Quality Magazine

Other Dimensions: Measurement Uncertainty

June 29, 2007

Editor’s note: This is the first in a three-part series on measurement uncertainty.

Nothing provokes heated discussion like uncertainty. We see it in the global warming-now relabeled as climate change-debate and scores of other issues. We demand precise yet simple answers to complex matters, as politicians know only too well. And if we can find someone else to blame, that’s even better.

In dimensional measurement, we use numbers rather than fear to make a point. But sometimes the irrational does creep in. For example: “I paid a gazillion dollars for that thing and you’re telling me it’s not accurate enough to measure these parts,” or “we can put a man on the moon but we can’t measure this better than ...”

These comments and others like them usually surface when there are disputes over measurements and the subject of measurement uncertainty enters the discussion.

Measurement uncertainty has been a part of every measurement ever made. The only thing new about it for most folks is the name, what it really means and how you deal with it. I will try to answer these questions in this series of columns as best I can, given the limitations of time and space.

Years ago if someone asked Charlie in the inspection department to check a part feature he would grab his trusty nondigital micrometer and would give a number. If there were a dispute, the next question would be, “How close do you think you are?”

The reply would be something like “close enough for artillery” or “within a couple of thou.” The first response would be sarcastically given when the worst known error was not significant for the tolerance involved.

The second response is similar to a measurement uncertainty statement of today. It meant that Charlie’s reading of size would be within a couple of thousandths of an inch of the true value for the part feature that he measured. It did not mean his reading of size was “out” or in error by that amount, only that it could be. His reading might have been the actual true value.

Charlie knew how accurate his micrometer was, but more importantly, he knew other factors besides the micrometer entered into the process and had a direct effect on how close his reading would be.

In the past such guesses were often surprisingly close to an uncertainty value that would be calculated today. The reason being that the person who made the statement was skilled at what he or she was doing, with enough experience to realize there was more to a measurement than a reading.

The problem with estimates from people like Charlie-or Charlene-in the inspection department is that they depend on skill, knowledge and integrity. Without all three elements they are worthless. And often an honestly given guess could be incorrect due to an oversight. Regrettably, the elements of skill and knowledge are missing in many workplaces today in favor of low-cost instrument readers and their managers who believe an unqualified reading is the same as a measurement.

Metrologists know otherwise and understand that a reading without an uncertainty statement is not worth much. They have developed a method of arriving at a value for measurement uncertainty in any given situation. It enables one to compare how a value was determined so when there is a dispute, both parties can compare this information and the treatment of it to see if the uncertainty claimed by both parties is realistic. The guesswork is dramatically reduced, if not totally eliminated to any significant extent.

How is this done? It’s a matter of knowing all the elements that will affect a given measurement under specific conditions and then processing that information in a methodical way. The tool used for this is the uncertainty budget-the subject of my next column.

The climate change folks use qualifiers such as “likely” or “very likely” based on percentage ranges while predicting the future of the planet. After reading their recent “report” I am not sure whether they are basing their fears on gambling odds or psychic predictions. Politicians are ready to tax and control on such specifically vague statements because they are masters at making them and the public believes it. Our colleagues in metrology would never let us get away with doing the same thing, so we’ll get into the details next time.