Introduction
Policy analysis provides information, evaluation, advice and advocacy for policymakers and is typically a species of naturalist or āscientificā or positivist social science. It has some distinguished proponents, including Yehezkel Dror (1968), Harold Lasswell (1971), and Aaron Wildavsky (1980). Among the multitude of definitions, Hill (2013: 5) is representative. He distinguishes between analysis of the policy process and analysis for the policy process. The former is descriptive and analytical seeking to explain the origins of policy and how it was implemented. The latter is prescriptive using the analytical techniques often associated with economics to provide information and evaluations for policymakers.
Naturalism refers to the idea that āThe human sciences should strive to develop predictive and causal explanations akin to those found in the natural sciencesā (Bevir and Kedar 2008: 503). For example, evidence-based policymaking is the latest fashion in policy analysis. It favours randomised controlled trials (RCTs). In brief, RCTs involve identifying the new policy intervention, determining the anticipated outcomes and specifying ways of measuring those outcomes. Following this, the investigator chooses control groups, whether comprised of individuals or institutions. The policy intervention is randomly assigned to the target groups with a designated control group. Using a randomly assigned control group enables the investigator to compare a new intervention with a group where nothing has changed. Randomisation is considered appropriate to eradicate the influence of external factors and potential biases (Cartwright and Hardie 2012; Haynes et al. 2012). With its roots in clinical trials, the influence of the natural sciencesā experimental method and the ambition to be āscientificā are clear for all to see.
This book rejects such naturalism and argues for decentred policy analysis rooted in an anti-naturalist epistemology. First, it rejects the naturalist thesis that we can explain actions by allegedly objective social facts about people. Meanings are largely irrelevant to mainstream political science and policy studies. Beliefs are, at most, intervening variables. Actions can be correlated with, and explained by, social categories such as class, economic interest or institutional position. These analytical moves suppress or deny human agency. Second, it rejects the naturalist thesis that the relation between antecedent and consequent in political explanation is a necessary causal oneāthat is it is law like as in the natural science. Political science seeks psychological or social laws, rather than historical narratives or understanding webs of meaning.
Naturalism has been widely criticised for its faith in pure experience. Political scientists recognise that we cannot approach objects from a theory-neutral position. They seem far less aware that the impossibility of pure experience also undermines the two theses just discussed. First, because people do not have pure experiences, they always construct their identities, interests and beliefs in part through their particular theories. Therefore, political scientists cannot explain behaviour by reference to given interests or objective social facts. Second, because social facts do not fix peopleās identities, interests and beliefs, we have to explain actions by referring to the intentionality of the actors. Therefore, political scientists cannot appeal to causal laws (and for a more detailed discussion see Bevir and Rhodes 2003, 2015b).
In sharp contrast, anti-naturalists argue that āconstitutive features of human life set it apart from the rest of nature to such an extent that the social or human sciences cannot take the natural sciences as a modelā. Instead, āthe relevant features of human action are that it is meaningful and historically contingentā (Bevir and Kedar 2008: 505). I develop these points in the next section.
Although naturalist policy analysis has paid some attention to narratives, 1 anti-naturalism underpins most narrative policy analysis, which has existed in the study of public policy since the 1990s. 2 It is one of the few subfields of political science where interpretive approaches have had some traction. Wagenaar (2011) provides an authoritative overview of the field, and there seems little point in covering the ground again. Instead, I focus on the key characteristics of our approach at Southampton: interpretive theory, decentring and fieldwork, especially ethnographic fieldwork. I discuss each in turn.
The rest of this chapter outlines the specific interpretive approach developed by Mark Bevir and R. A. W. Rhodes. It explains what we mean by ādecentringā and unpacks the ethnographic toolkit we employ. It provides brief descriptive summaries of the individual chapters but focuses on how each relates to the overall themes of the book. Finally, it discusses what we can learn from a decentred approach . I suggest that it delivers edification because it offers a novel alliance of interpretive theory with an ethnographic toolkit to explore policy and policymaking from the bottom-up.
Interpretive Theory
All political scientists offer us interpretations. Interpretive approaches differ in offering interpretations of interpretations. They concentrate on meanings, beliefs and discourses, as opposed to laws and rules, correlations between social categories or deductive models. An interpretive approach is not alone in paying attention to meanings. It is distinctive because of the extent to which it privileges meanings as ways to grasp actions. Its proponents privilege meanings because they hold, first, beliefs have a constitutive relationship to actions and, second, beliefs are inherently holistic. 3
First, an interpretive approach holds that beliefs and practices are constitutive of each other. As Clifford Geertz (1973: 5) famously claimed, social science needs to be ānot an experimental science in search of law but an interpretive one in search of meaningā. For example, when other political scientists study voting behaviour using attitude surveys or models of rational action, they separate beliefs from actions to find a correlation or deductive link between the two. In contrast, an interpretive approach suggests such surveys and models cannot tell us why, say, raising oneās hand should amount to voting, or why there would be uproar if someone forced someone else to raise their hand against their will. We can explain such behaviour only if we appeal to the intersubjective beliefs that underpin the practice. We need to know voting is associated with free choice and, therefore, with a particular concept of the self. Practices could not exist if people did not have the appropriate beliefs. Beliefs or meanings would not make sense without the practices to which they refer.
Second, an interpretive approach argues that meanings or beliefs are holistic. We can make sense of someoneās beliefs only by locating them in the wider web of other beliefs that provide the reasons for their holding them. So, even if political scientists found a correlation between a positive attitude to social justice and voting Labour, they could not properly explain peopleās voting Labour by reference to this attitude. After all, people who have a positive attitude to social justice might vote Liberal if they believe Labour will not implement policies promoting social justice. To explain why someone with a positive attitude to social justice votes Labour, we have to unpack the other relevant beliefs that link the attitude to the vote. To explain an action, we cannot merely correlate it with an isolated attitude. Rather, we must interpret it as part of a web of beliefs.
Third, human action is historically contingent. It is characterised by change and specificity . We cannot explain social phenomena if we ignore their inherent flux and their concrete links to specific contexts. Such historicist explanations work not by referring to reified correlations, mechanism or models, but by describing and locating contingent patterns of meaningful actions in their specific traditions. Historicists argue that beliefs, actions and events are profoundly contingent because choice is open and indeterminate. They question the possibility of either a universal theory or ahistorical correlations and typologies. In addition, they argue that if we are to understand and explain actions and beliefs, we have to grasp how they fit within wider practices and webs of meaning. Historicism promotes forms of understanding and explanation that are inductive studies of human life in its historica...