Civil society and contested elections in electoral autocracies: Dissent and caution in Uganda’s 2016 elections

ABSTRACT Civil society organizations (CSOs) respond differently to challenges from autocratic governments and opposition parties around the regulation and content of elections. Building on research on autocratic state regulation of CSOs, this article contributes a sharper focus on their horizontal relations to increase understanding of their diverse strategies of engagement. The article argues that even in contexts dominated by a heavy-handed state, relations between CSOs are especially important during elections, when they are most mobilized and motivated to build coalitions. The results of a study into how non-governmental associations (NGOs) and faith-based organizations (FBOs) engaged with electoral reforms around the Uganda 2016 general elections show how relations of cooperation, competition, or conflict between civil society organizations modify the effects of state regulation by adding to their incentives and capacities. The findings help explain the many ways CSOs engage with each other, the population, the state, and the opposition, including their submission to and protests against the autocratic order. Horizontal relations of conflict and competition among the FBOs prevented effective coalitions and facilitated submissive politics during Uganda’s 2016 elections. Relations among governance NGOs, however, were more cooperative and generated protest alliances, even though those alliances proved difficult to sustain.

engagement with autocratic elections? This article analyses the sources of such engagement through examining diverse civil society mobilizations around electoral reform. Setting out from documented civil society diversity in electoral autocracies, the study draws on insights about the significant role of selective state regulation on associational life and contribute to the literature by proposing a sharper focus on how relations among civil society organizations modify their incentives and capacities.
A growing literature demonstrates pronounced differentiation between the capacities and orientations of civil society organizations (CSOs) in electoral autocracies (Giersdorf & Croissant, 2011;Lewis, 2013;Toepler et al., 2020). State regulation is obvious and important in such contexts, shaping CSOs' scope for action. Autocratic governments typically repress contestation, co-opt potential critics, and cultivate consent (Haggard & Kaufman, 2016;Schedler, 2013), thus conditioning CSOs' incentives and capacities. While state regulation is important in shaping the patterns of civil society engagement, it does not, however, determine them. Research has shown the relative room for manoeuvre open to CSOs in electoral autocracies (Falkenhain, 2020) and how they express their political agency in different ways, from democratic dissent against the odds (Bunce & Wolchik, 2011;Hudáková, 2021) to mobilization for autocracy (Lorch, 2017;Sombatpoonsiri, 2020).
The significance of agency is further enhanced by electoral politics. Autocratic elections constitute focal points for struggles over government and regime legitimacy and activate mobilization and contention, at the levels of both party politics and civil society (Beaulieu, 2014;Bunce & Wolchik, 2011;Schedler, 2013;Trejo, 2014). But electoral mobilization is delicate. Elections create problems with coordination and collective action and often prompt political parties and CSOs to craft coalitions to address these problems and to dismantle the coalitions of other contenders. These observations suggest a need to carefully examine the dynamics of horizontal relations among CSO, not only vertical relations between governments and CSOs, to grasp the diversity of civil society engagement with autocratic elections.
Despite this, research on civil society in electoral autocracies has devoted relatively limited theoretical attention and systematic analysis to the role of relations between different CSOs. This article contributes to the literature by exploring how the combination of vertical and horizontal relations shape CSOs' incentives and capacities to engage with elections and influence patterns of support for, submission to, and resistance against political autocracy. Building on research on how governments selectively apply mechanisms of repression and co-optation to regulate CSOs and coalitions, the article analyses how their relations of cooperation, competition, or conflict contribute to shaping their interests and to modify the effects of government regulation.
The article demonstrates how, because of intensified mobilization and coalition building around elections, relations between CSOs matter even under heavy-handed state domination. It does so through a study of CSOs' engagement with regulations around the Uganda 2016 general elections. President Museveni of the National Resistance Movement (NRM) has been in power since 1986 and has sustained his position through a series of controversial elections, all disputed by the political opposition and questioned by independent observers. The firm presidential control of electoral regulation is a cornerstone of autocratic rule in Uganda and a deeply contentious issue. For many years, the political opposition has demanded electoral reforms towards a more level playing field, including an independent electoral commission and the removal of the military from the electoral process, but the government has steadfastly refused. This article analyses how important non-governmental organizations (NGOs), faithbased organizations (FBOs), and their coalitions in key sectors of Ugandan civil society contested and responded to electoral regulation and demands for reform around the 2016 elections in terms of supporting, submitting to, or protesting the autocratic order.

