Acceptance sampling has traditionally been a partner of statistical process control and control charts in the area of statistical quality control. Products are shipped around in batches or lots, and the idea behind acceptance sampling is that a batch can be declared to be satisfactory or unsatisfactory on the basis of the number of defective items found within a random sample of items from the batch. Thus acceptance sampling provides a general check on the “quality” of the items within a batch. Acceptance sampling can be performed by the producer of the products before they are shipped out.
In addition, sampling may be performed by customers who receive the products in order to check that they are receiving high-quality materials. Many companies are accustomed to performing acceptance sampling on their incoming raw materials before taking delivery of them. Notice that there is an important procedural distinction between statistical process control and acceptance sampling. The control charts used in statistical process control provide a real-time monitoring of a production process with the objective of avoiding the production of low-quality materials.
Changes in the process can be identified almost immediately so that corrective actions can be taken. In contrast, acceptance sampling is performed after production has been completed. It does not allow any monitoring of the production process, and whole batches of products can be wasted if they are found to be unsatisfactory.
Consider a batch of N items, each one of which can be classified as being either satisfactory or defective. When N is very large, a 100% inspection scheme of the batch in which each item is examined is generally too expensive and time consuming. Therefore, a random sample of n of the items is chosen, and the number of defective items x in the sample is found. The acceptance sampling procedure is based upon a rule whereby the batch is declared to be satisfactory