INFERENCES ON THE BASIS OF CONDITIONALLY NONINDEPENDENT DATA.

Abstract

The experiment was concerned with decision making in a tactical threat situation (simulated). More specifically, it was concerned with the ability of subjects to estimate the likelihood of hypotheses concerning the 'adversary's' intentions being true. The data, presented to the subjects concerning the enemy, were internally interrelated in a probabilistic sense. The purpose of the study was to evaluate the subjects' ability to exploit these interrelationships in making their estimates concerning the probable intentions of the enemy. The subjects' likelihood estimates as to the enemy's intent were compared with those computed using Bayes' theorem--a mathematical technique that, theoretically, makes maximum use of the data, their interrelationships and their relation to the hypotheses. In general, the estimates made by well-trained subjects agreed closely with those calculated from Bayes' theorem. The subjects apparently were able to exploit the probabilistic interrelationships in the data. Furthermore, the subjects' estimates reflected neither the conservatism nor the 'recklessness' suggested by other previous experiments. (Author)

Document Details

Document Type
Technical Report
Publication Date
Dec 01, 1965
Accession Number
AD0630662

Entities

People

  • David A. Schum

Organizations

  • Ohio State University

Tags

DTIC Thesaurus Topics

  • Conservatism
  • Hypotheses
  • Political Ideologies
  • Political Science

Readers

  • Military History / Militaries and War Studies
  • Statistical inference.
  • Systems Analysis and Design

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference