Abstract

To remain useful and relevant, measurement systems must be evaluated, maintained, and improved. Just as a process should be stable and capable, a measurement system should be characterized by reproducibility and repeatability. The key to assessing a measurement system is measurement sensitivity. There are various techniques for this assessment, depending on the circumstances in which a given measurement operates. First, when repeated measures are possible, measurement sensitivity may be evaluated by performing a measurement capability study using Cp and Cpk. Second, when there is a question about applying the measure across operators and calipers, a two-factor analysis of means can analyze statistical differences from situation to situation. Third, when repeated measurements are not possible, as in destructive testing, measurement sensitivity can be assessed using techniques such as low and high parameter settings and pooled standard deviations. Fourth, when it is necessary to select a measurement system from candidates having different scales, measurement comparison studies are appropriate.