Developing Political Strategies across a New Democratic and State Architecture

Under new democratic regimes, civil society organizations (CSOs) alter their political strategies to better engage public officials and citizens as well as to influence broader political debates. In Brazil, between 1990 and 2010, CSOs gained access to a broad participatory architecture as well as a reconfigured state, inducing CSOs to employ a wider range of strategies. This article uses a political network approach to illuminate variation in CSOs’ political strategies across four policy arenas and show how the role of the state, the broader configuration of civil society, the interests of elected officials, and the rules of participatory institutions interact to produce this variation. Data for this article’s analysis come from a survey of three hundred CSO leaders in the Brazilian city of Belo Horizonte. The survey identified the strategies they employed to promote policy change and direct resource allocation in the arenas of participatory budgeting, health care, social services, and housing. Sociographs generated from survey results reveal a distinct clustering within each policy arena of the strategies employed by CSOs, providing further support to the usefulness of the analytical framework.


Introduction
Civil society activists, across Latin America, now engage in demonstrations, incremental participatory venues, campaigns and elections, and oversight programs; they move inside and outside state and democratic institutions in the hopes of securing scarce public resources for their communities. 1 These activists develop political, personal, and policy networks to forge better connections to civil servants and public officials responsible for policy selection as well as service delivery. The dual reform movements of democratization and decentralization that spread across Latin America over the past thirty years have given activists a greater range of political and policy venues through which they can place their demands (McNulty 2011;Baiocchi, Heller, and Silva 2011;Snyder 2001;Grindle 2007;Lavalle, Acharya, and Houtzager 2005;Herrera 2017;Alvarez et al. 2017). How have civil society leaders adapted their strategies to take advantage of this new democratic architecture? Given the diversity of strategies that researchers have identified, a second question helps to account for these variations: How does the local context shape civil society leaders' strategies?
This article develops a political network approach to account for the wide range of the political activities within and across different policy arenas (health care, housing, social services, basic infrastructure). Four factors most significantly affect the strategies employed by civil society leaders: state formation, the historical development of civil society, party politics, and specific rules that determine who can participate in participatory venues. The process of state formation conditions whether a state has sufficient resources, personnel, and competencies to fulfill required constitutional and legislative obligations; civil society organizations (CSOs) are more likely to seek access to state agencies that have greater resources, more civil servants, and broader competencies. If the state is weakly established and without resources, activists must turn to other outlets, such as protest and party politics, to advance their claims. Second, the density of civil society and preexisting political repertories strongly condition CSOs' development of new strategies. When there are greater numbers of participants, CSO leaders have a broader base of resources that supports the development of new strategies: a greater number of participants increases CSO leaders' ability to overcome participation fatigue, it helps them to have representatives present in multiple venues and demonstrates to party and elected officials that their movement has the potential to mobilize citizens as voters and/or as protesters. In addition, we also expect that CSOs working in a civil society field with a diverse set of preexisting political repertories would be more likely to develop political strategies that permit them to engage in multiple state and democratic arenas.
The third factor involves party politics. CSOs and social movements are situated within a broader sociopolitical system that includes party politics. CSOs' willingness to work within state and participatory venues is strongly conditioned by their alliances with party leaders. When CSOs deem that elected officials and their political appointees are more sympathetic to their demands, it is more likely that these CSOs will work within state and democratic venues. Of course, as the degree of support decreases, it becomes more probable that CSOs will protest and use contentious politics. Finally, participatory institutions across the region use different rule sets to incorporate citizen participants. When these rules are more open, greater numbers of citizens will participate, thus altering civil society leaders' calculus regarding their choice of strategies. This analytical framework permits a broader understanding of the changing nature of democratic politics because it incorporates the vital roles of the state, civil society, political parties, and institutional rules.
Across the region, Brazil stands out for having created an extensive participatory governance architecture, a decentralized policy system in which municipalities deliver services, and a mobilized citizenry. Citizens have many opportunities to exercise voice and vote in a series incremental policy-making processes, including public policy management conferences, policy councils, and participatory budgeting (Abers and Keck 2013;Avritzer 2009;Pires and Vaz 2012;. If the mantra of many Brazilian CSOs during the 1980s, as Brazil returned to democratic rule, was "autonomy" (Weffort 1989;Dagnino 1994), the mantra evolved into "co-governance," whereby CSO leaders became responsible for mobilizing their communities and engaging in incremental policy-making within state-sanctioned venues. De-monopolization of access points into the state is transforming the basic tenor of state-society relations, thus establishing new forms of citizen engagement and accountability (Heller and Evans 2010;Smulovitz and Peruzzotti 2000;Wampler and Touchton 2016;Fox 2015). A key finding of this article is a clustering of political activities within different policy arenas. Housing activists, for example, are likely to make strategy choices similar to other housing activists but quite distinct from activists working in the social assistance policy arena. The implication of this finding is that there are historical processes and institutional rules specific to each policy arena that induce CSO leaders to act similarly within a policy arena but differently across distinct policy arenas.

