法庭科学的表述与法律证明

500··Evidence Science V ol.22 No.4 2014

法庭科学的表述与法律证明

［澳大利亚］ 安德鲁•立格特伍德 著 汪诸豪 译*

【摘 要】 法庭科学评价意见在法庭上陈述的方式应当与法官所要求、陪审团实际应用的刑事证明程

序相兼容。这并非一个数值性的归纳过程，而是在探求排除合理怀疑证明标准下的“最佳解释推理”。面

临的问题并不是控方主张的数学概率问题，而是在全面考量了法庭上出示的所有证据后，控方主张是否

为唯一可解释假说的问题。为确保陪审团仍然能专注地运用这一法律证明标准，笔者主张，控方提出的

评价性法庭科学证据不应当以似然比的形式在强调若控方主张为真便更有可能认定证据，而应聚焦于本

方证据对辩方有利的解释范围并在被告被定罪之前排除掉所有这些解释的合理可能性。

【关键词】 法庭科学证据；法律证明；排除合理怀疑证明标准

【中图分类号】D915.13

【文献标识码】A

【文章编号】1674-1226（2014）04-0500-11

Forensic science expressions and legal proof. Andrew Ligertwood1 Translated by Wang Zhuhao2,3.

1. Emeritus Fellow in Law, The University of Adelaide. 2. Collaborative Innovation Center of Judicial Civilization, Beijing 100088; 3. Key Laboratory for Evidence Science (China University of Political Science and Law), Ministry of Education, Beijing 100088.

【Abstract 】Evaluative forensic science opinions should be presented in a form that can be accommodated within

that process of criminal proof demanded by judges and applied by juries. This is a non-mathematical inductive process

that seeks ‘the inference to best explanation’ to a standard of proof beyond reasonable doubt. The question posed is

not the mathematical probability of the prosecution hypothesis but whether, having regard to all the evidence properly

before the court, the prosecution hypothesis is the only explicable hypothesis. To ensure that a jury remains focused on

applying this legal standard it is argued that evaluative forensic science evidence tendered by the prosecution should be expressed not as a likelihood ratio that emphasizes the greater likelihood of finding evidence if the prosecution’s

contentions are true, but in a form that emphasises the ranges of explanations for the evidence consistent with the

defence case and that must be excluded as a reasonable possibility having regard to all the evidence before the accused

can be convicted.

Forensic science evidence, Legal proof, Beyond reasonable doubt【Key Words】

*本文系立格特伍德教授于2014年7月31日在第四期“证据法学”高级研讨班上的讲演稿。

安德鲁·立格特伍德, 澳大利亚阿德莱德大学法学院荣誉资深研究员，中国政法大学证据科学研究院外国专家咨询委员会成员，国际证据科学协会副主席。

汪诸豪, 国家“2011计划”司法文明协同创新中心证据法学创新团队成员、中国政法大学证据科学研究院讲师。

2014年第22卷（第4期）501

··

一、引言

本文旨在聚焦出示法庭科学证据的法律语境；解释普通法是如何看待审判过程中法庭科学家的

1角色以及普通法对法庭科学家在表述证据时的要求。 在“R 诉T ”案2中，在对所用数值缺乏明确实

证论证的情况下，审理法院否决了一位法庭科学家使用数值似然比来推算其评估意见。此判决一出，法庭科学家的角色定位在澳大利亚被推上了风口浪尖，引起了业界的强烈反应。一群法庭科学家和律师站出来回应称，似然比是法庭科学家推算并表述证据性依据的最合理方法。3 该比率表达了争议中的法庭科学证据（即是否检控方而非被告方的论点为真）之认定可能性。

自从DNA 证据被允许作为似然比起，其以一种“惊人的方式”4偏向检控方，在缺乏对DNA 证据可靠性能产生质疑的证据情况下，陪审团似乎除了判定被告有罪之外别无选择。作为一名律师，我对法庭科学证据在法庭上的表述心存顾虑。

毫无疑问，绝大多数案件中的统计结论是正确的，即便该结论未经陪审团应用数学概率规则计算得出，然而，我仍然要说，这种结论的得出未运用法律上所要求的推理过程，即没有经历刑事案件中的证明决策过程。在我看来，似然比的概念与刑事案件中的法律证明概念二者难以兼容。

我的论证范围将限制在刑事案件中的法律证明。

概括来说，我认为刑事证明应要求检控方在法庭上出示的所有证据被证明到排除合理怀疑的程度。这就要求陪审团认真考量所有证据，认定其与检控方的主张一致，并排除证据可被解释为被告无辜之假设的任何合理可能。

