Accuracy & precision: ever second-guess a measurement system? Perform a GR&R.

Author:Maze-Emery, Elizabeth

Measurement system tools vary across industries and applications, yet the term measurement system can apply to every circumstance for which there is a measurement check and a decision made based on that measurement.

Some industries use sophisticated optical tools while other applications use simple gages. Regardless of the measurement tool used in your measuring system, it is important to ensure the tool provides sufficient accuracy and precision as required.

Accuracy refers to the centering of repeated tests while precision refers to the closeness of repeated tests. To illustrate this concept, refer to the target below.

Consider the innermost center as the actual value. Measurement system tools provide a measure close to the actual value.

Figure 1 illustrates accuracy by showing repeated measurements centered on the actual value. Figure 2 illustrates precision as the repeated measurements are tightly packed together, although not necessarily centered.



When a new measurement tool is introduced, it is a good idea to first verify that the tool will measure as expected. A good "rule of thumb" in selecting an accurate measurement tool is to evaluate the tolerance specification of the gage to be no greater than 1/10th of the measured value.

For instance, if the target thickness is 20mm, then choose a tool accurate to 2mm or less. Even with this selection, it is a good idea to verify the tool tolerance is acceptable. How to do this?

What if we do not know the tolerance of the measurement tool or critical gage? What if we have a measurement tool in place but do not believe the measurement system is working like it should? In our Quality Toolbox, there is a method to check the measurement system. We can perform a gage repeatability and reproducibility exercise, also known as a Gage R&R (GR&R).

This first part of the GR&R exercise is hands-on and requires two to four persons to use the measurement tools to generate a data set. Choose up to 10 samples or parts varying across the measurement range. Be sure to include maximum and minimum expected values within the range of the measurement tool. Label the samples to keep track of which is which.

Have each person, or appraiser, use the calibrated measurement tool to measure each sample and record the data values. Mix up the samples and have each sample measured two or three times by the same person collecting additional data values. One appraiser doing three repeats...

To continue reading