Comparison of Group Judgment Techniques with Short-Range Predictions and Almanac Questions
An experiment designed to discover whether the results of laboratory studies dealing with general (almanac) information are relevant to the applied case when the true answer is unknown. Using short-range prediction questions as subject matter, the experiment indicates that, in general, Delphi procedures are at least as effective with short-range prediction as they have been for almanac material. Eight groups, of about 20 each, of upperclassmen and college graduates were given short-range prediction questions to answer in a two-round Delphi exercise. Satisfactory answers were obtained for 32 of the 40 questions. Correlations between standard deviation and accuracy, and between group self-rating and accuracy, were significantly higher for the prediction than for the almanac questions. Half the groups generated estimates of the three quartiles of the distribution; the other half generated point estimates. No significant difference was observed between these two kinds of estimates.