Principle 6: Variability (Item 5)
All systems and processes exhibit variability, which impacts
on predictability and performance.
Reducing your special cause variation is the first step in improving
your processes. If your processes are not under control (i.e. the special
causes are not under control), you cannot begin the major improvement
effort - reducing your common cause variation.
The example of the wobbling pen introduces the concepts of special
and common cause variation. The `normal' wobble is due to the common
causes of the system (e.g., an unsupported arm). A special cause was
the push.
Faults inherent in the system are called `common causes' of trouble.
Faults from fleeting events are called `special causes'. Deming estimated
that 94% of problems are from common causes – and belong to the
system (and so are the responsibility of management – Principle
4 `To Improve the Outcome, Improve the System') and only 6%
are due to special causes.
People often make the mistake of assuming that every event (defect,
mistake, accident) is attributable to someone (usually the nearest
at hand), or is related to some special event. This is very rarely
the case.
| |
Variation is due
to |
| |
|
Special cause |
Common cause |
| Assign blame to |
People and events |
OK |
The very common mistake of blaming
people when the system fails (and is management's problem to fix) |
| |
The system |
The much rarer mistake of trying
to find a system problem when the problem is not in the system |
OK |
Examples of special causes include:
- The push given in the wobbling pen example.
- An employee is absent because they were hit by a bus, or their
child is ill.
Employee absence is a good illustration of the difference between
special and common cause. "When Mary was away on Tuesday, Fred
had to do her job and made several mistakes". Mary's absence
was a special cause of the mistakes. However, what about this. "On
average 10% of our staff are absent each day. When staff are absent,
other people, who do not have the training or knowledge of what the
absent person was doing, have to do their work. Mistakes are always
happening". Those mistakes are due to system problems and are
common cause.
Forget blaming any person. The problem is much, much, much more likely
to be in your systems and you should look there first. Find what is
causing the problem. Unfortunately, this approach is still very unusual.
It is much easier to say – "we found the problem, sir, and
he (or she) has been moved on. It won't happen again." However,
unless you find the cause (be it hiring policy, inadequate training,
inadequate knowledge to do the job, processes that cannot work) the
problem will happen again. In a few months time, you will be moving
the next poor sap on.
In 1998 in Sydney, a train tipped over when
its driver failed to slow at several warning signals and went into
a corner too fast. The driver was injured. No one else was on board
the train. There was a public outcry and howls for the driver to be
dismissed and castigated as "a dangerous person who should never
again drive a train". The employer took a very unusual position
and stated. "We learned a lot from this accident. We had put
too much pressure on the driver by making him work alone and forcing
him to comply with an impossible timetable. In those conditions, he
could not actually see the warning signals because he was busy with
other demands. This was clearly very dangerous. We have now changed
the warning signals so they appear in the driver's cabin and are audible
– not only visual. We have changed the timetable and the staffing
numbers so that the work is now possible to do." In other words,
the employer took the very unusual position of recognizing common
cause (although there was only one event) and fixing the system problems.
When the output of any system is measured, it shows variation around
an average result. We are interested in two components:
- the difference between the average measurement and the target
value
- the `range' in the measurements
What can you do to improve, to reduce the variation? You can tackle
reducing `range' in the measurements as well as moving the system
average closer to the target (i.e., reducing the `difference between
the average and the target').
That is, make the system more `stable' by making it more `consistent'
and more `capable' by moving it closer to your target. In that order.
Improvement
(less variation and closer to target)
Improving a process follows the path shown in the diagram below.
- First achieve stability (statistical control) by eliminating the
special causes of variation – reduce the `range'
- Then improve the capability by working to reduce the common
causes of variation – move the system average closer to
the target.
|