Multi Attribute and Multiple Criteria Approaches for Determining Bayesian Acceptance Plans in Quality Control and Auditing

摘要

The development of a bicriterion model for acceptance sampling in quality control is discussed. Such a model allows explicit consideration of conflicting goals that are not considered in existing acceptance sampling schemes. A quality measure and a cost measure are used as the conflicting criteria. Two optimization procedures for solving the model are explored, the first of which involves the measurement of a decisionmaker's utility function and subsequent optimization via an implicit enumeration algorithm. The second method employs an interactive procedure. Also presented is a laboratory study that compares the interactive process to utility function management methods in a quality control setting. Claims that interactive procedures are easier to use and achieve a more satisfactory solution are supported. The claim that interactive procedures provide more insight into the relationships of the criteria is refuted. An interactive procedure under uncertainty is developed and illustrated for the bicriterion case. Efficiency under uncertainty is defined, and conditions that guarantee efficiency are proven.