The Advocacy Coalition Index: A new approach for identifying advocacy coalitions

Policy scholars have increasingly focused on collaborative and competitive relationships between stakeholder coalitions. The Advocacy Coalition Framework (ACF) in particular has directed scholarly attention toward such relationships. ACF defines advocacy coalitions as groups of actors who share beliefs coordinate their action. However, previous research been inconsistent defining measuring coalitions, which hampered comparative theory building. We present a method called the Index, measures belief similarity coordination action manner that makes it possible to assess extent are found policy subsystems, whether subgroups resemble how individual contribute coalition formation. index provides standardized for identifying can be applied research. To illustrate effectiveness index, we analyze two climate change namely Finland Sweden, shown differ terms association with coordination. demonstrate performs well different types most formation, those involved cross-coalition brokerage. ????????????????????????????????????ACF????????????????????????????????????????????????????????????????????????????????????????????????“??????”?Advocacy Index????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????cross-coalition brokerage?????? Los especialistas en políticas se han centrado cada vez más las relaciones de colaboración y competitividad entre coaliciones partes interesadas. El Marco la Coalición Defensa ha dirigido atención los académicos hacia tales relaciones. La define defensa como grupos actores que comparten creencias coordinan su acción. Sin embargo, investigación anterior sido inconsistente definición medición coaliciones, lo obstaculizado comparativa construcción teorías. Presentamos un método llamado Índice Defensa, mide similitud coordinación acción una manera permite evaluar hasta qué punto encuentran subsistemas políticas, si subgrupos parecen cómo contribuyen individuales. formación coaliciones. índice proporciona estandarizado para identificar pueden aplicar comparativa. Para ilustrar efectividad del índice, analizamos dos cambio climático, saber, Finlandia Suecia, demostrado difieren términos asociación con coordinación. Demostramos el funciona bien identificación diferentes tipos subsistemas, así aquellos involucrados intermediación Political rarely able impose will others by acting alone. need political other gives rise phenomenon politics. According Weible et al. (2016, p. 1), politics results when people or organizations “mobilize about what government should not do an issue.” is one prominent theoretical frameworks relation public general, hundreds empirical applications, initially US (e.g., Henry al., 2011; Weible, 2006) more recently Europe Ingold, Nohrstedt, 2009) elsewhere (Henry 2014). built upon this wealth studies further development Jenkins-Smith 2014; 2019). With its emphasis enabled understand engage another, aim translating positions into policy. employed various methods identify A basic premise “actors hold similar core act concert—[…] first condition formation sufficient second” (Sabatier, 1998, 115). Accordingly, many early applications assumed would automatically follow. After assumption was critiqued being unrealistic (Schlager, 1995), shifted social network analysis, verifying co-existence (Sabatier & 2007). Some apply analysis examine based existence then also aligned (Ansell 2009; DeBray Gronow Ylä-Anttila, 2019; Ingold Gschwend, Ocelík Sabatier 2007; Tindall 2020; Wagner 2018). Others proceeded reverse order, grouping (Fischer, Variation measured perceived strength framework affords considerable applicability (Weible 2019); hand, may weakness, particularly light increasing calls regarding Varone, 2012) systematic building informed If defined differently each case, comparing becomes difficult, presents obstacles development. For example, major book concludes authors “are unable draw direct comparisons across countries attributes given methodological approaches” (Ingold 2016, 253). Drawing lessons from thus difficult if there excessive variance approaches used It therefore necessary develop “common complementary cases” In addition, analyses either related action, but they failed determine why case outcome. solution propose paper Index (ACI). ACI cross-case comparisons. Furthermore, exist degree subgroup within subsystem fulfils homophily conditions ACF, helps instances brokerage, often serve catalyst 2012). next section, review premises practical presented literature argue improve existing approaches. introduce ACI, do, use compare subsystems Sweden. conclude, discuss index's benefits limitations detail suggest lines future perceives process competition among united shared (Jenkins-Smith 2014, 2017; Sabatier, 1988; Jenkins-Smith, 1999). both coordination; activities 1998). Of central importance here so-called beliefs, refer normative values perceptions problems at hand. These “the primary perceptual filter allies opponents” 2005, 183). “some working together achieve objectives” 2007, 196). (2019) contend minimal beliefs. Thus, characterized mere described (or “disconnected coalitions” according Weible), true actions once observe opponents beginning pool resources form alliances, begin seeking because “to remain without invite defeat” 1988, 140). This alliances strengthened devil shift, tendency perceive one's hostile powerful than 140; 1987). reinforces original impulse assemble can, principle, precipitate situation every actor tied coalition. versions were criticized making assumptions invariably follow reacted turning