moving policy theory forward: connecting multiple stream ...howlett/documents/howlett_et... · ing...

Australian Journal of Public Administration, vol. 00, no. 0, pp. 1–15 doi:10.1111/1467-8500.12191

RESEARCH AND EVALUATION

Moving Policy Theory Forward: ConnectingMultiple Stream and Advocacy CoalitionFrameworks to Policy Cycle Models of Analysis

Michael HowlettSimon Fraser University and National University of Singapore

Allan McConnellUniversity of Sydney

Anthony PerlSimon Fraser University

The stages/policy cycle, multiple streams, and Advocacy Coalition Framework (ACF) ap-proaches to understanding policy processes, all have analytical value although also attract-ing substantive criticism. An obvious direction for research is to determine whether themultiple streams framework and the ACF can be refined and applied to other dimensions ofpolicy-making set out in the policy cycle model. This article argues that extending and mod-ifying Kingdon’s framework beyond the agenda-setting stage is best suited to this endeavour.Doing so makes it possible to bring these three approaches into alignment and enhances ourunderstanding, although retaining the core insights of each.

Key words: Kingdon, policy cycle, stages approach, multiple streams, advocacy coalition framework

Introduction: Policy ProcessConceptualising and Policy Studies

A pivotal feature of policy studies since themid-1980s has been the development and useof several different analytical frameworks tohelp capture the main characteristics and dy-namics of policy processes (Pump 2011). Theseframeworks are oriented toward moving beyondthe particularities of policy-making processesin such a way as to guide investigators and helpboth students and practitioners make sense ofthe complex set of socio-political activities thatconstitute policy-making as well as its outputsand outcomes (Althaus et al. 2013; Cairney2013; Howlett et al. 2009). However in their

present state, these models contain contradic-tory elements and their use has led to manystudies and scholars focusing upon or promot-ing one model over another in a process of ‘du-eling analytical frameworks’.

The longest-standing such conceptual frame-work is the notion of the policy process be-ing constituted by sequential, cyclical, phases,or ‘stages’ of governmental problem-solving.Although many authors, including most no-tably those involved in the development of theAdvocacy Coalition Framework (ACF) model(Sabatier 1991) called for the supersession ofthe stages approach and this has been triedin different ways, it lives on as the domi-nant heuristic applied to public policy-making

C© 2016 Institute of Public Administration Australia

2 Howlett, McConnell and Perl xxxx 2016

(Howlett et al. 2009). This longevity is dueat least in part to a normative preference formore logical modes of policy-making on thepart of many policy scholars who support aproblem-solving perspective on the subject (Al-thaus et al. 2013). But it is also very mucha result of other factors such as the simplic-ity of the framework and its capacity to dealwith the multiple activities and the many tasksinvolved in policy-making, from problem def-inition to policy outcomes and evaluation.Rejecting criticism of The Australian PolicyHandbook for recycling Lasswell’s cyclicalheuristic for understanding policy deliberations(Everett 2003), Bridgman and Davis (2003:102) wrote in this journal that ‘Policy is a seriesof interlocking steps, a dialogue between pro-cedures and substance, between public debateand private analysis’, a view which fits nicelywithin the cycle ‘heuristic’ (Jann and Wegrich2007).

The problem-oriented, multiple task-orientedstages approach can be contrasted with theother current major models of policy-makingthat focus on particular dimensions of the sub-ject. One prominent example of these alter-natives is the ‘multiple streams’ frameworkand its ‘garbage can’ perspective on policy-making dynamics found in the work of JohnKingdon (1984) and his devotees (Zahari-adis 1995; Zohlnhofer et al. 2015). Studyingpolicy-making through the lens of several semi-independent ‘streams’ of events and actors in-teracting with each other to define and controlthe policy agenda stresses its constant com-plexity, its occasional chaos, and sometimeshighly contingent nature – facets sometimeslost in the cycle approach (Colebatch 2006).Similarly, a well-known and self-described al-ternative model to the stages approach, the ACFput forward by Paul Sabatier and his colleagues(Sabatier 1987, 1988; Sabatier and Jenkins-Smith 1993) centres on policy formulation ac-tivities and the roles played by actors sharingcommon beliefs in articulating and promot-ing specific definitions of problems and themeans to solve them. This framework focusesattention upon the role of ideas, learning, andcoalition behaviour in policy-making. It con-tains both instrumental and non-instrumental

components and brings to policy studies anemphasis on the significance of these aspectsof policy-making, which orthodox models ofpolicy cycles tend to ignore or downplay.

Although some scholars have urged a sim-ple direct extension of the multiple streamsapproach to cover all policy-making activities,Kingdon’s model cannot simply be transplanteddirectly to explain non-agenda setting dynam-ics. Despite the occasional reference to otherstages, its overwhelming focus is on agenda-setting and an ‘idea whose time has come’(Kingdon 2011: 1). It is also unclear that theACF in the form developed by Sabatier andothers can offer satisfactory insights into im-portant aspects of policy-making such as themechanics of the ratification or rejection ofpolicy options, or the administrative politics ofprogram implementation.

Rather than engage in a process of contrast-ing these frameworks with each other and con-ducting empirical analyses intended to proveor disprove their superiority, a worthy ambitionfor researchers is to determine whether or notKingdon’s framework or the ACF can explainas many aspects of policy-making as the policycycle model and thereby replace it as the mostgeneral overall depiction of policy-making, orwhether they in fact deepen our insight into thevarious stages of the policy process, servingthereby to supplement rather than replace thecycle model.

As this article will argue, if they are to ad-vance thinking about policy-making, both theMultiple Streams Framework (MSF) and ACFapproaches to understanding policy processneed revision if they are to apply to the post-agenda setting and post-formulation activitiesinvolved in policy development and implemen-tation. Specifically, this article argues that areconciliation of streams, advocacy coalition,and cycles models only becomes possible onceit is recognized that neither the multiple streamsmodel nor the ACF, as presently constituted,can deliver fully functional frameworks capableof understanding the entirety of policy-makingactivity and behaviour. That is, neither alterna-tive on its own can match the analytical scopeand range of the policy cycle approach. Ratherthan being understood as conceptual rivals,


Moving Policy Theory Forward 3

therefore, we argue that each model can bringcomplementary and cumulative insights intohow policies are made and thus yield furtherappreciation of the policy process as a whole,enhancing the utility of the cycle frameworkby addressing astute and longstanding criti-cisms of its lack of clear and realistic agency,oversimplified depictions of policy delibera-tion, and obscure drivers of change (Colebatch2006; Sabatier 1991). In other words, a com-bination of elements from each model can ad-vance both policy thinking and the policy cy-cle framework’s application, more usefully thandoes the existing penchant toward the continuedreinforcement of the duelling frameworks idea.

