A template for constructing Bayesian networks in forensic biology cases when considering activity level propositions

Abstract

The hierarchy of propositions has been accepted amongst the forensic science community for some time. It is also accepted that the higher up the hierarchy the propositions are, against which the scientist are competent to evaluate their results, the more directly useful the testimony will be to the court. Because each case represents a unique set of circumstances and findings, it is difficult to come up with a standard structure for evaluation. One common tool that assists in this task is Bayesian networks (BNs). There is much diversity in the way that BN can be constructed. In this work, we develop a template for BN construction that allows sufficient flexibility to address most cases, but enough commonality and structure that the flow of information in the BN is readily recognised at a glance. We provide seven steps that can be used to construct BNs within this structure and demonstrate how they can be applied, using a case example.

Evaluation of forensic genetics findings given activity level propositions: A review

“…We direct the reader to [60] for explanations of the structure and terminology of BNs. Taylor et al [61] give a step by step method that can be used to construct BN for forensic biology findings that are being evaluated given activity level propositions. In their work they describe a process that starts by identifying the propositions of interest, identifying all the activities that would be required under either (or both) of the propositions, then considering all the results that would be expected, and argumentatively connecting them with the propositions regarding posited activities through considerations such as transfer and persistence.…”

“…This is achieved by evaluating the probability of obtaining different strengths of evidence in favour of the prosecution or defence propositions if either proposition is true. The advantage of carrying out such an evaluation at the beginning of the case is that evaluation is carried out in a manner that can then be used later on when results are obtained (although, as noted by Taylor et al [61], using BNs, the examination of exhibits may bring up unexpected findings that require fine tuning of the evaluation after examination). The process thus emphasises the thinking about the value of findings before examinations are carried out, thus avoiding the evaluation being influenced by the results actually obtained (i.e., avoiding being 'findings-led').…”

Bayesian networks for the interpretation of biological evidence

“…This means the court is left to consider these issues unaided by the forensic scientist and there has been increasing debate as to whether or not forensic scientists should be helping the court to address such higher level propositions. Addressing the complexities of the propositions further up the hierarchy is often challenging (Biedermann & Hicks, ; Cook et al, ; Jackson et al, ; Kokshoorn et al, ; Taylor, Abarno, et al, ; Taylor, Biedermann, et al, ; Willis et al, ). Bayesian networks are a tool that has been demonstrated to be effective in carrying out the task of evaluating evidence at both the source and activity level, and will be of assistance to forensic biologists in moving forward.…”

“…The programs discussed include Hugin, Netica, SamIam, and Bayesian network Toolkit by Microsoft, Ergo, and GENIe. The most commonly used software to construct Bayesian networks for forensic science problems currently is Hugin (http://www.hugin.com; Aitken et al, ; De Wolff et al, ; Fenton & Neil, ; Gittelson, Biedermann, Bozza, & Taroni, ; Madsen, Lang, Kjærulff, & Jensen, ; Mortera et al, ; Taylor, Abarno, et al, ; Taylor, Biedermann, et al, ). Construction of networks in these programs is simple and flexibility enables models to be altered and tested in real time.…”

“…However, when applied to a real‐life case, the number of parameters can be considerable, resulting in a structure that may become extremely complex, often making these template models impractical to use. To assist in consistency, Taylor, Biedermann, et al () provide a template using a step‐by‐step guide for constructing Bayesian networks for propositions in forensic biology, allowing flexibility to address a range of cases.…”