就检控方的主张是否正确来认定证据之可能性不是决定性的刑事证明，从法庭上合理出示的所有证据中排除掉被告清白的任何合理可能才是决定性的。为了确保陪审团能完成这项任务，我建议，非同于以似然比的方式在法庭上呈现来支持检控方，法庭科学证据应该以与检控方主张一致的方式在法庭上提出，并应通过表明该证据所含被告无辜之可能性来对其证明力进行解释。如此，陪审团能在“排除合理怀疑”地认定被告有罪之前，恰当地排除掉其面前所有证据中各种关于被告可能无辜的解释。

我是从务实而非哲学或规范性角度来谈自己的主张。在哲学和规范性层面，有关证明性质的问题就如同知识的本质问题一般复杂和富有争议。而我的出发点是，普通法上有一套刑事案件证明的特殊路径，所以当务之急是法庭科学家们能够按照法律的要求来提供他们的证据。在此，我并不是要论证法律规定的合理性，也不是在提倡我们当前的法律程序没有进一步完善的空间。我此刻只是在主张，当前普通法上要求陪审团成员和法官以某种特定方式去进行刑事证明，因此法庭科学证据需要符合（或者说是要融入进）这套特定的证明方式。

二、普通法刑事证明的性质

刑事案件中法律证明的本质是什么呢？看似有两种基本路径。

一方面，有些人认为，不存在所谓的知识确定性，法律证明和其他任何证明类型一样，都是基于证据来评估可能性的问题，而这些可能性可以数值的方式进行表述，通过数学概率逻辑规则进行1 See Ligertwood, Can DNA evidence alone convict an accused?， 33 Sydney Law Review，487 (2011). Also see Ligertwood and Gary Edmond entitled, Expressing evaluative forensic science opinions in a court of law， published in 2012 in Law and Science. 这两篇文章中能查找到进一步的参考素材。

2［ 2010］ EWCA Crim 2439（2010年10月26日）。

3 特邀编辑：《表述评价意见：立场声明》《科学与正义》2011年第51卷第1页。，

4 沃克SC 先生在“福布斯诉女王”［2010］ HCATrans 120（2010年5月18日）案中使用了该词。

502··Evidence Science V ol.22 No.4 2014评估。当可能性无法通过实证进行精确测算时，事实认定者们可以对评估后的可能性进行主观认定，以判断是否达到了法律上所要求的证明程度。因此，证明可以被定义为是一种精确的逻辑，最终可以用数值进行表述，所需要达到的证明程度可以根据错判将会造成的后果严重程度来进行设定。

另一方面，也有许多人认为，确定性或许的确不可能实现，但法院从来没有将法律证明视同为一种可以用数值来进行终极表述的理念。更确切地说，证明被视作为一种事实认定者对所有呈现在法庭上的证据进行合理解释的探求过程。鉴于法院受理案件通常涉及的是对世界上发生事件的决断，证明就是对能解释法庭上出示证据的某种诉说、描述、故事之探求过程。定罪所要求证明到的程度同样取决于后果的严重性，但非数值可以表述。在民事案件中，证明过程是对某种大于50%可能性解释的探求。在刑事案件中，这种对（证据）解释的探求以事实认定者能够排除合理怀疑作为终结。

在此，我的探讨并不涉及哪种证明路径为“正确”或者更佳，我只想强调，在普通法法庭上，上述第二种路径是通行的做法。证明标准并不以计数的方式进行定义。但更为重要的是，虽然法院也许允许专家法庭科学证据以某种统计的方式提出（至少，根据“R 诉T ”案，当该法庭科学证据被视为具有充分的实证基础时），但法院并不一定会认同该些数据统计与案件中其他证据相关联的逻辑。

最为明确的是，法院并不乐衷于使用数值来表达该逻辑，拒绝在刑事案件中指示陪审团数学规则或可用于计算出示证据的数值效力。当证据被允许以某种数值似然比表述时，法院并不会指示陪审团该如何运用贝叶斯定理来计算该证据对于被告犯有指控罪行可能性方面的效力。人们会问，在这些情况下，陪审团又如何能理性地评价似然比的效力呢？

