Skip to main content
v2026.11,986 entries · CC-BY 4.0

Advocacy Coalition Framework: Beliefs, Coalitions and Policy Change

The advocacy coalition framework explains policy change through belief-sharing coalitions; covers belief tiers, subsystem types and citations.

Written and maintained by CASRAI Editorial Board

Published

On this page

The advocacy coalition framework (ACF) is a theory of the policy process associated with Paul Sabatier, whose 1988 paper in Policy Sciences set out the framework. It explains policy change by looking at coalitions: groups of actors inside a policy subsystem who share policy beliefs, coordinate over time, and compete with other coalitions to shape government decisions. Researchers use it to study why policy stays stable for years and what lets it change.

What is the advocacy coalition framework?

The ACF treats a policy subsystem, such as one policy area in one jurisdiction, as the main unit of analysis. Inside it, actors with varied backgrounds but similar policy beliefs form advocacy coalitions. A review in Frontiers in Political Science describes coalitions as actors in a subsystem who coordinate over time to influence policy, citing Sabatier (1988) and Jenkins-Smith and Sabatier (1994), and notes that they aim to gain political dominance in the subsystem and then implement policies that reflect their beliefs. Eleanor Brooks, applying the framework to EU pharmaceutical policy, puts the core of the policy-making process as a competition between coalitions of actors with different belief systems.

The foundational statement is Sabatier (1988), An advocacy coalition framework of policy change and the role of policy-oriented learning therein, published in Policy Sciences 21, pages 129 to 168. A later review of 80 applications by Weible, Sabatier and McQueen (2009) found that the framework applies to varied substantive topics and geographical areas, and that the most commonly tested hypotheses concern policy change, learning and coalition stability.

Belief system tiers: deep core, policy core and secondary beliefs

The framework sorts beliefs into three tiers, and the tiers are what make coalitions identifiable and policy change explainable. The data table below sets the tiers side by side with the subsystem types discussed later.

Deep core beliefs

Deep core beliefs are fundamental values and axioms that span policy areas. Brooks describes them as an actor’s underlying personal philosophy, for example the relative value placed on freedom, power, health and knowledge, and says changing them is akin to religious conversion. Because they are so stable, they rarely explain a single policy dispute directly.

Policy core beliefs

Policy core beliefs have a subsystem-specific scope: core orientations and value priorities for one policy area. Brooks gives the balance between market and state and the role of expert participation in policy as examples, and says they are unlikely to change within a single legislative cycle. Ingold and colleagues (2025) call policy core beliefs specifically relevant for policymaking, and operationalise them by asking actors to rate policy core beliefs as high, medium or low in importance. Their example beliefs are economic efficiency, environmental compatibility, public health and social equity.

Secondary beliefs

Secondary beliefs concern the instrumental methods for reaching policy goals. Brooks describes them as resembling practical decisions about how to implement policy core beliefs in a given issue, and identifies them as the beliefs most susceptible to change. This is why one coalition can change an instrument, such as a regulatory tool or programme design, without giving up its goals.

Who belongs to a coalition?

Coalitions are not limited to elected officials. The Frontiers review lists members as parliament members, bureaucrats, industry groups, researchers, journalists, party representatives, think tanks and grassroots organisations, with citizens as supporters. Brooks likewise lists government bodies at several levels, business, civil society, journalists, researchers and analysts, and notes that different coalitions may control different governmental units.

The review also distinguishes principal members from peripheral members, who may switch coalitions, and describes coalitions of convenience in nascent subsystems. Ingold and colleagues add that internal coalition brokers introduce actors to each other and seed coordination, and cite the devil shift (Sabatier, Hunter and McLaughlin, 1987) as a related concept. For research administrators, the practical point is that university researchers and scientific bodies appear in this literature as coalition members, not neutral observers of policy.

Policy subsystems: unitary, collaborative and adversarial

The Frontiers review distinguishes three subsystem types. A unitary subsystem has one dominant coalition. A collaborative subsystem has several coalitions that negotiate with one another. An adversarial subsystem has competing, polarised coalitions. A subsystem can move between types: the review cites the Lake Tahoe Basin, which shifted from adversarial to collaborative between the 1980s and the 2000s (Weible et al., 2011b, as cited there). The same review notes that belief homophily is less predictive of coordination in contested subsystems, and that “contested” is under-defined in the literature.

How does the ACF explain policy change?

The framework expects stability within a subsystem most of the time, because coalitions hold their deep core and policy core beliefs firmly. Brooks identifies two routes to change in her account. The first is external shocks: she writes that major change can only be brought about by an external event, such as a new government or a health crisis. The second is policy-oriented learning, defined as coalitions changing secondary beliefs in response to new information. Brooks adds that learning is more likely when conflict between coalitions is low and the issue is analytically tractable, and that her EU case showed only limited learning.

The framework also sets a time horizon. Ingold and colleagues quote Nohrstedt and colleagues (2023): the ACF claims that a subsystem should be observed for at least one decade or more to understand or explain coalition formation and maintenance, policy learning and policy change. That rule shapes study design directly. A one-year interview study of a young policy area cannot test the framework’s central claims about change.

How do researchers study advocacy coalitions?

The Frontiers review searched Web of Science and Scopus, found 742 papers and narrowed them to 45 documents. It reports that the literature uses social network analysis, longitudinal case studies and modelling. Brooks instead used documentary analysis plus a literature review, and states that her article illustrates the main tenets of the ACF using one case rather than testing the framework formally. One workable design, following that literature, has four parts: define the subsystem and its geographic scope, identify actors, measure their policy core beliefs, and compare coordination across them over a decade or more.

