From: Stan Hilliard
Date: 30 Oct 1998
Time: 19:55:34
Remote Name: 160.94.26.37
Hi Stan, These AQL plans are not necessarily statistically valid! As I see it, to be valid an acceptance sampling plan has to be capable of discriminating between a lot that attains a defined acceptable quality level and a lot that attains a defined rejectable quality level. Not only that, but to be statistically valid, the plan must also make this discrimination with probabilities that are known by its designer/user and judged by him/her to be satisfactory for that application.
I think that many users of Mil-Standard AQL sampling plans select an AQL and believe that this meets some "valid standard" of sampling just because it comes from a Mil-Standard. Wrong!
This same argument applies to ANSI/ASQC Z1.4 and Z1.9 -- both AQL plans.
Sincerely, Stan Hilliard, hilli004@tc.umn.edu