在民事和刑事案件中，证明的路径看似就是在寻找呈现在法庭上证据之“最佳解释”的过程。有人或许能够证明，民事案件中的证明路径可以演变为支持某种最终可用数值表述的逻辑。法院的确在谈有关满足某种可能性衡平检验，一种显示为假定超过50%数值可能性的标准。而且，在案件审理目标仅为解决普通民众间争议时（当案件错判的后果为同等程度的损失时），该数值标准可能是广为接受的。

但是在刑事案件中，将证明标准转化为某种最终可用数值进行表述的逻辑是非常困难的。该标准被表述为需要达到排除合理怀疑。陪审团被要求在判定被告有罪之前排除所有合理可能的被告清白解释。这在本质上是对确定性的一种美好愿望，法院在试图寻求避免判定无辜之人有罪这种无法容忍的可能性。若转化为数字化标准将等于承认这种错判可能性的存在。正是基于这个原因，法院不愿意在这条路径上走下去。（当然，在事实认定者为陪审团或为未经数学分析训练的法官时，或者在将数值分配对应给相关的可能性缺乏精确实证基础时，采用数值路径还有其他明显的实践操作问题。）

三、审判中法庭科学证据的表述

以下是法庭科学家和其他专家被召集出庭作证的背景情况。他们的任务不是要对事实下结论性判断，那是法官或陪审团的工作。反之，他们的角色是要辅助法官或者陪审团达到法律上所要求的确信状态。这种辅助可以通过三种方式实现。第一，提供需要专业知识才能作出的观测证据。其次，给事实认定者补充知识，以便其能够从法庭上出示的特定证据中得出更为明智的推理决策。第三，基于其专业知识来对特定现有证据的推理结论强度来提供评价意见（法院有权拒绝接受此类证据，正如其有权拒绝接受任何由专家作出的评价意见）。

上述第三类辅助方式——就推理结论提供评价意见——最具争议性，尤其是当证据以似然比形式提出用以支持刑事案件中的检控方时。即便法庭科学家可能会严谨地强调他们仅仅是在就检控方的某种特定假设基础上之特定证据认定可能性来提供意见，重大风险是负责判断是否认定该假设得到证明的陪审团会将认定证据的可能性与认定该假设的可能性相等同。

因此，存在风险的是，有关证据中所能得出之结论的专家意见有可能会被陪审团误解，或是会在不提出其他任何问题的情况下被采纳作为具有强烈证明力的有罪性主张证明，或是至少在不考虑那些

2014年第22卷（第4期）503

··

专家未被要求考察的法庭上其他证据情况下就采纳来作为证明。有心理研究5表明，与应用贝叶斯逻辑所得出的证明力相比，陪审团往往在主观上倾向于给以似然比形式呈现之证据更高的证明力。6

虽然陪审团过高评价法庭科学证据之逻辑效力的风险可能被视作不高，而且法官可以指示陪审团最终的事实推理应由陪审团本身而非由专家来决断，但是必须要牢记的是刑事定罪量刑标准要求排除合理怀疑的证明，且这一过程必须要确保陪审团在达到满意的程度之前认真地排除掉所有合理怀疑的可能性。刑事审判的主要目的就是要确保严格地应用这一刑事证明标准，在合理可能的范围内尽量消除错判无辜者的风险。

显然，控制犯罪还有其他更为奏效的方法，但位于刑事司法之普通法体系中心位置的是对于无辜者的保护。在此大背景下，如果存在表述评估性法庭科学意见的其他方式，能更好地确保陪审团将工作集中在排除一切合理可能的清白假说上，那么依赖于陪审团指示就显得不合适了。

主张拒绝似然比方式表述的另一种回应是说，对抗制普通法审判鼓励每一位争辩者以最具说服力的方式来提供证据，而对抗方的角色就是通过交叉询问检控方证人并召集其他证人出庭作证等方式来反驳该证据，为其辩护主张提供有说服力的支持。

即便假定这些对抗性保护措施是有效存在的，我认为其仍然扭曲了刑事案件普通法审判的本质。这是控告方问责式审判，而非对抗制审判。检控方的控告必须要保证排除合理怀疑，而被告方需要获得充分的机会来对该指控和指控所基于的证据进行争辩。但是，检控方的职责是要通过合理方式（如通过有逻辑连贯性的论证）来说服事实审理者接受其指控；而不是通过那些能搅动事实审理者情绪或使其分心于理性工作的说辞来说服他们。

