When Rajesh & Co. Pvt. Ltd. has 12,000 purchase invoices in a year, no auditor can realistically verify every single one. That's where audit sampling comes in — it lets the auditor test a representative subset of items and draw a conclusion about the entire population. SA 530 is the standard that governs how this sampling must be done properly.
Audit sampling means applying audit procedures to less than 100% of items in a population, where every sampling unit has a chance of being selected, so the auditor can form a conclusion about the whole population. This is a 4-mark concept frequently tested — examiners love asking you to distinguish between sampling and non-sampling approaches.
There are two broad types: Statistical sampling uses random selection + probability theory to evaluate results (so you can mathematically quantify sampling risk), while Non-statistical sampling relies on the auditor's judgement — both are acceptable under SA 530, but the auditor must still use judgement sensibly. The key risks to remember: Sampling risk is the risk that the auditor's conclusion from the sample differs from the conclusion if the entire population were tested. Non-sampling risk arises from human error — wrong procedure, misinterpreting evidence — and has nothing to do with sample size.
For sample design, three things matter: purpose of the procedure, characteristics of the population, and sample size. A larger sample reduces sampling risk but increases cost — the auditor must balance both. When evaluating results, if a deviation or misstatement is found, the auditor first checks whether it's an anomaly (a one-off error that cannot recur) — if it is, it can be excluded from projecting error to the population, but the anomaly itself must still be addressed. If the projected misstatement exceeds tolerable misstatement, the auditor must take further action — extend the sample, perform alternative procedures, or modify the audit opinion. Remember: tolerable misstatement is set below materiality to allow for the possibility that multiple errors exist.