1 Introduction

The field of public policy seeks to advance knowledge that generalizes across place and time, that informs or reveals something about particular contexts, or both. It, thus, operates increasingly as a global phenomenon, with scholarship proliferating within countries and comparatively between them. This global enterprise poses multiple challenges. Among them is building a scholarly community that uses similar concepts in their research and then interrelates them to support communications and shared understandings. One of the ways this is accomplished is through the development of portable frameworks and theories.

The power of frameworks to support comparative research or comparative research programs lies in their vocabulary of concepts and how they relate and are applied around a common scope of inquiry (Ostrom 2005). Frameworks provide a common language for concepts and connect them in contextually bound interactions, as written in theoretically focused hypotheses, propositions, and principles. As frameworks and related theories are applied and tested using different methods and in varied contexts, we learn about their focal phenomena and subsequently update the frameworks and theories. This paper contributes to this special issue on “Recent Perspectives of Policy Research—How Far Do They Travel?” in German Political Science Quarterly by analyzing one such framework. Called the Advocacy Coalition Framework (ACF), it was developed in Western democracies and is applied in African countries. In this analysis, we identify the lessons gained from the application of one framework to different political systems outside its traditional context.

The ACF is one of the most prominently applied frameworks to study coalition formation, policy change, and policy learning. With roots in Western democratic contexts, it has been applied numerously across the globe. Past reviews of the framework have shown that it is vast in its application and includes single and comparative case studies (see Weible et al. 2009; Pierce et al. 2017; Nohrstedt et al. 2020). While these applications tend to be in Western democracies, an increasing number are in different contexts, as summarized with over 80 applications in China (Li and Weible 2021) and in South Korea (Jang et al. 2016). In this paper, we conduct a review of 27 ACF applications in African countries. We are interested in knowing how and where the ACF is applied in the African continent, whether theoretical and methodological foci resemble the global trend, and whether we can find nuances and specificities, particularly regarding which themes and lessons emerge.

This meta-analysis focuses on Africa for several reasons. First, the increasing role of nongovernmental entities (e.g., interest groups) shaping policy-making in Africa provides fodder for testing the ACF’s arguments about coalition formation and influence on public policy in parts of the continent (Lynch and Crawford 2011). In particular, the ACF’s roots in interest group pluralism from the United States offer a foundation appropriate for testing in varied African contexts and for learning whether its applications offer valid and new insights about African policy processes (Heclo 1978).

Second, although previous ACF meta-analysis (e.g., Weible et al. 2009; Pierce et al. 2017) identified few applications in Africa, there has been a steady growth in the number of scholarly works that use this framework. We build on past meta-analyses by analyzing and reporting emerging trends and suggesting ways for theoretical refinement.

Third, while we recognize Africa’s heterogeneity, as countries with ACF applications range from authoritarian to democratic political systems, the continent provides a pragmatic and relevant way to cluster applications, given its distinct position in global affairs. The findings also reveal important commonalities about the continent’s system structures. As we subsequently explain, these attributes project, for example, the presence of coalitions of governmental and nongovernmental entities across the continent.

2 Overview of the Advocacy Coalition Framework

Emerging in part from the pluralist research tradition in the United States, Paul Sabatier and Hank Jenkins-Smith established the ACF in the 1980s. The ACF’s theoretical breadth involves issues related to the political mobilization of individuals (and organizations) striving to influence policy-making, the patterns and tendencies of these individuals to learn about policy-related issues, and explanations of policy changes over time. Like all frameworks, the ACF operates from an ontology of concepts that form its language. This language supports communication among scholars in the ACF’s global research program and creates the lens to analyze policy processes. Thus, the initial step to take in learning about the ACF is learning its conceptual language.

Table 1 lists the essential concepts with succinct definitions in the ACF’s language. For more elaborative descriptions and definitions, we recommend Jenkins-Smith et al. (2018), Henry et al. (2022), and Weible et al. (2020).

Table 1 Summary of key terms in the Advocacy Coalition Framework (ACF)
Full size table

Given the concepts in Table 1, the framework then connects them. We can express such connections in multiple ways. Visually, we can depict them in a flow diagram (see Jenkins-Smith et al. 2018). Verbally, we can combine them in phrases as hypotheses for empirical testing (see Appendix, Table 4). Below, we integrate some—but not all—of them into a theoretical logic as summarized by this short, archetypical vignette.

