Overview More than simply theory, policy frameworks provide individuals with a blueprint for understanding how public policy is generated in a democratic society. Despite policy theories existing primarily in an ethereal sphere, the practical application and observation of policy formation can be seen in everyday contexts, provided one is willing to open their mind to identifying the factors that contribute to public policy formation. This assignment aims to provide you with an opportunity to read, synthesize, and relate established public policy theory frameworks within the context of practical application. For this assignment, you will need to read and select one of the policy theory frameworks outlined in the Weible and Sabatier (2017) text Theories of the Policy Process (4th Ed.). Although the book contains multiple policy theory frameworks, this assignment will require you to read and select a policy theory framework which both interests you and finds practical application through observing modern public policy issues. The first part of the assignment will require you to provide a synopsis of the policy theory framework selected, including pertinent information regarding the evolution of the policy theory framework itself (previous research, notable contributions, examples of how it has been utilized etc.), the strengths of the policy theory framework selected (such as ease of application, viability, or ease of comprehension/observation), and your own personal interpretation or conceptualization of how the policy theory framework unfolds. This part of the paper will require some creativity, as each student may conceptualize the framework differently or within different contexts. For example, I have noted before that I have conceptualized the Punctuated Equilibrium theory of policy formation as similar to the video game Pac-Man to illustrate periods of smaller policy change interspersed with large ‘punctuations’. Feel free to use descriptions, illustrations, or any other method you believe will help you relay how you conceptualize the policy theory framework you have selected! Following the policy theory framework synopsis, you will need to introduce and explain how the policy theory framework you have selected can be seen through modern or current public policy issues. This can be accomplished by outlining and addressing how components of the policy theory framework you have selected play out within the context of public policy. Additionally, you are welcome (as well as encouraged) to incorporate additional resources outside of the assigned readings if you believe they align with the core concepts contained within the policy theory framework selected. The Policy Theory Framework Analysis assignment should be between 6-8 full pages, using proper APA format for in-text citations and references as outlined. The Advocacy Coalition Framework: An Overview of the Research Program HANK C. JENKINS-SMITH, DANIEL NOHRSTEDT, CHRISTOPHER M. WEIBLE, AND KARIN INGOLD The study of policy processes brings focus to many questions of both theoretical and practical significance. Some questions concern policy change and stasis over time: What factors explain the likelihood of occurrence of major and minor policy change? To what extent is policy change affecting government agencies and procedures and, through them, broader public opinion? Another set of questions involves learning by actors: To what extent are actors learning from their experiences, from the experiences of others, or from scientific and technical information? What factors facilitate learning among allies and among opponents? Yet another set of questions centers on the behavior of actors who directly or indirectly attempt to influence policy processes by advocating for change or maintenance of the status quo: Under what conditions do actors form and maintain coalitions to achieve their policy objectives in a coordinated fashion? What are the characteristics of the network structures of these coalitions? To what extent, and in what ways, do opposing coalition actors interact? To answer these questions is to provide insight into some of the fundamental themes of governance and politics that ultimately affect the composition, dynamics, and course of society. The purpose of this chapter is to provide an overview of one framework that can help answer these questions: the Advocacy Coalition Framework (ACF).1 The chapter begins with a summary of the intellectual foundations of the ACF. It then provides an overview of the framework and its theoretical foci, including assessments of the hypotheses based on extensive empirical applications. The chapter ends by suggesting an ongoing agenda to continue the advancement of the ACF research program. INTELLECTUAL FOUNDATIONS OF THE ACF The ACF was created in the early 1980s by Paul Sabatier and Hank Jenkins-Smith. Chief inspirations were drawn from past shortcomings in policy process research, including a need to develop an alternative policy process theory to overcome the limitations of the stages heuristic; a need to provide theoretical insight into the role of scientific and technical information in policy debates; a need to shed light on ideological disagreement and policy conflicts; and a need to provide a comprehensive approach to understanding politics and policy change over time that went beyond traditional emphases in political science on government institutions (e.g., executive, legislative, and judiciary) and a few forms of political behavior (e.g., voting and lobbying) (see the discussion in Sabatier 1991). Establishing the framework took several years of effort that included conference papers written by Sabatier and Jenkins-Smith, ongoing data collection through interviews and surveys, and the creation of code forms for measuring belief systems and coalition stability over time.2 The foundations of the ACF were also influenced by debates in the philosophy of sciences that were still prominent in the 1970s and 1980s. Partly in response to Thomas Kuhn’s (1962) notion of scientific revolutions and paradigm shifts, Imré Lakatos developed his conception of the evolution of scientific research “programmes” in an effort to rescue the conception of cumulative, falsifiable science and the growth of knowledge (Lakatos 1970). A key contribution from Lakatos was the notion that scientific theories can be described as consisting of a “hard core” of unchanging and largely axiomatic propositions surrounded by a “protective belt” of auxiliary hypotheses that can be adjusted (or rejected) in response to potentially falsifiable evidence. This concept is a recognizable ancestor of the structure of belief systems in the ACF, which are characterized as hierarchically structured with a deep core of ontological and normative beliefs that are extraordinarily difficult to change, and “secondary aspects” of more specific propositions about how to effectively translate core beliefs into policy.3 A second important contribution was drawn from Lakatos’s related proposition that— given that Popper’s ([1935] 2002) ideal of falsifiable theories had proven implausible— theoretical progress was most readily evident in “progressive problem-shift.” As noted by Kuhn (1962), counterevidence need not displace a theory; empirical anomalies can persist for hundreds of years while a theory hangs on because (1) ad hoc defense of a theory was too effective to displace a vital theory and (2) a ready replacement theory was not available. Lakatos argued that ad hoc defense of a theory fails if it is persistently regressive—meaning ad hoc “adjustments” of theories to accommodate counterevidence do not add new theoretical content that can be (and eventually is) empirically verified. Lakatos also argued that healthy theories experience progressive problem- shift, wherein theoretical adjustments (i.e., new concepts and hypotheses added to the theory) not only address counterevidence but also add new empirical content that extends the explanatory reach of the theory. Hence, defense (or expansion) of a theory needs to be progressive to be scientifically legitimate. For Sabatier and Jenkins-Smith, this conception of scientific progress characterized the spirit of the theoretical growth of the ACF. The basic framework of the ACF (e.g., the assumptions and general subsystem dynamics) characterized the hard core while new propositions and theoretical logic (e.g., the addition of the concepts of coalition opportunity structures and endogenous pathways to policy change) occupy the auxiliary belt. Thus, revisions to the auxiliary belt were acceptable as long as they added new substantive theoretical content to the ACF. Indeed, the many additions to, and revisions of, key ACF hypotheses are reflections of this view of theory change and the growth of knowledge. Naturally, the assessment of whether these cumulative changes are truly progressive remains an open and important question. The earliest journal publications of the ACF began with Sabatier (1986), where the ACF was described as a synthesis of top-down and bottom-up approaches to implementation, and continued with Sabatier and Pelkey (1987), where the ACF was described as an approach to understand regulatory policymaking.4 The first overview of the ACF by Sabatier (1987) was published in Knowledge, and a nearly identical publication