By Steven Waches
The effective use of data to drive decision making requires adequate measurement systems. When interpreting data or the results of data analysis, we assume that data or results represent the process. However, excessive measurement error may result in inappropriate conclusions. Thus, it is critical to properly assess whether measurement systems are adequate for their intended use prior to their use. Only capable measurement systems should be utilized to support quantitative methods such as statistical process control, inspection activities, process capability assessment, hypothesis testing, data modeling, etc.
Important measurement system characteristics include discrimination, accuracy, precision (repeatability and reproducibility), linearity, and stability. Techniques exist to assess measurement systems for each of these important characteristics. Skipping such assessments can lead to the use of measurement systems that are not capable of monitoring process variation or, in extreme cases, even of distinguishing between conforming and non-conforming product. In short, validating measurement systems is an important pre-requisite to relying on data.
Measurement systems must be properly assessed to minimize risk and comply with customer and regulatory requirements. While most companies perform some aspects of measurement systems assessments (MSAs), such as gage repeatability and reproducibility studies, we often observe inadequate assessments of measurement systems.