Bitte wählen Sie ihr Lieferland und ihre Kundengruppe
A report of results of experimentation on the effectiveness of Delphi procedures, which incorporate anonymous response, iteration and controlled feedback, and statistical group response to elicit and refine group judgments where exact knowledge is unavailable. In spring 1968, Rand conducted 10 experiments using over 150 university students. Questions related to almanac-type information. Results showed that controlled feedback, compared with face-to-face discussion, improved the accuracy of group estimates, thus validating the use of Delphi techniques in areas of partial information. Insight was gained into group information processes. A meaningful estimate of the accuracy of a group response to a given question can be obtained by combining individual self-rating of competence on that question into a group rating. Adding this result to an observed relationship between accuracy and standard deviation makes it possible to attach accuracy scores to the products of a Delphi exercise. (Author)