In high-conflict situations, policy actors will coordinate their political behaviors among allies in an advocacy coalition to influence public policy, while their opponents will do the same. Coalesced by their policy core beliefs and showing stability over time, these advocacy coalitions engage in debate and argumentation, with most learning occurring among individuals within the same advocacy coalition. However, this usually leads to more reinforcement than to a change in their beliefs. In contrast, learning rarely occurs between coalitions, possibly leading to changes in beliefs. Some policy actors, particularly policy brokers, might facilitate cross-coalition learning and the possibility for agreement on policies. Given the friction in policy-making, advocacy coalitions need to exploit opportunities through internal and external events, the rare situations of cross-coalition learning, and, sometimes, negotiated agreements to achieve their policy goals.

This archetypical vignette was developed with Western democratic contexts. Therein lies the rub: Do the ACF’s theoretical arguments about advocacy coalitions, learning, and policy change apply to other countries? If not, what needs to be adjusted? Furthermore, are we imposing a Western lens on non-Western contexts in the form of scholarly neocolonialization?

In support of the ACF, this archetypical vignette has been found to generally hold outside of Western democracies, as documented in other meta-reviews in Jang et al. (2016) and Li and Weible (2021). Nonetheless, we must remain diligent in the power of a theoretical lens to understand policy process phenomena and to distort and limit what we see, and such evidence is far from conclusive.

Contributing to the ongoing dilemma about the portability and utility of a theoretical lens like the ACF in conducting comparative research and supporting comparative research programs, this paper analyzes how well the ACF (as expressed in this vignette) maps to policy processes in African countries.

3 Methods and Materials

This study is based on 27 empirical applications of the ACF in Africa, reflecting diverse policy topics and geographic spaces. While we identify several publications that could have been used in this paper, such as doctoral dissertations and policy analysis reports, we focused only on journal articles to ensure consistency with related works (Weible et al. 2009; Pierce et al. 2017).

These articles were obtained from Google Scholar and confirmed via Web of Science. Through this confirmation and the approach outlined below, we are confident to have included all ACF applications in African countries that were published in journals. We collected the articles in July and August 2021. In searching for articles, we used an innovative gathering strategy and picked each of the 54 African countries (e.g., Ghana) and matched it with the term “advocacy coalition framework.” For example, we searched “Ghana and advocacy coalition framework” in Google Scholar. We did not use any filters, such as dates, with the intention of “catching” any document that is available online. For example, if Google Scholar presented 10 pages for a search, we examined all the pages to collect the relevant articles. We repeated this process for all the remaining 53 countries in Africa. To check the reliability of the number of articles obtained, we used the same search terms in Web of Science. The results were identical, and in some cases, Google Scholar returned more results (e.g., for Ghana, Google Scholar produced four articles, while Web of Science returned only two articles that were included in this analysis). One reason for the differences in the results is that, while Google Scholar captures both peer-reviewed articles and other types of publications (e.g., journal article preprints and reports), Web of Science displays only peer-reviewed articles and articles registered in its database. Through its more inclusive search process, Google Scholar also netted peer-reviewed articles.

We developed a codebook for the assessment and comparison of the 27 articles. The codebook was informed by previous codebooks that were used for similar purposes, namely a systematic review of ACF applications (see Weible et al. 2009; Jang et al. 2016; Pierce et al. 2017). We added the categories “region” (including Southern Africa, West Africa, East Africa, and North Africa), “regime type” (following the Democracy Index, published every year by the Economist Intelligence Unit [eiu.com], for “democracy”, “incomplete democracy”, “hybrid system”, and “authoritarian regime”), and “key lessons from article.” These categories were useful for understanding the distribution of the applications across the continent. The intention for the latter was to make explicit insights from each of the articles with respect to finding thematic patterns within and across the articles that the authors of those articles might not have recognized. Finally, to foster reliability, the coding was frequently discussed among the authors to reconcile differences. See Table 5 in the Appendix for coding categories. We used the constant comparison method to analyze the data (Glaser 1969). We did so by following these steps: First, each study’s key lessons were summarized. The second step involved a thorough and iterative reading of the key lessons, comparing across the board, to develop broad-level themes that synthesized related lessons. This process was repeated until we generated the themes that are reported in the results.

