Providing Credible Evidence For Singular Causal Claims

Our 1st research questions:

What evidence helps predict if what you do will have the outcome you want-'Will doing X produce Y'?-or assess later if it has done so? Why does this count as evidence?

These are questions about 'singular causation' (contrasted with causal generalisation). Current answers are scattered in different places, they tend not to be theoretically well-grounded and are often contradictory. Our analyses suggest this is because the relation we are trying to find evidence for is not well understood. So to answer our primary questions, we aim to develop a rich theory of singular causation:
What's typically true when a cause succeeds (or not) in producing an outcome?

Singular causation is a central concern of philosophy-one with serious practical significance. Understanding how it works is essential in judging in practice whether an intervention leads to desired outcomes. Our project aims to improve the theoretical and practical understanding of singular causation. From that we aim to produce a framework for cataloguing evidence for singular causal claims that can help policy planners and evaluators (eg heads of schools, evaluators of development projects, those charged with implementing new child-protection programmes) in predicting for their local setting if an intervention can be implemented and succeed there, or evaluating later if it has done so.

We will use as a case study to test our ideas successes and failures in implementing the child protection programme Signs of Safety, whose Theory of Change includes wider systemic changes needed to support change in direct work with families. Munro has been conducting action research on its implementation in the UK and Ireland. There is also a body of research on its implementation and variable successes that we can use. A common problem is for the focus to be at the level of direct work with families without considering the organisational factors influencing front-line staff in ways that conflict with the work (eg recording requirements not matching the analytic structure of Signs of Safety; organisational messages giving lower priority to direct work with children than Signs of Safety specifies). Our study will produce guidance for senior managers to understand a fuller range of factors that may need changing. Similar implementation problems occur in other practice frameworks so the research will provide a general model of how to explore broader systems' influences.

These broader systems are labelled 'mechanisms' in philosophy, 'context' in methodology, where context is seen as a complex system where causes of different types studied at different levels of analysis interweave. Several new research methods have been developed (eg the Context-Mechanism-Outcome model from 'realist evaluation'; new work on 'complex social systems'). But there is insufficient analysis of the different types of mechanism that play a role, types that call for different methods of investigation studied in different disciplines.

We aim of to contribute to this debate by building on Cartwright's philosophical analysis of 3 types of causal mechanisms seen in philosophy and evidence-based policy research:
o The step-by step process by which a cause produces its effects
o The tendencies by which each step produces the next
o The underlying system that affords causal processes.

Our project will explore what methods are suited to studying the different kinds of contribution each type of mechanism makes in order to construct templates for theoretically well-grounded 'evidence-role maps' for causal prediction and evaluation. These are maps charting the kinds of information relevant for evidencing singular causal claims and indicating the role each plays. Understanding the different roles that different bits of information plays is crucial in evaluating what information to collect, the risks of missing information, and the import of the evidence all told.