Abstract

This abstract is an edited version of the author's original.

This paper describes the difficulty using Test Uncertainty Ratio (TUR) alone to assure the quality of calibrations. It investigates the relationship between TUR, confidence interval, and guardbands, and it describes how they can be used cooperatively to manage the calibration process. The risk of false test decisions is a function of TUR, confidence level, and a guardband factor (K), which is a multiplier of the specification limit (SL) to produce the test limit (TL). The author uses examines and compares the false accept and false reject risks associated with a number of strategies.

Using an economic model is probably the optimum way of picking guardband factors to minimize the consequences of false test decisions; however, this model is not likely to be widely accepted because of the difficulty in coming to agreement on the relative cost of false test decisions.