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.

Check

...

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.