Weight-of-Evidence for Forensic DNA Profiles (eBook)

David J. Balding, Christopher D. Steele (Autoren)

eBook Download: EPUB

2015 | 2. Auflage
John Wiley & Sons (Verlag)
978-1-118-81454-3 (ISBN)

Infos

The use of DNA evidence is widely used in the modern justice system. Statistical methodology plays a key role in ensuring that this evidence is collected, interpreted, analysed and presented correctly. This book is a guide to assessing DNA evidence and presenting that evidence in a courtroom setting.  It offers practical guidance to forensic scientists with little dependence on mathematical ability, and provides the scientist with the understanding they require to apply the methods in their work. Since the publication of the first edition of this book in 2005 there have been many incremental changes, and one dramatic change which is the emergence of low template DNA (LTDNA) profiles. This second edition is edited and expanded to cover the basics of LTDNA technology. The author's own open-source R code likeLTD is described and used for worked examples in the book. Commercial and free software are also covered.

Autor

David J. Balding, University of Melbourne, Australia; and University College London, UK. Christopher D. Steele, University College London, UK.

Textauszug

1
Introduction

1.1 Weight-of-evidence theory

The introduction of DNA evidence around 1990 was a breakthrough for criminal justice, but it had something of a ‘baptism of fire’ in substantial controversy in the media and courts over the validity of the technology and the appropriate interpretation of the evidence. DNA profiling technology has advanced since then, and understanding by lawyers and forensic scientists of the appropriate methods for evaluating standard DNA profile evidence has also improved. However, the potential for crucial mistakes and misunderstandings remains. Although DNA evidence is typically very powerful, the circumstances under which it might not lead to satisfactory identification are not widely appreciated. Moreover, new problems have arisen with low-template DNA (LTDNA) profiles, which can be subject to stochastic events such as drop-in and drop-out.

The report of Caddy et al. [2008] was commissioned by the UK Government in response to the controversy over the 2007 acquittal of a defendant charged with the 1998 Omagh bombing in Northern Ireland. It found the underlying science to be ‘sound’ and LTDNA profiling to be ‘fit for purpose’, while admitting that there was lack of agreement ‘on how LTDNA profiles are to be interpreted’. We find those phrases to be mutually incompatible. Fortunately, much progress has been made since 2008, but the international controversy surrounding the legal process arising from the murder of Meredith Kercher in Perugia, Italy, in 2009, in which LTDNA evidence played a central role, highlights the challenges that can arise. We aim in this book to present the fundamental concepts required for interpretation of DNA profiles, including LTDNA. We will initially focus on the general issues concerning the measurement of evidential weight, develop the weight-of-evidence theory based on likelihoods and discuss some alternative probability-based approaches. We will then apply the theory to forensic DNA profiling.

The primary goal of this book is to help equip a forensic scientist charged with presenting DNA evidence in court with guiding principles and technical knowledge for

• the preparation of statements that are fair, clear and helpful to courts, and
• responding to questioning by judges and lawyers.

The prototype application is identification of an unknown individual whose DNA profile was recovered from a crime scene, but we will also discuss profiles with multiple contributors, as well as paternity and other relatedness testing, and consider profiles that are subject to drop-out and other consequences of LTDNA and/or degraded DNA. We assume the setting of the United States, the United Kingdom and Commonwealth legal systems in which decisions on guilt or innocence in criminal cases are made by lay juries, but the general principles should apply to any legal system.

We will introduce and develop a weight-of-evidence theory based on two key tenets:

1. The central question in a criminal trial is whether or not the defendant is guilty.
2. Evidence is of value inasmuch as it alters the probability that the defendant is guilty.

Although these tenets may seem self-evident, it is surprising how often they are violated. Focussing on the right questions clarifies much of the confusion that has surrounded DNA evidence in the past.

It follows from our tenets that evidential weight can be measured by likelihoods and combined to assess the totality of the evidence using the appropriate version of Bayes' theorem. We will discuss how to use this theory in evaluating evidence and give principles for, and examples of, calculating likelihoods, including taking into account relevant population genetic factors.

No theory ever describes the real world perfectly, and forensic DNA profiling is a complex topic. The weight-of-evidence theory developed here cannot be applied in a naive, formulaic way to the practical situation faced by forensic scientists in court. Nevertheless, a firm grounding in the principles of the theory provides:

• the means to detect and thus avoid serious errors;
• a basis for assessing approximations and simplifications that might be used in court;
• a framework for deciding how to proceed when the case has unusual features;
• grounds for deciding what information a clear-thinking juror needs in order to understand the strength of DNA profile evidence.

Fortunately, we will see that the mathematical aspects of the theory are not too hard. Of course assessing some of the relevant probabilities—such as the probability that a sample handling error has occurred—can be difficult in practice, reflecting the real-world complexity of the problem. Further complications can arise, for example, in the case of mixed DNA samples (Section 6.5). However, the same simple rules and principles can give useful guidance in even the most complex settings.

