Measurement Systems Analysis - Revisited Basic Tools Taken from the Summer 1993 Newsletter
Abstract: While everyone would agree that knowing the variation of a measurement system is useful, how the analysis of that variation should be performed is sometimes the subject of controversy. This paper emphasizes that before jumping into the statistical analysis of any process, we must first understand the purpose (aim) of the process and the sources of variation which could affect it. Therefore it is important to know that activities requiring measurement systems fall into two general categories. The first category deals with the analysis of the process distribution, and requires variable measurements. The accepted guideline in this case is that the discrimination be less than 10% of the tolerance. The measurement system’s variation as measured by the repeatability and reproducibility (Gauge R&R) is a common industrial requirement that falls into this category. The second category involves control of the process variability and requires the ability to identify that a change has occurred. In this case the measurement can be variable, semi-variable or attribute. Since the objective of a measurement system analysis is to quantify the common cause variation, initial studies should verify the stability of the system. With this framework in mind, this paper provides a collection of methods to study the stability and variability of a measurement system.
Keywords: Measurement system analysis - Metrology - Gage repeatability and reproducibility (Gage R&R - Gauge R&R) - PDSA (plan-do-study-act) cycle - Individuals and moving range (X - mR) chart - Linear regression - sources of variation