This paper introduces a practical method for using subjective data to improve calibration interval estimates for test or measurement instruments. The procedure exploits the natural properties of conjugate prior distributions to reduce the most complicated aspects of typical Bayesian theory to a simple calculation. The approach is easily adaptable to existing interval calculation software, and it is even amenable to quick hand calculations. The basic Bayesian approach can apply the uninformed conjugate prior distribution to statistical calibration results data to form a posterior distribution. An arbitrary choice of a constructed data set can produce a wide variety of conjugate posterior distributions. Any one of these can serve as a prior distribution represented by the data set. This data simply combines with actual statistical data to form a practical Bayesian posterior distribution.
展开▼
