October 1997 // Volume 35 // Number 5 // Tools of the Trade // 5TOT2
Predicting the Future: Have you considered using the Delphi Methodology?
Frequently business, governmental agencies and organizations are faced with the problem of predicting or forecasting future events and relationships in order to make appropriate and reasonable plans or changes. Several methods exist for forecasting, one of which is called the Delphi technique. This approach has not been frequently used by Extension practitioners, but has potential as a tool for consensus building. The Tools for the Trade article provides an overview of the process and suggestions for implementation.
Extension professionals, attempting to gaze into their crystal balls and predict the future, often look for new ways to reach clientele or customers to determine where programs or applied research should focus. Unfortunately, it becomes more and more difficult to gather groups of people together to talk about the future because of work schedules and the time/cost of asking people to travel and meet together to talk about "what could be."
An under-used methodology combining quantitative and qualitative opportunities to explore the future is the Delphi. It's an old method, dating to the 1950s, developed by a team of researchers named Dalkey and Helmer. Today businesses, governmental agencies, and organizations are using Delphi methods to predict or forecast future events and relationships in order to make appropriate and reasonable plans or changes. Studies comparing the Delphi's results with other methods (Ulschak, 1983) confirmed effectiveness of the method related to generating ideas and use of participants' time.
Delphi is a group process and its goal is to help a group reach consensus. Rather than gathering people together for oral discussion, individuals provide written responses to questions. This is an advantage when persons possessing the knowledge and expertise to address the problem are not in close proximity. With the increasing growth of electronic mail, the technology can be adapted to facilitate the process. Anonymity of the respondents during the process is an important aspect of the Delphi and needs to be maintained if electronic mail is used.
Selection of who to include in a Delphi forecasting study is critical. The majority of Delphi studies have used between 15-20 respondents and run over periods of several weeks. The number of respondents was generally determined by the number required to constitute a representative pooling of judgments and the information summarizing capability of the research team. Large numbers of respondents generate many items and ideas making the summarizing process difficult.
Debecq, Van de Ven & Gustafson (1975) suggested using the minimally sufficient number of respondents. Dalkey, Rourke, Lewis & Snyder (1972) reported there was a definite and monolithic increase in the reliability of group responses with increasing group size. Reliability, with a correlation coefficient approaching .9, was found with a group size of 13.
WHO is invited to participate in a Delphi futuring exercise should be carefully considered. Randomly selecting participants is NOT acceptable. Instead, characteristics and qualifications of desirable respondents should be identified and a nomination process used to select participants. Because the group number will be small (12-15), the researcher needs to locate and target individuals who are "expert", have knowledge and experience to base their futuring activities upon, and are self-motivated. Delphi should not be used with groups that have difficulty in reading or expressing themselves in written communication.
In a Delphi study, the development and administration of questionnaires is interconnected. Note that multiple questionnaires are used. Participants in a Delphi agree to receive and respond to a series of questionnaires, usually at least three different rounds are used. The first questionnaire could take several forms, but would most likely be one or two open-ended questions related to a broad problem or issue. The second questionnaire is developed by the researcher based on the information collected during the first round. Being certain steps are taken to eliminate the chance of research bias is important throughout the Delphi process.
During the second round, the second questionnaire asks participants to review all items identified by the first round of the Delphi. From the original open-ended questions, a series of structured items are developed by the researcher. Participants rank-order items or use a lickert-type rating scale to establish preliminary priorities among items. Participants are invited to comment on their rationale for the rating and add additional items.
During the third, and any additional rounds, the Delphi respondents re-rate each item. To assist in their consideration, participants are provided with: (a) statistical feedback related to their own rating on each item, (b) how the group of participants rated the same item and (c) a summation of comments made by each participant. This feedback process makes the Delphi respondent aware of the range of opinions and the reasons underlying those opinions.
Delphi rounds of questionnaires continue until a predetermined level of consensus is reached or no new information is gained. A review of literature (Altschuld, 1993) found that in most instances three iterations were enough and not enough new information was gained to warrant the cost of more iterations.
There are many variations of the Delphi technique. The references included at the end of this article provide a starting point for learning more about the Delphi. Key points to remember when considering use of Delphi include:
- Futuring, policy decisions or planning for the future
- Consensus of a group is needed, but distance and schedules of participants make meeting difficult
- (3) A group of experts can be identified to compose the Delphi Panel
Altschuld, J. W., (1993). Delphi technique. Lecture: evaluation methods: Principles of needs assessment II. Department of Educational Services and Research. Columbus: The Ohio State University.
Dalkey, N.C. & Helmer, O. (1963, April). An experimental application of the delphi method to the user of experts. Management science, 9, 3, pp. 458-67.
Dalkey, N.C.; Rourke, D.L.; Lewis, R.; Snyder, D. (1972). Studies in the quality of life. Lexington, Massachusetts: Lexington Books.
Debecq, A.L.; Van de Ven; A.H. & Gustafson, D.H. (1975). Group techniques for program planning. Glenview, Illinois: Scott, Foresman and Company.
Ulschak, F. L. (1983). Human resource development: the theory and practice of need assessment. Reston, Virginia: Reston Publishing Company, Inc. Pp. 111-131.
Weaver, W.T. (1971). The delphi forecasting method. Bloomington, IN: Phi Delta Kappan, 52 (5), 267-273, pp 268