Delphi
The Delphi is an iterative method designed to achieve consensus among a panel of experts on a particular issue (Dalkey and Helmer, 1963; Linstone and Turoff, 1975). It was developed at the Rand Corporation, and over the past half-century has seen literally thousands of applications. The main steps of the "classical" version can be summarised as follows: a) the selection of a panel of experts. b) creation and editing of the first questionnaire. c) submission of the first questionnaire to the panel. d) calculation of a statistical synthesis of the responses (e.g. the median or the interquartile range). e) submission of a second questionnaire, in which experts are asked to provide (for each question) a second assessment, based on the provided statistical synthesis and reasons that may have been provided by participants for their answers. f) calculation of the second statistical synthesis and inclusion of the reasoning and comments in the questionnaire as well as specific spaces for counter-arguments. g) The iteration of steps b-f until a stopping criterion is reached. The stopping criterion can be, for example, a predefined number of iterations, or a consensus threshold.
The basic principles of the classical
Delphi procedure are: anonymity among participants, iteration of emerging group
opinions, asynchronicity, controlled feedback, and statistical synthesis.
Panel members are anonymous to one another and do not meet face-to-face. Anonymity allows the participants to express their opinions/judgements more freely, and social pressure to conform is reduced. Since the collection of judgements take place in several iterations, the panelists may revise and even modify their opinions more than once, after comparing them with the statistical synthesis of the group (iterative structure). Also, the experts work at a distance and at different times (asynchronicity) and this eliminates totally the pressure typical of any classic synchronous meeting. The controlled feedback (reasonings and comments) informs the panelists about the other participants' viewpoints and provides the opportunity to explain and/or change their opinions (controlled feedback). Lastly, a facilitator combines, at each round and for each question, the responses of the participants into statistical summaries (interquartile ranges, medians or means), which are circulated among the participants as responses of the group as a whole (statistical synthesis).
References
- Dalkey N.C., Helmer O.: An Experimental Application of Delphi Method to the Use of Experts. Manage. Sci. 9, 458-467 (1963)
- Linstone H.A., Turoff M.: The Delphi method: Techniques and applications. Addison-Wesley Publishing Company, Reading, Massachusetts, (1975)