Delphi Method

Overview

The Delphi method is an iterative process for consensus-building among a panel of experts who are anonymous one to another. In its traditional form, extensive questionnaires are distributed to the panel, responses are synthesized and used as feedback to the panel in the next round of questionnaires, for a series of rounds. Experts on the panel do not communicate directly with each other but rather only provide responses to the Delphi administrator. The method was first developed for the military by RAND in the 1940s, and is advocated as a means to achieve an optimally reliable expert consensus (Dalkey, 1963).

Key Concepts and Terms

Anonymity. By keeping the process of questionnaires and feedback anonymous, Delphi is intended to avoid undesirable group effects characteristic of face-to-face groups (ex., deference to high status individuals, deference to socially assertive individuals, groupthink, etc.). Critics, however, point out that anonymity could encourage irresponsible responses from panel members.
Legitimation. As in survey research, institutional support is often needed to gain access to the needed experts. Since experts are involved, there is often an appeal to advancing the state of research. Likewise, often there is a promise that research results of the study will be furnished to experts in a timely manner.

Assumptions

Potential for selection bias. Output from Delphi is only as good as the experts selected for the panel. Considerable commitment may be required from panel members, and this may affect the composition of the panel (ex., top experts may have the least slack time and choose not to participate). If experts are all professionals in the same area (ex., education), bias toward a professional agenda may emerge compared to selection of a different type of panel (ex., a panel of parents).
Potential for researcher influence. The researcher plays a critical role in synthesizing questionnaire responses and providing feedback to the panel. This process is inherently partly subjective and could introduce researcher bias.

Frequently Asked Questions

Isn't Delphi a very slow research method?

Bibliography

Dalkey, N. & Helmer, O. (1963). An experimental application of the Delphi method to the use of experts. Management Science 9(3):458-467.
Gordon, T. & Pease, A. (2006). RT Delphi: An efficient "round-less" almost real time Delphi method. Technological Forecasting and Social change 73: 321-333.
Landeta, J. (2006). Current validity of the Delphi method in social scineces. Technological Forecasting and Social Change 73: 467-482.
Linstone, A. & Turoff, M. (1975). The Delphi method: Techniques and applications. Reading, MA: Addison Wesley.
Rieger, W. R. (1986). Direction in Delphi developments: Dissertations and their quality. Technological Forecasting and Social Change 29: 195-204.
Sackman, H. (1975(. Delphi critique. Santa Monica, CA: RAND Corporation.
Scapolo, F. & Miles, I. (2006). Elicitng experts' knowledge: A comparison of two methods. Technological Forecasting and Social Change 73: 679-704.