Using subsampling to estimate the strength of handwriting evidence via score-based likelihood ratios

The likelihood ratio paradigm has been studied as a means for quantifying the strength of evidence for a variety of forensic evidence types. Although the concept of a likelihood ratio as a comparison of the plausibility of evidence under two propositions (or hypotheses) is straightforward, a number of issues arise when one considers how to go about estimating a likelihood ratio. In this paper, we illustrate one possible approach to estimating a likelihood ratio in comparative handwriting analysis. The novelty of our proposed approach relies on generating simulated writing samples from a collection of writing samples from a known source to form a database for estimating the distribution associated with the numerator of a likelihood ratio. We illustrate this approach using documents collected from 432 writers under controlled conditions.     

Score-based likelihood ratios for handwriting evidence

Score-based approaches for computing forensic likelihood ratios are becoming more prevalent in the forensic literature. When two items of evidential value are entangled via a scorefunction, several nuances arise when attempting to model the score behavior under the competing source-level propositions. Specific assumptions must be made in order to appropriately model the numerator and denominator probability distributions. This process is fairly straightforward for the numerator of the score-based likelihood ratio, entailing the generation of a database of scores obtained by pairing items of evidence from the same source. However, this process presents ambiguities for the denominator database generation - in particular, how best to generate a database of scores between two items of different sources. Many alternatives have appeared in the literature, three of which we will consider in detail. They differ in their approach to generating denominator databases, by pairing (1) the item of known source with randomly selected items from a relevant database; (2) the item of unknown source with randomly generated items from a relevant database; or (3) two randomly generated items. When the two items differ in type, perhaps one having higher information content, these three alternatives can produce very different denominator databases. While each of these alternatives has appeared in the literature, the decision of how to generate the denominator database is often made without calling attention to the subjective nature of this process. In this paper, we compare each of the three methods (and the resulting score-based likelihood ratios), which can be thought of as three distinct interpretations of the denominator proposition. Our goal in performing these comparisons is to illustrate the effect that subtle modifications of these propositions can have on inferences drawn from the evidence evaluation procedure. The study was performed using a data set composed of cursive writing samples from over 400 writers. We found that, when provided with the same two items of evidence, the three methods often would lead to differing conclusions (with rates of disagreement ranging from 0.005 to 0.48). Rates of misleading evidence and Tippet plots are both used to characterize the range of behavior for the methods over varying sized questioned documents. The appendix shows that the three score-based likelihood ratios are theoretically very different not only from each other, but also from the likelihood ratio, and as a consequence each display drastically different behavior.     

Learning about Bayesian networks for forensic interpretation: an example based on the 'the problem of multiple propositions'

Both, Bayesian networks and probabilistic evaluation are gaining more and more widespread use within many professional branches, including forensic science. Notwithstanding, they constitute subtle topics with definitional details that require careful study. While many sophisticated developments of probabilistic approaches to evaluation of forensic findings may readily be found in published literature, there remains a gap with respect to writings that focus on foundational aspects and on how these may be acquired by interested scientists new to these topics. This paper takes this as a starting point to report on the learning about Bayesian networks for likelihood ratio based, probabilistic inference procedures in a class of master students in forensic science. The presentation uses an example that relies on a casework scenario drawn from published literature, involving a questioned signature. A complicating aspect of that case study - proposed to students in a teaching scenario - is due to the need of considering multiple competing propositions, which is an outset that may not readily be approached within a likelihood ratio based framework without drawing attention to some additional technical details. Using generic Bayesian networks fragments from existing literature on the topic, course participants were able to track the probabilistic underpinnings of the proposed scenario correctly both in terms of likelihood ratios and of posterior probabilities. In addition, further study of the example by students allowed them to derive an alternative Bayesian network structure with a computational output that is equivalent to existing probabilistic solutions. This practical experience underlines the potential of Bayesian networks to support and clarify foundational principles of probabilistic procedures for forensic evaluation.     

The use of the likelihood ratio for evaluative and investigative purposes in comparative forensic handwriting examination

This paper extends previous research and discussion on the use of multivariate continuous data, which are about to become more prevalent in forensic science. As an illustrative example, attention is drawn here on the area of comparative handwriting examinations. Multivariate continuous data can be obtained in this field by analysing the contour shape of loop characters through Fourier analysis. This methodology, based on existing research in this area, allows one describe in detail the morphology of character contours throughout a set of variables. This paper uses data collected from female and male writers to conduct a comparative analysis of likelihood ratio based evidence assessment procedures in both, evaluative and investigative proceedings. While the use of likelihood ratios in the former situation is now rather well established (typically, in order to discriminate between propositions of authorship of a given individual versus another, unknown individual), focus on the investigative setting still remains rather beyond considerations in practice. This paper seeks to highlight that investigative settings, too, can represent an area of application for which the likelihood ratio can offer a logical support. As an example, the inference of gender of the writer of an incriminated handwritten text is forwarded, analysed and discussed in this paper. The more general viewpoint according to which likelihood ratio analyses can be helpful for investigative proceedings is supported here through various simulations. These offer a characterisation of the robustness of the proposed likelihood ratio methodology.     

Two or three contributors of DNA mixtures? – Some practical considerations

The number of contributors is hard to determine in DNA mixture profiles. Here, we deal with the special but frequent case that either two or three contributors are possible. In fact, it might happen that two contributors can explain the number of alleles seen but that three contributors are necessary if a specific person of interest is to be included in the mixture. Then the likelihood ratio assuming two contributors will be zero while the likelihood ratio for three contributors may be large. We evaluate this situation and offer suggestions on how to arrive at an overall likelihood ratio. To exemplify our line of reasoning we use an example proposed by Biedermann, Taroni and Thompson.     

Estimating the time since discharge of spent cartridges: A logical approach for interpreting the evidence

Estimating the time since discharge of a spent cartridge or a firearm can be useful in criminal situations involving firearms. The analysis of volatile gunshot residue remaining after shooting using solid-phase microextraction (SPME) followed by gas chromatography (GC) was proposed to meet this objective. However, current interpretative models suffer from several conceptual drawbacks which render them inadequate to assess the evidential value of a given measurement. This paper aims to fill this gap by proposing a logical approach based on the assessment of likelihood ratios. A probabilistic model was thus developed and applied to a hypothetical scenario where alternative hypotheses about the discharge time of a spent cartridge found on a crime scene were forwarded. In order to estimate the parameters required to implement this solution, a non-linear regression model was proposed and applied to real published data. The proposed approach proved to be a valuable method for interpreting aging-related data.     

精神障碍者受审能力评定量表的编制

目的研制符合中国法律体系的精神障碍者受审能力评定量表。方法从法学要件着手,提取出15个条目,编制出初步的精神障碍者受审能力评定量表,从决断值、条目与总分相关、校正条目与总分相关、条目删除后的α值、条目共同性、因素负荷量6个方面对量表进行条目分析,并以Logistic回归方程与ROC曲线划定界值两种方式进行诊断效能的判断。结果各条目极端组的决断值18.390~46.763,各条目与总分相关系数0.639~0.952,校正条目与总分相关系数0.582~0.944,条目共同性0.377~0.916,因素负荷量0.614~0.957。共7个条目进入回归方程,总样本的回判正确率为96.0%。ROC拟合曲线分析出宜以33分作为量表划界分,与专家鉴定意见的重叠率为95.8%,敏感性为0.938,特异性为0.966,阳性似然比为27.67,阴性似然比为0.06。结论量表编制合理,各条目符合同质性检验要求,诊断性评价指标较好。     

A contribution to the discussion of probabilities and human hair comparisons

The paper by Gaudette and Keeping on "An Attempt at Determining Probabilities in Human Scalp Hair Comparison" in the Journal of Forensic Sciences (Vol. 19, No. 3, July 1974, pp. 599-606) has provoked considerable controversy. This paper highlights two of the sources of the controversy and shows how the probability, 1/4500, quoted by Gaudette and Keeping should be treated with caution. The necessity of the use of a likelihood ratio statistic is described. It is suggested that the hair examination form resulting from the responses to the questionnaire recently distributed by the authors and also the discussions at Quantico (Proceedings of the International Symposium on Forensic Hair Comparisons, 25-27 June 1985, Quantico VA) should be used to facilitate the collection of the data which will be necessary to enable a likelihood ratio statistic to be estimated effectively.     

A cautionary note on the evaluation of genetic evidence from uniparentally transmitted markers

The combination of the information obtained from lineage genetic markers, such as mitochondrial DNA (mtDNA) and the non-homologous region of Y-chromosome, with data resulting from meiotically recombining loci (diploid/autosomal or haplodiploid/X chromosome) into a single likelihood ratio has been recently proposed. In this work we challenge this proposal and demonstrate that while the genetic evidence obtained from loci which reshuffle at meiosis is appropriate for individual probability calculations, mtDNA and Y-chromosome data are not and, consequently, that joining the evidential value of the two types of markers is generally inconsistent and should be avoided. The assumption of non-involvement of relatives must be clearly and explicitly stated and its acceptance must be left to the court decision.     

