1
Knowledge and Policy in the Literature
How have theorists conceived of the relationship between knowing and governing? What can past research tell us about how policy-makers might use knowledge in their work and, indeed, what âknowledgeâ and âusing itâ might constitute in this context? In this chapter I show how these questions were a central concern for foundational authors in the policy sciences, and how in the last three decades they have been taken up with renewed vigour by authors writing on the relationship between evidence and policy-making. Both of these literatures point to the limitations of technocratic accounts of policy-making, indicating instead the importance of attending to social interactions between policy-makers and researchers, and the context in which policy-makers operate, to understand how and why policy-makers use knowledge as they do.
But the evidence and policy literature itself seems to be stuck in something of an intellectual rut, and to pursue these themes further requires marshalling resources from other literatures. This chapter revisits some core concepts in the philosophy of knowledge and draws together more recent writing on the nature of knowledge in organisations and work, to sketch an alternative conceptualisation of what knowledge and knowing might comprise in the practice of policy-making. In this account knowledge is characterised as being underdetermined and is realised in activity, in interaction with others in a particular setting and in relation to a particular task.
Policy-making as a knowledge problem
In the pervasive conception of policy-making as a process of decision making, as âwhatever a government chooses to do or not to doâ (Dye 1972, p. 2), public administrators are expected to draw on authoritative, expert knowledge claims to inform those decisions. In fact English civil servants are legally required to base their âadvice and decisions on rigorous analysis of the evidenceâ (Civil Service Commission 2012). And yet as early as the 1940s, writers in the policy sciences were pointing out the practical and political limitations of technocratic accounts of decision making. Herbert Simonâs work in this area emphasised the cognitive and contextual barriers to public administrators acting wholly rationally (Simon 1947). Writing against an image of âeconomic manâ, who assesses all available options and chooses those which will most efficiently deliver some agreed objective, Simon describes the real âadministrative manâ as limited by practicalities which mean only a few of these options âever come to mindâ (Simon 1976, p. 81), and that he cannot have âa complex knowledge and anticipation of the consequences that will follow on each choiceâ (Simon 1976, p. 81). In this state of âbounded rationalityâ, administrative work is driven by a logic of âsatisficingâ, rather than âmaximisingâ (Simon 1957, p. xxvi). The administrator accepts that âthe world he perceives is a drastically simplified model of the buzzing, blooming confusion that constitutes the real worldâ, and is content to work with âa simple picture of the situation that takes into account just a few of the factors that he regards as most relevant and crucialâ (Simon 1957, pp. xxvâxxvi). Administrative man seeks a course of action that is âgood enoughâ (Simon 1957, pp. xxvâxxvi).
A similar conclusion was reached by Charles Lindblom, who identified the time, information, resource and intellectual constraints with which administrators worked, and the political and legal context for that work, as major barriers to ârational-comprehensiveâ models of decision making (Lindblom 1959). In practice, he argued that administrators are engaged in âmuddling throughâ, in which the âthe test of a âgoodâ policy is typically that various analysts find themselves directly agreeing on a policy (without their agreeing that it is the most appropriate means to an objective)â (1959, p. 81). Taking up the subject 20 years later Lindblom advised policy-makers to abandon aspirations to âimpossible feats of synopsisâ (1979, p. 518), and instead to develop more systematic versions of existing practices which involved making small incremental steps in policy, enabling administrators to gather âknowledge about the probable consequences of further, similar stepsâ (1959, p. 86). Lindblom also argued that politics could be made more âintelligentâ through attention to improving (rather than curbing) âpartisan analysisâ in which âparticipants make heavy use of persuasion to influence each other; hence they are constantly engaged in analysis designed to find grounds on which their political adversaries or indifferent participants might be converted to allies or acquiescentsâ (1979, p. 524).
The importance of social and political context to understanding how policy-makers use knowledge in their work is also emphasised in writing on âpolicy learningâ (Freeman 2006). In a seminal contribution to policy studies in the 1970s, Hugh Heclo sought to challenge dominant accounts of the policy process as one of conflict resolution, and to highlight the âpolitical learningâ involved in governing by attending to how governments âpuzzleâ as well as âpowerâ (1974, p. 303). For Heclo, learning is what governments do in response to changes in their environment; it is a process that comprises a ârelatively enduring alteration in behavior that results from experienceâ (1974, p. 306). It is shaped by three forces: (i) individuals; (ii) organisations and their inter-relationships (the administration, political parties, committees and interest groups); and (iii) past policies. In his study of the evolution of income maintenance policies in Britain and Sweden, Heclo identified the importance of civil servants in the learning process, as a permanent fixture on the scene, and also the actors to whom âhas fallen the task of gathering, coding, storing and interpreting policy experienceâ (1974, p. 303). These policy-makers puzzle on societyâs behalf. He also emphasised the important role played by ânetworks of policy middle menâ (1974, p. 311), who are at the âinterfaces of various groupsâ and have âaccess to information, ideas, and positions outside the normal run of organisational actorsâ (1974, p. 308). Notably, Hecloâs account incorporated learning about both the substance, and the process, of governing; policy-makers are concerned not just with what was done by past governments, or by foreign administrations, but also with how it was done.