Duelling Policy Frameworks: Stages,Multiple Streams, and Advocacy Coalitionsin Historical Perspective

The policy cycle approach is arguably the mostenduring conceptual construct in the policy sci-ences (Burton 2006; deLeon 1999; Weible et al.2012). Although the stages or cycle frame-work has many detractors (e.g. Colebatch 2006;Sabatier 1991) who have argued that it presentsan idealized image of sequential policy-makingactivity rarely encountered in practice, it hasretained a significant role in many policy sci-entists’ examination of policy-making for wellover six decades. This longevity and continueduse and re-use in the face of a series of chal-lengers cannot simply be ignored. The reasonsfor this must be carefully analysed and theirimplications brought forward into any attemptto integrate or synthesize aspects of alternativemodels and frameworks.

The origins of the cycle model date from theearliest works on public policy analysis, andparticularly those of pioneering scholars in thepolicy sciences such as Harold Lasswell. Theanalytical goal of simplifying the complexityof public policy-making by developing and ap-plying metaphorical accounts of its fundamen-tal processual and cyclical nature originatedin Lasswell’s earliest works (Lasswell 1956,1971). Based on his direct observations ofpolicy-making processes in the United States,he deconstructed the policy-making process

into several discrete stages – seven in thiscase including intelligence gathering, promo-tion, prescription, invocation, application ter-mination, and appraisal – corresponding withthe sequence of tasks involved in conceptualiz-ing the creation of and outputs of government.

Lasswell’s work was highly influential andformed the basis for many later approaches andnumerous permutations of his original stagesframework (e.g. Brewer 1974; Lyden et al.1968; Simmons et al. 1974). This idea of asequence of policy tasks and behaviour re-ceived varying treatment in the hands of dif-ferent authors following in this tradition. Laterstudies attempted to retain the parsimony andexplanatory power of a multi-staged processmodel of policy development although refin-ing it into a smaller number of distinct stagesassociated with applied problem-solving ac-tivity; such as, in Brewer’s (1974) case, in-vention/initiation, estimation, selection, imple-mentation, evaluation, and termination. Themodel ultimately evolved into the now ubiqui-tous ‘cycle’ construct of five main ‘stages’ ofpolicy-making: from agenda-setting and policyformulation through decision-making to policyimplementation and evaluation (Althaus et al.2013; Howlett et al. 2009; Jann and Wegrich2007).

Importantly, the later changes to the modelintroduced by Brewer and deLeon (1983) andothers added a dynamic component to the orig-inal stages approach by incorporating feed-back processes into it: thus presenting policy-making not just as a ‘staged process’ but as anon-going iterative one: a ‘policy cycle’. Thisinsight inspired several new versions of thestages perspective in the 1970s and 1980s. Themost well-known were set out in popular text-books by Charles O. Jones (1984) and JamesAnderson (1975), which adopted an explicitlyproblem-solving orientation toward the subject.Each model put forward slightly different inter-pretations of the names, number, and order ofstages in the cycle but retained the same funda-mental staged-feedback cycle architecture andproblem-solving focus (Howlett et al. 2009).

Although the cycle model remains prominentin the policy sciences pedagogy and research,it now competes with alternative approaches to



understanding the policy process (Sabatier andWeible 2014). These more recently developed,but still more than 30-year-old frameworks, viefor the crown of most favoured conceptual ap-proach to explaining public policy.

One such approach has received far-reachingand enduring attention from scholars bothwithin and beyond the policy sciences is the‘multiple-streams’ model most popularly asso-ciated with John Kingdon’s (1984, 2011) workon US congressional agenda-setting. In thismodel, quasi- or semi-independent ‘multiplestreams’ of political, problems and policy (so-lutions), events, and activities periodically flowtogether across realms.

The problem stream in this model containsperceptions, opinions, and attitudes held by var-ious members of the public and policy commu-nities (Mukherjee and Howlett 2015). Thesepertain to whether some problems are essen-tially public in nature and cannot be resolvedthrough private initiative and thus require gov-ernment action, as well as perceptions aboutthe merits of past government efforts to resolverelated challenges. The policy stream carriesrecommendations from researchers, advocates,analysts, and others in a policy community ex-amining problems and using their (sometimesself-proclaimed) expertise to propose prospec-tive solutions to them (Voß and Simons 2014).The political stream is stocked with contex-tual attributes such as the composition of ideasand values comprising national ‘moods’ andthe power shifts produced by legislative andexecutive turnover following events such aselections and cabinet shuffles that rotate thecomposition of policy-makers and affect im-portant events through the composition of polit-ical and legislative timetables (Stimson 1991).Kingdon’s operative idea was that in certain cir-cumstances, sometimes driven by institutionalevents such as budgetary or legislative dead-lines, or by focusing events such as airplanecrashes or earthquakes, these streams wouldjoin together to provide a window of opportu-nity for entrepreneurs to move their preferredissues and solutions onto government agendas(Birkland 1997, 1998). Although the exact tim-ing of some of these occurrences might be for-tuitous, at other times they would be more or

less predictable, such as immediately followingan election (Howlett 1997).

Kingdon’s ideas about policy streamstouched a chord in the policy sciences and werequickly seized upon and used to describe andassess case studies such as the nature of US for-eign policy-making (Woods and Peake 1998);the politics of privatization in Britain, France,and Germany (Zahariadis 1995; Zahariadisand Christopher 1995); the nature of USdomestic anti-drug policy (Sharp 1994); thecollaborative behaviour of business and envi-ronmental groups in certain anti-pollution ini-tiatives in the United States and Europe (Lober1997); and the overall nature of the reform pro-cess in Eastern Europe (Keeler 1993), amongother subjects.

In this understanding, policy developmentdid not occur automatically or spontaneouslyin response to a social problem as Lasswellseemed to suggest. Rather it emerged in amore complex and contingent fashion as theresult of the interaction and intersection of thethree streams, which led to certain issues be-ing taken up by governments and not others,defining their agendas and future activities. Inthis framework, important policy events occurat critical junctures, thanks to the initiative ofspecific kinds of actors – policy entrepreneurs– who link together policy problems, solu-tions, and their surrounding political aspects;leveraging various kinds of ‘focusing events’and ‘windows of opportunity’, which providethe possibility of generating the initiative andmomentum needed to begin a policy process(Mintrom 1997; Mintrom and Norman 2009).

