The process of analyzing gage variability falls into two categories:
- Often highly structured, involving an examination of the gages themselves for sensitivity to temperature changes, magnetic fields and other factors. These are the easy ones.
- Its source is in gage operators themselves, who may have different levels of training, experience, fatigue and even attitude.
Collecting data offers clues to sources of variability. But when this disciplined analysis fails to uncover real reasons for variability, it may be time for the Sherlock Holmes of variability to look at operational definitions — often the most overlooked consideration when evaluating variation among measurement devices.
Elementary, my dear Watson? Perhaps. Nonetheless, these definitions can lead to levels of variation in gage output if they are vague or nonexistent. “In the opinion of many people in industry, there is nothing more important for transaction of business than use of operational definitions. It could also be said that no requirement of industry is so much neglected.”1
Because we are used to somewhat loose definitions of tasks with the expectation that “everyone knows how to do this,” it’s easy to forget the importance of clearly defined instructions for collecting data. It’s one thing to say “Put the groceries away,” and another to specify, “Put cold things in refrigerator, frozen food in freezer and canned goods in pantry,” if, for example, the operator is a child. Since everyday tasks such as this do not demand precise measurement, one can afford to be casual about the instructions, defining them in terms of the person carrying them out. In manufacturing and service environments, however, lack of clear and specific operational definitions can create chaos, rendering data that is produced meaningless and outcomes unclear.
If inspectors are asked to identify defective devices, each will have his or her own sense of “defective.” If they have a clear understanding of a specific characteristic of interest (inaccurate measurement, for example) as well as the method for measuring it and the decision criteria that are to be considered, they are more likely to arrive at the same conclusions about what constitutes defective.
Sometimes an operational definition may appear to be appropriately focused and clear:
“When measuring the part, hold the gage firmly and tighten the thimble firmly. Measure to 1/16″.”
Sounds good, right? But do you know how tight the thimble really should be? Or whether to round up if the measurement is close to 1/16″?
How about this approach to an operational definition:
- Setup: Start the gage lab with all eight lights on. The temperature in the lab must be between 73 and 75 degrees Fahrenheit, with 20 percent to 30 percent humidity. All parts to be measured must be in the lab for a minimum of 30 minutes prior to any measurements being taken, to assure uniform temperature.
- When measuring tubes up to 3″, hold the 0″-3″ micrometer at a right angle to the tube or use the gage fixture. The Anvil and Spindle will be perpendicular to the tube.
- Tighten the thimble until the slipping clutch clicks.
- Measure to the nearest 16th of an inch. Round up if the measurement falls between scales.
This detailed operation definition helps to assure that all operators approach the task in the same way, reducing the levels of variation among them.
In another example, the directive, “Gages must be checked at regular intervals” invites chaos. What does the “checking” entail? It might be only glancing at the inventory to make sure gages are in the right place. And “regular intervals” could mean anything from every hour to once a year on your birthday. When operators are left to create their own definitions and understandings, the outcomes are not reliable.
Establishing an operational definition for “checking” may include a description of the instrument that is used (Naked eye? Camera? Historical record?). It may entail actually picking up the instrument, or taking additional steps to weigh it or assess its accuracy. Without specific information, one might assume that “checking” might mean just verifying that it is in the inventory. In the same way, “regular intervals” may express a variety of meanings; it is far more accurate and easy to interpret if the definition offers specific time intervals. “Every three or four days” of checking would be far less predictable than “once a month, on the final work day of the month,” for example.
In manufacturing fuel gages, it may be important to check the position of the needle on the gage and to define where it should be when the gage is measuring “empty.” We’ve all had fuel gages that indicated “empty” when the needle was actually above or even below the “E” on the gage, and the fuel tank actually was empty. We simply become used to the exact point—perhaps only after running out of gas a time or two. The manufacturer, on the other hand, needs to have assurance that a fuel gage is consistent and predictable in its announcement of “empty.” An operational definition might indicate that the needle must be touching or covering the “E,” for example, or that it will indicate an empty container only when the needle reaches a point below, and not touching, the “E.”
Not only are operational definitions essential to establishing a measurement system, but they also provide a diagnostic tool. When a system appears to be changing, the cause may be a change in the ways in which operational definitions are used. Whenever a system is unstable, operational definitions and their use should be evaluated for their impact.2
You may find that this evaluation brings with it the “Eureka!” moment that Sherlock himself experienced.