正如上文提及的心理学研究所建议的，相比于逻辑推导出的证明力客观值，普通人可能会在主观上给予似然比更高的证明价值，这种风险是实实在在存在的。绝对重要的是，在提出证据时应明确要求应用刑事证明标准。而在（似然比）这种模式下，无辜解释的可能性依旧存在，但却主张可基于在法庭上出示的其他证据来排除掉这些无辜可能性。

在此背景下，我只想主张，法庭科学家应该准备好不仅仅在法庭上称当检控方的解释比被告方的清白解释更为合理时，某证据就应该得到认定。这种表述伴有一种事实审理者风险，即法官或陪审团可能会跳转到偏向于检控方解释的过早结论。这种根据检控方论点来认定特定证据的可能性不应该是审理者应该考虑的问题。其必须要决定的是检控方的论点是否已得到了排除合理怀疑程度的确立。

若除了科学家在解释证据与检控方的被告有罪假说一致之外，也有科学家能以同样的频率解释证据与被告的清白假说一致，那将更有利于支持我的观点。如此一来，事实审理者能够立即感受到证据支持被告清白说法的发生率，理解被告无辜的可能性依然存在，并在所有恰当呈现在法庭面前的证据基础上来询问是否存在与被告无辜说法相一致的合理可能解释时考虑进这些因素。

四、举例：DNA

举例来说，在含有DNA 证据的案件中，科学家与其说DNA 样本属于被告方的可能性比该DNA 属于随机选择之人的可能性大一百万倍，不如说该DNA 样本与被告方相匹配，但是该DNA 并非为某特定个人所专有，然后再表述该种DNA 在普通人群中出现的预期频率（随机匹配的几率）。

如果事实审理者被告知该DNA 出现的预期频率是一百万分之一的话，那么很容易就可以意识到，在澳大利亚这个两千万人口的国家有可能还有人拥有该DNA ，因此排除犯罪人之一涉嫌为被告的可能性。如果预期的出现频率是十亿分之一，事实审理者就会意识到，要在澳大利亚找到与该5 例如，参见乔纳森·柯勒：《DNA 证据证明价值的表达：频率、似然比和错误率》《科罗拉多大学法律评论》1996年，

第67卷，第859，878~9，880页。（报告他所做的实验：“似然比与频率”）。

6 虽然法院在“Aytugrul 诉R ”案中，在未通过当事方恰当呈交和检验的情况下，不愿意采用这项研究（不愿对其进行司法认知）。

504··Evidence Science V ol.22 No.4 2014DNA 相匹配的其他人，与其说不是完全不可能，但机率也极低，无需更多进一步的证据就可以排除掉这种可能性。

但是，无论出现频率高低，事实审理者的任务仍然应聚焦于考虑剩下的证据，以便在作出有罪裁决之前排除该DNA 为被告人之外其他人所留下的任何合理可能性。（而且，理所应当还应该考虑并排除在配对DNA 证据收集和分析过程中发生人为错误的统计可能性。这些人为错误的可能性或许比随机配对的机率还要高。）

五、法院对法庭科学证据表述的控制

尽管以上主张说如是，有关法院可以对法庭科学证据表述所进行的控制方面依旧存在着一些法律问题。通过法院行使自由裁量权来排除相关但对被告方有不公正偏见的证据，可以说算是一种有效的控制。该问题在高等法院受理的“Aytugrul 诉R ”案7中进行了讨论。

高等法院被要求决定其自由裁量权是否要排除表述为某种排他百分比的DNA 证据（与被告相匹配的该种DNA 样本不会和99.9%的普通人相匹配），依据是存在风险陪审团可能会赋予这种（即便是准确的）表述过高的证据性证明力。新南威尔士州刑事上诉法院大法官麦克莱伦参照上文中提及的心理学文献8，认同应基于该理由来行使其排除自由裁量权。

在高等法院，法官们显然不愿意运用其自由裁量权来排除经准确表述的数值相关证据。然而，高等法院认定，即便运用其自由裁量权，该上诉仍然无法获得胜诉，因为陪审团会赋予该排他百分比过高证明力的可能性并未得以正式确立，而且即便可以确立这一点，该证据在频率上的相关性也已经向陪审团进行过解释，审判法官已经明确地指示陪审团只有在排除掉与被告无辜相关的所有可能性解释之后才能定罪，因此（澳大利亚高等法院认为）不存在陪审团会赋予该排他百分比过度证明力的风险。

上诉人因此未能说服高等法院在该案中存在司法不公的风险。然而，该多数判决的确承认了存在某些情况，数值表述可能会使得陪审团分心于其应做的工作，也就是去考量证据清白解释的可能程度以及在定罪前排除一切合理怀疑。