4 Results

We report the results of the 27 applications following these categories: summary of ACF applications, analysis of ACF theoretical components, and methods and data.

4.1 Summary of ACF Applications

The summary of applications is reflected in Fig. 1, comprising study location, topical foci, number of published articles, institutional affiliation of authors, and geographic location of publishers. The applications reflected in Fig. 1a generally represent the different regions of the continent, namely Southern Africa (9), West Africa (11), East Africa (7), and North Africa (1)Footnote 1. But the number of applications based on each country differs significantly. For example, Nigeria has the highest number of applications (five), followed by South Africa and Ghana (four each), and Kenya (three). The remaining countries (e.g., Uganda, Zambia, Mozambique) have only one or two applications each. In total, the ACF had been applied to 15 of the 54 African countries at the time of this investigation (summer 2021). Also, 26 applications focus on the national level, while one application (Runkle et al. 2013) focuses on the subnational level.

From the 15 countries represented in the sample, only one is considered a full democracy (Mauritius), and three are so-called incomplete democracies (South Africa, Ghana, and Tunisia) where the full enforcement of all necessary democratic institutions is still missing (see Appendix, Table 3). Most countries are of hybrid style, combining democratic institutions with authoritarian regime elements (see Appendix, Table 3 again). And four out of the 15 are considered authoritarian regimes: Egypt, Ethiopia, Mozambique, and Zimbabwe. We can conclude at this stage that studying ACF applications on the African continent means having the opportunity to learn about the ACF’s applicability to diverse political contexts and systems. We acknowledge that further important factors, such as the duration of independence, the former colonial regime, the degree of ethnic and language heterogeneity, and other cultural variables would be interesting to take into consideration.

The topical foci identified in the applications reflect nine policy areas, as shown in Fig. 1b. Of these topical foci, environmental and energy policy were associated with the highest number of applications (eight), followed by agricultural policy (six), health policy (five), and civil society and governance (one). The remaining policy areas (transportation policy, information policy, domestic violence, budget policy, and education policy) have only one application each.

Figure 1c depicts the frequency of ACF applications over time. We observe that the number of ACF applications has been increasing in the past 3 years. With the first application in 2001, the years between 2002 and 2007, 2010, and 2018 record no application. Overall, the highest number of articles occurred in 2019 (five), followed by 2020 (four), and 2021 and 2013 (three). While 2021 had not ended at the time of writing this paper, we expect the number of applications to increase to reflect the trend in the last 3 years.

As shown in Fig. 1d, we analyzed the distribution of authors’ institutional affiliations and publication outlets. We observe that about 41% of authors are affiliated with African institutions, while 59% of the authors are affiliated with foreign institutions (mainly in North America and Western Europe). Also, the majority (63%) of authors’ institutional affiliations are universities; the remaining 37% are distributed across different institutional types, dominated by research organizations such as the African Centre for Technology Studies in Kenya and the CSIR–Forest Research Institute of Ghana. In terms of publication outlets, about 90% of the applications in this paper were published in foreign journals, while African outlets represent 10%.

Fig. 1
figure 1

Summary of Advocacy Coalition Framework applications (n = 27) in Africa. a Where are the studies located (number of applications)? b What are the topical foci? c When were the studies published? d Where are the authors and publishers of the studies? Notes: One application compared Ghana and Uganda, another compared Nigeria and Zimbabwe, and another compared Egypt, Tunisia, and Lebanon. Each comparative study was counted once in the analysis. All countries on the African continent analyzed in these studies are indicated in (a); Lebanon is not. The figure in (a) was created by Bing Maps. In (b), agriculture policy includes biosafety and agriculture policy; domestic violence interweaves with wage policy; health policy includes nutrition, health, maternal and child health, public health, and alcohol policy; and environmental and energy policy includes urban water, land, acid mine drainage, climate, timber, tree, and forest policy. In (d), each author is unique and tagged by their affiliation; each publisher is unique (Policy Sciences and Forest Policy and Economics each published two articles)

Full size image

4.2 Methods and Data Collection

Sources of data collection and methods of the 27 applications are summarized in Fig. 2. Figure 2a represents the data sources used in the applications. Documents and reports represent 45%, followed by interviews (36%). Participant observation and news media each constitute 10%. In Fig. 2b it becomes evident that the majority of the applications use qualitative methods (67%), followed by mixed-method approaches (15%). Those with unspecified methods constitute 15%, while only one (4%) uses solely quantitative methods.