Although this is a powerful and parsimoniousway of conceptualizing and understanding themany different kinds of actors and activitiesthat go into problem definition and the be-ginning of policy deliberations in government,what has often been lost in the discussion andapplication of this model in the period fol-lowing Kingdon’s work, is that in its originalversion, the framework was not used to inter-pret all aspects of policy-making. Rather, King-don defined his task quite narrowly and onlysought to explain how issues moved onto gov-ernment agendas and became targets for action,rather than how, for example, solutions were



decided upon and implemented or put into ac-tion (Barzelay 2006; Guldbrandsson and Fos-sum 2009; Howlett et al. 2014).

Another prominent contender for the analyt-ical crown of policy studies that also devel-oped at around the same time is the ACF. Asis well known, the ACF was developed by PaulSabatier and Hank Jenkins Smith in a land-mark series of articles beginning in the mid-1980s; around the same time that Kingdon’swork appeared (Sabatier 1987, 1988). UnlikeKingdon, who did not engage directly with ri-val theories of policy-making, Sabatier did sowith gusto. He offered a trenchant critique ofthe policy cycle framework, for example, whichhe condemned for lacking ‘ . . . a causal theory[because it] . . . contains no coherent assump-tions about what forces are driving the processfrom stage to stage and very few falsifiable hy-potheses’ (Sabatier 1991: 145). Although find-ing some continuing use for the cycle approachas a ‘stages heuristic’ that could help illustratethe different activities which went into policy-making, Sabatier argued that the cycle frame-work ‘lacked agency’ and misled by suggest-ing a more linear and logical progression ofpolicy activities existed than could be observedin practice. The ACF essentially posited thatfocusing on the beliefs motivating collectiveaction within a subsystem would generate asuperior understanding of the conflict inher-ent within policy-making by comparison to the‘actorless’ vision of the staged approach.

Although Kingdon drew his empirical evi-dence for interpreting agenda-setting from thedeliberations of the United States Congress,Sabatier grounded his search for a more sat-isfactory way of understanding policy dynam-ics in the policy formulation work of Amer-ica’s state and federal bureaucracies. The prin-ciples of the ACF were induced from extensivecontent analysis of public input records intofederal rule making on environmental and nat-ural resource policies, and associated Congres-sional hearings as well as case studies of policy-making in areas such as agriculture and the en-vironment in California (Zafonte and Sabatier1998).

Whereas Kingdon’s units of analysis for dis-covering the causes of stasis and change on the

policy agenda were the heterogeneous forcesand factors that converged upon Congress,Sabatier and his colleagues focused on politicalactors as the drivers of policy development. Butrather than rely on the classic vehicle of plural-ist group interaction as a mode of collectiveaction (Truman 1971), or the amorphous is-sue network concept that had been proposed byHeclo (1977), Sabatier and Jenkins Smith cre-ated the ACF, an analytical structure in whichlike-minded actors formed competitive teamswithin each policy subsystem, contending toeither change policy formulation or maintainthe status quo (see Figure 1).

Like Kingdon, Sabatier postulated a muchmessier policy process than typically envi-sioned by problem-centered cycle theory, onein which duelling coalitions of actors vied tohave their policy-related ideas adopted in prac-tice. In this framework, the gravitational pullthat draws actors into a particular coalition isexerted by core beliefs, grounded in deeplyheld normative values about the way the worldworked, or should work, but communicated andcontested through what Sabatier (1988: 145)termed ‘near (policy) core beliefs’ regardingthe subsystem’s legitimate goals. The politi-cal rivalry between these coalitions over timeserved to establish the contours and contentof policies. Policy-making was thus much lessabout a sequence of problem-solving activitieson the part of disinterested actors, than abouthow coalitions formed, engaged their competi-tors, and about how that process establishedhegemony over problem definitions and policyalternatives (Weible 2005; Zafonte and Sabatier1998).

Although helpful in specifying who was in-volved in policy-making and how they inter-acted, however, the strength of the ACF for-mulation came at the expense of ignoring thedecision-making process and reverting to a pre-Lasswellian ‘black box’ in which the inputsformulated by a successful coalition some-how were melded together to produce policyoutcomes. This is apparent in the formula-tion of the ACF framework of policy-makingpresented above (see Figure 1) whereby deci-sions simply emerge from the contestation ofadvocacy positions and implementation exists



Figure 1. The ACF Framework

Source: Sabatier (1998: 102)

only as a policy output. Policy evaluation wasseen to exist as a form of learning (Bennett andHowlett 1992) in which feedback from outputsaffects subsequent inputs, but this vision wasnot clearly linked to the ACF belief structuresset out above.

For most of the time, the different origins,foci, and language used in these three contend-ing ‘meta-theories’ have precluded their rec-onciliation, let alone synthesis or integration(Cairney 2013). In a separate article, however,we have urged scholars to reconsider this ri-valry and re-assess the utility of connectingthese conceptual frameworks together by ex-tending Kingdon’s ‘streams’ approach to otherstages of policy-making such as formulation,decision-making, implementation, and evalua-tion (Howlett et al. 2015). The logic and rea-sons for doing so are set out below, along with

the elements of an integrated framework capa-ble of uniting these two conceptual constructswith the ACF model; combining the strengthsof each and reaching beyond their individuallimitations in better describing, explaining, andunderstanding policy-making activity.

Reconciling the Three Perspectives onPolicy-Making

As noted above, many researchers have sug-gested proceeding through a simple extensionof the multiple streams frameworks to encom-pass all phases or activities of policy-makingwithin a three-stream model (politics, pol-icy, and problem) (Barzelay 2006; Zahariadis2003). However, this is not a simple matter.Kingdon’s core assumption of the existence of



only three streams, for example, is impossibleto apply to other stages of policy-making be-yond agenda-setting such as implementation orevaluation when other central actors such asprogram administrators and policy analysts areactive. To accomplish a reconciliation of theconceptual pillars of the multiple-stage and cy-cle models, we have argued elsewhere, that afive-stream framework which retains the con-ceptual architecture and analytical vocabularydeveloped by Kingdon, offers a more com-prehensive and capable framework for captur-ing the full range of policy-making dynamics(Howlett et al. 2015).

