The Delphi method, run online with Mesydel

Consult experts, stakeholders or citizens in writing, at a distance, over several rounds. Each one reads the others’ arguments without knowing who wrote them, and takes a position again. What comes out is a collective answer, or a clear view of where and why people disagree.

Plan your Delphi with us See the studies conducted with Mesydel

Three principles make a Delphi

Anonymity

Everyone reads the arguments without knowing who wrote them: rank, prestige and the loudest voice weigh less.

Rounds

The panel answers, discovers the others’ answers, and takes a position again, round after round.

Controlled feedback

Between two rounds, your team decides what the panel sees: statistics, a synthesis, chosen quotes. That is what turns a repeated survey into a Delphi.

A qualitative and quantitative method: open and closed questions, from about ten to several hundred participants, on questions that have no absolute answer.

A Delphi, round after round

  1. Round 1 Open: bring out the ideas
  2. Analysis by your team Controlled feedback: statistics, arguments, chosen quotes
  3. Round 2 Assess: rate and justify
  4. Analysis by your team Controlled feedback: statistics, arguments, chosen quotes
  5. Round 3 Review: take a position again
  6. Results consensus and dissensus

Two or three rounds are usually enough. Each round loses some participants: Mesydel shows who took part in which round, so you can remind the right people.

Survey, focus group or Delphi?

Criterion Survey Focus group Delphi
Interaction between participants none direct, face to face indirect, through the feedback
Anonymity between participants yes no yes
Number of rounds 1 1 session 2 to 4
Size of the panel large, representative 6 to 12 from 10 to several hundred
Constraints of place and time few strong few
What you get a distribution of opinions a group dynamic arguments, how they evolve, consensus and dissensus

A Delphi is not a repeated survey: without feedback between the rounds, there is no Delphi.

When a Delphi is the right choice

The method comes after the question: we tell you honestly when another one fits better, and often combine several.

A Delphi fits

  • Uncertainty or foresight: no established answer, possible and desirable futures to explore
  • A conflictual subject, or power relations in the group (the boss’s opinion weighs too much in meetings)
  • Scattered participants, incompatible agendas
  • A need for collective learning: everyone discovers the others’ arguments
  • Preparing an acceptable decision: involving those it concerns, building common ground

Another method fits better

  • The answer is factual: look it up, don’t vote on it
  • You need statistical representativeness: run a survey
  • You want to observe the interaction itself: run a focus group
  • No time to analyse between the rounds: a Delphi needs work between each round

Consensus, and dissensus

Opened to stakeholders and citizens, the Delphi has become a tool of participation as much as of forecasting, and its aim has moved from consensus to the quality of the deliberation: positions are explained and tested. Showing clearly where and why participants disagree is a result in itself, and can lead to reframing the problem.

Mesydel shows how the answers are distributed; it does not compute a consensus for you. You set your rule before round 1, and your results stay true to it.

Where it comes from

1950s, the RAND Corporation

For American defence: forecasting and setting priorities from experts’ opinions, bringing views together without face-to-face confrontation, away from the pressure of rank, prestige and the loudest voice.

Since the 1970s, lay expertise

The panel opens to the people concerned, whose expertise is their experience of a situation (Linstone & Turoff, 1975): their perceptions, their representations, the expertise of those who use a service.

The name comes from the oracle of Delphi: the method was born to foresee the future.

The pitfalls, and how to avoid them

Forced convergence
It is easier to fall in line than to defend a minority view. Ask everyone for their arguments, not only those who stand out.
Fatigue
More rounds mean more drop-outs, and sometimes a consensus for show. Two or three rounds are often enough.
Selective drop-outs
Those who leave are not chosen at random. Mesydel shows who took part in each round, and lets you remind exactly those who have not answered.
Consensus is not truth
A group that shares a culture can be wrong together. Diversify the panel, and report the dissensus as carefully as the consensus.

A case: evaluating the decree on public reading

For the Fédération Wallonie-Bruxelles, the evaluation of the 2009 decree on the missions of public libraries, required by the decree itself. All the librarians of the Federation were consulted with Mesydel, over one year.

  1. Prepare 20 exploratory interviews and the official texts
  2. Round 1 December, about 25 questions, mostly open 59 % took part
  3. Analyse tags, read in pairs, then as a team
  4. Round 2 end of March, extracts of round 1 set side by side 104 of 190 answered (54 %), only 7 fewer than in round 1
  5. Deepen 8 case studies, 5 focus groups
  6. Report 2 successive reports

The Delphi gave the widest view; case studies and focus groups then went deeper.

Plan your Delphi with Mesydel

Access to the platform

You run your study on Mesydel, on your own. We train your team if you wish.

With our support

You keep the lead; we help you choose the panel, build the questionnaire, prepare the feedback between rounds or analyse the answers.

Contact us Tell us about your question: we reply with a proposal.

Mesydel in a minute and a half

Further reading

79 articles, reports and theses came out of studies conducted with Mesydel, or refer to it: the method at work, from medicine to public policy and the environment. Only documents made public are listed.

Browse the publications

Contact


Do not hesitate to contact us to let us know about your project.