From: Stan Hilliard
Date: 01 Jul 2003
Time: 17:13:25
Guru said: "I want to know what exactly is "discrimination"? How would you quantify it in statistical terms."
Discrimination in acceptance sampling refers to the ability of a sampling plan to detect a difference between a lot that is RQL% defective and another one that is AQL% defective. In other words, discrimination is the ability of a sampling plan to detect an off grade lot, should it occur. Better, it is the size of the difference. The smaller the difference (RQL-AQL), the more discriminatory. RQL means the same as LTPD.
The oc curve reflect the ability of a sampling plan to discriminate between RQL and AQL. the steeper it is, the better it can discriminate. See: samplingplans.com/usingoccurves.htm