The Delphi method, named after the Oracle of Delphi, was developed at RAND Corporation in the 1950-1960s by Olaf Helmer and Nicholas Rescher. The objective was to forecast the effects of technology on warfare while avoiding some of the limitations of other group forecasting methods. Delphi structures the group communication process so that the group is able to tackle a complex problem while minimizing the typical biases and distortions that arise in group processes. The method stands as a good example of the benefits of structuring procedures when judgmental approaches to forecasting are used.

Early on, the Delphi method was applied to science and technology forecasting, but later was applied more broadly, especially to public policy issues, as well as to business forecasting. The method is not limited to forecasting; it has also been used for tasks such as evaluating possible budget allocations, delineating the pros and cons associated with potential policy options, distinguishing and clarifying real and perceived human motivations, and exposing priorities of personal values and social goals. In all applications, the method aims to converge on a shared view among independent experts about an outcome or a question using surveys and questionnaires over several rounds, after which of each feedback is provided.

From a methodological perspective, Delphi is a judgmental approach, using the structured judgment of experts, with feedback provided to the participants at each of several rounds. Delphi is appropriate when there is insufficient data available to apply quantitative methods or precise analytical techniques, and when large (discontinuous) changes are not expected. In this Models & Methods article, we will look at the conditions for the use of Delphi, principles for using it effectively, and its limitations.

Click the View button above to download a three-page summary of the model (PDF format), complete with a ManyWorlds commentary that discusses the effectiveness of the model.