Activists in Action across Multiple Venues
Three vignettes illuminate the multiple venues used by community activists in pursuit of both narrow and broader public goods. CSO leaders need to master the technical minutia of legalistic policy-making processes so that they may convince government officials that their proposals can be implemented. Elsewhere, they must also connect strongly to government officials to build political support to show that their projects should be implemented; this is often in the forms of working on electoral campaigns or holding public demonstrations. They also use public venues to engage in contentious deliberation to pressure government officials to implement their policy preferences.
Augusto, a combative delegate of the Belo Horizonte Participatory Budgeting (PB) meeting, approached me at a downtown bus stop. After catching up on local PB news, we discovered that we were heading to the same meeting at the social service council. Augusto explained that he had just been at a municipal office (infrastructure), in an effort to resolve a sewage problem in his community. Because he was already downtown, Augusto decided to attend the social service council meeting to make contact with government officials regarding a different problem.
Augusto didn't say a word during the social service council meeting-very different from his role at regional PB meetings, where he is often the first to raise contentious issues and complain about government inaction. There is a different deliberative dynamic at social service council meetings. Educated professionals dominate the discussion, so there is considerable distance between the professionals' technical language and the knowledge base of community leaders. Augusto is representative of community activists who work in both participatory venues and government offices to secure social services and public works for their communities; these leaders must learn a range of skills to secure their policy goals.
At a PB meeting in the central region, I met Patrocina, a community leader from a local favela. I crossed paths with her the following week at a municipal housing council meeting. She was relatively new at community organizing and policy-making, so she felt she needed to attend multiple meetings to gain information, knowledge, and skills. Patrocina was trying to master the technical language associated with policy-making while also attempting to organize her neighbors to attend participatory meetings. When I visited her home, she showed me with immense pride the sewage lines that had been built through the PB process; she was participating because she wanted to see if it would be possible to tear down their existing houses and build apartment buildings. The policy-making processes created by the government greatly expanded the number of opportunities for her to directly interact with government officials. Patrocina was using the participatory institutions to establish policy and political connections among her neighbors, other CSO leaders, and government officials.
Julio, an active community leader in the Alto Vera Cruz favela, was deeply involved in multiple participatory institutions, a community organization, electoral campaigns, political party politics, and cultural events. He is a full-time community leader, receiving a small monthly stipend from the Communist Party of Brazil (PC do B) in addition to earning money as an event organizer. When Julio was a teenager in the early 1990s, he organized community dances that were well attended by local teenagers. Leaders from PC do B noticed his organizing skills and encouraged Julio to attend PB regional meetings to secure municipal funding for his cultural events. Due to his success in PB regional, Julio became very involved in community organizing, serving as an important bridge between an aging activist population and younger people. Julio's community and political organizing allows him to wear many different hats-party official, communist ideologue, cultural coordinator, aide to a municipal legislator, community leader-and he is able to build a dense local network and a broad network across the municipality.
These vignettes illuminate the diversity of strategies that CSO leaders use to advance their interests. They must work in a variety of venues and master different skills-from reading technical reports to organizing protests-in order to advance their organizations' interests.

Alternative Explanations
Three explanations are commonly used to explain social movements' and CSOs' activities: political opportunities, contentious politics, and political cultural.
The political opportunity approach provides an explanation based on a combination of both structure and agency (McAdam, McCarthy, and Zald 1996;Tarrow 1998). The structural argument identifies key economic and political institutions that frame the environment in which social movements operate. The agency is based on the ability of citizens to take advantage of political "opportunities" through which to advance their agenda. This approach emphasizes properly identifying the political opportunities that permit social movements to effect change; it also emphasizes identifying moments of high mobilization rather than incremental politics over an extended period. Thus, the real limitation to this approach is a strong focus on these moments of high action and an inability to sufficiently capture the day-to-day workings of social movements. The analytical framework developed in this article is better situated to account for the ongoing, day-to-day political and policy management that is now occurring.
A second explanation prominent in the social movement field is a focus on contentious politics, which seeks to show how politically marginalized actors use political disruption and political protest to advance their policy agenda. Contentious politics is a particularly important tactic used by social movements (Tarrow 1998). However, one drawback of the literature on contentious politics is that it doesn't sufficiently account for the multiple other avenues of political action undertaken by social movement leaders. This article demonstrates that CSO leaders are now engaged in a wide variety of political activities. CSOs in Brazil and across Latin America use contentious politics as well as participation in electoral campaigns, participatory institutions, and direct contact with government officials over policy adoption and implementation.
The role of political culture is a final explanation that scholars have long used to account for differences regarding the political behaviors and attitudes of civil society actors. Robert Putnam and colleagues (1993), in their classic Italy study, argue that differences in the density of civil society offer the best explanation for political, economic, and social change. Evelina Dagnino (1994) draws from new social movement theory and Gramsci to demonstrate how political contestation over the values that undergird a political system is vital to understanding how and when social change can occur. Dagnino's work overcomes the path dependency determinism of Putnam's work, but her work is limited by an unclear explanation regarding the incentives and motivations that induce CSO leaders to work in different political and policy environments. The analytical framework developed in this article incorporates Dagnino's key insights regarding the importance of political contestation over constitutionally guaranteed rights, but moves beyond the political cultural approach to better capture specific incentives that encourage CSOs to select different strategies.

