Developing the Technology of Probabilistic Inference: Aggregating by Averaging Reduces Conservatism.

Abstract

A relatively large body of research indicates that people are conservative processors of probabilistic information. Recent attention has focussed on two possible explanations of this phenomenon. The misaggregation hypothesis depicts conservatism as an inability to properly combine the information in sequence of data. The other explanation suggests conservatism is the result of a response bias: the avoidance of extreme odds or probability judgments. Two experiments explored the use of a specific response, average certainty that was devised to thwart conservatism caused by either a response bias or a certain form of misaggregation. Use of appropriate instructions and response scales made the average certainty judgments good subjective assessments of the arithemetic mean likelihood ratio which could then be used in the appropriate form of Bayes' Theorem to calculate posterior odds. These judgments seemed unlikely to be affected by a response bias since extreme responses were not needed. In addition, research has suggested that people are more likely to aggregate information by averaging than by adding or multiplying, so misaggregation may be exhibited only in specific forms of aggregation and may not be present in averaging.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Aug 01, 1977
Accession Number
ADA048569

Entities

People

  • David A. Seaver
  • Lee C. Eils Iii
  • Ward Edwards

Tags

Communities of Interest

  • Biomedical
  • Human Systems

DTIC Thesaurus Topics

  • Biological Sciences
  • California
  • Contracts
  • Department Of Defense
  • Engineering
  • Human Factors Engineering
  • Information Processing
  • Information Systems
  • Military Research
  • Operations Research
  • Probability
  • Psychology
  • Regression Analysis
  • Slope
  • Social Sciences
  • Systems Science
  • United States

Fields of Study

  • Psychology

Readers

  • Adaptive Control and Estimation with Uncertainty in Dynamic Systems.
  • Artificial Intelligence
  • Systems Analysis and Design

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference