ABSTRACT
Manufacturers often conduct various tests on samples from individual lots to infer the expected level of reliability of a product in customer applications. In addition, the stressing of consecutive groups of samples from production may be employed to monitor the reliability of a running process. Such results may be plotted on statistical control charts to verify that the manufacturing process is “under control.” However, as such studies can be costly and time consuming-and often destructive to units-it is important that ef—cient sampling designs be selected to provide the necessary information while using the minimum quantities of product. This chapter covers the implementation of various types of sampling plans for attribute data, the associated risks, the operating characteristic (OC) curves, and the choice of sample sizes. We discuss the applications of various discrete distributions, namely the binomial, geometric, negative binomial, Poisson, and hypergeometric. The calculation of con—dence limits for distribution parameters is treated. We introduce Fisher’s exact test for comparing proportions when sample sizes are limited. Finally, we look brieµy at the application of statistical process control charting for reliability. We illustrate how spreadsheets can be used effectively to perform calculations and thereby provide solutions to many problems related to discrete distributions.
