Sampling risk

The probability of incorrectly rejecting or accepting a particular lot based on a sample. In acceptance sampling you make a decision to accept or reject an entire lot based on the results from inspecting a sample from that lot. Because your decision to reject or accept the lot is based only on a sample, not on data from the entire lot, you risk rejecting "good" lots and accepting "bad" lots.

For example, you receive a shipment of 10,000 microchips. You have two criteria: sample size = 200 and acceptable quality level (AQL) = 1.5%. If fewer than 8 of the 200 inspected microchips are defective, you will accept the entire shipment. If 8 or more are defective, you will reject the entire shipment.  

Like hypothesis testing, you can make two types of errors while accepting/rejecting a lot. Type I error is referred to as the producer's risk because the producer made a good lot yet it was rejected. Type II error is known as the consumer's risk because the consumer has received a shipment that contains more than an acceptable number of defectives and will result in more waste or rework than anticipated.

 

Good lot

Bad lot

Accept lot

Correct decision

Consumer's risk

Reject lot

Producer's risk

Correct decision

 

The producer's risk is represented by a and the consumer's risk is represented by b.

An operating characteristic curve (OC curve) quantifies these risks on a graph that allows you to choose the appropriate sampling plan for the risks you are willing to incur.