Delphi method

The Delphi method (also Delphi study, Delphi method or Delphi survey called ) is a systematic, multi-stage survey process with feedback and an estimation method that is used to the like can assess future events, trends, technological developments and as well as possible.

The namesake of the method is the ancient oracle of Delphi, which his listeners provided advice for the future.

History

The Delphi method was - developed by the American RAND Corporation in 1963 and has since been frequently when applied in a different form, for the determination of forecasts / trends and for other opinion formation in the context of system tasks - building on work done in the late 1950s. More and more, the procedure has developed into an evaluation process for subjects in which it can be determined whether there is a consensus on the issue (or whether this can be achieved ) or not. In Germany it was in the 90s, the then Federal Ministry for Research and Technology (BMFT ), which was the first Delphi studies on the development of science and technology in order. The German Delphi studies report on the development of science and technology (1993) and Delphi '98 survey. Future demand. Study on the global development of science and technology (1998) were carried out by the Fraunhofer Institute for Systems and Innovation Research (ISI). However, the validity of the results of Delphi studies is very limited. Typical are the very general statements on meta level ( eg: ' Sustainable energy solutions will become increasingly important ')

Method

In a Delphi survey of a group of experts will be presented with a question or thesis catalog of the respective scientific community. Respondents have the opportunity to assess the theses in two or more so-called rounds. From the second round of feedback is given, as other experts who responded anonymously to the rule. In this way, an attempt is made to counteract the usual group dynamics with very dominant people.

The writing obtained in the first round responses, estimates, results etc., are therefore listed and, for example, combined with a special averaging, percentiles or average value calculations and anonymously submitted to the experts again for a further discussion, clarification and refinement of the estimates. This controlled process of opinion formation usually takes place over several stages. The end result is an edited opinion group that contains the statements for themselves and information about the range of existing opinions.

The opinion formation process includes the following elements: generation, correction or partial adaptation or refinement, averaging or thresholding, often open fields for explanations. Interferences are eliminated by anonymization, the compulsion to writing and individualization. The strategy of the Delphi method consists of: concentration on the essential, multi-stage, partially backcoupled editing process and safer, broader statements by allowing statistical fuzzy -like results. A common problem is that the experts do not even change their opinion expressed in the following rounds despite anonymity, so that the additional benefit of further rounds is often small.

As a supplement to the Delphi method, for example, the cross-impact matrix method can be used. To elements of the Delphi method can also be found in the D2 method again. Combinations of scenarios are now also tested. From Delphi results, simple roadmaps can be derived, so that this combination is becoming increasingly popular.

One can find various forms of the Delphi method, which vary the method of estimation of something: the standard and the broadband method are some of the variants. Meanwhile, most of the procedures are carried out electronically. Real-time Delphi method ( with an immediate feedback of results) are a variant that is only possible electronically.

Standard Delphi method

In the standard Delphi method several experts to estimate a project - used, which may not coordinate with each other - or to forecast. The process is as follows:

A project manager prepares a project before, in which the individual sub - products are listed and prepares them in a single form.
The project presents the goals of the overall project and distributed one copy of the work be mailed to each expert. There is no discussion of the estimates.
Each expert estimates the work packages contained in the working form. None of the experts working with another expert.
All work forms will be collected and analyzed by the project manager.
If there are serious discrepancies so be uniformly commented on this by the project manager on all labor forms in terms of the deviation up or down. Each work form then goes to its original processor back.
The experts think as a function of the comments their estimates.
The loop described above is repeated until independently of each other ( within a tolerance range ) adjusts consensus in the estimates.
Of all the estimates, the mean values are calculated and presented as the final estimate.

The absence of any discussion has two aspects that a project manager must review the one hand is thus prevented that emerge due to an unwanted group dynamic trends and tendencies in the opinions that prevent good estimates under certain circumstances. On the other hand, group discussions could help to avoid deficits in the individual expert know -how and related misperceptions.

Frequently Delphi surveys are conducted in writing and separately, the questionnaire will be sent to the experts so by letter or email. The individual experts never see and know only after completion of all survey rounds, the names of the other respondents. This approach is more reliable than the gathering of all the experts in one room. If the final report once before, all the experts and other interested parties are invited to a symposium in general.

Wideband Delphi method

In the Wideband Delphi method several experts are used to estimate a project that can coordinate with each other. The process is as follows:

A project manager prepares a project before, in which the individual sub - products are listed and prepares them in a single form.
The project presents the goals of the overall project and distributed one copy of the work be mailed to each expert. There will be a discussion of the work packages among the experts, in the view of the individual experts are the other participants will be provided in relation to the overall project and the sub-tasks.
Then each expert appreciates the work packages included in his work form. None of the experts while working with another expert.
The project manager summarizes the single estimate of statements, but it establishes the data and differences do not. The results are distributed to all experts.
The project manager convenes together a new meeting with the experts and speaks to the largest discrepancies in the estimates. Each work form then goes to its original processor back.
The experts think as a function of the deviations listed their estimates.
The loop described above is repeated until independently of each other ( within a tolerance range ) adjusts consensus in the estimates.
Of all the estimates, the mean values are calculated and presented as the final estimate.

Through the interactions of experts mutually different views are taught what a consensus formation accelerated. Advantage of this method is on the one hand, the anonymity of the estimates: the experts are not confronted with their serious deviations of the estimates and may thus influence the estimated expenses in their favor. Strong deviations from average values are disclosed. Disadvantage of this method is the risk of formation of opinion by the group dynamic in which a necessary could have serious estimation variance subject to peer pressure. Another disadvantage is that due to several iteration loops for the opinion expressed by the total estimated effort can be quite extensive. The Wideband Delphi method is a useful technique for estimating large projects where complex architectures can lead each other to realistic values by a large panel of experts with the help of the interaction of the experts.

Criticism

The Delphi method attempts to reduce misperceptions of experts through multi-stage, sometimes applied to consensus design. Still can not avoid all the problems of the expert survey; by interviewing more than one person is subject to other restrictions.

The main criticism is the basic assumption that experts have insights that go beyond the industry standard and can be generated in the future knowledge combination. However, this assumption is not provable. On the contrary, many experts statements later turned out to be false. For example. was the saying of Thomas J. Watson led, the early head of IBM, who declared in 1943: "I think there is a world market for maybe five computers."
For the selection of experts, there are no criteria, the term " expert" is purely subjective and arbitrary, and is done by the selection of participants already have an influence on the result.
Themes and theories need to be formulated before they can undergo the two-step process. In some cases, theses, although received in the procedure itself, as a rule, however, other methods are needed to achieve this.
The theses must have a short, concise, but clear. This can be an advantage, but it forces the participants to concentrate on the essentials. Methodologically complex subject matter can be evaluated but only conditionally.
Experts focus by definition essentially on their expertise area. The interdependencies with other developments, which are important in large-scale studies esp. are often neglected or must be reworked.
If relevant conditions considered ( social developments in the prognosis of the technical development of mobile communications, for example ), so you can not rely on the fact that the respondents said this same reliable expertise possess as in their actual field of expertise.
Experts tend to overestimate the speed of developments. In particular, the diffusion rate of an innovation in society is overestimated quickly.
When interviewing a group creates a social situation. This can by authority, or because of personal struggles grave, distortion. So is not always clear whether a consensus (or disagreement ) is actually based only on the intensive questioning their own opinion. Anonymised in the feedback session may iA these problems not be completely avoided. When using questionnaires ( e- mail or post ) the Delphi method is explicitly used to circumvent this dominance. In presence rounds this is only possible under certain conditions.

Survey methodology Fuzzy logic Cross impact analysis Technology roadmap

226796