRAP: A common wrong-answer choice frames nonstatistical sampling as "less rigorous" or "not permitted." Both are false. Nonstatistical is fully acceptable; the auditor simply cannot attach a confidence percentage.
Common mistakes
- Treating statistical sampling as the standard's preferred method. AU-C 530 makes both methods acceptable. A wrong-answer choice that says "the auditor should have used statistical sampling because it is more reliable" is the trap. Statistical adds a quantified upper bound; it does not make the conclusion more correct.
- Applying MUS when understatement is the concern. MUS selects on recorded dollars. An understated item has artificially few dollars and a depressed selection probability. Unrecorded liabilities, omitted invoices, and cutoff errors that reduce balances are largely invisible to MUS. "Auditor suspects accounts payable are understated" → classical variables, not MUS.
- Sampling from the recorded population to test completeness. Completeness asks whether all events that should have been recorded were. Sample from the sales journal and you cannot detect omitted events: they are not in the frame. Sample from the underlying event population (shipping documents, receiving reports) and trace forward.
Bottom line
- AU-C 530 governs all audit sampling. Statistical and nonstatistical sampling are both acceptable when professional judgment is properly applied.
- Attribute sampling for tests of controls measures the deviation rate against the tolerable rate. Variables sampling for substantive dollar tests projects misstatement against tolerable misstatement.
- MUS / PPS when overstatement is the concern and balances are positive. Classical variables (mean-per-unit, difference, ratio) when understatements, zero balances, or negative balances are in play.
- Substantive sample size: risk of incorrect acceptance down raises it, tolerable misstatement down raises it, expected misstatement up raises it, population variability up raises it. Population size has minimal effect once large.
Exam shortcut
When a stem describes "many zero-balance or negative items" or "the auditor suspects understatement," the answer is classical variables, not MUS. When a stem says "all positive balances and overstatement is the concern," MUS is correct. These are the cleanest tells. For attribute decisions, the rule is sample-rate-plus-allowance vs. tolerable rate. If the upper deviation limit is below tolerable, rely; if at or above, do not.
The full lesson (about 4,486 words, 30 min read) adds 8 worked examples, all 5 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.