Overview - ISO 28596:2022 (Two-stage sampling plans)
ISO 28596:2022 specifies two-stage (double) sampling plans by attributes for deciding acceptance or rejection of a target population based on the proportion of nonconforming items. It covers lot inspection, auditing of internal control systems (ICS), tests of details for financial auditing, misstatements in accounting records, and acceptance testing in product/process audits. The standard is intended where inspection cost is high or where a possible second sample and related delay are acceptable.
Key topics and technical requirements
- Two-stage decision procedure
- Stage 1: draw sample n1 - accept if 0 nonconforming; reject if ≥ Re1; proceed to stage 2 if between 1 and Re1 − 1.
- Stage 2: draw an additional sample n2 - accept if combined defects ≤ Ac2, otherwise reject.
- Indexed sampling plans
- Plans indexed by tolerance proportion p0, nominal confidence level γ (0.7, 0.8, 0.9, 0.95 or 0.99), and Trust levels (low, mid, high) that reflect prior information and reduce sample sizes with higher Trust.
- Statistical safeguards
- Designed to protect against both erroneous rejection of tolerable targets (Type II) and erroneous acceptance of intolerable targets (Type I).
- Annexes A–K provide supporting theory: confidence intervals, operating characteristics (OC), conditional risks, average sample numbers, OC matching, search algorithms and software guidance.
- Optimization objective
- Sample sizes n1 and n2 are chosen to minimize expected sample size under the specified confidence and Trust settings, enabling rapid, economical decisions when population quality is very low or very high.
Practical applications
- Lot inspection and acceptance sampling in manufacturing and supply chains.
- Financial auditing: auditing ICS and substantive tests of details (e.g., accounts receivable, raw materials, booking records).
- Quality auditing, functional testing and acceptance testing in product and process audits where attribute data (conforming/nonconforming) is collected.
- Situations where prior inspection results or other prior information can be used to set Trust levels and reduce inspection effort.
Who should use ISO 28596:2022
- Auditors and audit managers (internal/external) needing statistically defensible sampling for ICS and tests of details.
- Quality engineers, QA/QC managers and compliance officers implementing acceptance sampling by attributes.
- Statisticians and analysts designing economical inspection plans that incorporate prior information.
- Regulators and procurement teams responsible for lot acceptance criteria.
Related standards
- ISO 2859-2 (sampling plans indexed by limiting quality)
- ISO 3534-1 and ISO 3534-2 (statistics vocabulary and symbols)
- Relevant auditing guidance such as ISA 530 (statistical sampling in audits)
ISO 28596:2022 delivers practical, statistically grounded two-stage sampling plans to reduce inspection cost and decision time while maintaining controlled risk - especially useful where prior information can be exploited to optimize sample sizes.