Two experiments were devoted to examining the effect ofrequesting distributional responses rather than point estimates. Subjects were asked to furnish<br>the three quartiles for each question, that is, the number for which there is a<br>25% chance that the true answer is less, the number for which there is a<br>50% chance the true answer is less, and the number for which there is a 75%<br>chance that the true answer is less. The three were called the low, mid, and<br>high estimates, respectively. Somewhat to our surprise, the subjects had no<br>difficulty making these presumably more complex estimates.
正在翻譯中..