Causation in Law and Statistics

We propose a project that will explore the divergent approaches to causation in law and statistics, identify the scope for learning across disciplines, and advance the state of the art in each. Context Lawyers use qualitative models and natural language to identify ‘causes of effects’ at individual level, while statisticians use mathematical models and data analysis to study ‘effects of causes’ at population level. Despite their differences, the two disciplines increasingly interact. Courts are being called upon to assess statistical evidence when determining issues of civil and criminal liability, while statisticians are developing formal models that seek to capture a wide variety of causal mechanisms with application to law. The challenge The interdisciplinary challenge is not to convert legal reasoning into statistical reasoning or vice versa, but rather to identify methods, concepts and paradigms from each field which can be used to enhance the coherence and effectiveness of the other. Thus, statistical concepts can be deployed to show that legal precedents which have accreted over time lack a clear diagnostic rationale and may result in inconsistent decision making. Conversely, law offers statistics a rich source of data and a practical context in which to apply models and techniques. In particular, statistics may help us to better understand the causal effects of legal interventions on social and economic outcomes, and clarify the factors influencing judicial decision-making. Statistics understands the causal inquiry as an inferential process for choosing among alternative hypotheses and interpretations under conditions of limited information. Legal reasoning would benefit, we believe, from engagement with this way of thinking about causation. Our overarching challenge is to map this development and spell out its implications for practice in areas where the two fields overlap. The potential gains are great: for law, enhanced consistency and fairness in legal decision making, and for statistics, a data-rich field of inquiry. Aims and objectives (i) To examine how far statistical theory can be used to improve the coherence of legal concepts relating to causation. (ii) To develop new and emerging statistical approaches to causal inference and apply them to issues of law. (iii) To apply statistical models of causation to study the factors influencing judicial decision making. Applications and benefits In addition to exploring the scope for conceptual mapping and translation of ideas across the two disciplines, we will undertake empirical research to explore how legal decision-making works. To this end we will use historical data on British and Irish workmen’s compensation cases and more recent data on the use of AI assistants by US courts deciding whether to grant bail. We will use causal models derived from the state of the art in statistics to explore how courts arrive at causal understandings, what difference legal rulings make to social and economic outcomes, and how far AI assistance can improve the decision making process. Our work will have the potential to lead to fairer and more consistent outcomes in civil and criminal cases. We will also be in a position to make an important contribution to the fast-developing debate over the use of AI-related computational techniques in legal analysis and decision-making.