The Political Engagement of Civil Society in Electoral Autocracies
An emerging field of research demonstrates that civil societythe realm and ensemble of associational lifein electoral autocracies is politically differentiated. As opposed to earlier expectations about its democratic and democratizing propensity (Diamond, 1994), civil society has been shown to encompass a wide range of actors with diverse orientations and practices in relation to autocratic governments, including mobilized or tacit dissent, acquiescence, loyalty, and active support (Giersdorf & Croissant, 2011;Härdig, 2015;LeVan, 2011;Lewis, 2013;Lorch, 2020;Mietzner, 2020;Toepler et al., 2020;Ziegler, 2010).
Just as CSOs' political activities vary within each context, their composition and effects differ among electoral autocracies. Some electoral authoritarian systems are characterized by hegemonic rule, where, overall, deferential and subjugated CSOs have a mere nominal presence and may even fulfil a functional role in anchoring autocracy (Durac, 2012;Wiktorowicz, 2000). In more competitive settings, CSOs enjoy more leeway, and assertive groups and movements have sometimes played important parts in crafting successful pro-democratic collective action (Bunce & Wolchik, 2011;Hudáková, 2021;Way, 2014).
Overall, though, opportunities for critical civil society engagement are constrained in electoral autocracies, and their scope for dissent is typically strongly conditioned by state regulation (Buyse, 2018;Gilbert & Mohseni, 2018). To secure domination, authoritarian state rulers rely on coercion and co-optation and draw on ideological resources to create legitimacy. Governments combine their resources of domination selectively in relation to different actors in society and structure the scope and content of associational life by empowering some groups and controlling or marginalizing others. They use law, policy, administration, technology, monetary resources, and violence to create, differentiate, and block opportunities, incentives, and capacities for collective action (Gandhi & Przeworski, 2007;Gerschewski, 2013;Haggard & Kaufman, 2016;Lewis, 2013;Reny, 2021;Schedler, 2013;Yabanci, 2019).
This article adds to this line of inquiry. Drawing on the Gramscian tradition, it conceptualizes civil society as a terrain of contestation. In any setting, civil society contains a diverse set of interest groups and movements based on historically rooted and contemporary social distinctions such as class, gender, ethnicity, and religion. These groups can be assumed to promote diverse interests, setting off relations of cooperation, competition, and conflict amongst them. Their divergent positions do not merely constitute parallel paths of engagement, they also stem from or cause the many lines of conflict, competition, and collaboration that permeate civil society (Bob, 2011;Foley & Edwards, 1996;Kopecký & Mudde, 2003).
The importance of such relations and patterns of social and political agency is augmented by elections. While one perspective holds that elections and legislatures in electoral autocracies serve largely as government control mechanisms (Gandhi & Przeworski, 2007), much of the literature underscores the contestation generated by autocratic elections and how it activates both political parties and CSOs (Beaulieu, 2014;Brancati, 2016;Bunce & Wolchik, 2011;Haggard & Kaufman, 2016;Schedler, 2013;Trejo, 2014;Tucker, 2007). Autocratic governments control electoral processes and have the motives, means, and opportunities to unlevel the playing field in a vast number of ways. Despite this, elections create uncertainty: while government subversion of genuine competition makes continued authoritarian control likely, competition itself and challenges to its subversion continue to make change possible (Schedler, 2013). Autocratic elections are manipulated but contested arenas of representation that constitute focal points for mobilization. They are critical events, delimited in scope and compressed in time. As Schedler puts it (2013, p. 149), they provide democratic grievances and master frames of protests, and energize both political parties and CSOs.
The dynamics of autocratic elections often revolve around struggles to (de)mobilize coalitions. Uneven access to power and resources makes it necessary for opposition forces in political parties and among CSOs to embark on extensive and intensive mobilization around elections. Given the autocratic circumstances, this comes with challenges around coordination and collective action (Haggard & Kaufman, 2016;Schedler, 2013). To overcome these, opposition activists often seek to create coalitions involving political parties and CSOs (Bunce & Wolchik, 2011;Trejo, 2014;Tucker, 2007). When successful, coalitions can increase incentives and improve capacities for people and organizations to oppose the autocratic power by expanding resources, multiplying opportunities for engagement, and creating greater access to channels and arenas of action or acting as 'force multipliers' (Haggard & Kaufman, 2016, p. 69). Successful coalitions can limit the effects of restrictive state regulation.
Whether an opposition attempt at coalition results in a sustained alliance with cohesion and capacity or in fragmentation and deactivation depends to a large extent on the force of government efforts towards demobilization through incorporation or fragmentation. Extensive and forceful autocratic state regulation of CSOs shifts the balance of influence and power among such groups, affecting their patterns of engagement. Groups that support or accept the government benefit and prosper, while coalitions of government critics are less likely to emerge or endure.
