Versatile formal representations of webs of evidence, and some of their current applications, include Wigmore charts (cases at law), Bayesian networks (complex DNA cases), and computerised systems such as Flints (Forensic-led intelligence system, for linking crimes and criminals). We are studying their logical foundations and inter-relationships, and attempting to extend their capabilities and applications.

The interpretation of evidence is grounded in assumptions, which may be unverifiable. Using econometric examples and techniques we are seeking to isolate these, and developing approaches to minimise sensitivity to them. In particular we are addressing issues of causal conclusions from observational data, and developing methods to construct and test underlying economic models under the weakest possible assumptions.

Problems of historical evidence include: Factoring out bias or point of view in historical sources; distinguishing between actuality and memories; and assessing the weight, validity and interrelations of evidence from archaeology (e.g. artefacts, architecture, material culture), inscriptions (including coins), eye-witness reports, and secondary or retrospective accounts of events. We are studying the potential of subject-blind Evidence Science (including Wigmore charts) to describe and ameliorate these problems.

Conservatism and other inadequacies of evidence processing, both by lay people and by professionals, are well established. We are conducting experimental studies to characterise the nature and source of such biases, and investigating the potential of decision-aids to overcome them.

The project is a joint initiative between the Department of Primary Care and Population Sciences and the Department of Computer Sciences. It aims to investigate how evidence from research is identified, interpreted, negotiated, and fed into the healthcare policymaking process. We will use a novel cross-disciplinary theoretical framework that draws on both medical sociology (in which the focus of analysis is the roles, relationships and interactions of individuals and groups) and the philosophy of argumentation (in which the focus is on how evidence is constructed, framed and rhetorically presented to support particular arguments in particular micro-political contexts). Using in-depth case study methods, we will explore how teams charged with "summarising the evidence on X" refine their brief and then seek out, evaluate, summarise and present evidence to a wider policymaking team. We will follow what happens to this secondary (i.e. evaluated, summarised and synthesised) evidence as it enters the policymaking process.

This project is an investigation of evidence in the natural sciences from the viewpoint of the history and philosophy of science. Historians have studied numerous instances of evidential disputes in the natural sciences over the centuries, and philosophers have made serious attempts to reach a general understanding of the nature of scientific evidence. Considerations of the "theory-ladenness" of observation have called into question the precise status of observational or experimental "facts" as evidence. The debates on scientific realism have highlighted the difficulties involved in establishing the truth of a theory even when we have relatively secure factual evidence in support of it.

This project seeks to study, in the context of the “Evidence” programme, how different disciplines can engage with each other critically, and to identify those features that enable or disable this. In this way it will develop a conceptualisation of interdisciplinarity that is most appropriate for large-scale social science investigation.

This project broadens the concerns of “Formal Tools for Handling Evidence” to address strategies for gathering of discovering evidence. There are many high-profile recent instances of the dangers of flawed, ineffective or unimaginative approaches to this. A mix of perspectives, experience and skills will be applied to develop new insights into the nature of enquiry and detection, and to construct formal representations and computational tools to address the complex evidential problems arising from this – including in particular the determination of effective questioning strategies. We will motivate and test our methods using specific cases from a variety of applied fields, with particular focus on Police Detection and Intelligence Analysis.

Is there a concept of evidence that applies universally? Are there specific or generic techniques for manipulating evidence that can be applied across disciplinary boundaries? These are questions that arise continually in the multidisciplinary research programme “Evidence, Inference and Enquiry: Towards an Integrated Science of Evidence” at University College London, supported by the Leverhulme Trust and the ESRC.