Civilian Protest in Civil War: Insights from Côte d’Ivoire

How does civilian protest shape civil war dynamics? Existing research shows that civilian protests against violence and war contribute to peace and restrain violence against civilians. There is less research on civilian protests that are at odds with peaceful conflict resolution, such as protests to salute armed actors, advocate against peace agreements, and oppose peacekeepers. This study develops a novel typology of wartime civilian protest that brings together protests to different ends, and theorizes the heterogeneous effects of protest on civil war dynamics. Using quantitative and qualitative evidence from new disaggregated and georeferenced event data from Côte d’Ivoire, the study demonstrates that—contingent on certain demands—protests were associated with violence against civilians, violence involving peacekeepers, and failed conflict resolution. These findings contribute new knowledge on how civilians shape the dynamics of civil war, and caution that nonviolent civilian action may not only be a force for de-escalation and peace.


C
ivilians are not passive victims and often use nonviolent tactics to advocate against violence and war (Dorff 2019, 286).One expression of civilian agency is wartime civilian protest-instances of public, collective, and predominantly nonviolent action by which noncombatants make conflict-related demands on armed actors within the context of armed conflict.
Existing research highlights that wartime civilian protests are common (Vüllers and Krtsch 2020), often large (Chenoweth, Hendrix, and Hunter 2019), and emerge even in the harshest of circumstances (Bamber and Svensson 2022;Jackson 2021;Svensson et al. 2022).For example, in 2012, women in Maiduguri in Nigeria protested to condemn Boko Haram and appealed to the rebels to "sheathe their swords and embrace peace" (The Daily 2012).During the Guatemalan civil war, social movements organized weekly mass-marches to denounce the government's human rights abuses (Dudouet 2021, 11).And, in Gao in Mali, hundreds protested against the jihadists' destruction of ancient tombs and ban on football (Svensson and Finnbogason 2021, 581).
Yet civilians can also be at odds with peaceful conflict resolution and transformation, and protest to cheer on armed actors, oppose peace accords, denounce peacekeepers, or advocate for violence.In 2016, hundreds of thousands of Yemenis rallied in Sanaa in support of the Houthi's goal of regime change (France24 2016).In 2018, civilians demonstrated in Afrin in Syria to endorse brutal pro-government militias (The New York Times 2018).In Sri Lanka, a pro-war movement led by Buddhist monks organized frequent protests against the 2002 peace process (Paffenholz 2014, 88).Likewise, thousands rallied in Donetsk in Ukraine in 2014 to welcome the pro-Russian separatist forces (Radio Free Europe 2014).These examples raise several under-researched questions: What different types of civilian protest occur in civil war?How do civilian protests shape civil war dynamics?And what are the consequences of civilian protests that advocate against peaceful conflict resolution?
There is less research on the causes and consequences of nonviolent civilian action to seemingly less peaceful ends (cf.León 2017;Nilsson et al. 2020;Paffenholz 2014;Svensson et al. 2022), as wartime social movements that "mobilize against a peaceful or transformative agenda" are "often overlooked" (Dudouet 2021, 6).Several reasons may underlie this omission, including a notion that more authoritarian preferences engender political apathy (Hellmeier and Weidmann 2020, 74), a lack of integrated data on wartime contentious politics (González and Vüllers 2020), and the moral complications of depicting nonviolent action as "uncivil" (Chenoweth 2021, 68-70).Nevertheless, this lacuna is surprising given that scholars recognize that civilians can both resist and support armed actors (Arjona 2016;Barter 2012;Wood 2003).
This article focuses on the most visible form of nonviolent civilian action, namely wartime civilian protest.To fill the gap in our knowledge of civil war, I first develop a novel typology of wartime civilian protest that brings together civilian protests to different ends.I then theorize how civilian protests to certain ends may have adverse consequences, present new disaggregated and georeferenced event data on civilian protests in the Ivorian civil war , and provide quantitative and qualitative evidence that civilian protests-contingent on certain demandswere associated with violence against civilians, violence involving peacekeepers, and failed conflict resolution.In doing so, I advance knowledge on civil war, nonviolent action, and contentious politics in three ways.