But coalition capacity also depends upon relations among potential coalition partners among CSOs and political parties. Government efforts to incorporate and fragment CSOs can be overcome by temporary alliances during critical junctures. However, if such relations are marked by rivalry or strongly diverging core interests, identities, and values, even short-term pragmatic coalitions may be complicated and difficult. Competition and conflict among groups can often promote their separate engagements with the state, including supportive protests, submissive co-optation, or signals of readiness to provide legitimation in exchange for influence in their spheres of interest and against groups seen as competitors.
Existing research thus establishes that CSOs' political agency in electoral autocracies is differentiated; their scope for critical engagement is constrained, but not closed, and strongly, but not fully, conditioned by state regulation; and that electoral politics underlines the significance of diverseincluding autocraticcivil society engagement through intensified (de)mobilization and coalition-building efforts. Relatively limited attention, however, has been paid to how relations between different civil society actors shape their prospects for coalitions and the ways in which groups and coalitions relate to autocratic elections. This article contributes to the literature with a study on how the combination of state regulation and relations of cooperation, competition, and conflict between CSOs influence their electoral engagement, alone and in coalitions, and shape various patterns of mobilization around electoral reform in terms of their support for, submission to, and resistance against autocracy.

Methodological Note
To explore the argument outlined above, the study examines how a range of actors, drawn from important sectors of associational life in Uganda (governance NGOs and FBOs) engaged with electoral regulation during the general elections of 2016. Among the relatively large set of electoral autocracies, Uganda is closer to the hegemonic than the competitive end of the spectrum, and increasingly so over time, but it is neither extreme nor exceptional. The 2016 election was selected for analysis as civil society engagement was more visible than before, allowing for an in-depth exploration of the detailed dynamics the study aims to capture.
NGOs arose as a key sector of Ugandan civil society in the 1990s. The country witnessed a rapid growth of registered NGOs, from about 200 in 1986 to 12,500 by the end of 2013 (Uganda National NGO Forum, 2015). Most of these worked mainly nonpolitically to deliver social welfare or social development activities. Over time, tensions emerged within the Ugandan NGO sector over how to relate to politics. The vast majority of so-called development or service-delivery NGOs refrained from engaging with controversial matters, while a small number of what came to be referred to as 'governance NGOs' decided to promote political advocacy. This article focuses on those NGOs. To capture their activities, the study examined civil society coalitions based on such groups and their work around elections using the example of the Free and Fair Elections (FFE) campaign.
Faith-based, especially Christian, organizations have long been a crucial segment of Ugandan civil society. Churches offer key mechanisms for reaching broad swathes of the population. Religious leaders are expected to address political matters, and they have done so, albeit inconsistently and in different directions. Given that Catholic versus Protestant affiliation constituted the most recognizable basis for political identification around and immediately after independence, this study examined Uganda's two main religious umbrella organizations: the Uganda Joint Christian Council (UJCC) and the Inter-Religious Council of Uganda (IRCU). The UJCC, founded in 1964, is limited to the Catholic, Anglican, and Orthodox churches, while the IRCU, in existence since 2001, is open to all faith-based institutions in the country.
The article draws on fieldwork carried out in 2018, a systematic analysis of news sources, and a thorough examination of the secondary literature. The fieldwork data includes 20 semi-structured interviews with relevant actors, including both participants and analysts. The interviewees were carefully selected representatives of the different organizations, academics, other political analysts, and journalists expected to provide deep insights. The strategic selection of groups and individuals was facilitated by my over two decades' experience of conducting research on related matters in Uganda. The interviews normally lasted about an hour and centred on key aspects of electoral regulation and civil society engagement. They were conducted in English, but an interpreter was always present to ensure that conversations could switch to Luganda whenever respondents wished. For safety and ethical reasons, the interviewees remain anonymous. Media material (from a systematic search in Factiva, a global news database, for reports beginning in 2005) and written documentation related to and authored by the organizations contributed to the analysis of the organizations' responses.
State Power, Civil Society, and Elections in Uganda, 1986Uganda, -2011 This section provides an overview of the Ugandan political context, outlines the main features of different forms of regulation of civil society, and concludes with a subsection that specifically addresses the 2006 and 2011 elections. When the NRM took power in 1986, Uganda had suffered a decade and a half of military rule, dictatorship, and civil war. In attempting to extend and stabilize its political base over its first decade in power, the new government incorporated political leaders of various affiliations, formal and informal representatives of sub-national regions, and CSOs. Among government supporters, the imposed restrictions on political pluralism and its replacement by the so-called no-party system was broadly accepted as a temporary measure necessary to achieve cohesion and stability (Carbone, 2008;Sjögren, 2013;Tripp, 2010).