data, hoping make verify sense since proven associated (Henry, Leifeld Schneider, 2012; Matti Sandström, 2011, 2013; 2005). does always similarity, Fischer (2014, pp. 351–52) asserted collaboration too rigid criterion Other analyzing While these struggled Originally, developed US, then, contexts. new contexts, researchers collaboration. Scholars interpreted findings positing consensual, role less pronounced (Gronow 2020). acknowledged overall mode interaction actors. (2008), adversarial bread butter forms possible. Subsystems unitary, comes down single, dominating coalition, cooperative well. offered no clear guidelines subsystems. Our helpful evaluates nature basis lines. mainly takes place whereas tends extend categorized (2019), ideal-typical members exclusively collaborate like-minded contrast, whose own. connection typology Adversarial naturally coexist single subsystem, our demonstrate. As third type, mention disconnected wherein against despite similarity. (2011) suggests intermediate groups, ties type challenges (Koebele, 2019) explain logic behind group. (Ingold, Satoh, 2014) rather frequently. There several functions groups. connected some members) brokers 450; Brokers typically moderate actual members. An brokerage context important mediating opposing compromises 1988). context, thus, foremost, connections clear-cut definitions brokers. aspect perspective, fact mediate any while assumptions, principal still so (Weible, 2008). identified ACI. conforms expectations and, real world correspond ACF. Finally, discussed above, calculated level entire compared processes case. Next, features work detail. combined measure line assumptions. Data include information needed. One obtaining data involves surveys questions concerning collaboration, coded documents, press releases, materials. unit observation relationship, nonexistence tie postulate ideal tend actions, establishing homophilous ties. extreme all collaborate, forged diverging ideal, imply ever entirely fulfill criteria (see term describe Weberian fulfills ACF's twin real-life them ideal. claim would, efficient achieving goals. Much like indicators Pearson's correlation), constellation empirically probable theoretically possible, one. By demarcating condition, divergence idea. Divergence occur ways. First, “cross-coalition” interactions 191; 473; 2008; 2009). Based terminology, adopt potential reasons exist. might strive mitigate conflicts 2019, 8). common help costs, politicization, implementation failures 39). second scenario, missing relative type. Whereas implicitly presumed interact, revealed 1995; 2008, 622). number increases, costly, prevent occurrence free riding. group size affects ability collective (Olson, 1971). lacking. simple example illustrates equation means practice. Imagine six favor (coded 1) 0); sake simplicity, undirected binary. three (A1, A2, A3) policy, remaining (B1, B2, B3) opposed it. Figure 1a exhibits ideal-type situation, aside A2 A3 B3. stored conventional adjacency matrix format. Likewise, actors' scores converted agreement taking absolute difference score pair subtracting obtained 1. cells latter 1 same 0 differ. take entry coordinates feature plot (Figure 1b). points located = x, call line, adhere only occurs point , represents A1 (x-axis) (y-axis), aligns collaborate. Point B2 however, consistent exists B3, tie, words, tie. diverges A3, cases indexed point's distance line. It's worth noting places equal weight ties, although could argued assume rarer option involve weighting calculating ratio question, posit precisely be. weighting, interpretations focus raw value account versus below. Until now, considered simplest scenario. idea extended directed, valued networks scaled attributes, consider direction degrees belief. consideration enables transition beyond binary classifications (i.e., not, different) offers realistic description real-world strong weak, range divergence. multiple considered. Many surveys, ask mutual exchange (e.g. exchange) better complex create setting cut-off treating set matrices multiplex network, applying clustering techniques generalized Louvain algorithm (Mucha 2010). matrix. Scaled vary Similarly, handled network's careful standardization directly final location graph. Whether region see P 1b) (P’) depends researcher standardizes scores. valued, must graph corresponds understanding position before selecting method. Once determined, established. Ties above discarded certain solely left network. discard A3-B3 obtain components, A1-A2-A3 B1-B2-B3, conceptualized cases, decompose clearly, between-component exist, depending values. Therefore, addition discarding established techniques, community detection (Blondel 2008) factions (Borgatti 2002), generally force even treated easily consist Alternatively, intention find Both like-minded/non-like-minded likeminded/non-like-minded actualized, respectively. subcomponents useful actors, literature. believe contributes offering operationalizations components section actor-level assumes 197). variation, distinction auxiliary made 197; Larsen 2006; Zafonte 2004). former members, peripheral regularly engaged coalition-related activity. effective tool high indicates actualized member. low 1, likely member, half, member score, opponents; 0.33. apt indicator transcends betweenness centrality), irrespective cross (cf. activated evenly, expected behave predominantly demonstrated higher reflecting specific roles subsystem.1 At level, CCH characterizing two-dimensional diagram. high, few far between. Members own, yield ratio. They inactive forming opponents. last active forging 2011)