We present the details of this enhancedframework below, and go on to extend its reachthrough considering the coalition activity ofthe ACF as a vehicle for transmitting political(and other) beliefs into sovereign ratification,and then implementation of the outcomes ofpolicy decisions. Although beliefs are impor-tant in affecting coalition behaviour, and somemembers may remain wedded to their advo-cacy coalition regardless of the direction thata government decides to pursue, other policyparticipants can be drawn into a coalition bythe material and symbolic resources offered bygovernment once a course of action has beenchosen. Some recruits to a belief system maysupport coalition ideas more because of theirsupport for organisations and political leaders,than for the principles inherent in those ideas.

A First Step: Reconciling Multiple StreamsTheory with the Policy Cycle Framework

This analysis of the limitations of each modelsuggests a strategy and means through whichthe limitations of each can be surmounted bycombining key elements within an overall cy-cle rubric. As suggested above, the difficul-ties encountered in joining elements of thesetwo frameworks together, originate in how theframeworks were originally conceived. This istrue of the cycle analogy, which began as a staticframework, before developing later to incorpo-rate dynamic ‘feedback’ elements. Althoughgrowing in popularity, ‘stagist’ descriptionsof policy processes were often criticized forpresenting an assembly-line model of policy-

making without providing a clear descriptionor explanation for what policy actors actuallydid during the policy-making process, and whythey do what they did in any particular sequence(deLeon 1999; Sabatier 1991). The feedbackprocesses associated with the newer concep-tual design of a cycle overcame some of theseconcerns by subtly shifting the underlying pro-cessual analogy from mechanical action to amore ‘organic’ creation. Processes came to beviewed as the result of complex adaptive dy-namics rather than as the products of functionallinear logic (Howlett et al. 2009).

This emphasis on complex adaptive dynam-ics, of course, is not taken very far in thestages heuristic but is precisely where the mul-tiple streams model excels (Klijn 2008; vanBuuren and Gerrits 2008). Kingdon’s eclecticmix of streams, windows, and entrepreneurialmetaphors was designed to explain how policy-making began and how policy problems couldbeget solutions and vice versa, an oft-made ob-servation about policy-making that orthodoxstages thinking deemed anomalous (Beland andHowlett 2016).

As we have seen, however, the multi-ple streams framework requires substantive‘stretching’ beyond what had been envisionedby its creator to move from agenda-setting ac-tivity to encompass the entire policy process.Although the three-stream framework devel-oped by Kingdon is well suited to understand-ing that specific stage of policy-making, it re-quires augmentation to effectively encompassthe additional variables and activities affectingdeliberations, actions, and outcomes occurringover multiple stages or phases of policy-makingnoted above.

Reformulating the multiple streams ap-proach to take it beyond agenda-setting dynam-ics, as discussed below, enables a reconciliationof this model with that of policy stages. Devel-oping an ‘enriched’ policy stream frameworkcan fully engage with policy cycle thinking andcan overcome the limitations of both construc-tions,, although having the potential to create amuch more powerful conceptual apparatus foradvancing the understanding of policy-making.But mixing or integrating these two differ-ent frameworks is not a straightforward task.



Efforts to date to combine streams and cycles togenerate an improved understanding of the fun-damental nature of policy processes have oftenclouded matters by simply layering one frame-work on top of the other, leading to confusingor contradictory inferences. Our approach hereis not to layer, but to refine, adapt, and blend.

As Kingdon acknowledged, the idea of ‘pol-icy streams’ originated in earlier work by Co-hen, March, and Olsen (Cohen et al. 1972;March and Olsen 1979) into administrativedecision-making processes in complex, uncer-tain, and changing environments. Here it is im-portant to recall that these earlier authors hadnoted the existence of four, not three, streams.In their 1972 article, Cohen, March, and Olsenposited: a stream of choices, which arose fromthe decision-making inputs of policy actors; astream of problems, issues which had becomevisible in the public consciousness; a stream ofsolutions created by the alignment and resolu-tion of problems, remedies, and choices; and anenergy stream, comprising the time that partici-pants devote to deciding what to do, or not to do,about policy problems. Kingdon’s frameworkembraced both the conceptual structure of in-dependent streams and the substantive contentof two, i.e. problems and solutions (policy). Tothis, Kingdon added a politics stream to cap-ture the structural and cognitive dimensions ofauthority, whose power partially subsumed theenergy stream. And he omitted the stream ofchoices, the decision-making ‘occasions whenan organization is expected to produce be-haviour that can be called a decision’ (Cohenet al. 1972: 3), which Kingdon chose to repre-sent through a static metaphor (policy window)rather than a stream. Doing so is understandablebecause he was focusing on a single stage ofpolicy-making in which only a discrete choicecould occur: whether or not to have an item en-ter an official governmental policy discussion.

Determining exactly how many ‘streams’ ofevents exist across all dimensions of policy-making and identifying how they operatethrough all stages of the policy process, canprovide a logic of stream intersection that canbe applied to movements between each stage ordecision point. A version of the streams frame-work that is promising in this respect involves

thinking about policy-making as a sequence ofphases – much as in the stages-cycle frame-work – in which critical ‘confluence and distri-bution’ points among policy streams are linkedto specific ‘stages’, in the same way as var-ious tributaries to a river merge at differentpoints into the river as it makes its way down-stream (Howlett et al. 2015). This frameworkbegins with the classic Kingdonian articulationof problem, policy, and political streams affect-ing agenda-setting but adds in new process andprogram streams that feed into specific con-junctures where the existing streams intersector coalesce as the policy process unfolds (seeFigure 2).

In this five-stream framework, each conflu-ence point brings something new (new actors,new tactics, new resources) joining the flow ofpolicymaking events. Where each stream inter-sects, the merger point represents a ‘window’ inKingdon’s sense, and yields a different config-uration of policy inputs that generate a distinctpolicy pattern through each particular juncturemuch as the ‘rounds’ style of policy-makingtheories have suggested (Klijn and Teisman1991; Timmermans 2001).

In this way of thinking, the first confluencepoint occurs in agenda-setting much as King-don suggested, when the three problem, poli-tics, and policy streams coalesce temporarily inthe typical ‘policy window’ fashion that he de-scribed. This intersection creates a new policyprocess stream that becomes the main or cen-tral pathway upon which other streams subse-quently converge. In turn, critical junctures arecreated that set up the future impetus for pol-icy deliberations and establish the initial condi-tions, which animate subsequent policy processadvances (or retreats) essentially becoming the‘choice’ stream mooted by Cohen, March, andOlsen.