Fig. 2
figure 2

Summary of data and methods of Advocacy Coalition Framework applications (n = 27) in Africa. a Where were data obtained? b What methods were used?

Full size image

4.3 Theoretical Components and Hypotheses

Guided by in-depth analysis of the 27 articles, we explore the extent of application of the theoretical concepts in Table 2.

Table 2 Advocacy Coalition Framework (ACF) theoretical components across studies
Full size table

Table 2 indicates that the majority (51%) of the applications used in this paper examine the concept of advocacy coalitions. Only three applications explicitly test coalition hypothesis 1 (see Appendix, Table 4) about the stability of coalitions, with two offering confirmations and one offering refutation. Twelve of these applications focus on the theoretical argument that coalition formation is based on policy core beliefs expressed on a specific topic, though without always examining secondary beliefs (coalition hypothesis 2; see Appendix, Table 4). These applications mostly confirm this hypothesis (10 out of 12). For example, Chekole (2020), Funke et al. (2021), and Harris (2019) confirm the hypothesis that policy core beliefs provide the glue for coalition formation. Less common among the applications, Beverwijk et al. (2008) and Heinmiller et al. (2021) failed to confirm the second hypothesis.

The applications that focus on advocacy coalitions typically identify two coalitions that strove to influence public policy (e.g., Mockshell and Birner 2015; Mambulu et al. 2015). The categories of coalition members include government (31%); nongovernmental and civil society organizations (12%); academia, scientists, and research organizations (12%); consultants (10%); business (8%); international organizations (8%); interest groups, religious organizations, and farmer groups (8%); journalists (4%); environmental groups (4%); and other (2%). These findings are important for at least two reasons. First, they reflect a well-established position in the policy process literature about the diversity of policy actors that operate at different levels and seek to influence public policy. Public policy-making, then, is no longer the exclusive preserve of governmental organizations.

However, the five ACF studies in authoritarian countries show that mainly government, parliament, and public administration representatives dominate public policy-making in this regime type. Chekole (2020), studying urban land administration processes in Ethiopia, finds two opposing coalitions solely consisting of administration bodies, with the government playing a broker role. Also, Chikoto-Schultz and Uzochukwu (2016), analyzing civil society organizations in Nigeria (hybrid) and Zimbabwe (authoritarian), came to the conclusion that the policy processes are still largely dominated by the state apparatus and that ministries act as sole policy entrepreneurs. Hoppe and Kesseem (2019) also conclude that in authoritarian Egypt, governing nongovernmental organizations (NGOs) occur under a regime, rather than a policy issue logic; Beverwijk et al. (2008) confirm similar findings when showing belief stability in a nascent educational subsystem in Mozambique. On a positive note, those studies used standard concepts and techniques of stakeholder, belief, and coalition identification (Pierce et al. 2017): This shows the ACF’s portability to different political, cultural, and geographic contexts. In addition, they show that the “actors’ landscape” is lately changing in authoritarian systems as well, where (international) NGOs and donor institutions take their position in the subsystem (see Koivisto 2014).

Second, and related to the last point above, this relatively high number of international organizations potentially suggests that these organizations are actively involved in shaping policy processes in Africa. These organizations are diverse and focus on different issues. For example, Nwalie (2019) found Save the Children International focusing on health policy in Nigeria. Similarly, Harris (2019) identified the International Fund for Agriculture and the UN Development Program contributing to food policy in Zambia. Foli and Béland (2014) explain that ideas provide a mechanism through which these international organizations influence policy-making in Africa. Also, for some observers, these organizations are influential on the continent because they bridge important capacity and financial resource deficits in policy-making (Jaycox 1993).

