Evaluating evidence in linked crimes with multiple offenders

Area of Science:

• Forensic Science
• Probability Theory
• Legal Reasoning

Background:

• Previous work established a Bayesian network framework for evaluating evidence with a single suspect in multiple offenses.
• Current legal and forensic practices often struggle with complex scenarios involving multiple unknown offenders.

Purpose of the Study:

• To extend the Bayesian network framework for evaluating evidence in cases involving multiple offenders.
• To address new questions regarding offender distinguishability, evidence source attribution, and offender count.
• To demonstrate the impact of these factors on posterior probabilities using a mock case.

Main Methods:

• Implementation of a Bayesian network framework.
• Construction of appropriate Bayesian networks for different multi-offender scenarios.
• Analysis of a mock case example to illustrate differences in conclusions.

Main Results:

• Subtle differences in multi-offender situations lead to substantially different posterior probabilities.
• Bayesian networks effectively model the influence of various pieces of evidence on hypotheses.
• The framework helps identify and prevent potential pitfalls in evidential evaluation.

Conclusions:

• Bayesian networks are valuable tools for guiding expert and legal reasoning in complex forensic cases.
• The extended framework enhances understanding of evidential value and hypothesis probabilities with multiple offenders.
• While direct court presentation is cautioned against, Bayesian networks offer significant analytical benefits.