After this critical agenda-setting process hasoccurred, in many jurisdictions the politicalstream separates from the problem and pol-icy streams as specific sets of subsystem ac-tors such as policy analysts and stakeholdersorganized in advocacy coalitions contribute todeliberations and propose policy alternatives(Craft and Howlett 2013). This mobilizationof ideas about what to do continues until a



Figure 2. Five-Stream Framework of the Policy Process

second critical point occurs once these actorshave blended policy problems and solutions to-gether creating a configuration of alternativechoice possibilities. The contents of this rangeof choices provide the basis for a new phasewhen the politics streams returns to connectwith the process stream creating momentumtoward a decision.

The third critical point occurs if and when adecision is taken and a policy then requires im-plementation. At this point the ‘policy’ streamseparates from the main flow, which is com-prised of the process, politics, and problemstreams, and is now joined by a program streamcomposed of the actors and interests workingto calibrate new program instruments (and in-tegrating or alternating them with establishedones) to generate new outputs. This programstream’s sustained focus on policy implemen-tation embodies a good part of what Cohen,March, and Olsen had in mind for their en-ergy stream, where policy actors’ time commit-

ment to a relationship between problems andsolutions was encapsulated. Once these out-puts have accumulated for a time, the ‘policy’stream rejoins the other streams when evalua-tion occurs.

This connection between streams and cy-cle frameworks can encompass qualitativelydifferent kinds of policy-making occurring ateach intersection point, depending on exactlywhat inputs each stream brings to a particu-lar moment in the policy-making process. Suchan openness to diverse inputs offers the kindof insight into policy-making that draws uponthe analytical strength and persistence of thestages-cycle idea, although escaping the limita-tion that the absence of program or policy feed-back streams created for earlier efforts to linkthe stages framework with ‘streams’ thinking.Unlike the stages framework, however, whichis not immediately clear about why each stageoccurs or in what order different parts of thecycle come about, the five streams framework



not only explains how and why this happensas new actors, ideas, and interests merge intoan existing policy flow, but also explains thedifferent patterns of policy-making that resultfrom the presence or absence of relevant inputsat different critical junctures.

This way of thinking integrates the actors andbehaviours Kingdon identified in his agenda-setting study, but does so well beyond that taskof policy-making. Among other things, it ad-dresses many of the concerns Sabatier (1991)and others have voiced about the ‘actorless’ or‘agentless’ character of many orthodox cycleinterpretations and uses.

A Second Step: Reconciling the ACFFramework with the Revised Multi-StreamsPolicy Cycle Framework

Even the five-stream framework would remainvery much a bloodless analytical construct,however, without taking the step of specifyingnot only who the actors are within each stream,but also what they bring to the conjuncture withother streams and how they operate and interactto produce the policy outcomes and character-istic features of the policy process (Mukher-jee and Howlett 2015). It is here that the ACFapproach can inspire conceptual advances bygrounding policy-making in the ideas and be-liefs held by key actors about what is to bedone at particular moments of policy-making.Connecting the focus of the ACF on ideas tothe cycle-stream framework set out above, al-though expanding it to explain each stage ofpolicy-making, adds further insight into pol-icy processes and outcomes. It also generates asuperior framework for understanding policy-making, beyond either of the other two, eitheralone or in combination.

As with the Multiple Streams model, theACF should not, as many scholars includingSabatier have alleged, be thought of as a re-placement for policy cycle analysis but ratherconsidered, as deLeon argued, a refinement ofand advance upon the policy cycle (deLeon1999). Following advocacy efforts throughoutthe stages of policy development reveals impor-tant insights into the interplay between beliefsand behaviour within a policy formation pro-

cess, which strict adherence to the ACF model,as originally formulated, does not allow.

This relationship is reciprocal, with othermodels helping cover weaknesses in the ACFformulation. With respect to Kingdon’s model,for example, it is important to note that the ACFdoes not explain very well when and how thepolicy agenda changes. Although the logic ofthe model suggests that change occurs when acoalition that had not influenced policy formu-lation previously is able to exert its influenceto overcome the preferences of the previouslydominant coalition, or a coalition learns to alterits own behaviour, this activity is not conceptu-alized very well. Rather such shifts in subsys-tem structure and activity were often attributedby Sabatier et al. to exogenous shocks (e.g.economic crises, electoral realignments, or nat-ural disasters), which were premised to be un-predictable and thus remained under-examined(Sabatier 1987, 1988). But to give such out-of-system events a critical location in the frame-work undermines its explanatory power or atbest limits its applicability in times of ‘nor-mal’ policy-making. Connecting the ACF andMultiple Streams models with the policy cycleheuristic uses the strengths of each to help fillthe gaps in the others.

Explaining agenda-setting in terms ofstreams and subsequent policy developmentthrough the behaviour of coalitions and theirinteraction within and between subsystems (al-though engaging with ideas about policy prob-lems and solutions) helps shed light both onthe links between agenda-setting and formula-tion activities as well as their translation intodecision-making and implementation stages ofpolicy-making (Wilder and Howlett 2014).1 Inthis sense it is very useful to introduce ACF-type subsystem thinking and concepts into thefive-stream framework, which in itself (as wehave demonstrated above) can be integratedwith the stages/cycle approaches. Doing soallows us to conceive of streams interactingthroughout multiple phases of the policy pro-cess, not in a ‘bloodless’ and impersonal fash-ion, but in a vibrant politicized manner as com-peting coalitions of interests vie for dominancenot least in trying to ensure the primacy of theirbeliefs and ideas. In particular, it brings to the



fore in a more holistic fashion three modes ofthinking, which otherwise would not have beenpossible in the hitherto largely mutually exclu-sive duelling frameworks of cycles, streams,and coalitions.

We are using the ACF perspective in sucha way that we can conceive of a stream be-ing shaped to some degree by competing coali-tions of interests within a policy subsystem.The problem stream, for example, is shapedto a substantial degree by the outcome of theinteractions by competing coalitions of inter-ests (from scientific experts to lobby groupsand public servants) who seek others’ accep-tance of their authoritative definition of the‘policy problem’. Similarly, the process streamis shaped by coalitions who will often contestthe most appropriate process that should followfrom consultation with citizens or key interests.Such routes may for example include whetherto opt for a referendum or purely an executivedecision on a high-profile issue, as well as con-cerns about the best administrative practicesto follow in implementing policies (Mukherjeeand Howlett 2015).