Analytical Framework
The historical development of the state conditions the strategies used by CSO leaders, due to the degree to which there is an administrative state capable of implementing public goods. The establishment of a state bureaucracy capable of delivering public services induces CSO leaders to focus their political energies on gaining access to civil servants and elected officials who set policies and control those resources (Abers and Keck 2013). CSO activists use formal policy-making institutions, such as participatory institutions and public hearings, as well as informal processes, such as direct contact and developing personal networks, to pursue their interests within the broader state apparatus. A state that has resources and the capacity to implement public goods will induce CSO leaders to align themselves with civil servants and elected officials who control the state. CSO leaders attempt to redirect the state in the hope that state resources and authority will be used in their community.
In addition, the development of a larger state administrative apparatus, which began in many Latin American countries during the 1940s and 1950s, is also associated with expansion of public employee unions (French 1992). In the current democratic period, unions often represent the interests of their members by working across multiple formal institutional venues as well as within civil society. Union members often align themselves with social movements operating in civil society, altering the configuration of civil society.
Second, the historical development of civil society within each policy arena strongly affects the political strategies employed by CSOs. A combination of the established repertories of political engagement and the membership base of the CSOs helps to explain their strategies (Tarrow 1998;McAdam, McCarthy, and Zald 1996). By repertories of political engagement, I refer to the willingness of CSOs to utilize different tactics such as holding public demonstrations, attending public hearings, or engaging in participatory venues (Tarrow 1998;Holston 2008;Alvarez 1990). The membership base is a way to conceptualize the density and demographic profile of the CSOs. Density matters, especially in democratic settings, because larger numbers of activists can help amplify a CSO's voice and allow CSO members to attend different official functions (i.e., attend public hearings, go to protest events) (Putnam 1993).
The demographic characteristics of CSOs' membership base (actual as well as potential) also have a major impact on their strategies. When the potential beneficiaries of proposed policies are also active CSO members, there is a greater likelihood that a broader proliferation of strategies will be employed. When a small professional elite that won't directly benefit leads the CSO, there will be a narrower set of political strategies. For example, middle-class professional-led CSOs that advocate on behalf of "pro-poor" policies are likely to act very differently than CSOs that are led by individuals who will personally gain from policy change.
Third, the political coalition in charge of the executive branch (president, governors, and mayors) often sets the tone for how different participatory institutions will be used (Keck and Abers 2013; Wampler 2015; Baker, Ames, and Renno 2006). It is well-documented that strong presidential systems are a feature across Latin America, and this phenomenon clearly extends to Brazilian municipalities. Executives have significant leeway in the amount of support they extend to new democratic institutions; they can open or close different entry points into the local state through their funding priorities, their administrative staffing, and willingness to work with CSOs.
Fourth, the rules that regulate who can participate in participatory institutions as well as over what they deliberate also significantly affects the strategies employed by CSOs (Fung 2006). Rules that promote more open participation induce greater numbers of ordinary citizens to assume leadership positions and to have greater levels of contact with government officials. Rules that are more restrictive are more likely to draw in a smaller number of professional CSOs. In addition, participatory venues that are highly technical in nature are also more likely to draw in professional CSOs and discourage ordinary citizens from participating.
These four factors-state, civil society, elected political coalition, and institutional rules-have independent and interactive effects on the configuration of civil society organizing. Theoretically, this approach advances the debate over existing approaches. Most importantly, it incorporates both structure and agency to account for the ongoing political strategies that are used during the process of building out incremental politics. The structural approach is rooted in the long-term state and civil society development. Agency is then rooted in the specific incentives provided by participatory institutions as well as the support extended by elected political leaders. This approach moves beyond the traditional political opportunity approach because it accounts for the normal political activities that characterize so much of civil society organizing.
Second, the analytical framework helps to account for the clustering of political activities as revealed by the use of the political network approach. The analysis of the empirical evidence, presented below, illustrates that CSO leaders demonstrate similar patterns of behavior within each specific policy arena. This draws attention to the vital importance of moving beyond single case studies, as well as to how specific institutional arrangements help condition action.

New Institutional Architecture
Brazil experienced five key institutional changes during the 1980s and 1990s that altered how citizens and activists engage the state and democratic processes. First, the transition from authoritarian military regime to representative democracy in the mid-1980s included both the military regime's efforts to hand control over to civilians and the extensive mobilization of labor unions and social movements during the 1970s and 1980s (Stepan 1989). The second major institutional change was the promulgation of the 1988 Constitution, the "citizens' constitution" (Avritzer 2009), which formally extended political and social rights to all citizens.
The third institutional change was the decentralization of the federal system, especially with regard to service delivery (Montero and Samuels 2004). Municipalities spend roughly 15 percent of all public resources, with states spending around 35 percent. Increased municipal-level control over service delivery heightened citizens' and CSOs' attention to the state as they seek to gain access to the social rights formally controlled by the 1988 Constitution.
The fourth change was establishment of an extensive participatory architecture that permits citizens and CSOs multiple opportunities to be involved in ongoing deliberative policy-making processes (Avritzer 2009;Pires and Vaz 2012). This participatory architecture (councils, conferences, participatory budgeting) has clearly altered the institutional terrain on which citizens and CSOs seek to engage the state. By 2015, there were over fifty thousand public policy management councils ; see also Barreto 2011). Between 2000 and 2010, roughly 50 percent of Brazil's population lived in a city using participatory budgeting . CSO leaders' selection of political and policy strategies thus partly stems from how citizens are included in these processes. PB programs allow for open participation, which is then often followed by an internal process in which citizens elect PB delegates. Policy councils are not based on open participation, but on a prior selection of representatives to them. Thus, PB begins with an open-participation model whereas councils are based on a more limited number of representatives.
The final institutional change was the expansion of the public goods provided by the state during the first decade of the twenty-first century. The economic boom, accompanied by President Luiz Inácio Lula da Silva's interest in social programs to help poor citizens gain access to basic social rights, provided the resources and political will to enhance what the state was able to provide to citizens (Schneider 2016). This altered the political calculus of CSO leaders, as they were more actively looking to the state for public goods.
In sum, the expansion of rights and the decentralization of social service delivery induced CSOs to target the local state to secure access to public resources. There is now a significant expansion of signaling methods (voting, protests) as well as venues (councils, conferences, government offices) that involve ongoing contact and interactions among citizens, CSOs, and government officials.

