Warning: When calibration isn’t enough

Identifying measurement problems

As any mechanic will tell you, making sure that your car is in top-notch condition—as admirable as this practice is—will not assure that no one will run into you or that you will not go off the road as you talk on your cell phone. True, failing to have brakes checked periodically can indeed cause accidents; but even brand-new brakes won’t stop a car if the operator fails to put his or her foot on the brake pedal.

In the same way, calibration of gages and other measuring devices, while it represents critically-important practice, will not prevent measurement problems that are due to other factors, rather than to the accuracy of these devices.

Most organizations that use measurement systems have established processes to calibrate equipment on a regular basis. This practice is fundamental to good business. Generally, measurement equipment is uniquely identified and calibrated at a regular frequency against nationally-recognized standards. Compliant organizations maintain appropriate records to demonstrate conformity to ISO 9000 and other standards.

Even the most conscientious company with the best possible measurement system can run into issues when it comes to day-to-day assessment of its products, however. Just because the measurement system is calibrated does not mean that the results are going to be accurate. Other potential causes of variation exist, and these are often more significant than an out-of-calibration piece of equipment. The main cause may lie with the operator or tester. Does every tester in the organization complete the measurement in exactly the same way? Do apparent minor differences in testers’ work procedures matter?

A white paper by Jacqueline Graham, Ph.D. examines the ways in which data can be summoned to recognize faulty techniques that testers may have developed over time.

The following chart, a conventional Shewhart control chart, indicates changes in test results over time. Understanding whether the periods of poor control are related to production or to the approaches used by testers is critical to improving the process.

Graham’s white paper demonstrates ways in which data can be examined to discover variation among testers, between shifts, or in training of testers. Measurement systems analysis will reveal answers to questions about the accuracy of the testing data, discover bias issues, and reveal variation in operators’ approach to testing. The white paper explains the use of bias or ANOM charts to review results.

A measurement systems analysis (MSA) study is a kind of experimental design. It consists of a series of measurements made of the same product, using the same equipment, by different testers. When this is completed, the results clearly indicate where the measurement issues lie.

To learn more about ways to collect and analyze data related to issues of variation among testers, download this free white paper entitled “Warning: Is calibration enough? How to identify measurement problems.”

In the meantime, be sure to get your brakes checked. And stay off that cell phone while you’re driving.

Originally published in the May 2010 edition of Quality eLine, our free monthly newletter.

Subscribe now!