There exist other theories of weight of evidence based on, for example, belief functions or fuzzy sets. The theory based on probability presented here is the most widely accepted, and its philosophical underpinnings are compelling [Bernardo and Smith, 2009, Good, 1991]. So whatever is actually said in court in connection with DNA evidence, it should not conflict with this theory.

There has been debate about the appropriateness in court of using numbers to measure weight of evidence. We only touch on this argument here (Sections 6.3.4 and 11.4.5). It is currently almost universal practice to accompany DNA evidence by numbers in an attempt to measure its weight (but see Section 11.4.6), and so we focus here on issues such as which numbers are most appropriate in court, and how they should be presented.

1.2 About the book

Chapters 2, 3 and 11 are not scientifically technical and, for the most part, are not specific to DNA evidence. We therefore hope that lawyers dealing with scientific evidence, and forensic scientists not principally concerned with DNA evidence, will also find at least these chapters to be useful. Courtroom lawyers ignorant of the weight-of-evidence theory described in Chapters 2 and 3 should be as rare as theatre critics ignorant of Shakespeare, yet in reality, we suspect that few are able to command its elegance, power and practical utility.

We first set out the weight-of-evidence theory informally, via a simplified model problem (Chapter 2) and then more formally using likelihoods (Chapter 3). In Chapter 4, we briefly survey DNA-based typing technologies, starting with an introduction to autosomal1 short tandem repeat (STR) typing, emphasizing possibilities for typing error, then moving on to other DNA typing systems, digressing briefly to discuss fingerprint evidence and finishing with some newer evidence types: methylation, RNA and phenotype and ancestry prediction from DNA. Next, we survey some population genetics theory relevant to DNA profile evidence (Chapter 5). These two chapters prepare us for calculating likelihoods for DNA evidence, which is covered in Chapters 6 (identification) and 7 (relatedness). In Chapters 8 and 9, we extend identification inferences to LTDNA profiles and give a brief introduction to our freely available software for LTDNA profile evaluation, likeLTD. Chapter 10 discusses some alternative probability-based approaches to assessing evidential strength: none of these methods is recommended but each has its merits. In Chapter 11, we discuss some basic fallacies in the evaluation of DNA profile evidence and briefly review the opinions of some UK and US legal and scientific authorities.

1.3 DNA profiling technology

For the most part, we will assume that the DNA evidence is summarised for reporting purposes as the lengths of STR alleles at multiple autosomal loci (typically 10–25), reported as the number of tandem repeats of a DNA motif, usually 4 base pairs (bp) in length. The final result at four of the loci might be reported as

STR locus:

D18

D21

THO1

D8

Genotype:

14,16

28,31

9.3,9.3

10,13

in which each pair of numbers at a locus indicates the numbers of repeats on the individual's two alleles. Although whole repeats are the norm, partial repeats sometimes occur (Section 4.1.1); the profile represented here is homozygous for a THO1 allele that includes a partial repeat: the 9.3 indicates 9 full copies of the 4 bp DNA motif plus a 3 bp partial repeat.

Autosomal STRs now form the standard DNA typing technology in many countries. Both the 13-locus Combined DNA Index System (CODIS) and 15-locus European standard set (ESS) were derived by expanding earlier, smaller sets of STRs, and both have been superseded by larger sets. Most of Europe currently uses a 16-STR system developed from the ESS, which includes the amelogenin sex-identifying locus. The GlobalFiler® set from Life Technologies-Applied Biosystems offers 22 STRs, amelogenin and a Y-indel marker [Life Technologies, 2014], while the PowerPlex® Fusion System from Promega offers 23 STRs and amelogenin [Promega, 2014]. The different systems have many loci in common, but...

Produkt-Details

• Erscheint lt. Verlag: 18.5.2015
• Reihe/Serie: Statistics in Practice
• Reihe/Serie: Statistics in Practice
• Reihe/Serie: Statistics in Practice
• Sprache: englisch
• Themenwelt: Mathematik / Informatik ► Mathematik ► Statistik
• Themenwelt: Mathematik / Informatik ► Mathematik ► Wahrscheinlichkeit / Kombinatorik
• Themenwelt: Medizin / Pharmazie ► Gesundheitsfachberufe
• Themenwelt: Studium ► 2. Studienabschnitt (Klinik) ► Rechtsmedizin
• Themenwelt: Technik
• Schlagworte: biometrics • Biometrie • Biowissenschaften • Criminal Law • Evidence, Forensics, DNA profiling, statistical analysis, legal system • Forensic Science • Forensik • Law • Life Sciences • Rechtswissenschaft • Statistics • Statistik • Strafrecht
• ISBN-10: 1-118-81454-1 / 1118814541
• ISBN-13: 978-1-118-81454-3 / 9781118814543