Belo Horizonte
This article narrows the focus to the city of Belo Horizonte, Brazil, to illustrate how CSOs adapted their political strategies to take advantage of the new participatory architecture and expanded state. Five successive leftist and center-right city governments in Belo Horizonte, the state capital of Minas Gerais, institutionalized a transformative political project intended to promote social justice and popular participation. A political coalition led by the Workers' Party (leftist during 1980s and 1990s, centrist in 2000s) and the Brazilian Socialist Party (a moderate, center-right party) enacted a series of policy reforms in which they sought to include citizens' voices in decision-making and to direct scarce resources to lowincome neighborhoods. Over an eighteen-year period (1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010), government leaders adopted a series of public venues to provide citizens with access to government officials, allow citizens a voice in the policymaking process and, finally, give citizens a vote over policies and programs.
The methodological approach employed is comparative, but within a single major city. Belo Horizonte has a population of 2.5 million (the metropolitan regime is roughly 5 million), and it has more than forty municipal-level councils, and several hundred community-based councils (e.g., local health care posts, schools, parks). There are more than five thousand positions to which citizens are elected (e.g., local health care council; participatory budgeting delegate) (Wampler 2015, 79). There are also three different PB programs, which involved over 150,000 participants during the 2008-2009 policy-making cycle (Wampler 2015, 80). A comparative analysis of four policy arenas in a single city provides the methodological advantages of holding the city's long-term sociopolitical development constant while allowing a more precise identification in how state formation and the historical development of civil society affect the shape of political life.

Methodological Approach: Political Network Analysis
Citizens, CSO leaders, and government officials at Brazil's local level constantly interact (Abers and Keck 2013;Avritzer 2002;Baiocchi, Heller, and Silva 2011;Wampler 2007. A network-based approach allows us to more systematically identify the clustering of activities utilized by CSO actors. Social network analysis (SNA) seeks to contextualize individuals' actions by analyzing individuals in their sociocultural and political environment. SNA has largely been centered on the connections among individuals (Granovetter 1973;Wasserman and Faust 1994;Borgatti and Everett 1997;McClurg and Young 2011). Within the discipline of political science, there has been a recent turn to network theory to better account for individuals' voting decisions (Baker, Ames, and Renno 2006) and how they join social movements (Diani 1992), participate in voluntary organizations (Ansell and Gash 2008), exercise power (Lazer 2011), or form political networks (Knoke 1994).
A political network analysis (PNA) is based on the core idea that the "basic units of any complex political system are not individuals, but positions or roles occupied by social actors and the relations or connections between these positions" (Knoke 1994, 7; see also McClurg and Young 2011). CSO leaders have leadership roles within their communities and organizations, which lead to work within a variety of institutions (state agencies, democratic venues). PNA allows for the "mapping" of the political activities of CSO leaders, who are their organizations' representatives in a wide range of institutional and extra-institutional spheres. PNA draws heavily on a relational perspective, which analyzes how actors are embedded in specific institutional, social, cultural, and political structures (McClurg and Young 2011;Baiocchi 2005;Baiocchi, Heller, and Silva 2011).
It was been well established that CSO leaders from poor communities work within participatory institutions to deliberate over and help select specific policies (Baiocchi, Heller, and Silva 2011;Avritzer 2002;Wampler 2007 and. In these new venues they attempt to secure desperately needed resources for their communities and work within these institutions to expand their ties to political leaders and key bureaucrats. These CSO leaders often face two significant hurdles as they seek to gain public goods and state support for their communities: a lack of the personal connections and networks that have long been a hallmark of access to political power in Brazil, and grave informational asymmetries vis-à-vis professional civil servants and full-time political operatives. "The most important feature of a community power structure is its multiple networks of interorganizational information and resource exchange" (Knoke 1994, 138). CSO leaders use participatory institutions to expand their policy networks as well as to improve their technical knowledge, allowing them bring information from the community into the state and vice versa. PNA creates the opportunity to account for variation across the different policy arenas.

Methods and Evidence
Our research team administered a survey to three hundred CSO leaders to assess their political activities and then analyzed the results using a political network analysis approach. The data are two-mode or affiliation data, which means that they capture the relationship between the individual and an event or institution. The data are binary, which means we know whether or not the individual participated in an event or an institution in a given time period. Affiliation data has both relational and individual-level characteristics (Borgatti, Everett, and Freeman 2002).
The four cases were selected because they offered considerable variation in state development, civil society participation, and participatory institutions. Participatory budgeting was selected because it was a pioneering case of direct citizenship engagement, extensive state and government involvement, and robust civil society engagement. The housing policy arena was selected because it had extensive civil society engagement in multiple participatory institutions. The social service policy arena was selected because it is a relatively new formal policy arena, with many programs and institutions being established during the new democratic period. Finally, health care was selected because the civil society field is one of the strongest and best established across Brazil, and because the right to health care was one of the most important that governments sought to extend to citizens.
The "CSO Leader" survey included one hundred PB delegates, one hundred social service CSO leaders, fifty housing leaders, and fifty health care leaders.
• Participatory budgeting regional: "PB delegates" are elected within PB regional to monitor the implementation of small and mid-sized projects that were selected within the process. We obtained a list of 1,200 PB delegates and out of this list 100 respondents were randomly selected. • Housing: Housing núcleo leaders were included in the survey. A housing núcleo must organize one hundred families, all of whom had lived in Belo Horizonte for at least two years, and had a monthly household income of less than US$300. Fifty housing núcleo leaders were interviewed from the 172 housing núcleos formally registered with the municipality. • Health care: Municipal and regional (submunicipal) council members were included in the survey. We drew upon available lists of council members at the municipal and regional levels (nine regions). The council members were elected members representing CSOs (not including unions) within the council. The fifty respondents were randomly selected from a list of five hundred individuals. • Social services: Community organizations formally registered with the Municipal Council for Social Services (Conselho Municipal de Assistência Social, CMAS) were included in the survey. Out of a total of 686 associations that were registered at the time of survey, 100 associations were included in the survey.
Two types of CSOs are missing from the survey. First, we were unable to systematically include evangelical associations. Our research team had numerous contacts with evangelical leaders, but they were unwilling to provide the necessary contact information. Second, small organizations that have limited contact with government officials or public institutions are not included in the survey. These organizations often have short life spans and face significant difficulties maintaining participation.
There are a total of eight "institutional" or "event" nodes, which are the events or institutions that the CSO leaders reported actively using in the past two months. There are three hundred "individual" nodes, representing each individual respondent. The data do not link individuals to each other (Augusto to Patrocina) or to the same events (Demonstration on X day in Y location). Rather the data reveal the types of activities utilized by citizens (e.g., Augusto participated in a demonstration in past two months).