Assessment of forensic findings when alternative explanations have different likelihoods—“Blame-the-brother”-syndrome

Assessment of forensic findings with likelihood ratios is for several cases straightforward, but there are a number of situations where contemplation of the alternative explanation to the evidence needs consideration, in particular when it comes to the reporting of the evidentiary strength. The likelihood ratio approach cannot be directly applied to cases where the proposition alternative to the forwarded one is a set of multiple propositions with different likelihoods and different prior probabilities. Here we present a general framework based on the Bayes' factor as the quantitative measure of evidentiary strength from which it can be deduced whether the direct application of a likelihood ratio is reasonable or not. The framework is applied on DNA evidence in forms of an extension to previously published work. With the help of a scale of conclusions we provide a solution to the problem of communicating to the court the evidentiary strength of a DNA match when a close relative to the suspect has a non-negligible prior probability of being the source of the DNA.     

Uncertainty in probabilistic genotyping of low template DNA: A case study comparing STRMix™ and TrueAllele™

Two probabilistic genotyping (PG) programs, STRMix™ and TrueAllele™, were used to assess the strength of the same item of DNA evidence in a federal criminal case, with strikingly different results. For STRMix, the reported likelihood ratio in favor of the non-contributor hypothesis was 24; for TrueAllele it ranged from 1.2 million to 16.7 million, depending on the reference population. This case report seeks to explain why the two programs produced different results and to consider what the difference tells us about the reliability and trustworthiness of these programs. It uses a locus-by-locus breakdown to trace the differing results to subtle differences in modeling parameters and methods, analytic thresholds, and mixture ratios, as well as TrueAllele's use of an ad hoc procedure for assigning LRs at some loci. These findings illustrate the extent to which PG analysis rests on a lattice of contestable assumptions, highlighting the importance of rigorous validation of PG programs using known-source test samples that closely replicate the characteristics of evidentiary samples. The article also points out misleading aspects of the way STRMix and TrueAllele results are routinely presented in reports and testimony and calls for clarification of forensic reporting standards to address those problems.     

An Objective Corruption Risk Index Using Public Procurement Data

In order to address the lack of reliable indicators of corruption, this article develops a composite indicator of high-level institutionalised corruption through a novel ‘Big Data’ approach. Using publicly available electronic public procurement records in Hungary, we identify “red flags” in the public procurement process and link them to restricted competition and recurrent contract award to the same company. We use this method to create a corruption indicator at contract level that can be aggregated to the level of individual organisations, sectors, regions and countries. Because electronic public procurement data is available in virtually all developed countries from about the mid-2000s, this method can generate a corruption index based on objective data that is consistent over time and across countries. We demonstrate the validity of the corruption risk index by showing that firms with higher corruption risk score had relatively higher profitability, higher ratio of contract value to initial estimated price, greater likelihood of politicians managing or owning them and greater likelihood of registration in tax havens, than firms with lower scores on the index. In the conclusion we discuss the uses of this data for academic research, investigative journalists, civil society groups and small and medium business.     

Evaluation and presentation of forensic DNA evidence in European laboratories.

The first part of this article describes Bayes' theorem and its practical application to scientific evidence. Secondly, statements proposed by European DNA laboratories are presented, analysed and commented upon. Finally, the authors reiterate their proposal of adopting a likelihood ratio framework justified by the usefulness of the approach in practical context of DNA cases. Note that this framework underlines the main questions the experts have to answer at trial: to what degree does the evidence support the prosecutor's hypothesis? and to what degree does the evidence support the defence's hypothesis?     

A probabilistic approach to the joint evaluation of firearm evidence and gunshot residues

The present paper addresses issues that affect both the separate as well as the joint evaluation of firearm evidence (i.e., marks) and gunshot residues (GSR). Mark evidence will be used as a basis to discriminate among barrels through which a bullet in question might have been shot whereas GSR will be used to draw inferences about the distance of firing. Particular attention is drawn to the coherent handling of uncertainties associated with the various parameters considered within each item of evidence. The proposed analysis relies on a probabilistic viewpoint that uses graphical models (i.e., Bayesian networks) as an aid to cope with the complexity induced by the number of variables considered. The paper discusses how an approach based on a probabilistic network environment can be used for the formal analysis and construction of arguments. Emphasis is made on the gain of insight into structural dependencies that may be uncovered when the evaluative process is extended beyond single items of scientific evidence.     