During these years, the structure of civil society was characterized by revival and recomposition, and patterns of state-civil society relations were marked by a combination of legitimization and co-optation (De Coninck, 2004;Dicklitch, 1998;Oloka-Onyango & Barya, 1997). Organizations with a long history, such as trade unions, were reactivated, and new types of groups, such as donor-dependent development NGOs, emerged; the latter became the numerically dominant segment of associational life (De Coninck, 2004;Dicklitch, 1998). Many groups embraced the policies and state-building efforts of the NRM and some also benefitted from it. The women's and labour movements were given apportioned parliamentary representation, and most faith-based institutions steered close to the government (Kassimir, 1998;Okuku, 2003;Ward, 1995).
The first decade of NRM rule culminated successfully for the government in the promulgation of the new constitution in 1995 and the 1996 general elections, and the period was marked by reasonably good relations between and among the state and CSOs. In the late 1990s, however, that consensus was challenged by growing discontent with the government and accusations of increasing corruption and repression (Sjögren, 2013;Tripp, 2010). To critics of the no-party system, the 2001 elections showed that this mode of politics had outlived its role, and calls were raised for political pluralism. Although multiparty elections were reinstated in 2005, the presidential two-term limit was abolished at the same time (Makara et al., 2009).
The pattern of increased contestation and conflict between the government and various opposition groups continued after the introduction of political pluralism. As the NRM's incorporation project of consensus and legitimation based on economic performance and security was increasingly challenged, the emphasis in the government's general strategies for political domination shifted from cooperation and legitimation to co-optation and coercion (Carbone, 2008;Khisa, 2019;Reuss & Titeca, 2017;Sjögren, 2018;Tripp, 2010).
The latter tendency also increasingly characterized the regulation of associational life. During the 1990s, the handling of firmly incorporated groups, such as the women's movement and trade unions, shifted from cooperation to co-optation and these groups' began to lose their voices and influence (Barya, 2001;Goetz, 2002;Muriaas & Wang, 2012). The gradual increase in coercive domination (Dicklitch & Lwanga, 2003;Tripp, 2004) contributed to a more cautious approach by many CSOs before the 2006 elections (Kjaer & Olum, 2008). New and stricter legislation to regulate NGOs, debated for years, finally became law in 2006. The Non-Governmental Organizations Registration (Amendment) Act of 2006 contained clauses that stipulated supervision of NGO by security organs (Nassali, 2017, p. 283) and was followed up by the NGO Registration Regulations in 2009 and the National NGO Policy of 2010.
Religious institutions in Uganda are also formally regulated in several ways, sometimes differently from NGOs. The more established faith-based institutions, such as the Catholic, Anglican, and Orthodox churches and their organizational expressions are registered as legal entities under the Trustee Incorporation Act, as are the Uganda Muslim Supreme Council and various other religious umbrella organizations. This gives them more leeway and independence than NGOs as well as tax exemptions. In practice, the government has deployed a combination of co-optive and coercive strategies to divide-and-rule 1 to contain the behaviour of faith-based institutions and to tap into their vast networks of followers (Downie, 2015).
On balance, the legal and administrative regulation of CSOs became stricter, and the main targets were the relatively small but growing number of critical NGOs and their potential recruits (Human Rights Watch, 2012;Nassali, 2017, p. 284). The main faiths were not affected by these legal changes. The intensified dissent and diversification of political opinions and engagement in Ugandan society from the late 1990s found expression in civil society, too. As political contradictions came to the fore and cleavages were more clearly articulated, relations between and differences among CSOs became more distinct. Divisions among religious organizations became more evident (Alava & Ssentongo, 2016;Downie, 2015;Isiko, 2019), and tensions emerged between a politically vocal minority of anti-corruption and governance NGOs on the one hand and the vast majority of development and service-delivery NGOs on the other around political engagement and perceived complicity. This created a structural divide among NGOs that has continued to guide patterns of both self-understanding and (non) political agency (Dicklitch & Lwanga, 2003;Kjaer & Olum, 2008).
Electoral Regulation, Demands for Reforms andCivil Society Organizations, 2005-2011 Following the return of multiparty politics in 2005, Uganda is in many ways a typical electoral autocracy, albeit with increasingly repressive characteristics. Democratic institutions and processes are formally in place but not respected in practice. Multiparty elections are held but widely considered neither free nor fair. Elections in Uganda are administered by the Electoral Commission (EC), whose commissioners are appointed by the president and do not enjoy security of tenure. Executive control over the legislature and the judiciary is pronounced, the political opposition is harassed, and the media and associational life are frequently (though selectively) subject to repressive regulation (Tripp, 2010). Many of these features were strikingly evident around the general elections of 2001 (nominally conducted under the no-party system), 2006, and 2011, all of which were, to various degrees, marred by glaring irregularities and state-sanctioned violence against the opposition (Sjögren, 2018). These and other violations of the constitution and the law were acknowledged in the Supreme Court rulings on the opposition's petitions against the presidential elections in 2001 and 2006; despite this, on both occasions a divided Supreme Court ruled against the petitioners to uphold the declared victory of Yoweri Museveni (Murison, 2013).