Policy change was examined in 23% of the applications investigated in this paper and constitutes the second most studied ACF component in African countries. The ACF posits four main pathways to policy change in its first policy change hypothesis (see Appendix), namely internal shocks, external perturbations, negotiated agreements, and policy-oriented learning. Five studies explicitly tested this hypothesis and all confirmed it, and most identified internal and external events as the leading pathways to policy change. For instance, Heinmiller et al. (2021) examine policy change in Ghana’s information policy subsystem using news media data. The study found external shocks, such as change in government, adjustment in public opinion, and passing of the Special Prosecutor’s Bill, as the foremost factors that led to policy change. Similarly, Kingiri (2011) found that external shocks, such as renewed attention to biotechnology’s potential contribution to food security, influenced policy change in Kenya’s biotechnology policy subsystem. Thus, these studies confirm ACF’s theoretical postulations about policy change, broadly speaking. None of the applications explicitly analyzed policy learning as the theoretical emphasis, but some studies (e.g., Kingiri 2011, 2014) highlight information acquisition as a form of learning that contributed to policy change in Kenya’s biosafety policy subsystem.

The ACF conceptualizes policy-making as conflict based, meaning that coalitions struggle and compete over influencing policy agendas. Because of this, policy brokers exist to help resolve disputes among conflicting coalitions. Despite their importance, few studies are explicit about the role of policy brokers in the policy-making process. Only a few applications (13%; Table 2) explicitly identified policy brokers and their role in resolving conflicts among competing coalitions (e.g., Marfo and Mckeown 2013; Nwalie 2019). These studies broadly support ACF’s theoretical postulations about the role of policy brokers in arbitrating conflicts among coalitions. A slight deviation from what is typically found in the literature is that, as Nwalie (2019) and others report, some policy brokers represent international organizations operating in Africa. This observation potentially signals a category of policy brokers who can be called “international brokers” and who might be influential in policy-making processes in Africa. Because the ACF views the policy-making process as conflictual, we speculate that these “international brokers” will likely be influential in resolving policy disputes.

One of the areas often studied using the ACF is the role of experts or scientists (and their associated expert-based information) in the policy process (see Jenkins-Smith 1990; Weible 2008). Of the applications analyzed in this paper, three (corresponding to 8%; e.g., Funke et al. 2021) explicitly study the role of experts and expert-based information. The results reported in those applications largely confirm ACF’s assumptions about the role of experts in the policy-making process by showing the tendency for advocacy coalitions to use expert-based information as a means of persuasion in policy disputes. Some articles (e.g., Nwalie 2019) explore how resources such as money, skillful leadership, and proximity to policy-makers can be used to influence policy. However, given a thin literature on advocacy coalitions, expertise, and resources in the ACF literature in general, we are unable to draw strong implications from such articles into a wider context.

Overall, our findings are generally consistent with past ACF applications and support the ACF’s general theoretical logic as expressed in the previously mentioned vignette and its hypotheses (see Appendix, Table 4).

5 Discussion

One important theme is the growth of ACF applications in Africa, based on a comparison with previous meta-reviews. This growth can be observed in the number of applications and extension to other policy areas to which the framework has been less frequently applied beforehand. For instance, Weible et al. (2009) identify only one application on forest policy in South Africa (Tewari 2001). Henry et al. (2014) report six applications (e.g., Ainuson 2009; Runkle et al. 2013). While our search suggests that the number of cumulative applications had increased by 2017 (we found 16), Pierce et al. (2017) report three (Beverwijk et al. 2008; Kingiri 2011, 2014). We suspect that the use of Google Scholar in our study, which tends to be more inclusive, accounts for the difference in the number of articles. The applications we analyze reflect a growing uptick in the areas of environmental policy (eight), agriculture policy (six), and health policy (five). Overall, the fact that we identified 27 applications that use the ACF in Africa indicates growth in ACF applications in Africa. Aside from this observation, we discovered that scholars used the framework to analyze topics that have received less attention within ACF’s research program. Notable examples include nutrition policy (Harris 2019), alcohol policy (Mambulu et al. 2015), maternal and child health policy (Okeke et al. 2021), information policy (Heinmiller et al. 2021), and health policy (Nwalie 2019). In short, the numerical and topical expansion of the ACF suggests growing interest among scholars to use the framework to explore policy issues in Africa, thus cementing the relevance of the framework to analyze a range of policy topics.

