Adjustment or Tampering? Watch out!

There may be a world where there is no variation, but that it is not the one we live in. In our world everything is varying all the time. Heraclitus a coule of hundred years BC made the observation that one cannot touch the same thing twice. A quote often attributed to him is, “You can’t step in the same stream twice.”

Variation is ubiquitous.This is bad news for people who are trying to meet some targeted performance. Perhaps a manufacturer is trying to produce identical piece parts. Maybe a golfer is trying to reach a given performance goal on every round of golf. They are doomed. Not only will they not succeed, they cannot succeed.

But a lot of managers are either unaware of this reality or feel that they can somehow overcome it. So when the outcome does not match their perfect target, they take action; they adjust.

Dr. Lloyd S. Nelson, one of the nation’s top industrial statisticians, developed an experiential way of demonstrating what happens when different types of adjustments are made. He did so using a funnel and his demonstration is known as “The Nelson Funnel Experiment.”Statistical stability of a process indicates that if left unaltered, the process will continue to produce outcomes that show a stable pattern of variation.

If the process is not statiscally stable, it’s performance is unpredictable. We will assume our starting point is a stable process. A landing zone is prepared and a target shown on it. An elevated funnel is placed over the target (as best as one can) and a marble is dropped through the funnel and it’s proximity to the target is noted.Our statement about variation indicates that no matter how well we line up the funnel over the target, not every marble dropped will end up on the target.

There will be variation.

Nelson’s funnel experiment examines four strategies for adjustment. They are Rule 1: No adjustment is made. The process is left to run. Rule 2: Every time a marble lands away from the target, the funnel is moved from its current position an equal distance from the target, but in the opposite direction. Rule 3: The funnel is moved back to being centered over the target and is then moved from the target in the opposite direction as the marble missed the target and an equal distance. Rule 4: Try to repeat the last performance by moving the funnel over the landing point of the last drop.

With the exception of Rule #1, every other adjustment has the effect of making things worse. Using Rule #1 leaves things unchanged.

Are helter skelter adjustments a problem? You bet.Rule #2 and Rule # 4 are everywhere. Rule #3 doesn’t make a lot of sense and one doesn’t see many examples of it in practice, but re-adjusting after every off-target event is a common occurrence and repeating the last performance is also seen often in the workplace. Using Rule #2 approximately doubles the amount of variation and using Rule #4 produces a random walk meandering off toward infinity.An example of Rule #2 would be making an adjustment to a machine every time a part fell outside the specification limit. And example of Rule #4 is worker training worker or using a sample from the last batch produced as the standard for the next batch (seen often in the paint or chemical industry).

Deming called it tampering. It is the adjustment of of a stable process after each occurrence; treating every event as though it were special.

It makes things worse.