Following on from this point, this samestream/coalition logic can be applied to allstages of the policy process. At the implemen-tation stage for example, there is likely to bea powerful advocacy coalition of interests thatshapes the process stream (broadly, the pro-cedure for implementation) and the programstream (the specific policy instruments andtheir calibration). The outcome may range fromclassic and idealized top-down implementationwhere the dominant coalition has secured littleor no wiggle room for ‘street level bureaucrats’and others involved in the implementation pro-cess, to a much more contested implementationprocess. In terms of the latter, competing coali-tions would vie over procedure (perhaps con-testing through the courts) and/or fine programdetail with perhaps very different interpreta-tions of which micro-instrument best ensuresachievement of goals such as equity, efficiency,or effectiveness (Howlett et al. 2009).

We can also conceive of innumerable inter-actions between stages/streams and coalitionsin different contexts. For example, at the evalu-

ation stage of a highly controversial policy, wecan imagine competing coalitions attemptingto secure the dominant narrative reflecting onall stages of the policy process. With referenceto each of the five streams, questions differentcoalitions’ members might address include: didwe address the right problem? Why was onepolicy solution chosen over an equally viableoption? Did key political interests promotingthe dominant policy option have the public in-terest at heart? Should we not establish a morerobust consultation process that took on boardthe views of stakeholders rather than ignorethem? Could we not commit more funding tothe program?

Utilizing a combination of stages, streams,and coalitions thus not only has the potentialfor providing a more integrated and holisticunderstanding of policy, but also its diversityand flexibility provides room for better encom-passing creativity, inventiveness, art, and craftwithin policy-making. These are all core at-tributes of strong policy analysis that are often‘modeled-out’ of policy studies and thinking(Wildavsky 1987).

Conclusion: Synthesizing and MovingForward

Current policy theory, as Peter John (2012) hasreminded us, relies heavily upon works cre-ated in the late 1980s and early 1990s, if notearlier, which despite much case study andcomparative research that has questioned orchallenged many of their assumptions and pre-dictions, have not moved very far beyond theoriginal ideas advanced in the writings of thisera.

As both John (2013) and Cairney (2013)have argued, one way out of the conundrumof multiple, competing frameworks attemptingto explain the same set of facts is to stop view-ing them as mutually exclusive or competitiveconstructs. This article engages in a processof conceptual adjustment and reconciliation,which such a viewpoint calls for, suggestingterms and ways though which these three meta-frameworks of policy-making processes can be



usefully connected to offset their criticisms andprovide a generally superior model of policy-making to any taken on its own.

Linking the streams and stages models to theACF frameworks move this process forward. Ithighlights the interactions between and amongstreams of events and retains the essence ofKingdon’s ‘fluid’ dynamics although better ac-commodating the full range of tasks involvedin policy development. Such a framework re-tains the basic thrust and vocabulary developedby Kingdon although combining it with themore comprehensive cycle framework, captur-ing many more of the activities that affect pub-lic policy-making and beyond.

Furthermore, rather than serving as a com-peting approach to understanding policy-making, the ACF model can also be seen tocontribute to the effectiveness of joining up themultiple-stage and multiple streams models byclarifying the nature of the actors, activities,and motivations at each stage and conjunctureof policy-making. Streams, windows of oppor-tunity, and critical moments are concepts thatbetter help conceptualize who subsystem actorsare and how they interact over the policy cycle,than the simple coalition structure developed inthe original ACF framework.

The research possibilities involved in work-ing with this new framework are enormous,once we begin to see the value in adapting andcombining the core insights of stages, streams,and coalition approaches, rather than seeingthem as mutually exclusive. A new synthe-sis allows us to meld together analytical ap-proaches that focus on different stages of pol-icy processes, the interplay of multiple forcesthat shape these processes, and the competitionbetween different sets of actors (and beliefs) asthey vie for influence.2

Single case studies focusing on specificphases of the policy process, or specific streamsand their interactions with others and in spe-cific contexts such as election campaigns, scan-dals, and new governments coming to powerwould be useful. For example, one can imaginethe value of a research study examining a con-troversial commission of inquiry (in effect, anevaluation stage) with competing coalitions ofinterests including scientific experts, citizens,

lobby groups, media, seeking to build coali-tions in an attempt to secure the dominant nar-rative, which runs through reflections on all thestreams and all the stages. Studies that straddlemultiple stages of the policy process would alsohelp explore policy dynamics over time withinparticular policy subsystems.

Cross-case comparison, within and acrosscountries, is also possible and useful. Let usimagine two countries with a similar policysubsystem X and a similar dominant solutionstream at the formulation stage driven by a pow-erful advocacy coalition. A research agendaaround the issue of why, in one country, thiscoalition fragments at the implementation stagewith an alternative coalition able to dominatethe political stream, whereas in another coun-try policy implementation is much smootherand the dominant coalition is routinely able tomaintain its dominance throughout all stagesof the policy process, would also help refineand advance policy process theory beyond itspresent state.

Endnotes

1. It is important to note in this regard thatSabatier and Jenkins-Smith (1993) positedmost subsystems would have two to four Ad-vocacy Coalitions competing to formulate, orreformulate, policy strategies, goals, and ob-jectives. They suggested these policy contestswould foster social learning that reinforcedcoalition dynamics through ‘relatively endur-ing alterations of thought or behavioral inten-tions which result from experience and whichare concerned with the attainment (or revision)of policy objectives’ (Sabatier, 1988: 133).Schlager (1995) suggests another important di-mension of social learning within advocacycoalitions, that of obtaining material benefitsfrom policy through the pursuit of collectiveaction. Such rewards offer both a positive re-inforcement to the successful advocates, whobelieve that they have done well by doing good,in addition to a motivation to members of the,as yet, unsuccessful coalition(s) who see thedividends from further collaboration.



2. This then sets the stage for possible furtherintegration with the IAD and PET frameworks,which describe aspects of implementation ac-tivity and empirically driven characterizationsof the outcomes of policy process behaviour;two other frameworks, which developed aroundthe same time. The former helps spell out manyaspects of policy design work undertaken inthe policy process while the latter joins withparadigm theory in describing outputs as ei-ther incremental or transformative (Kiser andOstrom 1982; Baumgartner and Jones 1991).

