Eighth International Conference on Forensic Inference and Statistics Program as of 24 February 2011

Preconference Short Courses

Probabilistic Reasoning for Judges and Lawyers

Statistical evidence and probabilistic reasoning today play an important and expanding role in criminal investigations, prosecutions and trials, not least in relation to forensic scientific evidence (including DNA) produced by expert witnesses. It is vital that everybody involved in criminal adjudication is able to comprehend and deal with probability and statistics appropriately. There is a long history and ample recent experience of misunderstandings relating to statistical information and probabilities which have contributed towards serious miscarriages of justice. This course is designed as a general introduction to the role of probability and statistics in criminal proceedings. It covers basic terminology and concepts, illustrates various forensic applications of probability, and draws attention to common reasoning errors. The course develops a logical narrative, starting with basic issues of terminology and concepts and then guiding the participant through a range of more challenging topics.

The course is taught by Colin Aitken, Professor of Forensic Statistics at the University of Edinburgh, who is Chairman of the Royal Statistical Society’s working group on Statistics and the Law. The Society, through this group, has just published a Practitioner Guide “Fundamentals of Probability and Statistical Evidence in Criminal Proceedings”, the first in a series of guides to be published on the general subject of ’Communicating and interpreting statistical evidence in the administration of criminal justice’. He is joined by Charles Berger, Principal Scientist at the Netherlands Forensic Institute. Dr. Berger spends much of his time on research and educational activities relating to the interpretation of evidence.

Statistical Methods for DNA Evidence

Although a match between the DNA profile of a crime scene sample and a person suspected of having left the sample can provide strong evidence linking the suspect to the scene, there is a need for care in quantifying the strength of this evidence. Evett and Weir laid out some Principles of Interpretation in their 1998 textbook “Interpretation of DNA Evidence,” which will provided to participants in this course. By basing the quantification on likelihood ratios, as endorsed by the US National Research Council in 1996 and covered in many textbooks since then, Evett and Weir can accommodate the complexities of population structure, relatedness and mixtures. The course will focus on basic principles, rather than detailed algebraic treatments, and will make reference to relevant cases. Recent developments will, however, be referred to – including Evett’s treatment of the hierarchy of propositions (Evett, Gill, Jackson, Whitaker and Champod, Journal of Forensic Sciences 47:520–530, 2002) and Weir’s treatment of lineage markers when the evidence profile is not found in a database (Buckleton, Krawzcak and Weir, Forensic Science International: Genetics 5:78–83, 2011).

Ian Evett is Consultant to the Chief Scientist of the Forensic Science Service in the United Kingdom. He is leading in the training of the members of the FSS Statistics and Interpretation Group. Bruce Weir is Professor and Chair of the Department of Biostatistics at the University of Washington. He teaches an online course in Forensic DNA Statistics for the University of Central Florida.

The Use of Bayesian Networks in Forensic Science

Forensic and legal disciplines at large can be considered as problem domains that consist of a number of entities or events among which various kinds of relationships, affected by uncertainty, are assumed to hold. These core aspects of situations of reasoning under uncertainty can be rigorously represented by Bayesian networks, that is graphical models based upon concepts from graph and probability theory. A network’s graphical structure accounts for possible relevance relationships between the various entities of the problem domain whereas probability expresses beliefs about the strengths of the assumed relationships. Newly acquired information about the problem domain (in legal contexts typically referred to as evidence) can be used to update the probability of the various specified events or hypotheses. Bayesian networks operate this updating according to Bayes’ theorem, the fundamental rule for assessing the value of evidence in forensic science. During the past 20 years, there has been a regular stream of publications on the use of Bayesian networks in forensic and legal theory and practice. These contributions converge in their opinions that Bayesian networks provide valuable assistance to their user in coping with inferential issues that are marked by uncertainty.

This course will provide basic knowledge, modeling capabilities and methodology. Specifically, this workshop intends to: familiarize participants with the basic concepts of Bayesian networks; draw attention on the ways in which uncertainty affects the coherent evaluation of scientific evidence and how this issue can be addressed by probability theory; and enable participants to recognize potential applications of Bayesian networks in their own field of expertise. Participants should plan to bring their laptop computers – software will be provided to them prior to the workshop.

The course will be taught by Franco Taroni, Professor, and Alex Biedermann, Assistant Pro- fessor, in the School of Criminal Justice at the University of Lausanne. They are coauthors of the 2006 textbook “Bayesian Networks and Probabilistic Inference in Forensic Science.”