Second, some of the applications use the ACF as a policy analysis tool. In this case, the ACF evaluates specific policies to make recommendations. To illustrate, Chekole (2020) used the ACF to appraise Ethiopia’s federal land administration and institutional processes. The study concludes that two legal frameworks and distinct institutions guide land administration in Ethiopia. While these findings are not necessarily novel, they reinforce well-established knowledge that implementation is often an interagency effort and that established rules can enhance policy execution. In the case of Tewari (2001), who studied forest policy in South Africa, the framework is used to develop strategies to resolve conflicts among warring coalitions and identify governance principles for sustainable commercial forestry. Along these lines, we found that Chikoto-Schultz and Uzochukwu (2016) used the ACF to examine civil society governance. Following these observations, we speculate that the ACF can support both the analyses of substantive policy issues (e.g., energy policy) as well as governance issues (e.g., the role of civil society in governmental agenda).

Third, ACF scholars are keen to test the comparative utility of the framework, in response to earlier criticisms leveled against it (Schlager 1995). The crux of the criticism was that ACF’s pluralistic and Western roots potentially hamper applications to other geographic contexts. The analysis of Henry et al. (2014) demonstrates the applicability of the framework to non-Western political contexts. Some of the applications used in this paper represent an increase in ACF’s utility for comparative analysis. For instance, Mockshell and Birner (2015) compare coalition formation in agricultural policy subsystems in Ghana and Uganda, the first being an incomplete democracy and the second being a hybrid of a democratic and an authoritarian system. In both countries, the discourse and beliefs outlined were very similar, and the subsystem was divided in a domestic and a donor coalition in which each comprised diverse stakeholders from different backgrounds (e.g., government agencies and academic institutions). This finding reinforces a well-established view within the ACF that individuals and organizations with diverse backgrounds constitute advocacy coalitions (Jenkins-Smith et al. 2018). This seems to hold true for the various political regimes present in the 27 ACF applications, ranging from full democracies (Mauritius) to authoritarian systems (e.g., Ethiopia), even though some actor types are more dominant in certain regime types than others.

Fourth, although the ACF has three main theoretical foundations of advocacy coalitions, policy change, and policy-oriented learning, various subtheoretical components are germane to the framework. While some subtheoretical components have been less explored, we identified several efforts to theorize these subelements. For instance, Funke et al. (2021) analyze the roles of experts and expert-based information in advocacy coalitions around acid mine drainage in South Africa. A major contribution of this study is that experts can use information to raise awareness about a problem to influence government policy. This finding cements the role of technical information as a tool of persuasion in policy-making, even with reported success. In another study, Malkamäki et al. (2021) study coalition formation based on broad policy core beliefs and specific policy instruments. They find that broad policy core beliefs and specific policy instruments foster coalition formation, but the latter is less strong; this suggests that actors who coalesce around specific policy instruments may be less inclined to advocate policy. This confirms the ACF’s arguments about the importance of policy core beliefs in advocacy coalitions compared to more narrow policy instruments (and what the ACF would classify as secondary beliefs).

Lastly, as a framework, the ACF supports the analyses of relationships among macro-level variables and can complement other analytical lenses to understand policy processes. In one instance, Mockshell and Birner (2020) combine the ACF and Narrative Policy Framework to examine agricultural policies in Senegal. They identify two coalitions, namely agricultural support coalition and agricultural support critique coalition, that advance different narratives to underline their policy positions. Another study (Harris 2019) uses the ACF in combination with policy transfer theory to study advocacy coalitions and the transfer of nutrition policy to Zambia. The study identifies a national coalition and an international coalition in Zambia’s nutrition policy subsystem. This study is significant because it highlights a potentially less studied strategy through which coalitions influence policy: policy transfer. In this case, the international coalition was found to transfer policy ideas from the organizations they represent.

We deem the analyses that combine different theories as particularly useful because they demonstrate ACF’s flexibility and can unearth essential insights that may not be observed using one framework alone. What we can learn from Mockshell’s and Birner’s (2020) study is the explicit analysis of the argument structure of coalitions elucidating the relationship between agents and their discourses in policy processes. Harris’s (2019) study highlights the connections between national and international contexts in policy-making and elucidates how the ACF can foster the analysis of policy questions that have both local and international twists. To our knowledge, none of the ACF applications combined this framework with a model.

6 Conclusion

This paper analyzes how the ACF, a theoretical approach developed in Western democracies, has been applied in African countries. The analysis of the 27 ACF applications published in journal articles and spread throughout the African continent reveals many interesting insights.

