Do That Again
©2020, George J. Irwin. All rights reserved.

Before you can improve a process, you’ve got to know what you’re up against. Part of this knowing is understanding whether it’s being measured correctly.

There are two aspects to measurement, which in Black Belt Land are usually called Repeatability and Reproducibility.

Repeatabilty, in the form of a question, is “Can I get the same result each time I measure the process?”

Reproducibility, also in the form of a question, is “Can multiple people get the same result measuring the same process?”

Seems easy enough, and sometimes it is. There’s a tool for this called Gage R&R – you can guess what the R&R stands for.

Let’s keep in mind, though, that measuring the process ranges in complexity from “close to none”—you perform a manual count of defects—to “very”—you have several precision instruments that are calibrated to very low tolerances, for example tiny fractions of an inch.

Here’s a simple example. Suppose that you’re one of three inspectors at the Zappies Cereal Factory. Once a day, each of you pull a case of 24 boxes of Zappies off the production line and weigh each box to check that its Net Weight is actually six ounces, give or take one-tenth of an ounce. You weigh each box in the case… twice. You keep a list of the weights each time. When you’re done, do the lists match? If so, congratulations, your process is repeatable.

Then, you take the cases that the other two inspectors did, and without looking at their results, you perform the same exercise. Did your results match theirs? If so, hurray, the process is reproducible, as you all got the same answers.

If a Gage R&R fails, then there’s likely something wrong with either the measurement technique or the actual measurers. For example, it could be that the scale used isn’t working right, and tells you the first time that the box weighs 6.3 ounces and the second time that it’s 5.8 ounces. Or it could be that one of the inspectors is reading 6.3 as 6.8 which makes that reading different than yours.

Note that the measurement being alright isn’t the same as the actual process being alright. The three of you could have been repeatable and reproducible in your efforts. But the boxes of Zappies could all have had a Net Weight of 6½ ounces instead of the 6 ounces they’re supposed to have. But you do know that you’re consistent about this finding. With that out of the way, it’s time to understand why there are too many Zappies in the boxes.

Gage R&R tests can get pretty extensive, and expensive. As with other tools in the toolbag, it’s important to balance the cost of conducting the test against the value of the information you’ll gain. If the accuracy of measurement process is in doubt, then by all means, find out whether you can “do that again.”