Preparation
- What question or set of questions does the Delphi aim to answer? It should be a topic where the informed opinion of multiple people adds value and where no single objective answer exists.
- Examples: prioritization of intervention areas, prospective analysis of trends, evaluation of alternatives, construction of shared indicators.
- A Delphi panel is not a statistically representative sample: it is a group of people with relevant knowledge, experience, or perspective on the topic.
- Diversity matters: including people from different roles, geographies, sectors, and experiences enriches the process.
- Recommended size: 10-30 panelists. More than 30 complicates analysis; fewer than 10 limits diversity.
- The first round is usually more open (open-ended questions or ranking lists) to capture the diversity of perspectives.
- Subsequent rounds are more closed (agreement scales, rankings) to identify convergences and disagreements.
Step-by-step instructions
Round 1: Opening
- The first questionnaire is sent to all panelists, explaining the process and guaranteeing anonymity.
- Panelists respond individually and without knowing others' responses.
- The coordinating team analyzes the responses, identifies emerging themes, groups similar ideas, and drafts a synthesis.
Round 2: Informed review
- Panelists receive the anonymous synthesis of the previous round along with the second questionnaire.
- Each panelist sees what the group thought collectively and can revise or maintain their position.
- They are asked to briefly argue if their position differs significantly from the group majority.
Rounds 3 and 4 (if needed)
- The cycle is repeated until sufficient convergence is reached or positions stabilize.
- Generally 2-3 rounds are enough; more than 4 tends to produce diminishing returns.
Final synthesis and feedback
- The coordinating team prepares a final report with the consensus reached, the grounded dissents, and the topics requiring further exploration.
- The report is returned to all panelists and used as input for decision-making or planning.
Materials
Platforms
For virtual sessions or collaborative documentation.
Practical recommendations
Anonymity is central to the method: if panelists know what each other said individually, social pressure distorts responses. Syntheses are always collective and anonymous.
Design short questionnaires: panelists are usually busy people. A questionnaire with more than 15-20 items dramatically reduces the response rate.
Allow sufficient time between rounds for analysis: rushing the processing of responses produces superficial syntheses.
The Delphi does not eliminate dissent: if divergent positions persist after several rounds, that is valuable information, not a process failure. Document and analyze those divergences.
Combining it with in-person or virtual sessions at the start (to present the process) and end (to discuss results) enriches the experience and the commitment of panelists.
In educational contexts: the Delphi can be used for teachers or students to collectively build assessment criteria, curriculum priorities, or community agreements.
Inspiration
- Strategic planning: What are the 5 most important priorities for our organization in the next 3 years?
- Program evaluation: What indicators should we use to measure the impact of this project?
- Foresight: What are the most important trends that will transform our sector in the next decade?
- Curriculum development: What competencies should graduates of this program prioritize?
- Public policy: What lines of action are most effective for addressing this problem in our territory?
- Single-round Delphi: For processes with less time available; loses the iterative effect but maintains anonymity and collective synthesis.
- Real-time Delphi: With small in-person groups; responses are given on paper or digitally, processed quickly, and the synthesis shared within the same session.
- Scenario Delphi: Instead of asking about priorities, possible future scenarios are presented and panelists rate the probability and desirability of each.
- Community Delphi: Participatory adaptation in which the panel includes both technical experts and representatives of affected communities, giving equal weight to both types of knowledge.
Consulted sources
- Dalkey, N. & Helmer, O. (1963). An experimental application of the Delphi method to the use of experts. Management Science, 9(3), 458-467.
- Linstone, H. A. & Turoff, M. (Eds.) (1975). The Delphi Method: Techniques and Applications. Addison-Wesley.
- Hasson, F., Keeney, S. & McKenna, H. (2000). Research guidelines for the Delphi survey technique. Journal of Advanced Nursing, 32(4), 1008-1015.
