The Value of Subjective Information: An Empirical Assessment

Abstract

Metrology engineers want technically correct answers. Managers want to make decisions that trade off cost against product value. Calibration personnel want their work to count. Calibration intervals drive measurement reliability, the calibration budget, and the value of every calibration. We affect the value of our entire calibration program when we decide how often to calibrate. Unfortunately, we don’t always have enough historical calibration results data to predict the best calibration interval with a high degree of confidence. Although Bayesian statistical theory provides a method for including independent data sources to supplement calibration results data, limited empirical evidence exists to assess how well Bayesian statistics predicts measurement reliability. The literature has no example that measures how well subjective information estimates measurement reliability