by Sabatier (1988) led to the symposium on the ACF in Policy Sciences.5 Jenkins-Smith’s work on policy analysis within a process characterized by advocacy coalitions was published soon after (Jenkins-Smith, 1990). Applications of the framework slowly accumulated, with a coedited volume by Sabatier and Jenkins-Smith published in 1993. The ACF has since become one of the most utilized frameworks of the policy process. The ACF has been the topic of five special issues in peer-reviewed journals, including Policy Sciences (Sabatier 1988), PS: Political Science & Politics (Sabatier 1991), Policy Studies Journal (Weible et al. 2011), Administration & Society (Scott 2012), and Journal of Comparative Policy Analysis (Henry et al. 2014). Weible, Sabatier, and McQueen (2009) and Pierce et al. (2017) conducted comprehensive reviews of over 240 English-language ACF applications from 1987 through 2014 that span the globe.6 Additionally, Sotirov and Memmler (2012) reviewed ACF applications in the context of environmental and natural resource issues. Country-specific reviews have been conducted for applications of the ACF in South Korea (Jang et al. 2016) and Sweden (Nohrstedt and Olofsson, 2016b). Jang et al. (2016) found 62 ACF applications in Korean, suggesting the ACF is frequently applied in other languages. One edited volume compares and contrasts advocacy coalitions and policy change on the topic of unconventional oil and gas development across seven countries in North America and Western Europe (Weible et al. 2016). In aggregate, these reviews confirm the portability of the ACF to different policy issues and governing systems, but they also expose areas where the ACF’s concepts and assumptions appear to be less applicable and problematic. We return to these issues below when discussing a research agenda for the future. The ACF offers a general foundation for single case studies and for comparative analyses of policy processes across a wide range of policy issues and governing systems. Although a single chapter cannot adequately summarize the development of the ACF over time, present all the intricacies of the framework, list all the hypotheses offered by various analysts, or synthesize the findings from various applications of the framework, we offer here a synthesis of the most recent developments in the ACF and encourage interested readers to explore past theoretical and empirical publications as cited herein. THE FRAMEWORK AND THEORETICAL EMPHASES Borrowing from Easton (1965), Laudan (1977, 70–120), Lakatos (1970), and Ostrom (2005, 27–29), the ACF is best thought of as a framework supporting multiple, overlapping theoretical foci.7 The purpose of a framework is to provide a shared research platform that enables analysts to work together in describing, explaining, and, sometimes, predicting phenomena within and across different contexts. The components of a framework include a statement of the assumptions, description of the scope or type of questions the framework is intended to help answer, and the establishment of concept categories and their general relations. Most importantly, a framework provides a common vocabulary to help analysts communicate across disciplines, from different substantive policy areas, and from different parts of the world. Akin to Lakatos’s hard core foundation of a research program, a framework should be fairly stable in its basic premises over time. Additionally, frameworks are not directly testable but provide guidance toward specific areas of descriptive and explanatory inquiry. This is an important reminder, given the misperception among some students and researchers that a comprehensive “test” of the ACF requires empirical assessments of all its components and relationships among them. Rather, a framework supports multiple theories, which are narrower in scope and emphasize a smaller set of questions, variables, and relationships. Theories provide more precise conceptual and operational definitions of concepts and interrelate concepts in the form of testable and falsifiable hypotheses or propositions. The theories within the framework are where students and researchers should attempt to test and develop descriptions and explanations. Theories are, hence, akin to Lakatos’s protective belt that can (and should) be subject to experimentation, adjustment, and modifications over time. Although hypotheses should ideally offer refutable expectations among concepts, a pragmatic reason for using hypotheses is to highlight the most important relationships that describe what, why, and how those concepts relate and when and where those relationships are expected to be evident (Whetten 1987). A SUMMARY OF THE ACF AS A “FRAMEWORK” A framework is best described by its assumptions, scope (type of questions), and basic categories of concepts and general relations for answering research questions. This section provides an overview of the framework of the ACF. Assumptions The policy subsystem is the primary unit of analysis for understanding policy processes. Policy subsystems are defined by a policy topic, territorial scope, and the actors directly or indirectly influencing policy subsystem affairs. Policy subsystems have several defining properties that help in interpretation and application (Sabatier 1998; Nohrstedt and Weible 2010). First, subsystems contain a large set of components that interact in nontrivial ways to produce outputs and outcomes for a given policy topic. These components range from physical and institutional characteristics to actor attributes, including belief systems and political resources. One of the purposes of a framework and its theories is to specify some of the most important subsystem components to study in attempting to help solve puzzles concerning the policy process. Second, policy subsystems demarcate the integrated and nonintegrated actors on a given policy topic. Policy subsystems do not involve all people interested and affected by the policy decisions. Indeed, given limited time and attention, most people do not engage in any subsystem and, for those who do, the number of policy subsystems where they are active is finite and usually small in number. Third, policy subsystems are semi-independent but overlap with other subsystems and are nested within yet other subsystems. For example, an energy policy subsystem in Colorado overlaps with a food policy subsystem in the same state and nests within a national energy policy subsystem in the United States.8 Fourth, policy subsystems often provide some authority or potential for authority. Such authority may exist in the enforcement and monitoring of policy, the legislative or legal processes, or the potential for new policies that may alter the status quo. Fifth, policy subsystems undergo periods of stasis, incremental change, and major change. The set of relevant subsystem actors include any person regularly attempting to influence subsystem affairs. Borrowing from Heclo (1978), the depiction of subsystem actors expands beyond traditional interpretations of the policy process that tends to focus narrowly on legislative committees, government agencies, and interest groups. Subsystems are affected by any actor directly or indirectly influencing subsystem affairs and may include officials from any level of government, representatives from the private sector, members from nonprofit organizations, members of the news media, academic scientists and researchers, private consultants, lobbyists, think tanks, and even members of the courts (Hjern and Porter 1981). The extent and consistency of involvement and influence of these actors, of course, varies. Individuals are boundedly rational, with limited ability to process stimuli, motivated by belief systems, and prone to experience the “devil shift.” The ACF conception of individuals is based on a modified version of methodological individualism, that is, change in the world is primarily driven by people and not by organizations (Sabatier 1987, 685). In the terms coalition beliefs, coalition behavior, and coalition learning, coalition is used metaphorically in reference to the individuals comprising the coalition. Indeed, coalitions do not learn, but rather the actors within coalitions learn. Furthermore, the modified version of methodological individualism in the ACF does not suggest that people’s behavior is independent of context. Indeed, the theory within the ACF would expect that people’s behavior is shaped by various contextual factors, particularly, the nature of relevant institutions, the intensity of conflict, and the perceived severity of threats posed by opponents. The ACF’s assumption that individuals are boundedly rational means that people are motivated instrumentally by goals but are often unclear how to achieve those goals, and they are limited in their cognitive abilities to process stimuli such as information and experience (Simon 1957, 1985). Additionally, given limited cognitive abilities, individuals simplify the world through their belief systems and are, therefore, prone to biased assimilation of stimuli (Munro and Ditto 1997; Munro et al. 2002). The ACF assumes that policy actors have a three-tiered belief system structure. Deep core beliefs are fundamental normative values and ontological axioms. Deep core beliefs are not policy specific and, thus, can be applicable