First, I provide a new conceptual tool for identifying different wartime civilian protest types and theorizing about their respective causes and consequences.Existing research views wartime protest as a form of noncooperation that civilians engage in alongside strategies like compliance or displacement (Barter 2012;2014;Jose and Medie 2015;Kaplan 2017).Scholars further distinguish different types of nonviolent actions.Arjona (2015) differentiates between full and partial resistance to armed groups based on the scope of resistance; Dudouet (2021) distinguishes wartime social movements with pro-peace, pro-change maximalist, and pro-change sectorial objectives; and Masullo (2021) organizes civilian noncooperation on a spectrum depending on the level of confrontation.These conceptual works highlight relevant variation in wartime civilian protest, yet tend to overlook protests with ends that are at odds with peaceful conflict resolution.In contrast, I develop a typology that covers the full range of civilian protests in four opposing pairs: pro-government and anti-government protest; pro-intervention and anti-intervention protest; pro-peace and anti-peace protest; and pro-reform and anti-reform protest.
Second, I provide new insights on the prevalence and characteristics of wartime civilian protest.Although several subnational datasets capture wartime civilian protest-in Colombia (Arjona 2015;Kaplan 2017;Masullo 2021), India (Krtsch 2021), and Syria (Svensson et al. 2022)-there is a lack of event data that capture protests to all different ends and provide sufficient detail to disaggregate protest types. 1 Arjona, Kaplan, and Masullo all limit their focus to civilian protests against violence and war, Krtsch (2021) records only rebel-organized strikes, and Svensson et al. (2022) document only protests against jihadist insurgents.My database of civilian protest in Côte d'Ivoire thus provides a first glimpse of the prevalence, characteristics, and agents of civilian protest, and allows for a comparison across protest types.These insights can inform broader scholarship on contentious politics, and helps provide a blueprint for future data collection efforts.
Third, I advance knowledge on the conditions under which civilian protests shape three outcomes of relevance for restraining and ending civil wars: violence against civilians, violence involving peacekeepers, and conflict resolution.These outcomes jointly cover a large body of existing scholarship, which gives me greater leverage to locate civilian protests in the broader civil war literature.All three outcomes are extensively studied in separate bodies of literature.Nevertheless, our knowledge of how civilian protest shapes these outcomes is less developed (Vüllers and Krtsch 2020, 11) and there is a dearth of research on how protests that oppose peaceful conflict resolution influence civil war dynamics (Dudouet 2021, 6).Hence, by theorizing and probing the adverse consequences of civilian protests, I demonstrate that civilian nonviolent action can also be a driver of escalation and violence, and call attention to a new variable that should be integrated in civil war research.

PREVIOUS RESEARCH
Scholars have long noted that civilians exercise agency in civil war (Kalyvas 2006;Wood 2003).A first strand of literature examines the causes and consequences of violent civilian action in civil war.One form of violent civilian action is the creation of civilian self-defense groups-that is, armed and autonomous forms of civilian counter-insurgent mobilization (Jentzsch, Kalyvas, and Schubiger 2015).The formation of self-defense groups is often driven by the threat of indiscriminate violence against civilians (Schubiger 2021).A related set of studies focus on how civilians use denunciations to manipulate armed actors to use violence against local rivals, a strategy which is prominent in contested areas characterized by limited information (Balcells 2017;Kalyvas 2006).Violent civilian action, however, falls outside the scope of this investigation, which focuses only on nonviolent action.
A second strand unpacks the causes and consequences of nonviolent civilian action in civil war (Avant et al. 2019;Chenoweth, Hendrix, and Hunter 2019;Dorff 2019;Masullo 2021) and other violent contexts (González and Vüllers 2020;Ley, Mattiace, and Trejo 2019).Existing research shows that demonstrations, strikes, protest marches, and sit-ins are more common in civil war than often assumed.About onefourth of all protests in Africa in 1992-2013 took place in areas with active armed conflict (Vüllers and Krtsch 2020, 5).Case study evidence from Afghanistan (Jackson 2021), Colombia (Arjona 2016;Kaplan 2017;León 2017;Masullo 2021), Mali (Svensson and Finnbogason 2021), and Syria (Svensson et al. 2022) paints a similar picture.Such nonviolent civilian action follows systematic patterns, and often constitutes a reaction to violence (Vüllers and Krtsch 2020) or unpopular wartime governance (Arjona 2016;Svensson and Finnbogason 2021).Moreover, nonviolent civilian action is more common when civilians can draw on strong institutions (Arjona 2016;Kaplan 2017) and social networks (Rubin 2020; Svensson 1 Triangulated report-based event datasets like the Social Conflict Analysis Database (SCAD) (Salehyan et al. 2012) and the Armed Conflict Location and Event Data (ACLED) (Raleigh et al. 2010) constitute a good starting point, but often provide insufficient detail about the protesters' stated demands to identify types.and Finnbogason 2021; van Baalen 2021) to overcome collective action problems.Additionally, studies show that nonviolent civilian action is more likely when the political opportunity structure favors mobilization (Bamber and Svensson 2022).
