From: Stan Hilliard
Date: 27 May 1999
Time: 16:09:37
When developing acceptance sampling plans for non-normal variables, I have found the following methods to work well.
1) Use the non-normal distribution to calculate the value of the mean that corresponds to the fraction beyond specification at AQL and at RQL.
2) Then calculate the means at AQL and RQL, and develop a sampling plan for the mean. See:
www.samplingplans.com/programtp414.htm
3) The Central Limit Theorem will assure that it works.
If you don't know the distribution, plot within-lot data empirically on normal probability paper. You will get a curved line, but that doesn't matter. Read off the number of standard deviations between the AQL and RQL percentiles and the 50 percentile(median). Then adjust this for the difference between median and mean -- determined from your data. Use these AQL and RQL means, as above, to develop a sampling plan for the mean.
Sincerely, Stan Hilliard, shilliard@samplingplans.com