In the 1980s and 1990s Peter Hall went on to develop a theory of âsocial learningâ as the âdeliberate attempt to adjust the goals or techniques of policy in the light of the consequences of past policy and new information so as to better attain the ultimate objects of governanceâ (Hall 1988, p. 6; cited in Bennett & Howlett 1992, p. 276). He illustrated his account with an analysis of the shift from Keynesianism to Monetarism in UK economic policy in the 1980s (Hall 1993). Here, Hall distinguishes three main orders of learning. Within the normal run of policy-making, there is first order change, which involves adjustments to the instruments of governing, and, less commonly, second order change, which involves selecting different policy instruments to reach a specified goal. Changing the goals of policy, and the way in which a problem itself is conceptualised, constitutes a third order change, which amounts to a paradigm shift (Hall 1993, pp. 278â80). The changes to macroeconomic policy he describes exemplify this third type of policy learning. Explicitly drawing on Thomas Kuhnâs theory of scientific paradigms, Hall posits that:
policymakers customarily work within a framework of ideas and standards that specifies not only the goals of policy and the kind of instruments that can be used to attain them, but also the very nature of the problems they are meant to be addressing. Like a Gestalt, this framework is embedded in the very terminology through which policymakers communicate about their work, and it is influential precisely because so much of it is taken for granted and unamenable to scrutiny as a whole.
(Hall 1993, p. 279)
According to Hall, significant paradigm shifts are more sociological than rational or scientific in character. Though such shifts might be prompted by the emergence of anomalies which cannot be explained within the existing policy paradigm, an apparently more important prerequisite for a shift in policy paradigm is a prior and significant change in the âlocus of authority over policyâ (Hall 1993, p. 279). In Hallâs account, puzzling is not sufficient to account for what he termed third order change; there must be powering too.
Together these authors find that cognitive and practical limits to knowledge gathering and processing, including the political nature of policy-making as a social process, mean that rational technocratic models of policy-making are necessarily implausible and impracticable. Alternative forms of practice emerge, in which the process of policy development is more social than scientific, infused with politics and power and populated with multiple actors in addition to the politician, the administrator and the researcher or expert.
This work notwithstanding, the model of policy-making as a rational decision-making exercise based on expert knowledge has remained curiously persistent in the academy as well as government itself. This vision has underpinned successive efforts in Britain and elsewhere to improve the functioning of the civil service, and has provided the anchor point for an ever-burgeoning body of literature concerned with assessing the extent to which policy-making is indeed informed by evidence. The next section summarises the main themes from the âevidence and policyâ literature, and outlines how they have informed the design of this research study.
The evidence and policy literature
The extensive research and commentary in this field can be organised into two loose groups. The first, which begins with the assumption of a rational-instrumental relationship between research and policy-making, is concerned with how to better enable the flow of knowledge from one domain to the other. The second follows more closely in the tradition established by the policy scientists discussed in the previous section of this chapter by attending to the social nature of the knowledgeâpolicy relationship, identifying the multiplicity of actors involved and describing the iterative, and socially mediated or constructed nature of their communication, as well as analysing the political character of evidence and evidence use.
Rational-instrumental models: Bridging the two communities
Early models of the evidenceâpolicy relationship assume policy formulation to be a rational process in which research findings might be applied to help solve societyâs problems. In this body of research, the distinction between empirical description and normative prescription is sometimes elided. After early empirical research found that linear, rational models of the way in which research moves into policy bear little resemblance to practice, subsequent authors in this tradition have concerned themselves with understanding the obstacles to such research use by policy-makers, and how these might be overcome.
