From: Stan Hilliard
Date: 08 Jan 2000
Time: 22:53:06
Greetings David,
How would this work for you?
1) NON-PATHOGEN PLAN: Use historical data to estimate how high the non-pathogen colonies will be before you can expect to have pathogens present. Use this value for the RQL of the mean count.
2) DETERMINE DISTRIBUTION: Calculate the within-lot standard deviation for each historical lot. Compare the SDs to the square root of the average count of the same lot. If they are close, the counts follow a poisson distribution. If the distribution follows the poisson, you could use the poisson option of program TP105 to develop a sampling plan.
If the data does not follow the poisson, use a sampling plan for the mean with known standard deviation. Since the mean is probably a function of the standard deviation, use the SD expected when the mean is halfway between AQL and RQL. The program TP414 will design this type of sampling plan. Use the option for the mean with a one-sided upper limit.
3) TO CHOOSE A SPECIFIC SAMPLING PLAN: Design a specific plan by developing a series of plans for non-pathogens having the same RQL mean. Do this by using alpha=beta=0.05; choose different AQL means and generate the sample size, decision limit, and oc curve for each plan. Compare these values to decide which plan to use for non-pathogens.
4) PATHOGEN PLAN: Use an attribute plan. With TP105, use the decision rule input option. With Ac=0, choose various values of sample size. The program will calculate AQL and RQL. Use these values to choose the specific plan.
Sincerely, Stan Hilliard