Before the 2006 elections the opposition had voiced demands for electoral reforms. The demands were based on the experience of the 2001 elections and centred on the creation of an independent electoral commission and the reconstitution of the military and security forces and their removal from the electoral process (Mwanguhya, 2005). 2 The government dismissed all such claims. The opposition repeated its demands before the 2011 elections, adding to them the reinstatement of presidential term limits (Makara, 2014, p. 117); a number of CSOs and networks also stressed the need to transform electoral regulation. Through the efforts of the opposition umbrella organization, Inter-Party Cooperation, some administrative reforms were passed by parliament, but the structure and operation of the EC was left intact (Makara, 2014, pp. 117-118). The elections turned out to be less violent than the preceding ones; instead, the contest was characterized by the vast amount of money and state resources used by the incumbent to distort the playing field (Gibb, 2012, p. 460;Helle & Rakner, 2014;Izama & Wilkerson, 2011, pp. 70-71;Sjögren, 2018). As in 2001 and 2006, President Museveni was declared the winner. The opposition disputed the process and outcome, but this time did not go to court to challenge the results.

Civil Society Organizations and Electoral Regulation: The 2016 General Elections
The 2016 general elections repeated the pattern. The opposition reiterated their demands for electoral reforms, which were rejected by the government (Abrahamsen & Bareebe, 2016). President Museveni was once more declared the winner in a contest whose process and outcome the opposition did not accept. 3 Yet again an opposition candidate, this time former Prime Minister Amama Mbabazi, went to court with a petition; yet again, the petition was to no avail. And once more a general election was followed by renewed calls for electoral reforms, centred on the same core issues as before: an independent electoral commission, stricter regulation of public and other resources during campaigns, and the removal of the military from the electoral process. The engagement of CSOs, however, was more visible than during previous elections. This section analyses how selective state regulation and relations among CSOs shaped mobilization around electoral reform in terms of support for, submission to, and resistance against autocracy.

Governance NGOs: Intensified Engagement by a Core Constituency
During previous election campaigns opposition parties had led in demanding reforms. While some CSOs had supported such calls, they did so as followers rather than leaders. Ahead of 2016, however, governance NGOs took a more active role, expressed through several campaigns as early as directly after the 2011 elections. One initiative addressed the issue of presidential term limits, while another, the Citizen's Electoral Reform Agenda, took a broader perspective on electoral regulation but was stopped in its tracks by security forces (Nassali, 2017, p. 287). In 2013 renewed efforts by a coalition of CSOs and political parties led to the formal launch the FFE campaign in early 2014. Among the groups and umbrella organizations spearheading FFE were the Uganda National NGO Forum, the Citizens' Coalition for Electoral Democracy in Uganda, the Great Lakes Institute for Strategic Studies, and later the Foundation for Human Rights Initiative.
As assessed by both activists themselves and independent observers, going into 2016 governance NGOs' political engagement was different. It was stronger and took more daring positions; it was more intense and offered new leadership; and it was broader and went beyond lobbying and press conferences to engage in community mobilization and work openly with political parties (interview 1, political journalist; interview 2, NGO leader; interview 9, NGO leader; Nassali, 2017). The FFE campaign traversed the country. It first held rallies and then returned for town-hall meetings during which the organizers collected views on electoral regulation from citizens. In November 2014, the campaign culminated in a national delegates' conference in Kampala, the outcome of which was the document Citizens' Compact for Free and Fair Elections, presented to the Speaker of parliament. The document was mentioned during parliamentary debates in 2015, but none of its recommendations were ever accepted (Nassali, 2017).
After the national delegates' conference, the energy of the FFE campaign faded and some of its bolder ideas were abandoned. There were plans to return to local communities with the Citizens' Compact for further engagement and to form the Democratic Front, a civil society counterpart to the opposition electoral coalition the Democratic Alliance. None of this materialized. Instead, organizations went their separate ways. Some returned to the specific work stipulated by their mandate and funding, while others worked on election-related matters but in smaller constellations, for example in the Citizen Election Observer Network (CEON) and in the Topowa campaign to promote voter registration (interview 2, NGO leader).