Civil Society Alliances
Participatory institutions provide a structure that encourages CSO leaders to engage broader networks outside their community and develop alliances because these leaders are in constant contact with one another. These institutions also help CSOs mobilize their own followers because the incentive structure encourages the distribution of specific benefits, helping CSOs to overcome hurdles associated with collective organization (McAdam, McCarthy, and Zald 1996;Ostrom 1990;Olson 1965). Information flows between CSO leaders and government officials as well as between CSO leaders and citizens, meaning that CSO leaders are intermediaries linking state and society. The bonds among CSO leaders allow them to share information and build a broader collective organization that can be used to apply pressure on government officials. To capture CSOs leaders' activities, the survey included three questions: 1. Does your association have weekly or biweekly meetings? Node label: "Attended CSO meetings" 2. Did you attend a public demonstration in the past two months? Node label: "Attended demonstration" 3. Did you attend a PB meeting or a council meeting (municipal or regional) in the past two months?
Node label: "Attended PB meeting" (PB delegates only) OR "Attended municipal council meeting" These three nodes tap into distinct types of activities. The first node, "Attended CSO meetings," addresses the extent to which CSO leaders participate in internal meetings. Holding meetings is vital to sharing information among members, reaffirming shared bonds among members, and refining their organization's mission (Wampler and Avritzer 2004;McAdam, McCarthy, and Zald 1996). The second node, "Attended demonstration," taps into the willingness of the CSO leader to engage in demonstrations and protest activities central to the success of social movements (Tarrow 1998;Mische 2008). Because there is a political risk for activists engaging in a public demonstration against the local municipal government, it is likely that these participants would have stronger "bonds of solidarity" with their fellow citizens and other CSOs (Alexander 2006). The use of contentious politics is carefully considered, as demonstrations can severely strain relationships with government officials and other power holders. The third node, labeled either "Attended PB meeting" or "Attended council meeting," addresses CSO leaders' recent participation in participatory institutions, demonstrating the importance of these new institutions to leaders' political strategies. We would expect that participatory democracy is more central to those policy arenas that have more robust forms of social movement mobilization.

CSO-State Connections
CSO leaders and citizens bring information about problems to the attention of government officials, who then must decide how to act. Government officials share information with CSO leaders concerning what types of resources are available and what steps must be taken by CSO activists to secure resources. CSO leaders develop connections to government officials and subsequently use these connections to build the broader frames that allow them to maintain the vitality of their organizations (McAdam, McCarthy, and Zald 1996).
CSO leaders often engage with the state and democratic institutions to secure a combination of narrow and broader public goods. Broader public goods can include projects that transform their region or community. These policy goods are secured through deliberative processes in which citizens have the legal right to vote on resource allocation. CSO leaders also use the connections forged inside of participatory institutions to secure specific goods that may benefit specific individuals. CSO leaders cultivate ties to public officials because they help overcome the basic collective action problems: Individuals are more likely to engage in collective action organizations if they can secure targeted, specific benefits.
In Brazil (and most other parts of Latin America), there has long been a middle-and upper-class bias associated with access to information. Low levels of education, limited readership of newspapers, and few formal links to government institutions have long made it difficult for the majority of Brazilian citizens to gain access to information. CSO leaders and citizens now have multiple venues through which they can gather information-public hearings, participatory governance meetings, government officials' formal presentations and workshops, and informal conversations with government officials-all of which help CSO leaders from poor communities overcome information asymmetries (Stiglitz 2007). Three questions capture these relationships: 1. Did you contact X municipal department in the past two months? Node label: "Contact with X department" (X = health care, housing, social services, or planning/PB regional) 2. Did you contact a municipal legislator in the past two months? Node label: "Contact with legislator" 3. Does your organization currently have a government contract to provide social services? Node label: "Active government contract" The first node, "Engagement X department," captures the CSO leaders' direct interaction with a state official in relevant administrative agencies (e.g., health care, housing). This taps into ongoing connections linking CSO leaders to public officials. The second node in this section, "Engagement with municipal legislator," taps into the participation of CSO leaders in an additional public venue that shows the worthiness of their organizations' demands as well as their ability to turn out their followers. The legislative branch serves as an entry point that establishes the means for obtaining constituency service. Finally, the third node in this section, "Active government contract," identifies CSOs that receive public resources to provide services to citizens. CSOs with formal government contracts are more likely to be more professionalized than those CSOs without contracts. There is a greater potential for co-optation of CSO leaders by government officials if the leaders rely on discretionary public funding to sustain their organizations.

