Building Social Policy
Research stream 1: Building social policies.
We do not just read off instructions for building a laser – which may ultimately be used to operate on your eyes – from knowledge of basic science. Rather, we piece together a detailed model using heterogeneous knowledge from a mix of physics theories, from various branches of engineering, from experience of how specific materials behave, from the results of trial-and-error, etc. Similarly, building a successful social policy equally requires a mix of heterogeneous kinds of knowledge from radically different sources. Trying to understand, describe and find ways to improve the technology of social science is a highly risky enterprise, in part because there is so little work of this kind done. High gain is the flip side of high risk: there are immense payoffs to understanding better how to put social science knowledge to practical use.
K4U will address 3 specific questions here:
- What knowledge do we need? Previous work suggests that far more emphasis needs to be put on theory and on understanding the underlying social systems that afford the causal regularities relied on for policy prediction.
- How should we use it? Here there will be 2 subprojects: 1. To advance practicable ways to construct “built-to-purpose causal narratives” [cf Reiss 2013, Lane 2008, Pawson 2013] that can produce reliable (enough) predictions. 2. To develop an account of “evidence synthesis by building a case” for synthesising “inhomogeneous” evidence (like the mix in legal trials). [cf Twining 1994, Suter & Cormier 2011].
- What role should values play? This requires seriously Rethinking values in science  at the science/society interface, from choice and framing of research topics through choices of modelling techniques to choices of how to operationalise and measure. [Douglas 2009; Atkinson 2001; Dupre, Kincaid and Wylie 2007; Reiss 2007].
 Rethinking values in science at the science/society interface, from choice and framing of research topics through choices of modelling techniques to choices of how to operationalise and measure. [Douglas 2009; Atkinson 2001; Dupre, Kincaid and Wylie 2007; Reiss 2007].