Determining optimum sample sizes with respect to cost for multiple acceptance quality characteristics.

摘要

There has been little and limited research done to determine the optimum sample size n for highway construction acceptance plans containing pay adjustment provisions. To optimize the sample size of Acceptance Quality Characteristics (AQCs) with respect to the cost to the agency, an understanding is required of costs of testing and of the cost consequences associated with making a wrong acceptance decision. This study developed an optimization model to determine the sample size. Actual state agency's data were used not only to check the validity of model's assumptions but also draw the conclusions. Among the key conclusions were: (1) The optimum sample size is generally small (n=3) for Percent Within Limits acceptance plans; (2) The degree of correlation among AQCs has a negligible effect on the optimum sample size; and (3) The highway agency tends to underpay contractors through use of its pay adjustment provisions. Some cautions are presented for state highway agencies planning to use small sample sizes. Agencies are encouraged to perform their own optimization calculations. By doing so, they will gain a better understanding of their acceptance plan systems and associated costs, and have greater confidence in applying economic decision analysis principles to minimize expected costs and optimize statistical acceptance risks.