The literature reviewed above focuses predominantly on nonviolent civilian action to tamp down on violence and war.Less attention has been given to civilian action seemingly at odds with peaceful conflict resolution and transformation.Recent scholarship shows that citizens also engage in nonviolent action to support authoritarian regimes (Hellmeier 2021;Hellmeier and Weidmann 2020) and undermine democratically elected governments (Chenoweth 2021, 76-7).Similar insights are yet to be incorporated into research on nonviolent civilian action in civil war.While several studies acknowledge that not all wartime civilian movements advocate for peace and provide illustrative evidence that it may provoke violence and undermine peace processes (Dudouet 2021;Nilsson et al. 2020;Paffenholz 2014;Svensson et al. 2022), there is little systematic evidence on the effects of such nonviolent action in civil war.One exception is León (2017), who examines the effect of pro-and anti-government protest on violence against civilians in Colombia and finds that supportive protests curb violence while hostile protests provoke violence.

A TYPOLOGY OF WARTIME CIVILIAN PROTEST DEMANDS
Expanding knowledge on wartime civilian protest hinges on scholars' ability to systematically identify protest events and classify protest demands.To this end, I expand existing conceptual work (Arjona 2016;Dudouet 2021) and develop both the overarching concept wartime civilian protest and a typology of protest demands.I define wartime civilian protest as an instance of public, collective, and predominantly nonviolent action by which noncombatants make conflictrelated demands on armed actors within the context of armed conflict.
First, wartime civilian protests are collective and public acts.Collective actions are carried out by a group of people who are motivated by some common interest and who work together to achieve a shared objective.Public acts are those that are overt and visible, meaning that the target is aware of the act.Such actions include, for instance, demonstrations, rallies, protest marches, sit-ins, strikes, manifestations, mass-meetings, and occupations, which are typically carried out in public and require at least a few participants.Acts that are covert or carried out by a single individual do not qualify as collective and public acts.
Second, wartime civilian protest is carried out by noncombatants.Noncombatants are individuals that are not involved in planning, leading, or participating in combat.Civilians that support conflict parties by providing administrative, logistical, or material support are not considered combatants.I recognize that distinguishing combatants and noncombatants can be difficult, and that the difference can be a matter of degree (Barter 2014, 10-2).Nevertheless-at the conceptual level-it remains important to distinguish those individuals that take no active part in hostilities from those that do.The definition does not preclude that armed actors are involved in organizing protests, for instance, through their political wings.Armed actor involvement can range from overlooking protests to hiring protest brokers to coercing participation, none of which implies that those participating should be seen as combatants.Moreover, ignoring protests organized by armed actors demands that analysts can distinguish armed actors and their affiliates from civilians and determine their degree of involvement, which is rarely possible across many events (Nilsson et al. 2020, 245-6).
Third, wartime civilian protest entails predominantly nonviolent action.Following Chenoweth, Hendrix, and Hunter (2019, 297), I consider protest actions predominantly nonviolent when the vast majority of participants are unarmed and do not intentionally threaten to or harm the physical well-being of people.Actions by civilian self-defense groups that use violence, hence, fall outside of the definition.Nonviolent action can sometimes unintentionally result in harm to people-for instance, due to repression by the target or spontaneous violence by protesters.Such civilian protest events are still considered predominantly nonviolent as long as nonviolent methods are the main tactic used during the event.2Thus, initially, peaceful demonstrations by unarmed participants that escalate into violence are considered civilian protests, whereas demonstrations by soldiers brandishing weapons are not.