to multiple policy subsystems. One way to conceptualize and measure deep core beliefs is by incorporating insights from cultural theory (Douglas and Wildavsky 1982; Ripberger et al. 2014; Jenkins-Smith et al. 2014; Trousset et al. 2015). Cultural theory offers four distinct orientations—hierarchs, egalitarians, individualists, and fatalists. Each of these orientations is buttressed by a set of “myths”—about human nature, society, and natural systems—that can serve both to justify the orientation and its values and to imply appropriate forms of social organization (Douglas and Wildavsky 1982; Thompson, Ellis, and Wildavsky 1990). Whereas cultural theory has demonstrated its utility as one way of conceptualizing and measuring deep core beliefs especially for comparative analyses, other ways certainly exist and could be developed. In contrast to deep core beliefs, policy core beliefs are bound by scope and topic to the policy subsystem and thus have territorial and topical components. Policy core beliefs can be normative and empirical. Normatively, policy core beliefs may reflect basic orientation and value priorities for the policy subsystem and may identify whose welfare in the policy subsystem is of greatest concern. Empirically, policy core beliefs include overall assessments of the seriousness of the problem, basic causes of the problem, and preferred solutions for addressing the problem (called policy core policy preferences). Secondary beliefs deal with a subset of the policy subsystem or the specific instrumental means for achieving the desired outcomes outlined in the policy core beliefs.9 Finally, the ACF borrows one of the key findings from prospect theory that people remember losses more readily than gains (Quattrone and Tversky 1988). Remembering losses and the tendency to filter and assimilate stimuli through belief systems result in the “devil shift,” where actors exaggerate the power and maliciousness of their opponents (Sabatier, Hunter, and McLaughlin 1987). The expected result is a noncollaborative attitude, growing mistrust, the protraction of conflict, and the obstruction of effective policy solutions (Fischer et al. 2016). Subsystems are simplified by aggregating actors into one or more coalitions. Depicting policy subsystems as consisting of any actor attempting to directly or indirectly influence subsystems affairs presents a dilemma for analysts: there might be hundreds of actors somehow involved in a policy subsystem. Also, analysts encounter subsystems at different levels of maturity; mature subsystems comprise relatively established and clearly differentiated coalitions, whereas nascent or emergent subsystems are characterized by ambivalence and unclear political positions. Simplifying assumptions must be made to describe and analyze.10 Analysts could organize subsystems by organizational affiliation, which provides important insight into the resources and strategies of actors in the policy subsystem, but the organizational level of analysis comes at the cost of realizing that the number of organizations involved in the policy subsystem is not many fewer than the number of actors. A more effective approach is to organize actors into one or more advocacy coalitions on the basis of shared beliefs and coordination strategies. By grouping and analyzing actors by coalitions, the analysts can simplify the hundreds of actors and their organizational affiliations into groupings that may be stable over time (Sabatier and Brasher 1993) and that are instrumental for understanding policy actors’ strategies for influence and policy change (Nohrstedt 2010). Aggregating actors into coalitions can follow the rule of first identifying actors sharing similar belief systems, and subsequently searching for a nontrivial degree of coordination among those actors (Henry 2011). It then also raises original questions such as the degree of cross-coalition interactions, intracoalition cohesiveness, and factors contributing to coalition defection (Jenkins- Smith, St. Clair, and Woods 1991). Policies and programs incorporate implicit theories reflecting the translated beliefs of one or more coalitions. Public policy can be conceptualized and defined in multiple ways (Birkland 2010, 8). Whereas some definitions can be simply stated and communicated, such as defining public policy as any inaction and action by government, other definitions are more nuanced and insightful. Lasswell and Kaplan (1950, 71), for example, describe policy as “a projected program of goal values and practices.” Notable from this definition, and from similar ones, is the insight that public policy consists of translations of the belief systems of the designers. In this regard, public policies represent the political maneuvering and negotiations not just among coalitions but also of causal theories (Pressman and Wildavsky 1973, xv; Mazmanian and Sabatier 1983, 5). Causal theory, when used to describe the implicit or explicit content of public policy, refers to the sequence of steps, a linking of anticipated events, or desired procedures that describe the reasoning for achieving outputs and outcomes of a public policy. Analysts applying the ACF should, therefore, interpret policies not just as the actions or inactions of government but also as the translations of belief systems as manifested in goals, rules, incentives, sanctions, subsidies, taxes, and other instruments regulating any given issue (Jenkins-Smith et al. 2014, 486). This interpretation of policy provides insight into why coalition actors advocate so intently over time and how they interpret public policies as bolstering or as being antithetical to their belief systems. Scientific and technical information is important for understanding subsystem affairs. In the previous assumption, belief systems were described as the mechanism for simplifying and interpreting the world. Belief systems are not, however, simply abstract representations of values and priorities but also encapsulate policy actors’ perceived causal patterns and relationships that shape the empirical world. A major source of this causal representation in a given context is scientific and technical information that can point to specific causal relations, problem attributes, and, sometimes, policy alternatives. To better understand policy processes is thus to understand how scientific and technical explanations are integrated into (or deflected from) belief systems, used in political debates and negotiations, and integrated with other forms of knowledge, especially local knowledge.11 Researchers should adopt a long-term time perspective (e.g., ten years or more) to understand policy processes and change. Policy processes are ongoing without beginning or end (Lindblom 1968, 4) and, thus, strategic behavior and learning of coalition actors, the reasoning and patterns of policy change, and assessments of the success or failure of public policy should be understood from a long-term perspective. The point has been misinterpreted to mean that a perspective of ten years or more is required to interpret policy processes through the ACF. This is too literal of an interpretation and often prevents interested analysts from applying the ACF even if the framework could help answer their research question. Some questions, for example, require intensive methods of data collection that preclude longitudinal data, such as an understanding of coalition structure using quantitative network analysis approaches (Henry 2011). Other datasets permit long-term perspectives, such as the multidecade perspectives taken by Albright (2011), Andersson (1999), and others to understand patterns of policy change. We also know that coalitions, though existing for decades, often take short-term perspectives as opportunities and constraints alter their immediate strategies (Jenkins-Smith, St. Clair, and Woods 1991). The general meaning behind this assumption is the recognition that understanding public policy requires focusing on temporal processes that characterize public policy over time. Scope A framework’s scope provides the set of general questions about the policy process that it helps the analyst answer. The traditional scope of the ACF includes questions involving coalitions, learning, and policy change. As suggested by the assumptions above, the framework is most useful for understanding these topics in high-conflict situations at the subsystem level of analysis. However, the framework has been applied in other settings, such as at the organizational level in collaborative settings (Leach and Sabatier 2005; Leach et al. 2013), a form of application to which we return when discussing future research agendas. General Conceptual Categories and Relations Flow diagrams are useful for identifying general categories of concepts and how they relate. Figure 4.1 presents a flow diagram depicting the policy process within the ACF.12 The policy subsystem is represented by the rectangle on the right illustrating a case with two competing coalitions representing their actors’ beliefs and resources. The two coalitions use various strategies to influence decisions by government authorities that affect institutional rules, policy outputs, and, eventually, policy outcomes. These decisions then feed back into the policy subsystem but also can affect external subsystem