Campaigns and Elections
Elections are at the core of representative democracy. Elections are also vital to participatory democracy because the political party in control of the executive branch has a tremendous impact on how the new institutions will function and how rights will be extended (Abers 2000;Abers and Keck 2013;Avritzer 2009;Wampler 2007). The constant interaction of CSO leaders and government officials in formal participatory venues, as well as in parallel, informal interactions, produces multiple opportunities for them to forge new alliances. CSOs leaders are well positioned to influence the voting behavior of ordinary citizens. In Brazil, where the larger political context is framed by weak party identification, lower voter information, and strong incentives to cultivate a personalistic vote, CSO leaders emerge as key intermediaries between candidates and voters (Baker, Ames, and Renno 2006).
Two questions address these nodes: 1. Did you attend a campaign rally in the 2008 municipal elections? Node label: "Attended campaign rally" 2. Did you speak with your association's members about a specific candidate? Node label: "Campaigned among members" The first node, "Attended campaign rally," is geared to show the willingness of CSO leaders to be involved in campaigns and elections. When CSO leaders attend these events, they attend as participants but also as representatives who mobilized others to attend. The second node, "Campaigned among members," taps into whether the CSO leaders sought to influence their members' votes, indicating whether CSO leaders are acting as intermediaries between voters and candidates.

Results
To assess the differences across the four public policy arenas, we begin with degree centrality, which measures the importance of the different political activities utilized by CSO leaders in each policy arena.
In Table 1, the three most central activities utilized in each policy arena are in bold text. In all four policy arenas, participation in weekly or biweekly CSO meetings was the most common activity. The density statistic (last row) measures the proportion of actualized connections out of the total possible number of connections. Table 1 shows that housing activists have the densest networks, which means that their leaders are involved in the most activities. PB delegates are the least active CSO leaders-a surprising finding since there is an implicit understanding in the broader PB literature that PB serves as a springboard for greater civic participation (Baiocchi 2005).
To visualize the differences among CSO activity, UCINET and NetDraw software were used to produce sociographs of the activities reported by CSO leaders (Borgatti, Everett, and Freeman 2002). The layout of the nodes was based on a geodesic "spring embedding technique" that locates the nodes "in such a way as to put those with smallest path lengths to one another closest in the graph" (Hanneman and Riddle 2005, chapter 4). The size of the event nodes is calculated based on the degree of the nodes, which is the number of connections to the individual-level nodes. In the sociographs below, the line represents the connections of each individual to the eight institutional or event nodes. We now turn to the four policy arenas.

Participatory budgeting (PB): My PB, my neighborhood
Belo Horizonte's PB process is now a routinized part of policy making. A biannual participatory process provides opportunities for citizens to engage in substantive deliberation, political theater, and policy selection (Wampler 2007. The first phase of PB, from 1994 to 2002, involved greater contestation and vitality among CSOs, which highlights the importance of the political coalition in control of the state. Over time, participatory budgeting institutionalized community-level participation into a single policymaking process. Public officials engage interested citizens and CSOs into a routinized PB process in which they vote on the allocation of small-to medium-sized infrastructure projects to be implemented by the municipal government (local state). Belo Horizonte's PB regional program directs high levels of resources to low-income communities-nearly 60 percent of PB resources go to the poorest communities-suggesting that specific institutional rules strongly affect mobilization and turnout (Wampler 2015, 113).
The sociograph in Figure 1 is drawn on the activities of one hundred PB delegates. At the core of their activity are CSO meetings involving their own members; these meetings engender the "bonds of solidarity" so vital to helping CSOs maintain their collective action (Alexander 2006;McAdam, McCarthy, and Zald 1996). Only 15 percent of the PB delegates work with a CSO holding an active government contract to provide social services, which means that these CSOs are not dependent on the state for their livelihood.
Importantly, the data shows that two-thirds of the PB leaders are politically active only in local PB, and that a third of PB activists are engaged politically across multiple venues beyond PB. The majority of PB delegates do not seek out other participatory or political venues to press their claims, which may be due to the efforts of government officials to organize PB and bring the policy-making process to their communities. As a result, political and policy focus is narrow: "My neighborhood, my PB." The CSO leaders involved in PB are best categorized as neighborhood organizers. PB doesn't serve as a springboard to encourage most PB delegates to be involved politically across a broader political spectrum. Thus, PB plays a much weaker role as a "school of democracy" than was expected or has been found elsewhere (Baiocchi 2005). This finding confirms a criticism of PB programs: that PB programs induce activists and citizens to narrow their political attention to their local community rather than inducing them to organize more broadly (Navarro 2003).
However, there is a second group of activists who engage in a number of activities. In the upper section of Figure 1, there is a clustering of activity around campaigns and elections, and another around demonstrations. These CSO leaders are involved in community organizing, electoral politics, and contentious politics, and they participate locally as well as more broadly as they seek to secure specific public policy goods for their communities. They are likely to have been positively influenced by the democratic and policy opportunities afforded by PB, and illustrate that PB can promote political awareness (Baiocchi 2005;Wampler and Avritzer 2004).
Returning to the analytical framework, we first find there was a local state capable of implementing basic infrastructure projects. Strong community-based organizations forged alliances with elected officials to use this local state capacity in favelas and poor neighborhoods. The open rules channeled participation among community leaders and citizens into PB, which may help explain why two-thirds of the PB leaders only engage in local issues. PB brought public officials directly into their communities, reducing the need for most CSO leaders to expand their participation beyond their local community. In sum, a capable state and an open participatory process focused on local issues encouraged CSOs to focus their political organizing locally as well.

