Continuous quality improvement
In many business and industrial contexts, and increasingly in government,
statistical analysis is an important part of long-term monitoring
and improvement of the performance of the system. This process
is often called continuous quality improvement.
The statistical part of the process again involves a feedback cycle of data
collection and analysis, aimed at improving aspects of the system. The following
diagram shows a common form of this cycle.

Data collection and analysis are an integral part of the Plan-Do-Check-Act
cycle.
Example: Shelf life of milk in a supermarket
A supermarket sells thousands of bottles of milk each week, but occasionally
gets complaints from customers about milk going off within a few days of sale.
How should this problem be addressed? A possible scenario is given below.
Plan
|
What are possible causes of the problem? And which are most likely
to give rise to the complaints?
- Problems before delivery of the milk to the supermarket?
- Delays in putting the milk in supermarket refrigerators?
- Poor cycling of the milk on the shelves so the oldest milk does
not get sold first?
- Customers not refrigerating their milk soon enough after purchase?
If possible, collect data about the complaints and the current system
to help with this planning stage.
|
|
Do
|
The cycling of milk on the shelves is thought to be the most likely
cause of the problems, so a daily inspection of milk is started with
the oldest milk being moved to the front each night. This new system
is trialled for 3 months and data are collected about...
- Complaints
- The age of milk on the shelves in random weekly surveys
|
|
Check
|
After 3 months, the supermarket analyses the data that have been collected.
- Has the incidence of complaints been reduced? If so, by how much?
- Is 'old' milk on the shelves less common now?
|
|
Act
|
If the trial has been successful and complaints have been reduced enough
to warrant the cost of moving milk each night, the change in the system
might be made permanent.
|
|
Plan
|
Has there been enough improvement? If not, was the correct cause of
the problem identified? Was the change in the system not effective in
attacking the cause?
- Try addressing another potential cause of the problem?
- Try addressing a different remedy for the cause?
|
|
Do
|
Try a different modification to the system, and collect data about
its effect.
|
|
|
The statistical process in other contexts
In many other contexts, a similar cycle is used to obtain 'improvements'
in the system. For example, biologists may want to increase numbers of an
endangered native bird. Why are numbers decreasing? Predators, disease, habitat,
...?
The Plan-Do-Check-Act cycle can be used to determine reasons for the decline
and obtain remedies.
A Plan-Do-Check-Act cycle underlies most practical applications
of Statistics.