affairs. One category of variables that condition subsystem affairs includes relatively stable parameters, which are the basic social, cultural, economic, physical, and institutional structures that embed a policy subsystem (Hofferbert 1974; Heclo 1974). Some concepts within relatively stable parameters are best conceptualized as external to subsystem affairs, such as the basic constitutional structure of the political system, whereas others can be internal to the subsystem, such as physical conditions of the subsystem. A second category of variables consists of dynamic external events, which includes relevant features external to the subsystem and prone to change. Examples include socioeconomic conditions, the state of subsystem-relevant technology, public opinion, the composition of governing coalitions (Burnham 1970), and spillover effects from other policy subsystems. The listings under relatively stable parameters and dynamic external events in Figure 4.1 are illustrative examples and are not exhaustive; clearly, other concepts can be placed in each category, such as crises and disasters under dynamic external events (Nohrstedt 2011; Jenkins-Smith, St. Clair, and Woods 1991). In between relatively stable parameters and a policy subsystem is an intermediary category of concepts concerning the nature of the long-term coalition opportunity structures that establish the degree of consensus needed for major policy change, the openness of the political system, and overlapping societal cleavages. Essentially, long-term coalition opportunity structures are some of the important byproducts of the relatively stable parameters on policy subsystems. Between external events and policy subsystems are the short-term constraints and resources of subsystem actors; this means that changes outside the subsystem provide short-term opportunities for coalitions to exploit. THEORETICAL EMPHASES Theoretical Focus on Policy Change One of the central objectives of the ACF is to contribute to the understanding of policy change and stability, and this has been the subject of considerable empirical investigation. Thanks to these contributions, we now have more detailed knowledge about the nature and causes of policy change within and across policy subsystems than we had just a few decades ago. What has provoked this focus is the recurrent observation that, although many public policies and programs remain stable over long periods of time, others are subject to periods of dramatic and nonincremental change (Sabatier 1988; Baumgartner and Jones 1993).13 For example, indicators of such policy change may include revisions in policy core components of governmental programs, termination of programs, or launching of new programs. Similarly to other theoretical perspectives on policy change (Baybrook and Lindblom 1963; Hall 1993; Rose 1993), the ACF focuses on the directionality of policy evolution and makes a clear distinction between minor and major policy change (Capano 2009, 2012; Howlett and Cashore 2009; Nisbet 1972). The level of change in a governmental program is defined according to the extent to which alterations deviate from previous policy. The ACF assumes that public policies and programs are translations of policy-oriented beliefs and can be conceptualized and measured hierarchically, like belief systems. Change in the core aspects, defined as “major policy change,” indicates significant shifts in the direction or goals of the subsystem, whereas change in secondary aspects (e.g., change in means for achieving the goals) is evidence for “minor policy change” (Sabatier and Jenkins-Smith 1999, 147–148). Advocacy coalitions often disagree on proposals related to these components, and policy debates therefore often revolve around diverging preferences regarding initiatives to either change or preserve governmental programs (Sabatier and Weible 2007, 195). Since the belief system categories differ according to their susceptibility to change, minor policy change should be not as difficult to achieve as major policy change (Sabatier 1988). For example, minor changes in administrative rules, budgetary allocations, statutory interpretation, and revision are relatively frequent and do not necessitate as much evidence, agreement among subsystem actors, or redistribution of resources. By contrast, because normative (policy core) beliefs are rigidly held and screen out dissonant information, major policy change is unlikely as long as the advocacy coalition that instituted the program remains in power. The ACF offers four conceptual pathways to policy change. The first is attributed to some external source (e.g., as might be found in the categories of dynamic external events or even relatively stable parameters from Figure 4.1). External shocks, or perturbations, include events outside the control of subsystem participants (in terms of their ability to influence underlying causes and triggers) and involve change in socioeconomic conditions, regime change, outputs from other subsystems, and extreme events such as some crises and disasters. These events increase the likelihood of major policy change but require one or several enabling factors (causal mechanisms), including heightened public and political attention, agenda change, and most importantly redistribution of coalition resources and opening and closing of policy venues (Sabatier and Weible 2007, 198–199). A key factor in this regard is mobilization by minority coalitions to exploit the event, for instance, by pursuing public narratives to attract attention to favored courses of action and by appealing to new actors (Sabatier and Jenkins-Smith 1999, 148; see also McBeth et al. 2007; Nohrstedt 2008). Because of the importance of these intervening steps, it has been hypothesized that significant perturbations external to the subsystem are one of the necessary, but not sufficient, paths for changing the policy core attributes of a governmental program (Sabatier 1988). Major policy change may also result from a second pathway based on internal events that (1) occur inside the territorial boundaries and/or the topical area of the policy subsystem and (2) are more likely affected by subsystem actors (Sabatier and Weible 2007, 204–205). Various types of internal events, including crises, policy fiascoes, scandals, and failures, are likely to influence beliefs and heighten attention to certain governmental programs (Birkland 2006; Bovens and ’t Hart 1996). Advocacy coalitions can be expected to engage in framing contests over such events and debate the severity of problems, their underlying causes, attribution of responsibility, and policy implications (Boin, ’t Hart, and McConnell 2009; Nohrstedt and Weible 2010). Internal events can be expected to confirm the policy core beliefs of minority coalitions and increase doubts about the core beliefs of the dominant coalition and bring into question the effectiveness of their policies. Whether or not internal shocks result in major policy change depends on the same mechanisms that mediate the effect from external shocks. A third source of minor policy change is policy-oriented learning, but this is likely to happen incrementally over longer periods of time. Following Caplan, Morrison, and Stanbaugh (1975) and Weiss (1977), Sabatier (1988) expects that policy analysis seldom influences specific governmental decisions but often serves an “enlightenment function” by gradually altering the concepts and assumptions of subsystem participants. In addition, learning can also facilitate major policy change, but this is more likely when learning takes place in conjunction with an external or internal shock (Nohrstedt 2005). A fourth pathway to policy change is through negotiated agreement among previously warring coalitions and may result in substantial change in governmental programs. Negotiated agreements may emerge in a variety of ways but are facilitated by collaborative institutions conducive to negotiation. Specifically, Sabatier and Weible (2007, 205–206) identify nine prescriptions fostering negotiation: a “hurting stalemate,” broad representation, leadership, consensus decision rules, funding, commitment by actors, importance of empirical issues, trust, and lack of alternative venues. The most important condition instigating negotiations is a “hurting stalemate,” which occurs when warring coalitions perceive the status quo as unacceptable and do not have access to alternative venues for achieving their objectives (Weible and Nohrstedt 2012, 132). A recent review of ACF case studies shows that among 161 empirical applications from 2007 to 2014, learning is the most frequently cited source of policy change (identified in 29 percent of the applications reviewed), followed by external sources and events (28 percent), negotiated agreements (14 percent), and internal events (6 percent) (Pierce et al. 2017). In summary, the original version of the ACF offered two hypotheses of policy change, focusing on external perturbations and power shifts. However, Weible and Nohrstedt (2012, 133) merge the four pathways to policy change into a single hypothesis: Policy Change Hypothesis 1. Significant perturbations external to the subsystem, a significant perturbation internal to the subsystem, policy-oriented learning, negotiated agreement, or some combination thereof is a necessary, but