Delphi

An introduction to the Delphi Method

Delphi is based on the assumption that collective expert judgement is a valuable method of forecasting in the absence of reliable data on which to base traditional forecasting methods. To overcome the problems of group dynamics the members of the expert panel are not brought together and remain anonymous. In its original form, developed by the RAND Corporation in the 1960s, a series of questionnaires interspersed with feedback were sent to panel members. Software is now available to run Delphi studies on-line.

Requirements

1. An administrator
2. A panel
3. A series of questionnaires
4. A means of communication between the administrator and the panel members

Procedure

1. Decide on the topic to be considered.
2. Clarify the purpose of the Delphi.
3. Set up the panel.
4. Depending on the purpose of the exercise the first round can take one of two main forms. It may either pose certain defined questions or be completely open, asking the panel, for example, what are the most likely changes they anticipate in the area of concern in the next 25 or 50 years.
5. In the second round the panel are asked to consider the list generated by the group and to say whether they think the items listed are likely to occur, or if the aim is to obtain an idea of when the developments may occur, to ask for dates. The value of circulating the list comes from the much wider set of ideas generated by the group than each individual alone would create.
6. The quantitative Delphi should continue until the forecasts have reached stability. Each round of a non-quantitative Delphi should be given a clear function.

Examples

Delphi has been used by several national Foresight Programmes including the first UK programme.

Advantages

• Delphi may be the only way to predict emerging developments where there is no empirical database.
• Useful where there is little evidence that more conventional quantitative methods provide more accurate long-range forecasts.

Disadvantages

Subject to the usual problems of judgement such as

• Bias
• Overconfidence

References

Expert Opinion in Forecasting: The role of the Delphi Technique Gene Rowe and George Wright in Principles of Forecasting: A handbook for researchers and practitioners J Scott Armstrong ( editor) Kluwer, Boston 2001 pages 125-144