Testing an Integration of Control Theories: The Role of Bonds and Self-Control in Decision Making

Since the redefinition of self-control (Hirschi, 2004) social bonding measures have been utilized as predictors or indicators of revised or decisional self-control. This approach departs from the prior literature which has hypothesized a selection effect of self-control on social bonds. To reconcile this discrepancy and explain the relationship among two self-control measures (attitudinal and revised self-control), an alternate model is proposed that explains the process of self-control on refrainment from offending at the situation level. Surveys using the vignette method for a hypothetical drunk-driving decision-making task were collected from large samples of young adults and inmates. Path analysis is used to model bonds and self-control as determinates of decisional self-control, thus indirectly influencing self-reported estimates of drunk-driving likelihood. Decisional self-control, attitudinal self-control and social bonds also directly affect intentions to drive drunk. Overall, the proposed model is supported. A strong direct effect of attitudinal self-control on drunk-driving likelihood remains while controlling for decisional self-control. There are relatively smaller indirect effects of social bonds and attitudinal self-control on drunk-driving likelihood, through their effects on decisional self-control. These findings support the need for the conceptual separation of bonds, attitudinal and decisional self-control as well as increased attention to differential effects of self-control.     

Decision theoretic properties of forensic identification: underlying logic and argumentative implications

The field of forensic science has profited from recent advances in the elicitation of various kinds probabilistic data. These provide the basis for implementing probabilistic inference procedures (e.g., in terms of likelihood ratios) that address the task of discriminating among competing target propositions. There is ongoing discussion, however, whether forensic identification, that is, a conclusion that associates a potential source (such as an individual or object) with a given item of scientific evidence (e.g., a biological stain or a tool mark), can, if ever, be based on purely probabilistic argument. With regard to this issue, the present paper proposes to analyze the process of forensic identification from a decision theoretic point of view. Existing probabilistic inference procedures are used therein as an integral part. The idea underlying the proposed analyses is that inference and decision are connected in the sense that the former is the point of departure for the latter. As such the approach forms a coordinated whole, that is a framework also known in the context as 'full Bayesian (decision) approach'. This study points out that, as a logical extension to purely probabilistic reasoning, a decision theoretic conceptualization of forensic identification allows the content and structure of arguments to be examined from a reasonably distinct perspective and common fallacious interpretations to be avoided.     

The evaluation of evidence in the forensic investigation of fire incidents (Part I): an approach using Bayesian networks

The forensic investigation of the origin and cause of a fire incident is a particularly demanding area of expertise. As the available evidence is often incomplete or vague, uncertainty is a key element. The present study is an attempt to approach this through the use of Bayesian networks, which have been found useful in assisting human reasoning in a variety of disciplines in which uncertainty plays a central role. The present paper describes the construction of a Bayesian network (BN) and its use for drawing inferences about propositions of interest, based upon a single, possibly non replicable item of evidence: detected residual quantities of a flammable liquid in fire debris.     

论声纹鉴定意见的表述形式

本文针对目前国内外讨论比较热烈的声纹鉴定意见表述问题进行了评述.首先介绍了实践中正在使用的听觉分析法、声谱比对分析法、声学分析法、听觉-声学分析法和说话人自动识别五种鉴定方法,指出了各种方法的优缺点;然后对现存的二元判决、可能性等级、似然比和英国立场声明四种鉴定意见表述形式进行了介绍和评析,通过分析发现,上述四种意见表述形式都存在一定的问题,实践中选择何种形式表述鉴定意见要综合考虑其科学性、逻辑性、现实性和可行性等多种价值选项;最后认为解决该问题的根本方法是各相关领域的专家应加强在鉴定方法上的合作性的基础研究.     

Human Rights and Unfair Dismissal: Private Acts in Public Spaces

This article addresses the termination of employment because of the conduct of the employee in her leisure time, in the light of the right to private life. It explores the impact on the retention of employment of activities taking place outside the workplace and outside working hours, and argues that the approach of UK courts and tribunals, which is based on a primarily spatial conceptualisation of privacy, is flawed. A fresh approach to privacy, resting on the idea of domination, is proposed, which is sensitive to the particularities of the employment relationship. Considering the fairness enquiry in dismissal, it argues that off‐duty conduct may lead to lawful termination of employment only if there is a clear and present impact or a high likelihood of such impact on business interests; a speculative and marginal danger does not suffice. It further proposes that a particularly meticulous test is appropriate when certain suspect categories, such as the employees' sexual preferences, are at stake.     

Assessment of approximate likelihood ratios from continuous distributions: a case study of digital camera identification

A reported likelihood ratio for the value of evidence is very often a point estimate based on various types of reference data. When presented in court, such frequentist likelihood ratio gets a higher scientific value if it is accompanied by an error bound. This becomes particularly important when the magnitude of the likelihood ratio is modest and thus is giving less support for the forwarded proposition. Here, we investigate methods for error bound estimation for the specific case of digital camera identification. The underlying probability distributions are continuous and previously proposed models for those are used, but the derived methodology is otherwise general. Both asymptotic and resampling distributions are applied in combination with different types of point estimators. The results show that resampling is preferable for assessment based on asymptotic distributions. Further, assessment of parametric estimators is superior to evaluation of kernel estimators when background data are limited.