not sufficient, source of change in the policy core attributes of a governmental program. There has been strong support for the first policy change hypothesis. Many find support for at least one of the pathways (Barke 1993; Bischoff 2001; Green and Houlihan 2004; Tewari 2001; Kübler 2001; Dudley and Richardson 1999). One challenge in testing this hypothesis is the occurrence of one of the pathways without a change in policy (Weible, Sabatier, and McQueen 2009; Sotirov and Memmler 2012). Another challenge is explaining minor policy changes after an external shock (Burnett and Davis 2002; Davis and Davis 1988). Critical in testing the first hypothesis about policy change is to understand how a coalition can capitalize on (or exploit) the opportunity, which ultimately involves attempts to either preserve the status quo or seek policy change. This has led some analysts to focus heavily on coalition resources and strategies following various external events and developments (Smith 2000; Ameringer 2002; Albright 2011; Ingold 2011; Nohrstedt 2005, 2008). The second hypothesis relates coalition influence in the subsystem, major policy change, and nested policy subsystems: Policy Change Hypothesis 2. The policy core attributes of a government program in a specific jurisdiction will not be significantly revised as long as the subsystem advocacy coalition that instated the program remains in power within that jurisdiction—except when the change is imposed by a hierarchically superior jurisdiction. There is strong to partial support for Policy Change Hypothesis 2 (Sotirov and Memmler 2012). Studies that confirm the logic of the second policy change hypothesis include Ellison (1998), Olson, Olson, and Gawronski (1999), Elliot and Schlaepfer (2001), and Kübler (2001). However, this second policy change hypothesis has been tested but a few times. One of the next steps in studying policy change will be to focus on developing best practices for documenting and explaining policy while accounting for context. Too many studies of policy change apply different methods of data collection and analysis, with the result that comparison across cases is difficult. In addition, studies adopt slightly different definitions of policy, which complicates the task of comparing drivers of policy change across governing systems. Advocacy Coalitions Advocacy coalitions are defined by actors who share policy core beliefs and who coordinate their actions in a nontrivial manner to influence a policy subsystem. In studying coalitions, analysts typically focus on a range of topics, from the structure and stability of coalition actor belief systems to the formation and maintenance of coalitions over time. The traditional hypotheses about advocacy coalition include the following: Coalition Hypothesis 1. On major controversies within a policy subsystem when policy core beliefs are in dispute, the lineup of allies and opponents tends to be rather stable over periods of a decade or so. Coalition Hypothesis 2. Actors within an advocacy coalition will show substantial consensus on issues pertaining to the policy core, although less so on secondary aspects. Coalition Hypothesis 3. Actors (or coalitions) will give up secondary aspects of their belief systems before acknowledging weaknesses in the policy core. Coalition Hypothesis 4. Within a coalition, administrative agencies will usually advocate more moderate positions than their interest group allies. Coalition Hypothesis 5. Actors within purposive groups are more constrained in their expression of beliefs and policy positions than actors from material groups. From these hypotheses, evidence to date largely confirms Coalition Hypothesis 1 about the stability of coalitions over time (see Pierce et al. 2017 review paper). To assess the stability of coalitions, most of these studies use coded legislative statements (Jenkins-Smith, St. Clair, and Woods 1991; Jenkins-Smith and St. Clair 1993; Sabatier and Brasher 1993; Zafonte and Sabatier 2004; Pierce 2011; Nohrstedt 2010), with a few studies using survey and interviews (Weible, Sabatier, and McQueen 2009; Ingold 2011) and discourse analysis (Leifeld 2013). Important in these studies is the documentation that although coalitions are generally stable over time defection is not uncommon and membership often changes. Analysts have documented a range of reasons for defection or change in coalition composition, such as extreme coalition actors defecting to prevent the adoption of “balanced” policies (Munro 1993, 126); major internal or external events that switch allegiances, especially elections (Jenkins-Smith, St. Clair, and Woods 1991; Zafonte and Sabatier 2004; Pierce 2011); and strategic decisions by coalition actors to achieve shortterm political objectives (Nohrstedt 2005; Larsen, Vrangbaek, and Traulsen 2006). To further develop Coalition Hypothesis 1, the next steps must develop and test a range of theoretical rationales for the stability or instability of coalitions. The testing of Coalition Hypotheses 2 and 3 has resulted in only a few confirmations (Weyant 1988; Elliot and Schlaepfer 2001; Kim 2003) but many falsifications and, at best, findings of partial support (Barke 1993; Jenkins-Smith and St. Clair 1993; Sabatier and Brasher 1993; Olson, Olson, and Gawronski 1999; Sobeck 2003; Larsen, Vrangbaek, and Traulsen 2006; Ingold 2011; Zafonte and Sabatier 2004). There are at least two interpretations for the mixed support for Coalition Hypotheses 2 and 3. The first interpretation involves variation in conceptualizations and measurement of belief systems in establishing coalitions. If this interpretation is correct, there needs to be a concerted effort to clarify the theoretical distinction between policy core and secondary aspects as well as methodological guidelines for measurement. Olson, Olson, and Gawronski (1999), for example, found it difficult to isolate policy core beliefs from secondary aspects. The second interpretation points to a faulty or imprecise model of the belief system and overall logic of Coalition Hypotheses 2 and 3. To put it simply, even if analysts could adequately measure and distinguish policy core and secondary aspects, perhaps Coalition Hypotheses 2 and 3 are wrong. Although we are not in a position in this chapter to reject both hypotheses, we underscore the mixed support for them and draw attention to a need for better approaches in conceptualizing and measuring belief systems in the ACF. The fourth and fifth hypotheses are rarely tested in the ACF. Evidence supporting the Coalition Hypothesis 4 remains mixed, with some evidence offering confirmation (JenkinsSmith, St. Clair, and Woods 1991; Jenkins-Smith and St. Clair 1993) and others providing only partial to no support (Sabatier and Brasher 1993). The most important confirmation for Coalition Hypothesis 5 remains Jenkins-Smith, St. Clair, and Woods (1991) and JenkinsSmith and St. Clair (1993). The implication from this assessment is clear enough: there is a need for renewed testing and development of Coalition Hypotheses 4 and 5. Although it is not a traditional hypothesis, a large number of studies have tested the expectation that coalitions form on the basis of shared beliefs, known as the Belief Homophily Hypothesis. Studies confirming this hypothesis can be found in a number of publications, including Weible (2005), Matti and Sandström (2011), Henry (2011), Ingold (2011), and Leifeld (2013). Whereas the results tend to confirm the Belief Homophily Hypothesis, the findings raise two new implications for studying coalitions under the ACF. The first implication is the presence of other factors, outside of beliefs, that affect coalition formation and stability. These other factors include, but are not limited to, perceived influence or resources of others (Weible 2005; Matti and Sandström 2011), interests (Nohrstedt 2010), and trust (Henry, Lubell, and McCoy 2011). The second implication is that coalitions are shaped more by sharing opponents than by sharing beliefs (Henry, Lubell, and McCoy 2011). Research on the Belief Homophily Hypothesis supports the argument that beliefs remain a major factor in forming and maintaining coalitions, but other factors clearly have an effect, and the precise role of beliefs in shaping coalitions needs theoretical refinement. The traditional hypotheses about coalitions highlight some of the theoretical logic about coalitions and many of the most important concepts. However, the theoretical argument about coalitions is broader than is articulated in the listed hypotheses and sometimes includes additional concepts and their interrelations, some of which are summarized below in four categories. • Dominant and minority coalitions. Although some subsystems exhibit advocacy coalitions steeped in conflict marked by long periods of ongoing one-upmanship, other subsystems exhibit a “dominant” coalition that largely controls (most likely through resource superiority) subsystem politics and policy, and either a “minority” coalition vying for influence or the absence of any coordinated opposition. Even though a number of studies have documented the stability of dominant advocacy coalitions in steering a policy subsystem, the attributes of various coalitions remain underdeveloped, particularly the comparison of beliefs, resources, strategies, and