Intuitive Prediction: Biases and Corrective Procedures
Abstract
This paper presents an approach to elicitation and correction of intuitive forecasts, which attempts to retain the valid component of intuitive judgments while correcting some biases to which they are prone. This approach is applied to two tasks that experts are often required to perform in the context of forecasting and in the service of decision making: the prediction of values and the assessment of confidence intervals. The analysis of these judgments reveals two major biases: non-regressiveness of predictions and overconfidence. Both biases are traced to people's tendency to give insufficient weight to certain types of information, e.g., the base-rate frequency of outcomes and their predictability. The corrective procedures described in this paper are designed to elicit from experts relevant information which they would normally neglect, and to help them integrate this information with their intuitive impressions in a manner that respects basic principles of statistical prediction.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 1977
- Accession Number
- ADA047747
Entities
People
- Amos Tversky
- Daniel Kahneman