Social service: Third-sector service delivery organizations
The defining feature of CSOs in the social service policy arena is the presence of formal government contracts-77 percent of CSOs have service delivery contracts. The leaders of these CSOs have regular contact with civil servants, attend internal meetings, and have high levels of education (75 percent have college or more). The state outsources social service delivery to CSOs that have the technical and professional skills to comply with complex bureaucratic regulations. The social service CSOs are professional associations most accurately described as third-sector organizations. Figure 2 demonstrates that the social service policy arena is focused on formal, institutionalized processes involving internal CSO management and direct engagement with state agencies. These professional CSOs are directly involved with public officials to ensure they are fulfilling the requirements of their current contract as well as working to secure future contracts. Figure 2 also demonstrates that these CSO leaders have limited engagement with electoral politics, participatory institutions, and protest politics. Public policy management councils in Brazil incorporate citizens, CSO and union representatives, service providers, and government officials into policy-making processes. Most councils are charged with the responsibility of approving their sector's annual budget and monitoring policy implement. The social service council focuses on legalistic and technical issues, which discourages CSOs interested in debating broader policy issues within participatory councils. Potential recipients of social services (service users, clients) are very weakly organized, which means that they are not present in participatory venues. High hurdles to organizing among the poor account for why well-educated professionals take the lead in acting on behalf of potential social service recipients. The activities at the center of Figure 2 are the result of third-sector CSOs' alliance with government officials. Government officials outsource service delivery to professional organizations and the CSOs rely on government funding to support their organization. The roots of this outsourcing extend back to the development of the social service policy arena that began under President Getúlio Vargas in the 1930s (Cunha 2009). A consequence of this outsourcing over the twentieth century was very limited state development, which means that it is not possible to "mobilize the state" because the state is an empty shell in this policy arena (Abers and Keck 2009). Government officials now find it more advantageous to outsource services for a range of reasons: lower cost, the absence of unions, increased policy flexibility, and the opportunity to provide contracts to political allies.
Service contracts tend to be short term and are awarded based on a combination of professional competence and at the discretion of government officials. CSOs seeking state contracts must demonstrate that they are professionally competent, but CSO leaders must also maintain strong ties with elected officials. This helps explain why these CSO leaders had the lowest engagement in public demonstrations.
In sum, it is the absence of an administrative state and the great difficulty in organizing potential beneficiaries that best explains the shape of democratic politics in this policy arena. The state outsources service delivery to professional NGOs, which then lack incentives to politically mobilize beneficiaries. Civil society is thus dominated by third-sector professional organizations that deliver these services. Social movements comprised of potential beneficiaries of these policies are weakly established, and there are very few union members in this area. Mayoral administrations have long awarded service delivery contracts to their preferred professional organizations, but increased political scrutiny is now imposing additional rules that require greater professionalization. One consequence is that the social service policy council is now a technically oriented policy council. Thus, this policy arena is dominated by professional third-sector NGOs that are primarily responsible for service delivery. The establishment of new participatory institutions has not induced many of the CSOs in this field to be actively involved in participatory governance or in electoral politics.

Health care: Council democracy
Social movements in the health care policy arena led the struggle to establish public policy management councils (conselhos) and expand social rights, most notably the right to health care, during the 1980s and 1990s (Jacobi 1989;Sugiyama 2012). Prior to the 1988 Constitution, the Brazilian state provided minimal resources to meet the health care needs of the poor majority, but it did provide health care services for unionized workers. The 1988 Constitution produced two significant changes in the field. First, universal access to health care was a newly gained social right in the 1988 Constitution. Second, the government decentralized the provision of health care; municipalities are in charge of providing these services. Civil society mobilization revolved around a municipal-level state charged with building a health care system (Sistema Único de Saúde or SUS) as well as around council-based democracy.
The sociograph in Figure 3 shows that health care CSO leaders use a variety of activities. The municipal government is responsible for providing health care via SUS, so it is no surprise that these CSO leaders seek out policy and political venues through which they can interact with government officials.
Interestingly, there are two distinct clusters of political activities. The cluster on the left represents extensive "participatory engagement" as these leaders are involved in demonstrations, councils, and direct contact with public health officials.
Demonstrations are at the center of the activity in this policy arena because they are organized by CSO leaders' labor union allies. Demonstrations are a common part of the "repertories of action" that developed within the health care field during 1980s and 1990s. CSO leaders also know that holding public demonstrations is a politically viable option because the government cannot cut off funding, since most funds are transferred by the federal government. This is in sharp contrast to the PB, housing, and social service fields, where discretionary funding makes up the majority of municipal spending.
The cluster on the right side of the sociograph captures activities associated with representative democracy as these leaders are intimately involved with campaigns and mobilizing their CSO base-they hold local meetings, work with political parties, and strive to get out the vote.
To account for the different strategies, we turn first to state formation. During the 1950s and 1960s, Brazil extended health services through public and private insurance to workers in the formal economy. Although these employees represented a minority of the population, there was the development of unions of doctors, nurses, and other medical professionals. Following re-democratization, the Brazilian federal government established the Universal Health Program (SUS) and greatly expanded spending in this area. The state thus has an active role in the provision of health care, which explains why it is a target of political organizing by CSOs.
Second, the health civil society field was among the most active in Belo Horizonte. Demonstrations and protest politics were part of union activities as well as among the CSOs working to improve health care service. Third, the mayoral administrations encourage community organizations to be directly involved. Mayor Celio de Castro (1997-2002 was a medical doctor committed to promoting universal health care. Finally, the participatory institution rules encouraged community-based leaders and union officials to have a more prominent place in decision-making. The development of two distinct clusters of political activities results from two separate trends. First, some CSO leaders are choosing the political contestation and deliberation path, which leads them to work within the councils and to attend public demonstrations. The second path is the more traditional path of community leaders, which is to work on electoral campaigns.