How did state regulation and relations among NGOs shape their incentives and capacities for engagement? The pattern of more restrictive state regulation of associational life discernible in the period leading up to 2011 continued over the next five years. In the months after the 2011 elections, the opposition organized mass protests against what was described as general misrule (Golooba-Mutebi & Sjögren, 2017). The government's harsh legal response to these and other protests and to the general discontent building in urban centres was the Public Order Management Act (POMA), 2013. The POMA regulates freedom to assemble and demonstrate and gives the Inspector General of Police a wide-ranging mandate to prevent or dissolve public meetings. While not targeting political parties or CSOs as such, the law imposes restrictions on public expressions of dissent, sends distinct messages about the benefits of keeping a low profile, and has been deployed against government critics. 4 Another expression of this pattern was the new Non-Governmental Organizations Act of 2016, which repealed earlier legislation and regulations. The Act created a new NGO Bureau with extended surveillance powers and introduced a new set of obligations for NGOs in relation to both the NGO Bureau and district and sub-county monitoring committees. Again, although the letter of the law applied equally, its spirit and implementation were targeted. The revised NGO legislation had been hinted at for many years, and its introduction in parliament in 2015, less than a year before the 2016 elections, was seen by many as a response to CSOs' activism around electoral reforms through the FFE campaign (Nassali, 2017, p. 284). 5 Service-delivery development NGOs were not targeted.
These changes in legal and administrative state regulation interplayed with relations among NGOs and reshaped their incentives and capacity for engagement. As early as the late 1990s, distinct divisions had emerged among NGOs about whether to engage in political activism, with the majority taking a cautious line. The government's gradual shift from the early 2000s towards coercion of critical organizations and co-optation of others reinforced this tendency, and the acceleration of such oppressive regulation further polarized NGO patterns of engagement. While many organizations withdrew, those that did not become more committed and became a cluster of galvanized and politicized NGOs operating within an increasingly restrictive formal framework. Greater pressure and higher stakes created a stronger need among activists to unite, resulting in alliances amongst themselves and closer engagement with politicians (interview 9, NGO leader; interview 11, NGO leader). One aspect of such horizontal collaboration was the tendency to organize in unregistered coalitions to escape the restrictions of formal regulation: most campaign vehicles were informal in that sense and could host and allow for separate organizations' intensified resistance (interview 3, NGO leader).
Informal networks, however, are difficult to sustain and under pressure tend to generate intense but episodic activism. They require oxygen in the form of electoral or otherwise intense politics, and while they benefit from engagement with political parties, such interactions are complicated as parties and movements often have different and sometimes competing aims and modes of operation. The new unity and political engagement among activist NGOs eventually subsided. Reasons why the Citizens' Compact was not taken back to citizens and why the TDF did not take off varied. Some organizations withdrew into normal NGO work and political silence after having been warned off by the government (Interview 4, NGO leader; interview 8, scholar, Makerere University). Funding became more difficult. Both funders and some organizations felt that 'when consultations were over in 2014, the project was over.' (Interview 11,NGO leader). Thus, horizontal relations of close collaboration were undercut not only by state intimidation, but also by the organizational logics of mandates and modes of operation. The limits to collaborative resistance also had structural roots: the politically radical parts of civil society were composed of NGOs who lack the financial and organizational resources required for mass mobilization.

Faith-based Organizations: Peace and Caution
Faith-based organizations' engagement with electoral reforms around the 2016 elections was more cautious and diverse than that of governance NGOs. Broadly speaking, religious institutions in Uganda engage through three main mechanisms: by denomination or more commonly through ecumenical umbrella organizations (the UJCC and the IRCU, analysed below) or through views expressed by individual members of the clergy. A small number of clergy members participated in or supported the FFE campaign as individuals, but the umbrella organizations kept their distance.
While both the UJCC and IRCU are criticized by governance NGOs and the political opposition for being non-confrontational on political matters, the IRCU is seen as the more cautious of the two, while the UJCC occasionally speaks out (interview 1, political journalist; interview 9, NGO leader). The organization has a history of participating in election work: in 2006 and 2011 it observed the election under the auspices of DEMGROUP. After the 2011 elections, however, it left DEMGROUP, which subsequently disbanded partly because of internal disagreements. The UJCC had voiced concern about what it regarded as DEMGROUP's pro-opposition partisanship (interview 6, UJCC representative; interview 7, DEMGROUP official), but when CEON was created as the new civil society vehicle to observe the 2016 election, the UJCC decided not to take part and observed the 2016 elections on its own (interview 6, UJCC representative).
Both the UJCC and the IRCU followed the general trend in Uganda in 2013 and 2014 by joining widespread calls for electoral reforms (Kasozi, 2014;Namutembi & Ssenyonga, 2013). The government, however, put pressure on the IRCU (Kaaya, 2014) andin a distinct example of the interplay between state regulation and intra-civil society relations the IRCU soon decided to withdraw from closer engagement with the FFE (Nassali, 2017, p. 288): 'We contacted the FFE and told them they made a mistake. Civil society needs to be independent' (interview 5, IRCU representative). Its main engagement with the elections was to organize a televised debate between all presidential candidates (Alava & Ssentongo, 2016, p. 683). While the UJCC remained active, it restricted its engagement in the elections by a self-imposed neutrality. For instance, although the UJCC preliminary election report contained critical observations, these were undercut by individual statements to the contrary and situated within an overall narrative of peace (Alava & Ssentongo, 2016, pp. 682-683) elaborated at the expense of demands for electoral integrity. Academic analysts and activists have assessed the contribution of FBOs to electoral reform as a self-imposed balancing act, characterized by sending mixed messages, playing safe on controversial issues, and preaching peace (Alava & Ssentongo, 2016;interview 2, NGO leader;interview 3, NGO leader;interview 10, NGO leader).
