2. Scientific Reasoning and Decision Making

  1. Franco Taroni1,
  2. Silvia Bozza2,
  3. Alex Biedermann1,
  4. Paolo Garbolino3 and
  5. Colin Aitken4

Published Online: 9 APR 2010

DOI: 10.1002/9780470665084.ch2

Data Analysis in Forensic Science: A Bayesian Decision Perspective

Data Analysis in Forensic Science: A Bayesian Decision Perspective

How to Cite

Taroni, F., Bozza, S., Biedermann, A., Garbolino, P. and Aitken, C. (2010) Scientific Reasoning and Decision Making, in Data Analysis in Forensic Science: A Bayesian Decision Perspective, John Wiley & Sons, Ltd, Chichester, UK. doi: 10.1002/9780470665084.ch2

Author Information

  1. 1

    School of Criminal Justice, University of Lausanne, Switzerland

  2. 2

    Department of Statistics, University Ca' Foscari, Venice, Italy

  3. 3

    Faculty of Arts and Design, IUAV University, Venice, Italy

  4. 4

    School of Mathematics, University of Edinburgh, UK

Publication History

  1. Published Online: 9 APR 2010
  2. Published Print: 9 APR 2010

ISBN Information

Print ISBN: 9780470998359

Online ISBN: 9780470665084

SEARCH

Keywords:

  • Bayesian networks;
  • decision making;
  • probabilistic model building;
  • scientific reasoning;
  • statistical inference

Summary

In this chapter arguments are given for justifying the acceptance of the probability calculus as a standard of coherence for degrees of belief, and the acceptance of the principle of conditionalization as a rule for coherent updating of degrees of beliefs. Informal discussion is also given as to how this concept of coherent reasoning under uncertainty can be extended to scientific thinking in general and to statistical and forensic science inferences in particular. Bayesian networks can be naturally extended to represent decision problems by adding decision nodes and utility nodes. Decision nodes have no probability tables associated with them, because it is not meaningful to assign probabilities to a variable under the control of the decision-maker, and arrows pointing to decision nodes do not carry quantitative information; they only indicate that the state of the decision node’s parents is known prior to the decision.

Controlled Vocabulary Terms

Bayesian statistics; decision making; inferential statistics; probability sampling