activities. • Overcoming threats to collective action. One of the critical theoretical arguments that has yet to be sufficiently developed involves how coalitions overcome threats to collective action (Schlager 1995). Actors form coalitions and overcome threats to collective action on the basis of three rationales (Zafonte and Sabatier 1998; Sabatier and Weible 2007, 197). First, similar beliefs among allies reduce the transaction costs for coordination. Second, actors are involved in policy subsystems at different levels of intensity and, thus, some engage in weak forms of coordination (sharing information) and others in strong forms of coordination (jointly developing and executing shared plans). Third, actors often experience the devil shift and, therefore, exaggerate the costs of inaction and the need for action (Sabatier, Hunter, and McLaughlin 1987). • Principal and auxiliary coalition actors. Network analysis techniques have shown that some coalition actors are more central to a coalition than others and that sometimes actors rarely interact with their allies. To account for this variation in coalition membership, a distinction is made between actors who are principal and those who are auxiliary to a coalition (Larsen, Vrangbaek, and Traulsen 2006; Silva 2007; Zafonte and Sabatier 2004; Weible 2008). Principal actors are expected to be more central and consistent coalition members, whereas auxiliary actors are expected to be on the periphery, involved intermittently or sometimes only for a short period of time, and therefore not as regularly engaged in coalition-related activities. • Resources, strategies, and activities. Coalitions are marked not only by shared beliefs and coordination patterns but also by their resources. These resources include formal legal authority to make policy decisions, public opinion, information, mobilizable supporters, financial resources, and skillful leadership (Sabatier and Weible 2007; Weible 2007; Nohrstedt 2011; Ingold 2011; Albright 2011; Elgin and Weible 2013). Resources are an important contribution that provide the theoretical leverage for understanding the capacity of a coalition to make strategic decisions and engage in various activities to influence policy subsystems. Overall, the support for the study of coalitions is strong for Coalition Hypothesis 1 and for the Belief Homophily Hypothesis, mixed for Coalition Hypotheses 2 and 3 that involve the hierarchical belief systems in the ACF, and mostly untested for Coalition Hypotheses 4 and 5. Several underdeveloped areas within this theoretical emphasis involve the role of coalition resources, strategies, and activities; the role and type of coalition members; the testing of the argument involving the collective action rationale for the formation of coalitions; and the continued development of dominant and minority coalitions. Policy-Oriented Learning Policy-oriented learning is one prominent pathway within the ACF for the explanation of policy change and plays a central role in belief change and reinforcement of members of advocacy coalitions. If it has always been of central focus within the ACF, it is possibly still the most intractable concept to study (Bennett and Howlett 1992; Levy 1994). Policyoriented learning is defined as “enduring alternations of thought or behavioral intentions that result from experience and which are concerned with the attainment or revision of the precepts of the belief system of individuals or of collectives” (Sabatier and Jenkins-Smith 1993, 42). Learning is associated with changes in beliefs systems of coalition members that include not only the understanding of a problem and associated solutions but also the use of political strategies for achieving objectives (see May 1992). Some of the important questions in the study of learning include: What belief system components change or remain the same through learning? What contexts foster learning by coalition members? How does learning diffuse among allies and possibly among opponents? What is the role, if any, of policy brokers in facilitating learning among opponents? The theory underlying learning in the ACF emphasizes four categories of explanatory factors. • Attributes of forums. Forums are the venues where coalitions interact, debate, and possibly negotiate. Jenkins-Smith (1982, 1990, 99–103) makes the theoretical argument about how the attributes of a forum, essentially the forum’s institutional arrangement, affect the extent that learning occurs among allies and opponents. A couple of the most important attributes defining a forum are the degree of openness in participating (open vs. closed forums) and the extent that participating actors share a common analytical training and norms of conduct. • Level of conflict between coalitions. Level of conflict relates to the extent that actors perceive a threat to their policy core beliefs from their opponents’ objectives or actions. Jenkins-Smith (1990, 95–97) and Weible (2008) essentially argue for an inverted quadratic relationship between level of conflict and learning between members of opposing coalitions, which has been called “cross-coalition learning.” At low levels of conflict, there is little cross-coalition learning because coalition actors attend to other subsystem affairs. At high levels of conflict, there is also little crosscoalition learning because coalition actors defend their positions and reject information that disputes their belief systems. At intermediate levels of conflict, opposing coalitions are threatened just enough to attend to the issue and remain receptive enough to new information to increase the likelihood for cross-coalition learning. • Attributes of the stimuli. Attributes of the stimuli relates to the type of information and experience coalition actors are exposed to. Jenkins-Smith (1990, 97–99) argues that analytically intractable phenomena involve uncertainty, low-quality data, and, hence, variation in interpretation and high levels of disagreement. The more intractable an issue, the lower the level of cross-coalition learning expected. • Attributes of actors. Attributes of the individual actors include their belief system, resources, strategies, and network contacts. Given the importance of belief systems in filtering and interpreting information, for example, the expectation is that coalition actors with extreme beliefs are less likely to learn from opponents than are coalition actors with more moderate beliefs. Additionally, some actors can serve as policy brokers who primarily seek to mitigate the level of conflict and help opponents reach agreements (Sabatier and Jenkins-Smith 1993, 27). There are no predetermined criteria defining who can or cannot be a broker within a subsystem; indeed, a broker could be affiliated with any organization type, from academia to government to the private or nonprofit sector. One important role for brokers is facilitating learning among opponents (Ingold and Varone 2012). These four attributes can be found in the following five hypotheses on policy- oriented learning within the ACF: Learning Hypothesis 1. Policy-oriented learning across belief systems is most likely when there is an intermediate level of informed conflict between the two coalitions. This requires that: (1) each has the technical resources to engage in such a debate, and (2) the conflict is between secondary aspects of one belief system and core elements of the other or, alternatively, between important secondary aspects of the two belief systems. Learning Hypothesis 2. Policy-oriented learning across belief systems is most likely when there exists a forum that is: (1) prestigious enough to force professionals from different coalitions to participate and (2) dominated by professional norms. Learning Hypothesis 3. Problems for which accepted quantitative data and theory exist are more conducive to policy-oriented learning across belief systems than those in which data and theory are generally qualitative, quite subjective, or altogether lacking. Learning Hypothesis 4. Problems involving natural systems are more conducive to policyoriented learning across belief systems than those involving purely social or political systems because in the former many of the critical variables are not themselves active strategists and because controlled experimentation is more feasible. Learning Hypothesis 5. Even when the accumulation of technical information does not change the views of the opposing coalition, it can have important impacts on policy— at least in the short run—by altering the views of policy brokers. Studies of policy-oriented learning have not always supported these hypotheses. A good number of studies have documented learning at both secondary (expected) and policy core (not expected) levels of the belief system (Sabatier and Brasher 1993; Eberg 1997; Elliot and Schlaepfer 2001; Larsen, Vrangbaek, and Traulsen 2006). These results echo the mixed support for Coalition Hypotheses 2 and 3. That is, the hierarchical belief system of the ACF— especially the distinction between policy core and secondary aspects—is not finding strong support in many of the hypothesis tests. In support of Learning Hypothesis 3, Sotirov and Memmler (2012) find in their review of the literature that a handful of studies show that learning was limited when data were lacking or were of qualitative