All things in life vary. We make decisions daily based on the variation around us. The variation we expect is due to common causes. When variation is larger than expected, we look for special causes. The key is to understand the information in the variation, the difference between common and special cause. Actions required to reduce special causes of variation are totally different from common causes. Attempts to reduce the common cause will increase the problem, countering efforts to reduce special cause effects.
In the figure above, where no adjustment was made after each drop, the variation is less. What we see is the result of variation in a stable system. The pattern is regular, with no dramatic outliers. This is characteristic of common cause variation.
Next, we move the funnel in response to the location of the last drop. The intent is to reduce the pattern of variation. The tampering pattern occurred when, after each drop, the funnel was moved to counter the last drop. That is, if the drop occurred say, at top of the pattern (12 o’clock), the funnel was moved toward 6 o’clock for the next drop, etc., etc.
Each pattern can be represented by a bell-shaped curve, with most data points near the center and fewer near the edges. We say the point of inflection is one standard deviation from the center. This distance is the common “Sigma” we often refer to in our work.
The Normal Distribution Curve
Using the Normal Distribution
A distribution signifies where points are likely to occur. In our marble-drop, most fell near the X, with fewer at the edges. A “Normal” distribution shows this quite well.
The popular Normal Curve describes many real life examples, with neat math properties. With the Normal, we can make strong predictions about how data points occur in real processes.
If we know the parameters of a stable process: Average (Mean) and Standard Deviation (spread). Observations outside 3-sigma limits are unlikely.
We can say, with some confidence, points outside the limits are due to special causes. Take Action!
Data points inside these limits are due to common causes, beyond our control. LEAVE THEM BE!
Any attempt to adjust a stable process always results in greater variation. Again, we call this “Tampering.”
Common vs. Special
How does an understanding of variation help us work smarter? The only way common cause variation can be reduced is by a change in the process. Management is responsible.
This is a smart division of labor: Common Cause = Management Control, Special cause = Worker Control
And this is the ONLY way it will work!
For example, take a stable system (Harry’s performance on the cleaning process which he has done for many months). The chart above is a measure of the particle count (smaller is better) measured each day after Harry’s efforts.
John, Harry’s supervisor, sees the particle count fall as shown on Day A. So John gives Harry some kudos! But it then gets worse, (see point B). So John kicks a little tush and sure enough, the performance improves.
What kind of supervisor does John become? What works best, Kudos or Tush? Why not? It has worked for months! (years???)
This funnel example illustrated the frustration that results when all variation is blamed on a single incorrect cause. The horizontal position of the funnel was not the cause of the variation. What we missed for this example was the proper analysis of data to clearly identify the information in the variation.
There is a powerful difference between data and information!
Variation Take-Away – 1
Variation has two types of causes:
Common Causes: Those inherently part of reprocess, that contribute to relatively small, apparently random shifts in outcomes hour after hour, day after day.
Special Causes: Factors that drive variation above that inherent in the system, arising because of special; circumstances.
Common cause variation is difficult if not impossible to link to any particular source. Special cause variation is called “assignable” because it can be tracked down to an identifiable source.
Variation Take-Away – 2
An economic balance must be found to minimize… – Looking for assignable causes when they do not exist. – Overlooking assignable causes when they do exist.
The funnel powerfully illustrates how missing the information in variation can ruin even the best efforts.
Misunderstandings: (what are the problems here?)
– “you’ve done it once, that proves you can do it every time.”
– “You must perform above average all the time.”
Variation Take-Away – 3
A stable process (one with only common causes present) is said to be in a state of statistical control.
The cause system for variation remains constant over time.
In Statistical Control, variation in outcomes is predictable within statistically established limits. Knowing this, managers/supervisors can: a) Avoid blaming people for problems beyond their control. b) Avoid spending money on new equipment not needed. c) Avoid wasting time looking for causes for trends when nothing has changed.
We began this section with the idea of “Smart-Think,” in an attempt to broaden our view of problems and procedures we bump into daily. And each time the issue was what our “mind-set” is in relation to the world around us.
Consider the following puzzle to further broaden your outlook: Arrange the six toothpicks to form four identical triangles
Think about 3-dimensions…
Which center square appears larger? Think of a product as an object surrounded by a cluster of advertising, marketing, and sales promotions, like the squares that surround the identical figure in the center above. When these promotions and programs are understated (smaller), the quality of the product (the inner square) appears greater than it really is.
This is a powerful tool for those who craft advertising dollars for their product. For example,
When you have a problem, you can write a challenge statement, study it for a while, then leave it, change it, stretch and squeeze it, and restate it until you feel the challenge is centered, then you are ready
Write it as a definite question, beginning “In what way might I…?
Vary the wording of the challenge by substituting synonyms for key words.
Stretch the challenge to see the broader perspective.
Squeeze the challenge to see the narrow perspective.
Divide it into sub-problems.
Solve the sub-problems.
Keep asking “how else?” and “why else?”
“Lincoln was not great because he was born in a log cabin, but because he got out of it.”
James Truslow Adams