在我看来，高等法院不愿意否定使用排他百分比来表述DNA 证据的证明力与其支持寻求通过排除包含于证据之中的合理可能清白解释来降低错判无辜者风险的证明过程之间并不和谐。虽然DNA 配对频率表述与此证明路径尚且兼容，但排他百分比和似然比表述就很难与此证明路径相融合了。如果是这样，这种表述只会给事实审理者造成困惑，甚至会导致该种表述被赋予过高的证明力。并不需要接受心理学文献就可以得出这一点。实际上，这源自于普通法刑事证明的本质。

六、结语

无论你是否认同我所主张的刑事案件中法庭科学证据表述之合理模式，我希望本文提供了一个基于普通法刑事证明本质的角度，以及法庭科学家有责任表述其证据的法律证明背景。说到底，问题在于法庭科学家如何能最有效地推进刑事审判的目标，即排除错判无辜被告人的合理可能风险、所有支持该判决的证据均已排除合理怀疑、实现准确判决。

（收稿：2014-05-20，修回：2014-08-15）

（责任编辑：刘鑫）

7

8［ 2012］ HCA 15。 参见前引5。

2014年第22卷（第4期）505

··

Forensic science expressions and legal proof

Andrew Ligertwood1

1. Introduction

The object of this paper is to emphasise the legal context within which forensic evidence is presented; to explain how the common law sees the role of forensic scientists in the court process and the demands that this puts upon forensic scientists when it comes to expressing their evidence.1 This role has been brought to a head by the decision in R v T 2 where the court disapproved of a forensic scientist using numerical likelihood ratios to calculate his evaluative opinion in the absence of clear empirical justification for the numbers used. In strong reaction to that decision, a group of forensic scientists and lawyers have responded that the likelihood ratio is the most appropriate method for forensic scientists to calculate and to express evidential support.3 This ratio expresses the likelihood of finding the forensic evidence in question if the prosecution’s contentions, as opposed to the defence contentions, are correct.

As a lawyer I have been concerned about the expression of forensic evidence in courts since DNA evidence has been permitted as a likelihood ratio that appears so ‘mind-bogglingly’4 favourable to the prosecution that, in the absence of evidence raising the possibility that the DNA evidence is unreliable, the jury would appear to have no option other than to convict the accused.

No doubt in most cases not a bad statistical conclusion, even if not calculated by the jury by applying rules of mathematical probability, but, I will argue, a conclusion that fails to approach the decision of proof in a criminal case via the process of reasoning demanded by the law. In my view the notion of a likelihood ratio sits uneasily within the notion of legal proof applied in criminal cases.

My argument is confined to legal proof in criminal cases.

In summary I argue that criminal proof demands proof of the prosecution case, having regard to all the evidence properly before the court, beyond reasonable doubt. This requires the jury carefully to consider all the evidence, find it consistent with the prosecution case and to exclude any reasonable possibility of that evidence being explicable by an hypothesis consistent with the accused’s innocence.

The likelihood of finding evidence if the prosecution case is correct is not decisive of criminal proof. See Ligertwood, Can DNA evidence alone convict an accused?， 33 Sydney Law Review，487 (2011). Also see Ligertwood and Gary Edmond entitled, Expressing evaluative forensic science opinions in a court of law， published in 2012 in Law and Science. References to further source materials can be found in these papers.

2［ 2010］ EWCA Crim 2439 (26 October 2010).

3 Guest Editorial, Expressing evaluative opinions: a position statement，51 Science and Justice，1(2011).

4 The term is used by Mr Walker SC in argument: Forbes v The Queen［ 2010］ HCATrans 120 (18 May 2010) ［2050］. 1

506··Evidence Science V ol.22 No.4 2014It is the exclusion of any reasonably possible innocent account having regard to all the evidence properly before the court that is decisive. To ensure that the jury performs this task I suggest that, rather than forensic evidence being presented for the prosecution as a likelihood ratio, that it be presented as consistent with the prosecution case and that its evidential weight be explained by indicating the likely frequency of the evidence assuming innocent explanations for its existence. The jury can then go about its task of excluding these possibly innocent explanations having regard to all the evidence properly before it before finding the accused guilty ‘beyond reasonable doubt.’