or subjective nature (e.g., Weyant 1988; Elliot and Schlaepfer 2001; Nedergaard 2008), but the findings also show the same for situations for learning using quantitative data (Elliot and Schlaepfer 2001; Kim 2003). Studies have found that learning is more likely to occur with tractable issues, with intermediate levels of conflict, and with the availability of scientific and technical information (Larsen, Vrangbaek, and Traulsen 2006; Meijerink 2005; Elliott and Schlaepfer 2001). Providing indirect support for learning within the ACF, Leach et al. (2013) find forum structure, attributes of the individual learner, and level of scientific certainty affected belief change and knowledge acquisition. This is an area in need of renewed theoretical and empirical attention. One promising direction of research involves development of the policy broker concept (Ingold and Varone, 2012). This work finds support for Learning Hypothesis 5 and identifies some systematic evidence that certain actor types are more likely to play the broker role than others. These actors are not acting in an altruistic way: to engage in a brokerage role, they need a certain level of self-interest and an awareness of the potential benefits from policy compromise or the potential losses from the status quo. Across many of the applications of learning, the most pressing concern is the inconsistency in conceptualization and measurement of the concept. And similarly to the change hypotheses, what also needs to be addressed in this theoretical emphasis is a set of best practices for studying learning within and across advocacy coalitions. There must also be a fresh look at the factors that shape learning, including levels of conflict, attributes of the actor, the role of policy brokers, nature of stimuli, and characteristics of the forum. A FUTURE RESEARCH AGENDA The future trajectory of the ACF depends on the innovative and creative efforts of numerous analysts from around the world. Nonetheless, we offer a research agenda for analysts to consider in moving the framework forward. Reconsider the ACF’s belief system. Empirical applications of the ACF suggest that the belief system model needs to be further specified. There are many ways forward, including clarifying the distinction between policy core and secondary beliefs, combining the policy core beliefs and secondary beliefs into a single category under deep core beliefs, and drawing inspiration from other theories, such as the value-belief-norm theory (Stern 2000; Henry and Dietz 2012), cultural theory (Douglas and Wildavsky 1982; Jenkins-Smith et al. 2014), and Narrative Policy Framework (Shanahan, Jones, and McBeth 2011). Advance the theory and measures of learning. Despite its centrality to the framework, conceptual development of policy-oriented learning—including causes, kinds of learning, and implications—is among the least mature components of the ACF. Analysts are encouraged to undertake reexamination of this concept within the framework as well as the theoretical implications. Research in this domain should emphasize clear conceptualization and measurement of various products of learning and the processes by which it is encouraged and inhibited (Heikkila and Gerlak 2013). Refine the theory of coalition structures and coordination. The study of coalitions remains a staple of the framework, and significant advances in understanding coalitions have occurred over the past decade, particularly with network analysis techniques (Henry 2011). This effort should continue with special attention to the assumed hierarchy of belief homophily and coordination patterns among coalition members (Calanni et al. 2015; Ingold and Fischer 2014). It also needs to focus on the sources of stability of coalitions, with attention to the likelihood and reasons for defection by coalition members. Develop a hierarchy for coalition resources. The ACF assumes that access to and exploitation of various political resources are important for advocacy coalitions as they seek to influence public policy. Following Sewell (2005) and Sabatier and Weible (2007, 201– 204), we encourage efforts to identify a typology of political resources that includes formal legal authority to make policy decisions, public opinion, information, mobilizable troops, financial resources, and skillful leadership. Although coalition resources were long neglected in empirical research (Sabatier and Weible 2007, 201), recent studies have investigated how coalitions mobilize and exploit resources in the policy process (Albright 2011; Ingold 2011). These studies confirm that redistribution of political resources is an important step in explaining policy change. Meanwhile, as suggested by Nohrstedt (2011, 480), some resources are more important than others for coalitions to achieve influence, which is ultimately given by governing system attributes such as constitutional rules. For example, having coalition actors in positions of legal authority is a major resource because legislators are veto players whose agreement is needed for policy change (Tsebelis 1995; see also Sabatier and Pelkey 1987). Legal authority is also one defining element of a dominant coalition, which has more of its allies in positions of formal authority than do minority coalitions (Sabatier and Weible 2007, 203). Resources could therefore be hierarchically arranged with regard to their perceived usefulness and effectiveness to coalitions, which in turn raises challenges and questions for future research (Weible et al. 2011, 356–357). For example, under what conditions are some resources more important than others for coalitions to gain influence? Which strategies do coalitions utilize to select which resources to exploit? What is the relative importance of specific kinds of resources in different political systems? How does redistribution of resources influence policy change and learning? A related challenge is to advance approaches to operationalize resources by, for example, network analysis (Ingold 2011) and qualitative research (Mintrom and Vergari 1996; Nohrstedt 2011). Study venues and forums within policy subsystems. The focus of the ACF on policy subsystems has important impacts on conducting research. However, some notable applications of the ACF have focused on organizational-level analysis, especially in the area of collaborative partnerships (Leach and Sabatier 2005; Leach et al. 2013). For example, Leach and Sabatier (2005) applied the ACF in the study of watershed partnerships. These partnerships, however, do not encapsulate the entire policy subsystem but rather involve a single venue within the subsystem. As a result, the study of the partnership represents a selected sample of subsystem actors choosing to participate in the partnership. Such organizational-level applications of the ACF are encouraged because we gain a deeper understanding of how coalition actors learn from each other and negotiate agreements. Additionally, because coalitions seek to affect government decisions through venues, the choice of one venue over another remains an important topic of study. Use the ACF for comparative public policy research. Most comparative work on the ACF has been based on implicit comparison across political- institutional systems. Few empirical studies based on the ACF systematically compare policy subsystems, coalition behavior, and policy processes across political systems (Gupta 2012). One is an ACF study applied in seven countries (Canada, France, Germany, Sweden, Switzerland, the United Kingdom, and the United States) that compares the policies and regulations related to oil and gas development using hydraulic fracturing (Weible et al. 2016). Using different methods of data collection and analyses, the same research questions about advocacy coalitions and the propensity for policy change were asked and answered. The comparison confirmed the importance of subsystem properties for explaining differences observed across the seven countries (Ingold et al. 2016). Not only do basic institutional and constitutional arrangements of the political system decisively affect coalition formation and the propensity for policy change but also subsystem attributes (such as jurisdictional level, maturity, or autonomy) and issue characteristics (such as salience and potential threat to certain values within the belief system) do as well. We encourage future work in this direction, developing systematic comparisons of policy subsystems across countries to disentangle the factors accounting for advocacy coalitions, policy-oriented learning, and policy change. The expansion of applications to new countries is a trend that can inspire future comparison across systems. Comparative work obviously brings additional costs in terms of data acquisition and analysis but also important gains in terms of new insights regarding the role of political institutions and cultures in shaping the formation, maintenance, and behavior of advocacy coalitions in the policy process. Here is a gap waiting to be filled. Fruitful avenues for future comparative work involve vicarious policy-oriented learning (how coalitions learn from the experience of others) and policy transfer (how policies diffuse from one political system to another) (Bandura 1962; Dolowitz and Marsh 1996). Following the emphasis on coalition