From: Stan Hilliard
Date: 22 Jan 2001
Time: 22:38:38
Mil-Std-105D and Mil-Std-105E are almost identical. Mil-Std-105E has some additional oc curves that Mil-Std-105D does not. They apply only if you use switching procedures. They do not describe the protection for individual lots, but rather to the long-term average of many lots. Other than that, there are just some semantic changes -- like substituting the word "nonconforming" for "defective".
Both Mil-Std-105D and Mil-Std-105E are based on the AQL that you choose. To see the problems inherent in that approach, go to www.samplingplans.com/aqlprimer.htm.
I recommend that a more effective way to design a sampling plan is to choose two points on the oc curve: the producer's point and the consumers' point. See www.samplingplans.com/usingoccurves.htm
If you are thinking of using Mil-Std switching rules to adjust the sampling plan for the historical quality level, I suggest using sequential sampling plans instead. Mil-Std switching rules change the oc curve - and thus the error risks - when you switch. Sequential plans don't. Sequential sampling plans change the sample size and decision limits but they also adhere to the same oc curve regardless.
Additionally is the matter of assumptions. Mil-Std Switching rules are based on the assumption that the quality of the lot that you are testing currently is like recent previous lots. Sequential plans don't make that assumption. With sequential plans, the accept/reject decision depends only on the quality of the current lot.
Stan Hilliard