My argument is pragmatic rather than philosophical or normative. At a philosophical and normative level questions about the nature of proof are as complex and controversial as questions about the very nature of knowledge. My starting point is that the common law demands a particular approach to proof in criminal cases so it becomes imperative, so long as the law demands this approach, that forensic scientists present their evidence in a form that can be accommodated within it. I am not seeking to justify what the law does, nor to say that our current legal processes could not be improved. I want simply to argue that currently the common law demands that juries and judges approach criminal proof in a particular way and that forensic evidence needs to fit within it.

2. The nature of criminal proof at Common Law

What then is the nature of legal proof in criminal cases? There appear to be two basic approaches.

On the one hand it is argued that, certainty of knowledge being impossible, legal proof, like any other sort of proof, is a matter of assessing probabilities based upon evidence, and that these probabilities can be numerically expressed and assessed by using logical rules of mathematical probability. Where probabilities cannot be empirically measured precisely then fact-finders may use subjectively assessed probabilities to determine whether the required degree of proof has been achieved. Thus, proof can be conceptualised as a quite precise logic that can be ultimately expressed numerical terms, and the degree required for proof can be set according to the seriousness of the consequences that might follow in the event of the decision being wrong.

On the other hand it is argued that, yes, certainty may be impossible, but that courts have never regarded legal proof as a notion ultimately capable of mathematical expression. Rather, proof is seen as a matter of a search for that explanation that the experiences of the fact-finder suggest accommodates all the evidence properly put before the court. With court cases generally involving the determination of events that have occurred in the world, proof is the search for that account, that narrative, that story, which explains the evidence. The degree of conviction required for proof again depends upon the seriousness of the consequences but is not expressed enumeratively. In a civil case the search is for the more probable explanation. In a criminal case the search is for that explanation that can be concluded by the fact-finder beyond reasonable doubt.

Without entering into the debate about which approach is ‘correct’ or preferable, I just want to assert that it is the second approach that prevails in common law courts. Standards of proof are not conceptualized in enumerative terms. But furthermore, although courts may permit expert forensic evidence to be tendered in a statistical form - at least, according to R v T, where it can be regarded as having sufficient empirical basis - courts do not stipulate the logic by which those statistics might be related to other evidence in the case.

Most clearly courts shy away from mathematical expressions of that logic, and refuse to direct juries in criminal cases about how mathematical rules might be used to compute the effect of evidence presented

2014年第22卷（第4期）507

··

in numerical terms. Where evidence is permitted as a numerical likelihood ratio, juries are not instructed on how to employ Bayes Theorem to calculate the effect that this evidence has on a prior probability that the accused committed the crime alleged. One can only wonder how in these circumstances a jury does otherwise rationally assesses the effect of a likelihood ratio (however expressed).

It appears that in both civil and criminal cases the approach to proof is simply to seek the ‘best explanation’ of all the evidence before the court. One might make a case that the approach in civil cases could evolve to support a logic that can ultimately be expressed enumeratively. Courts do talk in terms of satisfaction on the balance of probabilities, a standard that appears to assume a numerical degree of anything over 50%. And such a numerical standard may generally be appropriate where the object is simply to settle a dispute between citizens who have the same to lose in the case of a wrong decision.

But in a criminal case the standard of proof is difficult to convert into a logic that can ultimately be expressed numerically. The standard is expressed in terms of satisfaction beyond reasonable doubt. Juries are asked to exclude all reasonably possible innocent explanations before convicting the accused. It is in essence an aspiration to certainty as courts seek to avoid the unacceptable possibility of the conviction of an innocent person. To convert to a numerical standard would recognise this possibility. For this reason alone courts are reluctant to go down this path. (And of course there are other obvious practical problems in taking a numerical approach where the fact-finders are jurors and judges untrained in mathematical analysis, and where there is no precise empirical basis for allotting numerical values to relevant probabilities.)

3. Expressing forensic evidence at trial

This is the context within which forensic and other experts are called to court to testify. Their role is not to determine facts conclusively, that is the task of judge or jury. Rather their role is to assist judge or jury in reaching the required state of satisfaction. This assistance may be given in three ways. First, through providing evidence of observations that could just not be made without the expertise in question. Secondly, through supplementing the knowledge of the fact-finder, be it judge or jury, so that it may reach a more informed decision on the inferences to draw from particular evidence before the court. And thirdly, through providing evaluative opinions based upon their expertise about the strength of the inferential conclusions they believe can be drawn from the existence of particular evidence (it always being the prerogative of the court to reject this evidence, and, of course, any evaluative opinions drawn by the expert).