opportunity structures (Kübler 2001; Sabatier 1998; Sabatier and Weible 2007), there is also a need to investigate how specific institutional attributes such as veto players, the required level of consensus, and system openness shape coalition interaction and policy change (Fischer 2015; Gupta 2013). Empirical research in these areas would yield important insights about the policy process and expose questions and areas for future research, including the role and importance of advocacy coalitions as a type of political organization actors exploit to coordinate strategies and gain influence. Although comparative analysis is a long-term challenge and will probably generate limited generalizability in the short term given the complexity of the policy process (Schmitt 2012), the ACF offers concepts and assumptions that should stimulate and facilitate comparative analysis. Focus on types of actors, including auxiliary and principal coalition actors, policy brokers, and policy entrepreneurs. Exceptional actors often play critical roles in policy subsystems. Some of these actors could be principal coalition actors but possibly auxiliary coalition actors. Other categories are policy brokers (Ingold and Varone 2012) and policy entrepreneurs (Mintrom 2009; Mintrom and Vergari 1996). From its earliest renditions, the ACF has suggested that brokers can play important roles in policy-oriented learning (Sabatier and Jenkins-Smith 1993), and empirical applications have provided some evidence on brokers’ impact on policy outputs (Ingold and Varone 2012; Ingold 2011). But further research is needed to theoretically and empirically refine the role of brokers in policy subsystems in general (Christopoulos and Ingold 2015) and in the design of learning mechanisms in particular. Policy entrepreneurs might also be critical players in maintaining coalitions and causal drivers of policy change, but few have analyzed this type of exceptional actor in ACF studies. Focus on nascent and mature policy subsystems. Most studies of the ACF focus on mature policy subsystems. In mature policy subsystems, policy actors have typically fortified their belief systems about the risks and benefits associated with an issue, they interact in stable advocacy coalitions, and conflicts among opponents have endured over time both within and across decision making venues. Sometimes, mature policy subsystems absorb new issues as they emerge on the political agenda, whereas on other occasions new issues provoke the formation of a new policy subsystem (Nohrstedt and Olofsson 2016a). Unfortunately, few scholars have studied nascent policy subsystems (Ingold, Fischer, and Cairney 2016; Stritch 2015; Beverwijk, Goedegebuure, and Huisman 2008). As a result, theoretical insights about nascent subsystems remain underdeveloped. Speculatively, nascent policy subsystems are likely to feature policy actors with ambiguous perceptions of the risk and benefits of a policy issue, unclear preferences for known policy solutions, and unstable alliances among allies and opponents. Studies on nascent subsystems could yield insights about the initial conditions of policy subsystem characteristics, the process of coalition formation, the establishment of interactions within and across coalitions, and the role of coalitions in agenda setting. A focus on nascent policy subsystems will allow scholars to adopt a prospective approach (e.g., how does the variation in initial conditions in nascent policy subsystems give rise to differences in conditions in mature policy subsystems?), help identify the reasons for nascent subsystem formation (e.g., in response to a crisis, a policy change, or other), and assist in investigating the propensity for future policy change (Weible et al. 2016). Expand our understanding of science and policy analysis in the policy process. The ACF was originally created to help inform the role of scientists and science in the policy process. Several recent publications address this area. Much of this work began with Jenkins-Smith’s (1990) theoretical and empirical efforts in studying the role of policy analysis in the policy process. Since then the effort has shifted mostly to the roles of scientists and technicians and scientific and technical information in the policy process (Jenkins-Smith and Weimer 1985; Weible 2008; Silva and Jenkins-Smith 2007; Silva, Jenkins-Smith, and Barke 2007; Weible, Sabatier, and Pattison 2010; Montpetit 2011; Lundin and Öberg 2014). This research strongly suggests that the use of science and policy analysis is driven by the level of conflict in the policy subsystem (Jenkins-Smith 1990; Weible 2008). The next step is to test these expectations under different conditions and develop a coherent theoretical explanation for the findings. Establish common methods of data collection and analyses for applying the framework, identify trade-offs in using different methods, and promote contextually based theoretical innovations. The ACF is a tool for comparative analyses of policy processes. To foster comparative work, there is a need to develop common methods of collecting and analyzing data given common research questions. Clearly, some methods of data collection and analysis are more suitable in some contexts than in others (e.g., online surveys vs. interviews). Similarly, other methods of data collection and analysis are feasible when directly comparing policy subsystems over time (e.g., newspaper content analysis). The best strategy is not to promote one method of data collection and analysis over another but rather to utilize the best methods given the research questions, contexts, and resources of the researchers. To support such an effort, researchers must recognize the trade-offs of different approaches and, ideally, combine more than one to capture their respective strengths and compensate for their weaknesses. Explore the need for theoretical refinement emanating from application in nontraditional settings. Underlying the need for comparative methods is a simultaneous need for contextually based theoretical development. The majority of empirical applications of the ACF involves cases of mature policy subsystems marked by high conflict in heavily democratized political systems (Weible et al. 2016; Pierce et al. 2017). Application of the ACF to nascent subsystems, to policy subsystems marked by low or moderate levels of conflict, or to policy subsystems within different types of political systems is less frequent and might require theoretical innovations and adjustments. Prior comparative work on the ACF outside the United States and Western Europe reports strengths as well as weaknesses; studies confirm the applicability of the ACF’s concepts and assumptions, and they identify limitations related to descriptive and explanatory validity (Henry et al. 2014). Although some of these limitations (as discussed above in this chapter) apply more broadly, scholars should also identify limitations that are related to the attributes of policy processes in (for example) hybrid or authoritarian regimes. Ascertaining how the ACF might be adapted to address this expanded array of contexts (without altering the framework’s axiomatic propositions) and how the ACF fares compared to alternative frameworks are important questions for the future.14 CONCLUSION Our intent in this chapter is to provide an overview of the ACF research program. The framework has attracted worldwide attention and scholarship over several decades, and we readily acknowledge that in this short chapter we have not been able to adequately incorporate all of the important theoretical and empirical contributions. To supplement this chapter, we highly recommend the excellent theoretical and methodological insights that can be found in Fischer (2015) on the role of institutions on coalition formation, in Sotirov and Memmler (2012) on the ACF in environmental and natural resource contexts, and in Leifeld (2013) on discourse coalitions. In addition, some of the best emerging work can be found in recent PhD dissertations (Gupta 2013; Valman 2014; Donadelli 2016). The continuing growth of ACF scholarship gives us some confidence that—over thirty years after its initial articulation—the framework is still undergoing progressive problem-shift. We conclude this chapter with a challenge. Although the ACF has spawned a fruitful research program on coalitions, learning, and policy change, we must raise the question: What ends will ACF research serve? Clearly, analysts applying the ACF must continue to use the best science available to improve and develop the framework and to seek answers to some of the most pressing puzzles about policy processes. But some analysts must also work toward developing the framework as a tool for informing and, possibly, improving actual policy processes. To what extent can the framework be used as a policy analysis tool for informing decisions (Nohrstedt 2013; Weible, 2007)? Can the logic of the framework help people strategically influence the policy process (Weible et al. 2012)? And can we eventually draw lessons from the framework to inform what may enhance (or undercut) the capacities of a policy process for the betterment of society? We do not have answers to these questions, but we encourage new and experienced analysts to take them on.
Purchase answer to see full attachment