A historical perspective is necessary to explain how selective state regulation and internal relations mutually reinforced the incentives and capacities of various FBOs to engage with elections. Faith-based civil society in Uganda is marked by a long trajectory of politicized religious factionalism. The 1900 Buganda Agreement established the role and status of Anglicans, Catholics, and Muslims in that order of influence, a hierarchy that was consolidated during the decades that followed. The Anglican Church of Uganda evolved into the church of the establishment (Ward, 1995), while the Catholic Church remained outside the circles of political power (Downie, 2015;Kassimir, 1998;Okuku, 2003). Party political identification before and after independence was perceived as closely tied to religious affiliation (although it actually overlapped and cross-cut ethnoregional identities): The Uganda People's Congress was regarded as the political vehicle of Protestants and the Democratic Party seen to promote Catholic interests. These links were for obvious reasons weakened by the no-party system, and NRM rule has encouraged the expansion of Pentecostal churches, with links to the first family and widely regarded as loyal to the government (Alava & Ssentongo, 2016).
While ties between political parties and specific faiths have been severed during NRM rule, the status distinctions between religious institutions are still relevant to how they relate to the government and to each other under umbrella organizations. The role of the Church of Uganda as the establishment faith with close ties to the government, (limited) financial resources of its own, and contentions with evangelical rivals has reproduced incentives and capacities that make it tend to steer close to power and keep a timid and government-loyal political profile (Downie, 2015;Okuku, 2003).
The Catholic Church has evolved along a different trajectory. Historically, it has been excluded from political power, with resulting grievances and a cemented outsider identity. Compared with the Church of Uganda, however, it possesses much greater material resources and enjoys stronger external links (Downie, 2015). Theoretically it thus has stronger incentives and a greater capacity to speak out on social and political matters. While it has done so somewhat more than the Anglican Churchfor example during the war in Northern Uganda in the 1990s and the early 2000sits track record is mixed. Individual bishops have often criticised the government, but they have not been able to rely on the consistent support of their church as the government has deployed a strategy of co-optation of less outspoken leaders in order to isolate the critical voices (Downie, 2015;interview 8, academic;interview 11, NGO leader).
The trajectories of the individual faiths have obvious consequences for religious umbrella organizations. The relations between them have long been characterized by competition and stratification, and the government has skilfully deployed regulatory strategies that shift among and within the faiths: it has kept one church close and the other at a distance, and it has co-opted some church leaders into loyalty while intimidating and isolating others (interview 2, NGO leader; interview 8, academic). While interfaith bodies can potentially create safety and strength in numbers, Ugandan faithbased umbrella organizations are no stronger than their weakest links. The features discussed here impose inherent restrictions on their incentives and capacities, and their common positions tend very often to be watered-down compromises. The IRCU's frequently observed caution can partly be explained by the effects of its gravitating towards a modest common denominator among its broad membership. The clergy issues sporadic critical pronouncements, but these are typically voiced by individuals and usually modified or undercut in following statements from the organization as a whole (interview 2, NGO leader; interview 3, NGO leader; interview 10, NGO leader).

Comparative Analysis
This study illustrates the diversity of engagement with autocratic electoral regulation among CSOs within and across sectors and demonstrates the relevance of analysing the combination of state regulation and horizontal relations among CSOs. The NGO sector at large has long been characterized by a structural divide regarding political engagement. The overwhelming majority of NGOs steers away from contentious issues and positions, leaving a committed minority of outspoken organizations to rely upon each other. Governance NGOs form the most cohesive and critically vocal group of Ugandan CSOs and the category that has most clearly articulated its resistance to autocracy. These vocal but few groups have by necessity developed horizontal relations of cooperation and been galvanized into relative cohesion, since their outspokenness has made them the target of repressive state regulation, necessitating their more closelyknit alliances and coalitions to survive.
FBOs, on the other hand, operate within the confines of separation structured by historically anchored status hierarchies, competitive relations, and differential treatment by governments. This combination of selective state regulation and frictional horizontal relations place restrictions on their scope for effective coalition building, both amongst themselves and in relation to others. Occasionally particular individuals or groups transcend their inherited restrictive conditions, both to cooperate and to challenge autocratic rule, but as organized entities FBOs mostly tread cautiously and submit to the existing order.
