The process of improving systems

Statistical analysis is often an important part of long-term monitoring and improvement of the performance of a system. This process is often called continuous quality improvement and is particularly important in business and industrial contexts. The approach taken to solving many biological, agricultural and health problems can also be considered to be a process that is aimed at improving the treatment of a disease, the yield of a crop, etc.

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: Saving a native bird

A government conservation department is concerned about declining numbers of a native bird and funding has been allocated to try to reverse this trend. A small team is put together to utilise the funding, and the following is a description of what they did.

Plan

What are possible causes of the problem? And which are most likely to have resulted in the decline in bird numbers?

  • Attacks by predators?
  • Climate change?
  • Disease?
  • Changes to the birds' habitat?

If possible, collect data about the birds and their behaviour to help with this planning stage, or some data may already exist in records.

Do

Attacks of eggs and newborn chicks by rats are thought to be the most likely cause of the decline, so the ecologists try rat-proofing several trees in which the birds nest. Number of surviving chicks in each nest from these and other trees that were not rat-proofed are recorded in the following season.

Check

The data collected about bird survival is now analysed to assess whether the number of chicks surviving in rat-proofed trees is higher than in the ordinary trees.

  • Has the survival rate been increased? If so, by how much?
     

Act

If the trial has been successful and the survival of chicks has been improved enough, a programme of poisoning or trapping rats might be introduced in breeding areas for the birds.

Plan

Has there been enough improvement? If not, was the correct cause of the problem identified? Was the change in the system 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, hospitals may be concerned that patients are taking longer to recover from heart surgery. What is the cause? Older patients, changes to surgical procedures, changes to care, ...?

The Plan-Do-Check-Act cycle can be used to determine reasons for the increase and determine ways to improve recovery times.

A Plan-Do-Check-Act cycle underlies most practical applications of Statistics.