References

Althaus, C., P. Bridgman and G. Davis. 2013. TheAustralian Policy Handbook, 5th ed. Sydney:Allen and Unwin.

Anderson, J. E. 1975. Public Policy Making. NewYork: Praeger.

Barzelay, M. 2006. ‘Introduction: The ProcessDynamics of Public Management Policymak-ing.’ International Public Management Journal6(3):251–282.

Baumgartner, F. R. and B. D. Jones. 1991. ‘AgendaDynamics and Policy Subsystems.’ Journal ofPolitics 53(4):1044–1074.

Beland, D. and M. Howlett. 2016. ‘How SolutionsChase Problems: Instrument Constituencies in thePolicy Process.’ Governance, forthcoming.

Bennett, C. J. and M. Howlett. 1992. ‘The Lessonsof Learning: Reconciling Theories of PolicyLearning and Policy Change.’ Policy Sciences25(3):275–294.

Birkland, T. A. 1997. After Disaster: Agenda Setting,Public Policy and Focusing Events. WashingtonDC: Georgetown University Press.

Birkland, T. A. 1998 ‘Focusing Events, Mobiliza-tion, and Agenda Setting.’ Journal of Public Pol-icy 18(1):53–74.

Brewer, G. D. 1974. ‘The Policy Sciences Emerge:To Nurture and Structure a Discipline.’ PolicySciences 5(3):239–244.

Brewer, G. and P. deLeon. 1983. The Foundations ofPolicy Analysis. Homewood, CA: Dorsey.

Bridgman, P. and G. Davis. 2003. ‘What Use is a Pol-icy Cycle? Plenty, if the Aim is Clear.’ AustralianJournal of Public Administration 63(3):98–102.

Burton, P. 2006. ‘Modernising the Policy Process:Making Policy Research More Significant?’ Pol-icy Studies 27(3):172–192.

Cairney, P. 2013. ‘Standing on the Shoulders of Gi-ants: How Do We Combine the Insights of Mul-tiple Theories in Public Policy Studies?’ PolicyStudies Journal 41(1):1–21.

Cohen, M., J. March and J. Olsen. 1972. ‘A GarbageCan Framework of Organizational Choice.’ Ad-ministrative Science Quarterly 17(1):1–25.

Colebatch, H. K. 2006. Beyond the Policy Cycle: ThePolicy Process in Australia. Crows Nest: Allenand Unwin.

Craft, J. and M. Howlett. 2013. ‘Policy AdvisorySystems and Evidence-Based Policy: The Loca-tion and Content of Evidentiary Policy Advice’.In S. Young (ed.), Evidence-Based Policy-Makingin Canada (pp. 27–44). Toronto: University OfToronto Press.

deLeon, P. 1999. ‘The Stages Approach to the PolicyProcess: What Has It Done? Where Is It Going?’In P. A. Sabatier (ed.), Theories of the Policy Pro-cess (pp. 19–34). Boulder: Westview.

Everett, S. 2003. ‘The Policy Cycle: Democratic Pro-cess or Rational Paradigm Revisited?’ AustralianJournal of Public Administration 62(2):65–70.

Guldbrandsson, K. and B. Fossum. 2009. ‘An Explo-ration of the Theoretical Concepts Policy Win-dows and Policy Entrepreneurs at the SwedishPublic Health Arena.’ Health Promotion Interna-tional 24(4):434–44.

Heclo, H. 1977. A Government of Strangers: Ex-ecutive Politics in Washington. Washington, DC:Brookings Institution.

Howlett, M. 1997. ‘Issue-Attention and PunctuatedEquilibria Frameworks Reconsidered: An Empir-ical Examination of the Dynamics of Agenda-Setting in Canada.’ Canadian Journal of PoliticalScience 30(1):3–29.

Howlett, M. 1998. ‘Predictable and UnpredictablePolicy Windows: Institutional and ExogenousCorrelates of Canadian Federal Agenda-Setting.’Canadian Journal of Political Science/RevueCanadienne de Science Politique 31(3):495–524.

Howlett, M., M. Ramesh and A. Perl. 2009. StudyingPublic Policy: Policy Cycles & Policy Subsystems.3rd ed. Toronto: Oxford University Press.

Howlett, M., A. McConnell and A. Perl. 2015.‘Streams and Stages: Reconciling Kingdon andPolicy Process Theory.’ European Journal of Po-litical Research 54(3):419–434.

Jann, W. and K. Wegrich. 2007. ‘Theories of the Pol-icy Cycle’. In F. Fischer, G. J. Miller, and M. S.Sidney (eds.), Handbook of Public Policy Anal-ysis: Theory, Politics and Methods (pp. 43–62).Boca Raton: CRC Press.



John, P. 2012. Analyzing Public Policy. Abingdon,Oxon; New York: Routledge.

John, P. 2013. ‘New Directions in Public Policy:Theories of Policy Change and Variation Re-considered.’ Paper presented at the InternationalConference on Public Policy, Grenoble, 26–28June, Available from bit.ly/1ovKxbF [Accessed12 February 2016].

Jones, C. O. 1984. An Introduction to the Study ofPublic Policy. Monterey, CA: Brooks/Cole.

Keeler, J. T. S. 1993. ‘Opening the Window forReform: Mandates, Crises and ExtraordinaryPolicy-making.’ Comparative Political Studies25(4):433–486.

Kingdon, J. W. 1984. Agendas, Alternatives and Pub-lic Policies. Boston: Little, Brown and Company.

Kingdon, J. W. 2011. Agendas, Alternatives and Pub-lic Policies, 2nd ed. Boston: Longman.

Kiser, L. L. and E. Ostrom. 1982. ‘The Three Worldsof Action: A Metetheoretical Synthesis of Institu-tional Approaches’. In E. Ostrom (ed.), Strategiesof Political Inquiry (pp. 179–222). Beverly Hills:Sage.

Klijn, E.-H. 2008. ‘Complexity Theory and PublicAdministration; What’s New? Key Concepts inComplexity Theory Compared to Their Counter-parts in Public Administration Research.’ PublicManagement Review 10(3):299–317.

Klijn, E. H. and G. R. Teisman. 1991. ‘EffectivePolicymaking in a Multi-Actor Setting: Networksand Steering’. In R. J. in ’T. Veld, L. Schaap, C.J. A. M. Termeer, and M. J. W. van Twist (eds.),Autopoiesis and Configuration Theory: New Ap-proaches to Societal Steering (pp. 99–111). Do-drecht: Kluwer.