The findings are even clearer when different types of groups are compared. Beginning with faith-based organizations, the study has focused on the Anglican and Catholic Churches; the observed tendencies towards cautious and diverse engagement are even more pronounced when evangelical churches and Muslim organizations are taken into consideration as some of these are vocal government supporters (Alava & Ssentongo, 2016;Downie, 2015). This wider context underlines how the perceived silence and inconsistency on political matters of many religious leaders and institutions has shaped horizontal relations between governance NGOs and FBOs by creating mistrust among NGO activists and making alliances between organizations from the two categories difficult to sustain (Downie, 2015, p. 59;Kaaya, 2014; interview 2, NGO leader; interview 3, NGO leader; interview 10, NGO leader).
Other sectors further illustrate the diversity of civil society political orientation. Service-delivery NGOs have consistently exhibited political caution. Most women's organizations kept a low profile in relation to the FFE campaign to avoid becoming drawn into partisan politics, although some individual leaders engaged more strongly. The National Organization of Trade Unions also distanced itself from the reform campaign, as one proposal was to abolish special interest group representation in parliament, including workers' representatives appointed by the trade union (Nassali, 2017, p. 288). Business associations such as the Uganda Manufacturers' Association have always been silent on political matters (interview 2, NGO leader; interview 10, NGO leader). Buganda Kingdom, a key non-state interest group, has been uneven in its electoral engagement over time, shifting from strongly expressed demands and contacts with the opposition by some of its representatives in 2011 to studious silence in 2016 (Sjögren, 2020).

Conclusions
Over the last decade the main global tendency of political regime change has been towards autocracy. While this has emanated mainly from attempts by governments and state rulers to undermine democratic procedures and practices, autocratizing tendencies have also been visible at the level of associational life. Interest groups and social movements express different political persuasions that range from defending democratic institutions, processes, and norms (Bunce & Wolchik, 2011;Hudáková, 2021) to resisting them (Lorch, 2017), and they occasionally contest one another (Mietzner, 2020;Sombatpoonsiri, 2020). This underscores the significance of examining relations among CSOs. This study examined how selective state regulation combined with cooperative, competitive, or conflict-ridden relations among CSOs created incentives and capacities for political engagement, illustrated by an analysis of how Ugandan CSOs related to the regulation of autocratic elections in 2016. The findings demonstrate a fuller understanding of the variety of CSOs' political engagement, including their support for, submission to, and resistance against the autocratic order.
The scope of the argument covers electoral authoritarian regimes. There are, however, reasons to believe that important differences within this regime category affect the impact of social mobilization on state power. In hegemonic electoral autocracies, such as Uganda, state control is harsher, coercion and co-optation as forms of domination are more strongly rooted, and the significance and causal effects of civil society mobilization, whether supportive or critical of the government, are overall more modest and circumscribed by state regulation. In competitive electoral autocracies the political opposition is stronger and the scope for autonomous mobilization of CSOs is wider, both vertically in relation to the state and horizontally in relation to other groups. While both differentiated state regulation and relations among CSOs matter anywhere, state regulation can be expected to be relatively more important in hegemonic electoral autocracies, pointing to avenues for further research.
While there has been an overall shift in emphasis from cooperation to cruder forms of coercion and co-optation in state regulation, as in Uganda over the last 20 years, there are still significant differences in the regulation of CSOs, their inter-group relations, and their political engagement. Even in contexts characterized by strong state domination, governments relate to interest groups in society different ways. Studies of how selective state regulation interplays with relations among CSOs thus contribute to more fine-grained analyses of how autocratic political orders are anchored in society. 8. Scholar Makerere University, 1 October 2018, Kampala 9. NGO leader, 1 October 2018, Kampala 10. NGO leader, 3 October 2018, Kampala 11. NGO leader, 3 October 2018, Kampala Notes 1. The use of divide-and-rule by stoking tensions between factions has been a particularly prominent government strategy in relation to Muslim organizations (Downie, 2015, p. 55). 2. The demands also included various legislative changes and stricter regulation of campaign spending and the use of government resources. 3. The 2016 elections were widely regarded at the time as the worst under NRM rule from a democratic point of view. Not only was both government abuse of resources and administrative irregularities more extensive than before, state-sanctioned violence rose again after the relatively peaceful 2011 elections (Sjögren, 2018). However, the 2021 elections were characterized by even more state-based coercion and worse malpractices (Abrahamsen & Bareebe, 2021). 4. The Act was deemed unconstitutional and annulled by the Supreme Court in March 2020. 5. After the 2016 elections the government cracked down hard on critical civil society organizations (Khisa, 2019, p. 348).