The Essence of Acceptance Sampling
Often, inspection is needed as one of the quality control technique in industry. While 100% inspection is the most straight forward and commonly used technique, it comes with a lot of undesirable impacts, such as high cost and time consuming that makes it impractical in most cases. Furthermore, it does not guarantee 100% compliance to the requirements. Therefore, it raises the needs for sampling plans that economically provide us with a reasonable amount of protection to ensure good quality.
Acceptance sampling uses statistical sampling that permits acceptance or rejection of a batch or lot of products based on a sample of units. It provides one rational means of verification that a lot conforms with the requirements of a predetermined standard. It is therefore regarded as an audit tool that stands between no inspection and 100% inspection.
A wide variety of acceptance sampling plans are available, that can be categorized by the type of data measured; attributes or variables. In acceptance sampling by attribute each item tested is classified as conforming or non-conforming. A sample is taken and if it contains too many non-conforming items the lot is rejected, otherwise it is accepted. On the other hand, acceptance sampling by variables is carried out by measuring a variable rather than classifying an item as conforming or non-confirming. Typically, the mean measurement from the samples will be compared with the acceptance criteria, for decision of lot acceptance or rejection. Acceptance sampling by variables is usually more complicated, but with the advantages of gaining more information and requires smaller sample size.
In acceptance sampling, Acceptance Quality Limit (AQL) is a quality level that is the worst tolerable process average when a continuing series of lots is submitted. It is largely used to as an index to the specific acceptance sampling plans in many standard procedures. Some of the well-known acceptance sampling standard procedures are documented and published in MIL-STD, ANSI/ASQ and ISO, for instance MIL-STD 105E, ANSI/ASQ Z1.4, and ISO 2859 for attributes; and MIL-STD 414, ANSI/ASQ Z1.9 and ISO 3951 for variables. Implementation of acceptance sampling in the industry is often following one of those standard procedures.
With acceptance sampling, two parties are usually involved: the producer and the consumer of the products. When specifying a sampling plan, each party wants to avoid costly mistakes in accepting and rejecting a lot. These are the producer’s risk (alpha risk) and consumer’s risk (beta risk) that is associated with each sampling plan. The Operating Characteristic (OC) curve of a particular sampling plan will show the risks of making a wrong decision as mentioned above. OC curve is a graph of the probability of accepting a lot against the various lot quality levels. It describes how well a sampling plan discriminates between good and bad lots. The diagram below shows an example of OC curve.
Nevertheless, it is important to understand that acceptance sampling alone does not improve quality. Product quality comes from proper product and process design, and process control activities. When such activities are effective, acceptance sampling is a redundant effort and an unnecessary cost.