# Acceptance Sampling - Purdue University

Acceptance Sampling and the evils thereof. 1 Rationale: (I dont buy the argument, but.) In situations where Producers and Consumers are willing to put up with shipments of goods which contain defects, the issue then becomes

how to keep Lots (shipments) with an excessive number of defects from being accepted. In order to keep Appraisal costs down, sampling schemes are devised to sample items from the Lot and then to accept or reject the Lot based upon the number of defective items found in the sample. 2 Notation N= number of items in the Lot.

n= number of items in the random sample of items taken from the Lot. D= number (unknown) of items in the Lot which are defective. r= number of defective items in the random sample of size n. c= acceptance number; if r is less than or equal to c then the Lot is deemed to be of Acceptable Quality, otherwise the Lot is rejected. 3

Acceptance Sampling Plan An Acceptance Sampling Plan is defined by: N= Lot size n=sample size c=acceptance number 4 Acceptance Probability and Risks

Assuming: p D N Pa ( p ) P(accept _ lot _ with _ p proportion _ defective) We can evaluate a sampling plan with its corresponding Operating Characteristic Curve, a.k.a. OC curve. 5

Calculating probabilities for the OC curve The probabilities used in calculating values for the OC curve are calculated using the Hypergeometric Distribution. Given that p D N then the probability that the number of defectives in the sample of size n is equal to r is given by N D D

n r r N n 6 Acceptable Quality Level-AQL

If the Lot has a proportion of defectives lower than the AQL, i.e. p AQL then the Lot is deemed to be of Acceptable Quality Level by Producer and Consumer. 7 OC curve and Producer and Consumer Risk

The Probability that we accept a Lot is Pa p P r c for a given proportion of defectives p D N which is calculated using the Hypergeometric Distribution as given. 8

Risks and the AQL If the proportion defective is less than or equal to the AQL, then the Producer runs the risk of having the Lot rejected which is: 1 Pa p , p AQL If the proportion of defectives is greater than the AQL, then the Consumer runs the risk of accepting a Lot of unacceptable quality which is: Pa p , p AQL

9 Suppose AQL=5% If the AQL=5%, we might think that the sample proportion which does not exceed 5% would indicate that the Lot is acceptable. Let us suppose our Lot size=1,000. If we plot Pa P( Accept _ Lot )vs. p proportion _ defective we can see how well various plans work.

10 How well do small sample sizes work? 11 They can work if If the Lots are either really good (very few defects), or really bad (many defectives).

The samples are representative, i.e. close to the true proportion of defectives. Representative is not the same as random. Random samples are chosen in such a way that every possible subset is equally likely. Only large samples are likely to be representative. 12 Producer and Consumer Risks

If n=20, the producers risk at the AQL is around 30%, i.e. a Lot of acceptable quality may be rejected. At just over 5% defective, the Consumers risk is about 70%. The plans work better as n increases. 13 When does random sampling start to work

well? As n increases, we would expect the Producers and Consumers risk to both decrease. How much do we have to sample to reach an acceptable level of risk for both of them. 14 Compare n=20, 100 and 240

15 More sampling clearly does work If we compare n=20 to n=100, there is a substantial improvement. If we compare n=100 to n=240, there is still an improvement, but the amount of improvement is not as great, i.e. there is a diminishing return.

16 Comparing large sample plans 17 Largest sampling plans comparison The largest sampling plan, n=500 has the best performance in terms of risk, but it is not

much better than n=100 or n=240. The only way to get no risk is to sample everything, or.. Invest in Prevention, then only use sampling to monitor the improved processes. 18

## Recently Viewed Presentations

• Effects of Osmosis on Life. Osmosis- diffusion of water through a selectively permeable membrane. Water is so small and there is so much of it the cell can't control it's movement through the cell membrane.
• Space Science. Moon Phases. Sun and Earth Rotations. Space Exploration
• BeagleBone Black - truly open-source hardware and software. Open source gives people the freedom to control their technology while sharing knowledge and encouraging commerce through the open exchange of designs and code
• Common Characteristics of Descriptive Studies Build on previous information Show relationships between variables Representative samples required Structured research plans Require substantial resources Conclusive findings Major Types of Descriptive Studies Descriptive Studies Consumer Perception And Behavior Studies Image Product Usage ...
• Unit 27 - Network Operating SystemsAssignment 1 - Task 1.1. You are required to evaluate types of NOS and NOS services of a selection of Network Operating Systems such as standalone, infrastructure based and cluster based systems.
• The intent is to use the Health Services Specialty Track as a means to capture the HSO mission-support areas. Examples: ... Minor / Ward Health Form ... (Medical Plan) for incidents and routine operations in collaboration with the Mission/Activity Safety...