Writing in 2001, John Landry and colleagues elaborated on Carol Weissâs (1979) description of âknowledge-drivenâ and âproblem-solvingâ models of research use in policy-making to characterise the two dominant, early models of the evidenceâpolicy relationship. The problem-solving, âdemand pullâ models identify policy-makers and other research users as more or less formal directors or commissioners of research (Landry et al. 2001, p. 335). Knowledge-driven or âscience pushâ models present advances in research as the principal stimulus for knowledge use by policy-makers and others, with authors focusing on how utilisation is effected by the particular attributes of the research (its âtechnical qualityâ, complexity, âdivisibilityâ and âapplicabilityâ [2001, p. 334]). For Landry and his colleagues, the weakness of these approaches lies in their failure to focus on the process by which research is taken up by policy-makers, including the active role played by the actors who are involved in the transfer, and the translation work required to make the research findings âusableâ. The authors build on the findings of existing empirical literature to argue that it is a lack of interaction between research producers and users that is the main cause of research being under-utilised.
A key characterisation, which is common to both of these models (knowledge-driven and problem-solving) and the call for greater interaction between policy-makers and researchers, was articulated by Nathan Caplan in the late 1970s as the âtwo communitiesâ thesis (Caplan 1979). In this account:
social scientists and policy makers live in separate worlds with different and often conflicting values, different reward systems, and different languages. The social scientist is concerned with âpureâ science and esoteric issues. By contrast, government policy makers are action-orientated, practical persons concerned with obvious and immediate issues.
(Caplan 1979, p. 459)
Variations on this set of assumptions can be found in much of the subsequent literature. Authors in this tradition have identified barriers to evidence use presented by the different priorities and values of policy-makers and academics respectively. They found that personal interaction between members of the two groups is the most important enabler of evidence use, and described or prescribed the importance of various bridging roles to serve as conduits between the two worlds.
For example, two systematic reviews of the factors influencing research use by health policy-makers published in the early 2000s found that, of the two dozen studies covered in each (with only a couple of overlaps), personal contact between researchers and policy-makers is the most cited enabler of evidence use, followed by the timeliness and relevance of research (InnvĂŚr et al. 2002; Hanney et al. 2003). Sandra Nutley and colleaguesâ comprehensive review of the field published in 2007 also concluded that models which focus on the interaction between researchers and policy-makers âintegrate our best current knowledge about the kinds of factors that seem to support the use of researchâ (Nutley et al. 2007, p. 119). A systematic review published in 2014, which incorporates studies from across the world on a range of policy areas including health confirmed that âthe quality of the relationship and collaboration between researchers and policy-makersâ was the single most cited factor influencing evidence use by policy-makers and other decision makers working in public policy (Oliver et al. 2014).
Work in this field has also drawn attention to the significance of âcontextâ for understanding how policy-makers use knowledge, which is by definition âdifficult to include in a deductive theory of knowledge utilizationâ (Landry et al. 2001, p. 346). The work of John Lavis and colleagues, also based in Canada, has called for greater attention to be paid to the type of policy under consideration, the different forms that âuseâ takes, and the impact of research in the context of other influences on policy development (Lavis et al. 2002). Back in the United Kingdom, Sandra Nutley and colleaguesâ 2007 review of the field also found that âcontext ⌠seems to be the key to whether and how research gets usedâ (2007, p. 89) concluding that:
Simple surveys of what seems to support or inhibit the use of research can only take our understanding of the research use process so far. This means we need to attend in more depth to the ways in which these different âfactors affectingâ interact, in complex and dynamic ways, in complex and dynamic contexts.
(Nutley et al. 2007, p. 89)
Relatedly, Katherine Oliver and colleagues find in their recent systematic review that the relatively narrow set of methodological approaches deployed in this field (surveys or interviews, with a majority focusing on the perceptions of researchers rather than policy-makers) means that there have been âfew studies exploring how, when and why different facilitators and barriers come into play during the policy-making processâ (Oliver et al. 2014).
So although this academic sub-field starts with simplified, rational models of research use, the conclusions of its more recent work prescribe attending to a multiplicity of actors, to the importance of interaction for research mobilisation, and to the significance of context in understanding how, why and when research is used. These authors set out an agenda for studying research use by policy-makers as a complex social process.
Research use as a social and political process
A second body of research, which has developed from and in parallel to the work described above, has been concerned with understanding the social structures and processes that shape research use by policy-makers. This work tends to be more critical and analytical than some of the early work on evidence in policy, often eschewing the rational assumptions and normative positions assumed by those authors. Instead of focusing on how to overcome obstacles to rational research use by policy-makers, these researchers have attended to the multiplicity of actors relevant to research mobilisation by policy-makers, and to querying the boundary between the âtwo communitiesâ of policy-makers and researchers narrowly conceived, to re-examining the form and qualities of the evidence that is the object of exchange, and to describing the various symbolic and polit...