Dec 31, 1969
An investigation of the effect on group accuracy of two variations in the Delphi procedures. In these exercises, 20 general information questions are answered by two groups of respondents, who, after receiving some form of feedback, may revise their answers. In the first variation, feeding back individual percentiles resulted in no improvement over feeding back the median and quartiles of the group response. On the other hand, in the second variation, adding a relevant fact to the median and quartiles information resulted in a statistically significant increase in numerical accuracy. The number of changed answers was also greater, suggesting that introduction of a relevant fact strengthens motivation for revision. For a number of military concerns, such as long-range technological development assessment or future threat evaluation, the expertise of a group of decisionmakers is relied on. The Delphi studies are an effort to improve such judgments through refined procedures.