Lasswell, H. D. 1956. The Decision Process: SevenCategories of Functional Analysis. College Park:University of Maryland.

Lasswell, H. D. 1971. A Pre-View of Policy SciencesNew York: American Elsevier.

Lober, D. J. 1997. ‘Explaining the Formation ofBusiness-Environmentalist Collaborations: Col-laborative Windows and the Paper Task Force’.Policy Sciences 30:1–24.

Lyden, F. J., G. A. Shipman, R. W. Wilkinson andP. P. Le Breton. 1968. ‘Decision-Flow Analysis:A Methodology for Studying the Public Policy-Making Process’. In P. P. LeBreton (ed.), Compar-ative Administrative Theory (pp. 155–168). Seat-tle: University of Washington Press.

March, J. G. and J. P. Olsen. 1979. Ambiguity andChoice in Organizations. Bergen: Universitets-forlaget.

Mintrom, M. 1997. ‘Policy Entrepreneurs and theDiffusion of Innovation.’ American Journal of Po-litical Science 41(3):738–770.

Mintrom, M. and P. Norman. 2009. ‘Policy En-trepreneurship and Policy Change.’ Policy StudiesJournal 37(4):649–667.

Mukherjee, I. and M. Howlett. 2015. ‘Who isa Stream? Epistemic Communities, InstrumentConstituencies and Advocacy Coalitions in Pub-lic Policy-Making.’ Politics and Governance3(2):65–75.

Pump, B. 2011. ‘Beyond Metaphors: New Researchon Agendas in the Policy Process.’ Policy StudiesJournal 39:1–12.

Sabatier, P. A. 1987. ‘Knowledge, Policy-OrientedLearning, and Policy Change.’ Knowledge: Cre-ation, Diffusion, Utilization 8(4):649–692.

Sabatier, P. A. 1988. ‘An Advocacy CoalitionFramework of Policy Change and the Role ofPolicy-Oriented Learning Therein.’ Policy Sci-ences 21(2/3):129–168.

Sabatier, P. A. 1991. ‘Toward Better Theories of thePolicy Process.’ PS: Political Science and Politics24(2):144–156.

Sabatier, P. A. 1998. ‘The Advocacy CoalitionFramework: Revisions and Relevance for Eu-rope.’ Journal of European Public Policy 5(1):98–130.

Sabatier, P. and H. Jenkins-Smith. 1993. ‘The Ad-vocacy Coalition Framework: Assessment, Revi-sions, and Implications for Scholars and Practi-tioners’. In P. A. Sabatier and H. C. Jenkins-Smith(eds.), Policy Change and Learning: An Advo-cacy Coalition Approach (pp. 211–36). Boulder:Westview.

Sabatier, P. A. and C. M. Weible (eds.). 2014. The-ories of the Policy Process. 3rd ed. Boulder, CO:Westview Press, a member of the Persus BooksGroup.

Schlager, E. 1995. ‘Policy Making and Collec-tive Action: Defining Coalitions within the Ad-vocacy Coalition Framework.’ Policy Sciences28(3):243–270.

Sharp, E. B. 1994. ‘Paradoxes of National Anti-DrugPolicymaking.’ In A. David, R. W. Cobb, and R.W. Cobb (eds.), The Politics of Problem Defini-tion: Shaping the Policy Agenda (pp. 98–116).Lawrence: University Press of Kansas.

Simmons, R. H., B. W. Davis and R. J. K. Chap-man. 1974. ‘Policy Flow Analysis: A Concep-tual Framework for Comparative Public PolicyResearch.’ Western Political Quarterly 27(3):457–468.



Stimson, J. A. 1991. Public Opinion in America:Moods Cycles and Swings. Boulder: WestviewPress.

Timmermans, A. 2001. ‘Arenas as Institutional Sitesfor Policymaking: Patterns and Effects in Com-parative Perspective.’ Journal of ComparativePolicy Analysis 3(3):311–337.

Truman, D. B. 1971. The Governmental Process;Political Interests and Public Opinion, 2nd ed.New York: Knopf.

van Buuren, A. and L. Gerrits. 2008. ‘Deci-sions as Dynamic Equilibriums in Erratic Pol-icy Processes.’ Public Management Review 10(3):381–399.

Voß, J.-P. and A. Simons. 2014. ‘Instrument Con-stituencies and the Supply Side of Policy Inno-vation: The Social Life of Emissions Trading.’Environmental Politics 23(5):735–754.

Weible, C. M. 2005. ‘Beliefs and Perceived Influ-ence in a Natural Resource Conflict: An Advo-cacy Coalition Approach to Policy Networks’.Political Research Quarterly 58(3):461–475.

Weible, C. M., T. Heikkila, P. deLeon and P. A.Sabatier. 2012. ‘Understanding and Influencingthe Policy Process.’ Policy Sciences 45(1):1–21.

Wildavsky, A. 1987. Speaking Truth to Power: TheArt and Craft of Policy Analysis. 2nd ed. NewBrunswick, NJ: Transaction.

Wilder, M. and M. Howlett. 2014. ‘The Poli-tics of Policy Anomalies: Bricolage and theHermeneutics of Paradigms.’ Critical PolicyStudies 8(2):183–202.

Woods, B. D. and J. S. Peake. 1998. ‘TheDynamics of Foreign Policy Agenda-Setting’.American Political Science Review 92(1):173–184.

Zafonte, M. and P. Sabatier. 1998. ‘Shared Be-liefs and Imposed Interdependencies as Deter-minants of Ally Networks in Overlapping Sub-systems.’ Journal of Theoretical Politics 10(4):473–505.

Zahariadis, N. 1995. Markets, States, and Public Pol-icy: Privatization in Britain and France. Ann Ar-bor: University of Michigan Press.

Zahariadis, N. and S. A. Christopher. 1995. ‘Ideas,Networks, and Policy Streams: Privatization inBritain and Germany’. Policy Studies Review14(1/2):71–98.

Zahariadis, N. 2003. Ambiguity & Choice in Pub-lic Policy: Political Decision-Making in ModernDemocracies. Washington, DC: Georgetown Uni-versity Press.

Zohlnhofer, R., N. Herweg and F. Rub. 2015. ‘The-oretically Refining the Multiple Streams Frame-work: An Introduction.’ European Journal of Po-litical Research 54(3):412–418.


moving policy theory forward: connecting multiple stream ...howlett/documents/howlett_et... · ing...

Documents