Housing: Full engagement across multiple arenas
The housing movement is the most active social movement in Belo Horizonte and activists participate across a broad range of local state institutions. The housing movement in Belo Horizonte was mobilized in the early 1980s, prior to the formal reestablishment of representative democracy; their activities are in alignment with James Holston's (2008) account of insurgent housing movements in São Paulo. Figure 4 demonstrates that CSO leaders in this policy arena are engaged in the greatest number of activities.
It is revealing that only a small minority of the CSOs held active government contracts, which means that these CSOs do not rely on the state to support their organizations. It is important to note that the two public policy arenas (housing and health care) in which CSO leaders are the least likely to have social service contracts are policy arenas in which CSO leaders are the most likely to use demonstrations. We can infer from these data that the presence of active government contracts reduces the scope of political activities available to activists.
We begin with civil society to explain why the housing field has the densest, most active, and most contentious forms of action. First, during the 1980s and early 1990s, housing-based social movements were extremely active, engaging in multiple land occupations across the city. These contentious land occupations (generally referred to as illegal invasions by property owners) forged a strong identity and bonds of solidarity among activists working in the housing field (Alexander 2006;Holston 2008). Thus, social movement activity prior to the establishment of the 1988 Constitution continues to impact CSO leaders' ability to sustain their Second, movement leaders work within multiple municipal agencies, two different participatory venues, the municipal legislature, and federal development banks to build support as they pursue their policy agenda. The policy and political complexity and large numbers of actors mean that CSO leaders have to position themselves within multiple venues. The density of their political activities is a response to the organization of the state. There are more public agencies involved in housing than in the other three areas, which leads CSOs to actively engage public officials at multiple points.
Third, mayoral administrations generally supported the housing movements, but they lacked the funding to build the necessary infrastructure. Housing movements and several mayoral administrations forged an alliance that led activists to mobilize political support through public demonstrations and testimony to the municipal legislature. Housing movement officials used their political activities to maintain pressure on mayoral administrations to better ensure continued support. Finally, the policy council rules encouraged the development of a partnership between the government and CSO leaders. During the periods of study (2003)(2004)(2009)(2010), there were political alliances between elected governments and CSO leaders elected to the council, which turned the council into a politicized body in which they promoted their mutual interests.
In sum, the housing public policy arena CSO leaders are active participants in multiple spaces as they respond to a fragmented state policy-making process. They must be everywhere because there are multiple agencies and elected branches of government that hold some type of relevant responsibility. Housing CSO leaders are most active in exercising political rights formally established in the 1988 Constitution as a way to access social and civil rights guaranteed by that constitution. The expansion of the number of places where housing activists could place their claims and the absence of specific budgetary resources dedicated to housing induces CSO leaders to be active across the state. Of course, this places a major burden on the housing movements because they juggle multiple new roles: as social movements, policy-makers, and public interest groups. Not all groups have the skills and knowledge to carry out this complex set of tasks.

Conclusion
CSO leaders in new democratic settings now employ a great diversity of political strategies in pursuit of their organizations' interests. Instead of a dichotomous relationship with state officials (protest or co-optation), CSO leaders utilize a wide range of strategies. The political network approach utilized in this article helps to illuminate the clustering of political activities by CSOs within specific policy arenas as well as the differences across policy arenas. Although this article focused on four policy arenas in one Brazilian city, the analytical framework is useful to understand democratic politics across Brazil and Latin America because it demonstrates how the long-term development of the state and civil society independently and interactively affects the strategies employed by CSOs in new democratic environments. Based on this analytical framework, we would expect that the long-term development of the state and civil society, as well as the specific institutional rules associated with democratization and participatory institutions, induce CSO leaders to behave in specific ways.
This article demonstrates that when a capable state and bureaucracy were created prior to democratization, CSOs in the democratic era were more likely to work within civil society as well as with public officials who control scarce resources. The interests of public officials and CSO leaders are complementary, inducing them to work together. Public officials seek out the mobilization strength of CSOs while CSO leaders seek out public goods and policy changes.
The presence of a strong state but limited CSO activity (for example, in the military and corrections sectors) leads to policy arenas that are the most immune to democratic pressures and the most resistant to policy and political reform. This combination of strong state and weak civic engagement helps to account for some of the most abusive and corrupt practices that persist across the region (e.g., military police violence and pattern of extrajudicial executions; horrific prison conditions; limited civilian control over internal military affairs).
Conversely, when there was a weak development of the state but an active, mobilized civil society, leaders were the most likely to pursue their interests in a wide range of formal and informal spaces. CSOs engage in contentious politics, lobbying, participatory policy-making, election campaigning, and so on. Civil society organizations are unable to mobilize a state that has limited capacity (Abers and Keck 2009). Rather, CSOs must engage in the politically contested work of building the necessary political support to invest in building state capacity.
When there is a weak state, and civil society is demobilized, professional NGOs, many of which hold service delivery contracts, gain prominence in the policy arenas because they act as both "state" and "civil society" to deliver services. Thus, when long-term state development is absent, and when civil society capacity is low, there is a greater likelihood that well-organized, professionalized groups will dominate the policy sector. We would expect professional lobbies or third-sector NGOs to have a much more active role in civil society formation.
These lessons are applicable beyond Belo Horizonte in two ways. Analytically, the key insight is that we must better incorporate state formation and state capacity in basic analyses of democratic politics. The study of participatory institutions, for example, often downplays the role of the state, but this analysis has demonstrated the vital importance of the presence of the state in producing actors (unions) as well as being the target of CSOs' strategies. The second important insight is that researchers need to be much more mindful of the variation across policy arenas, especially in decentralized and democratic settings. The analysis in this article demonstrates how basic differences in state formation and the configuration of civil society induce CSO activists to engage in very different behaviors. This analytical framework can be applied from the US-Mexico border region to environmental conflicts in the Amazon and Andes to better understand the strategies employed by CSOs as they pursue their organizations' interests.
Science Review, Comparative Politics, Comparative Political Studies, World Development, Polity, and Latin American Politics and Society.