It is this final form of assistance, providing evaluative opinions about inferential conclusions, that is most controversial, particularly where the evidence is given as a likelihood ratio in support of the prosecution in a criminal case tried before judge and jury. For while forensic scientists may be careful to emphasise that they are only giving an opinion about the probability of finding certain evidence on the assumption of a particular prosecution hypothesis, there is a grave risk that the jury in deciding whether to find that hypothesis proved will equate the probability of finding the evidence with the probability of the hypothesis itself.

The opinion of an expert about what conclusion he or she would draw from the evidence thus runs the risk of being misunderstood by the jury and either accepted as proof of a contention highly probative of guilt without further question or at least without regard to other evidence before the court that the expert has

508··Evidence Science V ol.22 No.4 2014not been asked to consider. There is psychological research5 that indeed suggests that juries tend to give evidence presented in the form of a likelihood ratio more evidential weight than it deserves in application of Bayesian logic.6

While this risk of jury overestimating the logical effect of forensic evidence may not be seen as great, and while a jury can be directed that the final inferences of fact are for it to determine and not for the expert to draw, it must be kept in mind that a criminal conviction requires proof beyond reasonable doubt and the process must ensure that the jury has carefully excluded all possibilities for reasonable doubt before reaching that degree of satisfaction. The very object of the criminal trial is to ensure rigorous application of this criminal standard of proof, to eliminate as far as is reasonably possible any risk of the conviction of an innocent person.

There are obviously other more cost-effective ways of controlling crime but it is the protection of the innocent citizen that lies at the heart of the common law system of criminal justice. In this context it would seem inappropriate to rely upon jury directions if there are other ways of expressing evaluative forensic opinions better suited to ensuring that the jury focuses upon its task of excluding all reasonably possible innocent hypotheses.

Another answer to the argument rejecting expression in terms of a likelihood ratio is that the adversarial common law trial encourages each contestant to put evidence at its most persuasive and the role of the opponent is, through cross-examining prosecution witnesses and calling other witnesses, to counter that evidence through expressing it in a way that is persuasive of the defence case.

Even assuming the effectiveness of these adversarial protections, this, I believe, is to distort the nature of the common law trial in criminal cases. The trial is an accusatorial trial not an adversarial trial. The State brings an accusation that it must justify beyond reasonable doubt and the accused is given every opportunity to dispute that accusation and the evidence upon which it is based. But the duty of the State is to persuade the trier of fact to accept its accusation by rational means, through arguments that have logical consistency; not to persuade through rhetoric that stirs the emotions or distracts the trier of factfrom its rational task.

And as the psychological research referred to above suggests, there is a real risk that the likelihood ratio may be given more weight than it logically deserves by laypersons. What is of absolutely fundamental importance is that evidence be presented in a form that makes clear the task in applying the criminal standard of proof. This is a form that emphasizes the possibilities of innocent explanations left open by the evidence but argues that these can be excluded having regard to other evidence before the court.

In this context I simply argue that forensic scientists should be prepared to put their evidence in a form that does not simply say that the evidence is most likely to be found if the prosecution explanation occurred than an innocent explanation. This expression leaves open the risk of the trier of fact, whether judge or jury jumping to a premature conclusion about the prosecution explanation. The likelihood of finding certain evidence given the prosecution’s contentions is not the question with which the trier is concerned. It must determine whether the prosecution’s contentions have been established beyond reasonable doubt.

5 See, for example, Jonothan J Koehler，On Conveying the Probative Value of DNA Evidence: Frequencies, Likelihood Ratios, and Error Rates，67 University of Colorado Law Review 859, 878–879，880ff (1996) (reporting on his experiment: ‘Likelihood Ratios vs Frequencies’).

6 Although the court in Aytugrul v R was reluctant to act upon this research (take judicial notice of it) without it being properly tendered and contested by the parties.

2014年第22卷（第4期）509

··

To assist with this my argument is that it would be much more helpful if the scientist simply explained that the evidence is consistent with the prosecution’s hypothesis of guilt, but that one would expect also to find it in this frequency in the case of innocent hypotheses. The trier of fact can then immediately appreciate the frequency with which the evidence may accompany innocent scenarios, understand that innocent scenarios remain open, and take these into account when asking whether, on all the evidence properly before the court, there is a reasonably possible explanation consistent with innocence.

