Providing Credible Evidence For Singular Causal Claims – Funded by Arts and Humanities Research Council (AHRC)
CHESS’s mission is to integrate work from the humanities and the social and natural sciences to understand better how to improve societies. In particular, to investigate what kinds of academic knowledge — ‘science’ in the German sense — can best inform policy and practice, what methods will produce this knowledge and how this knowledge should be put to use. Causal knowledge is clearly vital here. Past work at CHESS in this area has focussed on causal modelling and causal inference. Our current concern is with the use of knowledge to make better assessments of causal effects, especially the effects of social policies. How can you predict if a policy intervention will have its targeted effect in a particular case, or determine afterwards whether it has done so?
Our focus is on warrant for singular causal claims: claims that one thing was/will be the cause of another. Our project begins from the assumption that to provide evidence for something, you need a solid understanding of what that something is like. But, surprisingly, there is no well-developed account of singular causation available. So we are working to develop a rich well-grounded account of singular causation, identifying features useful for evidencing singular claims.
The current work is sponsored by the UK Arts and Humanities Research Council, with a grant to philosopher Nancy Cartwright (CHESS co-director, Durham) and child protection expert Eileen Munro (emeritus professor, LSE), with Dr Brendan Kelters working as postdoctoral research associate.
To aid in the collection and processing of evidence, we aim to provide user-friendly templates for what we call ‘evidence-role maps’: maps charting what role each piece of evidence plays in supporting an overall judgment. To assure that our account is sufficiently rich and detailed to constrain how to apply it practically we are developing our theoretical analysis with application to a case study, a child protection programme, Signs of Safety, with which Professor Munro has been involved as a researcher. We aim to construct evidence-role maps for the implementation of Signs of Safety that are of practical use to those designing implementation strategies in identifying and assessing evidence about whether or how it can be implemented successfully in a targeted context and what changes might be needed to increase the likelihood of success. This will serve as a test of our general template and an exemplar for other domains.
Evidential Pluralism, Social Science Methodology, and Policy-Relevant Research
Some social scientists are adherents of methodological monism, the belief that there is only one valid method for research on specific kinds of social science questions or even for social research in general. For instance those in the evidence-based medicine and evidence-based policy (EBM and EBP) communities take randomized control trials (RCTs) as the method for establishing causal claims, with even carefully stratified observational studies attempting to mimic RCTs as a poor second best. Others embrace methodological pluralism, which argues that it is permissible – and even necessary – to rely on multiple methods in social science.
Monists generally argue that we should base our social science claims on only the soundest type of evidence, especially where those claims will be used to affect people’s lives and well-being, noting ‘Garbage in, garbage out’. Pluralists tend to urge two arguments to the contrary. First, different types of evidence can be used to make up for weaknesses in other types. Second, a claim is not genuinely well established if there are outstanding questions about whether the central facts that are necessary for it to hold are in place. But there are always a variety of different kinds of facts necessary for even fairly simple social-sciences claims to hold, different kinds of facts that call for different methods.
The EBM+ group, which has been gaining traction in the UK, urges causal pluralism primarily on the first grounds. Note though that what they advocate is a limited kind of causal pluralism – ‘causal dualism’ — which demands both RCT evidence and evidence of a casual mechanism in order to secure a causal claim. The second reason plays a prominent role in the work of those who focus on the context and conditions that are necessary if the processes by which a cause is supposed to achieve its effects are to succeed.
This research project investigates the advantages and disadvantages of different kinds of methodological monism and methodological pluralism for different questions, and it analyses how different types of studies can be better combined. It is especially concerned with how issues of methodological monism and pluralism affect the ways social scientists conduct policy-relevant research and how the use of plural methods can be improved in this domain.
Professor Cartwright has been pursuing this project with political scientist Hilde van Meegdenburg and philosopher of social science Rosa Runhardt (both from the Netherlands), along with postgraduate researcher Gabriel Nyberg (from UCSD) and undergraduate researcher Shahzen Chauhan (from Durham). It is currently sponsored by summer grants for research assistants from UCSD and Durham.