Six-sigma projects and other factors have contributed to the increased popularity of gauge r&r studies. The DMAIC (define, measure, analyze, improve, control) problem solving framework commonly used in six-sigma projects starts by defining a problem followed by the measurement step. Measurement systems analysis is based on the notion that you need to determine the capability of your measurement systems. To put it another way, you need to measure your measurement system.
Gauge R & R is a form of measurement systems analysis, designed to determine the precision of a measurement system. R&R studies are not, in general, designed to directly determine the accuracy of a measurement system. Accurate measurements are centered close to the true value (of course, we usually don’t know the true value), while precise measurements are ones that are close to each other. This means, for example, that you can be accurate but not precise – an average near the true value makes for accuracy, but lots of variability makes for a lack of precision. In general, gauge R&R studies focus attention on the capability of the measurement system to get consistent results.
The first ‘R’ represents repeatability, also known as pure measurement error. If we measure a parameter on the same widget several times with the same measurement system and get about the same result each time, we can say that we have good repeatability. Measuring something over and over and getting about the same result each time does not make those results accurate, but it’s almost always good news. If we determine that measurements are not especially accurate but they are repeatable, then we can try to find the source or sources of the inaccuracy and work them out of the measurement process. For instance, we may have a calibration issue, and if we can fix that issue suddenly we have accuracy and precision combined.
In the R&R world, the second ‘R’ represents reproducibility, which is a measure of (for example) whether two or more operators can measure the same parts and get the same average results. Organizations such as AIAG and ASTM consider equipment and time as reproducibility factors. You might collect data across multiple program revisions, or run an experiment with different settling times or sampling rates, and consider these to be reproducibility factors. Repeatability and reproducibility, combined, form the total measurement system error (or gauge error). This may sound a bit awkward, but reproducibility is not the ability of operators (or equipment) to get the same level of repeatability. The repeatability results for different operators could be quite different but if the average measurements are comparable then the reproducibility is good.
What constitutes good repeatability or good reproducibility? The answer is, of course, “It depends.” Still, there are several, commonly used gauge metrics aimed at helping you decide whether your r&r results are acceptable. We’ll comment on that in an upcoming data blog.