Bayesian Analysis of Inspection Sampling Procedures Discussed by Deming.
Abstract
In Chapter 13 of Quality, Productivity, and Competitive Position, Deming discusses in detail inspection sampling relative to two different cost setups. Because of their practical importance, one of these models was examined in detail from a Bayesian point of view by Barlow and Zhang. A very general Bayesian treatment of these problems is provided by Lorenzen. A computer program is also provided by Lorenzen for the Bernoulli case as well as inspection sampling plans which reject the lot when the number of defective items is sufficiently large. However, as we shall see, simple rules of this form are not optimal for all of Deming's models. In particular, Deming's discussion of inspection rules when the finished assembly cannot be repaired by replacing defective units is of this type. Deming describes this model as value added to substrate. For example, if the unit is a bag of cement powder, then it cannot be recovered from the assembly after water has been added. The purpose of this paper is to analyze and contrast the two models considered by Deming from a Bayesian decision analysis point of view. We will analyze the models using influence diagrams. This is a relatively new technique for studying statistical problems. It will be useful for studying inspection sampling problems. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1986
- Accession Number
- ADA166189
Entities
People
- Richard E. Barlow
- Xiang Zhang
Organizations
- University of California, Berkeley