Two CASRAI guides cover methods that appear in this literature. Social network analysis is the standard way to map who coordinates with whom, and process tracing is a case-study method for testing causal mechanisms such as the link between an external event and a policy change. For broader context on policy as a research object, see what public policy is, and for a journal that publishes this kind of work, the Policy Studies Journal guide. Interview-based belief measurement is a form of qualitative research.

Assumptions and common mistakes

  • Assuming shared beliefs are enough. Schlager (1995) argued that shared beliefs are not sufficient to explain coordination, and Henry (2011) found shared beliefs not significant in California land-use planning, both as summarised in the Frontiers review. Test coordination directly rather than inferring it from belief similarity.
  • Studying too short a window. The ten-year rule exists because learning and change are slow. A nascent subsystem is a different case: Ingold and colleagues report that the few studies of nascent subsystems conclude that some ACF assumptions are not confirmed there.
  • Skipping subsystem boundaries. Coalitions are defined within a subsystem. State the policy area and jurisdiction before listing actors.
  • Treating every policy dispute as a coalition contest. The subsystem types include unitary and collaborative cases, so competition between coalitions should be shown, not assumed.
  • Leaving trust unmeasured. The Frontiers review notes there is no consistent measure of trust in the literature, so say how you operationalised it if you use it.

Critiques and open questions

The 2009 stocktake reports that the hypotheses tend to be confirmed, but questions remain about the membership, stability and defection of coalition members, about the causal mechanisms linking external events and policy change, and about the conditions that facilitate cross-coalition learning. Emerging areas it names are policy subsystem interdependencies and coordination within and between coalitions. Brooks makes a narrower point from practice: the ACF does not offer a comprehensive explanation of the debate she studied, and she finds it an uneasy fit to label the 2009 reassignment of pharmaceutical policy within the European Commission an external system event.

How to cite the advocacy coalition framework

Cite the original statement for the framework, a review for its state of the art, and a recent application or method paper for your own design. In APA 7th edition style:

  • Sabatier, P. A. (1988). An advocacy coalition framework of policy change and the role of policy-oriented learning therein. Policy Sciences, 21, 129–168. https://doi.org/10.1007/BF00136406
  • Weible, C. M., Sabatier, P. A., & McQueen, K. (2009). Themes and variations: Taking stock of the advocacy coalition framework. Policy Studies Journal, 37(1), 121–140. https://doi.org/10.1111/j.1541-0072.2008.00299.x

In text, use (Sabatier, 1988) or Sabatier (1988) for the framework, and (Weible et al., 2009) for the review.

Sources

Data: Advocacy coalition framework: belief tiers and subsystem types

Original compilation of the advocacy coalition framework's three belief-system tiers and three policy subsystem types, with a definition, an example and an observed stability or operationalisation for each, assembled from a 2025 systematic review in Frontiers in Political Science, a 2025 chapter by Ingold and colleagues, and a 2018 case article by Brooks in the European Journal of Public Health.

Period covered: 1980s to 2025 literature.

Advocacy coalition framework: belief tiers and subsystem types
ElementCategoryDefinitionExample or observed behaviourSource
Belief tierDeep coreFundamental values and axioms spanning policy areasRelative value of freedom, power, health and knowledge; changing it is akin to religious conversionWesche et al. 2025; Brooks 2018
Belief tierPolicy coreSubsystem-specific core orientations and value prioritiesMarket versus state balance; rated high, medium or low on economic efficiency, environmental compatibility, public health, social equity; unlikely to change in one legislative cycleWesche et al. 2025; Brooks 2018; Ingold et al. 2025
Belief tierSecondaryInstrumental methods for achieving policy goalsPractical implementation decisions; the beliefs most susceptible to changeWesche et al. 2025; Brooks 2018
Subsystem typeUnitaryOne dominant coalitionSingle coalition controls the subsystemWesche et al. 2025
Subsystem typeCollaborativeSeveral coalitions that negotiateLake Tahoe Basin after its shift from adversarial, 1980s to 2000sWesche et al. 2025
Subsystem typeAdversarialCompeting, polarised coalitionsLake Tahoe Basin before its shift to collaborativeWesche et al. 2025

Follow CASRAI

Research-administration guidance, standards updates and independent tool reviews.

Ask CASRAI · free to try

Ask about Advocacy Coalition Framework: Beliefs, Coalitions and Policy Change

2 free questions a day, no account.

An AI assistant for research: funder rules, compliance, methods and publishing. It links the sources behind its answers and labels answers that come from a web search.

Answers draw on CASRAI's guides and dictionary plus the federal and funder documents we index: Federal Register, Grants.gov, Regulations.gov and UKRI.

Works on this site and inside Claude, Cursor and the AI tools you already use.

Everything CASRAI publishes — this page, the dictionary, the guides and the news — stays open to read, with no account and no card.

Referenced across the research world

University of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logoUniversity of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logo
  • University of Cambridge logo
  • Columbia University logo
  • Crossref logo
  • University of Edinburgh logo
  • Harvard University logo
  • University of Oxford logo
  • Princeton University logo
  • Stanford School of Medicine logo
  • University College London logo
  • ORCID logo

View CASRAI adoption →