Glossary ยท approach
Acceptance Sampling
Statistical-OR practice of inspecting a random sample of n items from a lot and deciding to accept or reject the entire lot based on whether the number of defects is below or above an acceptance threshold c.
Acceptance Sampling PlansLot Acceptance SamplingSampling Inspection
Acceptance Sampling is the statistical-OR practice of inspecting a random sample of n items from an incoming or outgoing lot and deciding to accept or reject the entire lot based on whether the observed defect count is at most an acceptance number c. The plan (n, c) balances producer risk ฮฑ (probability of rejecting a truly good lot, typically 5%) against consumer risk ฮฒ (probability of accepting a truly bad lot, typically 10%). Single sampling (one draw of n items), double sampling (two-stage draw with an intermediate region), multiple sampling (3-7 stages) and sequential sampling (item-by-item decision per Wald 1947 SPRT) reduce the average sample number (ASN) at similar OC-curve protection. Foundational Dodge and Romig (1959) *Sampling Inspection Tables*; modern reference Schilling and Neubauer (2017) *Acceptance Sampling in Quality Control*; international standards ISO 2859 and ANSI/ASQ Z1.4 add switching rules (tightened โ normal โ reduced) for supplier-based calibration.
รrnek
A factory receiving 50,000-part lots sets AQL = 0.65% and LQL = 4%. Under a single-sampling plan with n=315 and c=4, the OC curve gives a 95% acceptance probability at p=0.65% and a 10% acceptance probability at p=4% โ the plan satisfies both producer and consumer risk targets, while the AOQL bounds the average outgoing quality.