The 27 applications cover 15 of the 54 African countries and include democracies, incomplete democracies, hybrid systems, and authoritarian systems. The applications spread across topics, with environmental and energy issues most frequently studied and with a relatively high frequency of studies in agriculture policy. Most of the applications are dedicated to the study of advocacy coalitions, followed by the analysis of policy change. Almost all applications confirmed the ACF’s hypotheses for the stability of coalitions, the importance of policy core beliefs, and the presence of at least one of its pathways for policy change. None of the 27 applications to an African country or constituency is dedicated to policy-oriented learning as a theoretical emphasis. Some studies (e.g., Kingiri 2011, 2014) explored learning, but only in the context of being one of the pathways to policy change. Other theoretical concepts studied include policy brokers, the role of experts, and resources. Indeed, only three of the traditional ACF hypotheses were deliberately and explicitly tested, suggesting that—along with learning—there remain more theoretical ideas to analyze. Yet we must also laud applications that apply the ACF in a way that suits the goals and interests and the researcher, even if it falls outside of the ACF’s traditional hypotheses. The flexibility of the ACF represents one source of its utility.

Overall, and similar to other global and regional ACF reviews (Weible et al. 2009; Henry et al. 2014; Pierce et al. 2017) also outside the Western context (Jang et al. 2016; Li and Weible 2021) or based in one particular Western country (Nohrstedt and Olofsson 2016), we can confirm that the most commonly focused topical area is environmental and energy issues related to coalition formation and development, and that most African applications are single-case studies based on qualitative methods. Interestingly, and except for only one study, the ACF applications in African countries analyze national subsystems, which likely reflects the predominance of their unitary forms of government (only Nigeria and Ethiopia in this sample of countries have federal systems of government). Perhaps also unusual compared to ACF applications in Western democracies, international organizations were prominent among coalition members, or even took the role of so-called policy brokers. This observation reflects the potential influence of external actors on policy processes on the continent (see also similar findings in South Korea in Jang et al. 2016).

Consistent with the policy transfer literature, we suspect that ideas could provide one mechanism through which these international actors might influence policy (Stone et al. 2020).

Besides coalitions, policy change was also frequently analyzed in the 27 studies, and the findings mostly confirm ACF’s theoretical expectations. Most studies identified internal and external events leading to policy change and, occasionally, policy learning. Of course, we do not know whether negotiated agreements and learning were underemphasized because the authors overlooked the phenomena or if it did not occur. Some of the nuanced findings, which are not explicit in past reviews, include the role of policy brokers in resolving conflicts among coalitions in fostering policy change.

Ultimately, regarding the ACF’s theoretical arguments, the applications of the ACF in Africa mostly confirm its arguments as phrased in some of its hypotheses and the vignette expressed earlier. For example, most studies identified opposing coalitions bound by their policy core beliefs, and thus identified a certain degree of conflict within the subsystem at study. Those that studied policy change found a previous external or internal shock and the presence of policy brokers as fostering policy change. Different from most ACF studies in Western democracies is the “landscape of actors” present in the subsystems. In the authoritarian regimes, state actors are the sole or dominant actors who bring change and act as entrepreneurs. And besides the national, international and donor organizations also play a crucial role in public policy-making. In a comparative research design, it will be interesting to triangulate these and other findings with data on the duration of independence, degree of cultural fragmentation, subsystem type and maturity, etc.

This meta-review of ACF applications in Africa shows that four of 27 applications compared subsystems across two or three countries. This observation reinforces arguments for using the ACF to conduct comparative public policy scholarship (Jenkins-Smith et al. 2018). This also aligns with an already observable empirical trend in ACF scholarship (Cairney et al. 2018; Kukkonen et al. 2018; Weible and Heikkila 2016) in which policy subsystems are compared across different countries.

Despite the consistency of the findings of applications of the ACF in Africa, questions and concerns remain about its suitability and use as a portable framework.