4. Example: DNA.

For example in the case of DNA evidence the scientist, rather than saying it is a million times more likely that matching DNA will be found if the DNA is that of the accused as opposed to a randomly selected person, can say that the DNA sample matches that of the accused, but that DNA is not unique to a particular individual, and then express the expected frequency of DNA of that profile in the population (the chances of a random match).

If the trier of fact is told that the expected frequency of that DNA is (say) one in a million it can easily see that there are likely to be other persons with that DNA profile in Australia, which has a population of 20 million, and exclude the possibility that one of those persons committed the crime alleged against the accused. If the expected frequency is (say) one in a billion the trier of fact can see that the chance of finding another person with that DNA profile in Australia, although not impossible is unlikely, and little further evidence would be required to eliminate it.

But whatever the frequency the trier of fact’s task remains focused on considering the remaining evidence to exclude any reasonable possibility that the DNA was left by a person other than the accused before reaching a verdict of guilty. (And of course it should also consider and exclude the statistical possibility of human errors having occurred in the collection and analysis of the DNA evidence of a match. These possibilities of human error may be higher than the chances of a random match.)

5. Court control over the expression of forensic evidence.

Despite these arguments there do remain legal questions over the control a court can exercise over the expression of forensic evidence. Control can arguably be effected through exercise of the court’s residuary discretion to exclude relevant evidence likely to be unfairly prejudicial to the accused. This issue was argued before the High Court in Aytugrul v R7.

The Court was asked to determine whether the residuary discretion demanded exclusion of DNA evidence expressed as an exclusionary percentage (the DNA sample whilst matching the accused would not match 99.9% of the population) - on the ground that there was a risk of the jury giving this expression (although accurate) greater evidential weight than it deserved. In the NSW Court of Criminal Appeal Justice McClellan, referring to the psychological literature mentioned above,8 was prepared to exercise the exclusionary discretion on this basis.

In the High Court there was an apparent reluctance to apply the discretion to exclude an accurate expression of the numerical relevance of evidence. However it held that even if the discretion could apply 7

8［ 2012］ HCA 15. See Supra note 5 above.

510··Evidence Science V ol.22 No.4 2014the appeal before them could not succeed because the likelihood that a jury would give excessive weight to the exclusionary percentage had not been formally established, and, even if it had been, because the relevance of the evidence had also been explained to the jury in frequency terms, and the trial judge had clearly directed the jury not to convict unless it excluded all possible explanations consistent with the accused’s innocence, there was no risk that the jury had given undue weight to the exclusionary percentage.

The appellant thus failed to persuade the Court that there was in the circumstances any risk of there having been a miscarriage of justice. However the plurality judgment does recognize that there might be circumstances where a numerical expression might distract a jury from its proper task, namely consideration of the extent of possibly innocent explanations for evidence and their exclusion before conviction.

In my view the Court’s reluctance to disapprove of the use of an exclusionary percentage to express the probative value of DNA evidence sits uneasily with their endorsement of a process of proof which seeks to mitigate the risk of convicting an innocent accused by eliminating so far as is reasonably possible innocent explanations for the evidence before the court. While expression of DNA matches in terms of a frequency is compatible with this approach it is difficult to see the place of expressions in terms of exclusionary percentages and likelihood ratios. If this is so such expressions can only confuse a trier of fact or be given greater probative value than they deserve. There is no need to accept the psychological literature to make this point. Rather it follows from the very nature of common law criminal proof.

6. Conclusion

Whether or not you find my arguments on the appropriate form of expressing forensic evidence in criminal cases convincing I hope I have provided a perspective upon the nature of common law criminal proof and the context of legal proof within which forensic scientists are obliged to express their evidence. In the end the question must be how the forensic scientist can most effectively promote the object of the criminal trial, accurate verdicts that exclude so far as is reasonably possible the risk of convicting an innocent accused, verdicts justified by all the evidence beyond reasonable doubt.

法庭科学的表述与法律证明作者：

作者单位：

刊名：

英文刊名：

年，卷(期)：安德鲁·立格特伍德(著)[澳大利亚]， 汪诸豪(译)， Andrew Ligertwood， Wang Zhuhao安德鲁·立格特伍德(著)[澳大利亚],Andrew Ligertwood(澳大利亚阿德莱德大学; 中国政法大学证据科学研究院)， 汪诸豪(译),Wang Zhuhao(中国政法大学证据科学研究院)证据科学Evidence Science2014(4)

本文链接：http://d.g.wanfangdata.com.cn/Periodical_flyyx201404009.aspx