Not all the theoretical arguments of the ACF were supported, and some were supported with caveats. For instance, Beverwijk et al. (2008) found that policy core beliefs were quite stable in Mozambique’s nascent educational policy subsystem. This finding deviates from prior expectations about the policy core beliefs being unstable as policy subsystems reach maturity (Jenkins-Smith et al., 2018). Similarly, while Malkamäki et al. (2021) found that policy core beliefs supported coalition formation in South Africa’s tree plantation policy subsystem, beliefs about policy instruments were not strongly associated with coalition formation. This potentially implies that different categories of beliefs might provide different degrees of cohesion for coalition formation.

Much of the ACF is not deployed in the applications in Africa. For example, notably missing are studies that focus on learning. More interesting, and this is also true across many country-specific applications, applications of the ACF concentrate primarily on the policy subsystem without incorporating the context setting; that is, they ignore the ACF’s “relatively stable parameters.” When we examine the type of political regime (from authoritarian to democratic), we see no noticeable difference in the ACF’s applications. While this supports the possibility of using the framework in diverse political settings (Li and Weible 2021), it also begs new research about how to incorporate context into the ACF. For example, the ACF’s relatively stable parameters refer to the cultural, institutional, socioeconomic, and biophysical elements that surround a subsystem. These relatively stable parameters affect the presence and amiability of decision-making venues (e.g., legislatures and courts) and discourse venues (e.g., news and social media). It dictates the geographic conditions and their historical distributions. It also dictates the diversity of cultural, ethnic, and racial distributions and abundances. All of this affects the nature of the policy subsystem. Thus, we might observe coalitions, but their most significant characteristics and behaviors over time might also be overlooked if some components of these relatively stable parameters are not considered. However, we continue to struggle with how to accomplish this task.

We should also take pause when drawing comparative insights based on counting commonly found phenomena. For example, outside of the most rigid authoritarian regimes or situations of overt or covert political suppression or oppression, we know that people and organizations mobilize to influence policy decisions. Of course, if our goal is to find individuals and organizations engaged in policy issues, we will find them in most contexts. Similarly, if we are studying policy change, of course we will likely find an example of an internal or external event preceding it or some instance of learning or negotiation among coalitions. Indeed, identifying instances of policy change without one of the ACF’s pathways is so unlikely that the hypothesis is not refutable.

One implication is to consider ACF research in terms of two generations of questions. The first generation pivots around the original hypotheses (as curtailed and integrated in the vignette earlier). The first generation of research confirmed most of the hypotheses about coalitions, raised issues about the conceptualization and measurement of coalitions, and showed that some hypotheses were rarely tested at all (e.g., Weible et al. 2020). Similarly, most past research confirmed that at least one of the pathways was always present before policy change. Although learning is studied the least among ACF’s focal theoretical areas, the results show reinforcement within coalitions and rarely belief change through intercoalition interactions. This first generation of research questions should continue, particularly in filling in some of the rarely studied areas of the ACF and improving conceptualizations and measurements.

Thus, through this first generation of questions, we see a tendency for confirming the core hypotheses from the ACF about coalitions, learning, and policy change. Moreover, we have learned from this first generation of questions that the ACF is portable but that its applications are sensitive to measurement issues and often overlook issues related to context and the relatively stable parameters.

What we need is a second generation of questions that dig a little deeper and that might capture local nuances. These questions might include the following: How does the broader context (i.e., relatively stable parameters) affect the policy subsystem? Who is involved and represented and who is not in policy subsystem affairs? What political behaviors and strategies are observed, and what resources are deployed? Who has influence, and who does not? What are the strategies of successful policy brokers? What is the nature of the analytical debates (i.e., public argumentations and narratives), and which ones are persuasive or not? How are policy conflicts overcome, circumvented, or resolved? Answering these questions requires a deeper dive into our cases and moves us to research designs that document and describe and get more into dynamics, mechanisms, and short-term and long-term patterns. It also opens the door for original hypotheses about them and perhaps the development of subvignettes or more tailored or specific flow diagrams.

The motivation of this study was to understand and assess the portability and usefulness of a traditionally Western theoretical approach as given in the ACF, and how it has been applied in African countries. Yes, indeed, we find the ACF is portable to the African context. Yet we caution against its blind application and deployment in Africa or in any context. The emergent pattern is one of confirming the ACF’s first-generation questions and theoretical arguments (i.e., vignette). What we are missing is deeper and meaningful second-generation questions about policy processes. We invite scholars within the ACF’s research program to begin to devote more attention to the second-generation questions.