Fourth, wartime civilian protest involves articulating conflict-related demands on armed actors.I conceive of armed actors as any state, nonstate, or international armed organization that uses armed force to influence an intra-state contest over political power.Armed actors, hence, include the government, pro-government militias, rebel groups, external intervention forces, and United Nations (UN) peacekeeping missions.A broad inclusion of possible targets is important because understanding how civilians shape civil war calls attention to all actors capable of influencing the armed contest.Conflict-related demands relate to an armed actor's political demands, involvement in the war, use of violence, or wartime governance and behavior.Demands on governance and behavior are only considered conflict-related when they are explicitly linked to the actor's involvement in the war.Hence, civilian calls to escalate fighting, engage in negotiations, or provide financial restitution to war victims are considered conflict-related, whereas general civilian calls for higher wages are not (Chenoweth, Hendrix, and Hunter 2019, 297).
Finally, wartime civilian protest takes place within the context of armed conflict-that is, within a state where a nonstate armed actor is using violence to challenge the state.I consider states to be embroiled in armed conflict as long as there is an expectation of continued violence, meaning that the nonstate armed actor remains armed and the main incompatibility unresolved, even if the number of conflict deaths remains low.The focus on an expectation of continued violence is warranted by the observation that fighting is not omnipresent in armed conflict-affected countries.Instead, civil war is characterized by a broken state monopoly on violence and a situation of dual power (Kalyvas 2006).

The Typology
Wartime civilian protest can be organized around a diverse array of demands.To conceptualize this variation, I develop a conceptual typology of wartime civilian protest demands. 3Focusing on stated demands provides a relevant classification basis because it goes to the heart of what protesters (claim they) want (González and Vüllers 2020, 310;White et al. 2015, 473).Both conflict resolution (Wallensteen 2015) and civil resistance scholars (Chenoweth and Stephan 2011;Nepstad 2015) stress the importance of demands for understanding contentious politics and resolving conflict.Moreover, existing research demonstrates that protest demands affect protest outcomes (León 2017;Svensson et al. 2022).Finally, differentiating civilian protest demands is important because it enables scholars to link protests to specific outcomes.After all, the best way to measure a social movement's success is by examining whether it reached its stated objectives (Chenoweth and Stephan 2011).Thus, although other dimensions-like protest tactics and origin-are also important, a focus on protest demands allows scholars to better assess the consequences of wartime civilian protest.
First, I distinguish civilian protest demands based on the issue at stake.An important insight in the civil resistance literature is that not all campaigns aim to fundamentally alter the prevailing political order (Chenoweth and Stephan 2011, 69;Nepstad 2015, 72) or resist an armed actor's rule altogether (Arjona 2015, 183-4).Instead, some nonviolent campaigns advocate for radical change like regime change or secession, whereas other movements have more limited goals that aim to change existing political institutions, particular policies, or alter the target's actions.Bringing this insight to wartime civilian protest, I distinguish between four categorically different issues at stake: (1) the prevailing political order, (2) foreign intervention, (3) peace efforts, and (4) armed actors' wartime governance and behavior toward civilians.
Second, I identify two contrasting subtypes of each civilian protest type based on whether protesters articulate opposition to or support for the issue at stake.Existing conceptualizations tend to focus on protests in which participants demand change rather than oppose change, meaning that protesters who advocate against regime change or secession, or in support of foreign intervention and continued war, risk being overlooked.However, a relevant insight in the civil resistance literature is that social movements are often opposed by countermovements that also use nonviolent tactics, but seek to resist or reverse an original movement's demands for change (Chenoweth 2021, 78).Examples from authoritarian states like Russia, Iran, and Turkey demonstrate that citizens also take to the streets to demonstrate their support for the regime and fend off anti-regime challengers (Hellmeier 2021;Hellmeier and Weidmann 2020).Drawing on this insight, I therefore distinguish between protest demands that entail opposition to the issue at stake and those that imply support for the issue at stake.
Table 1 brings these insights together in a typology of wartime civilian protest demands that yields four general types (and eight subtypes) of protest: alignment protest, intervention protest, peace-related protest, and reform protest.The typology categorizes distinct civilian protest demands and the different types are, therefore, mutually exclusive.However, since protest events can encompass multiple demands-either on the same target or on multiple targets-a single protest event can comprise different types of demands.Thus, the analytical categories should be viewed as heuristic devices that reflect protest events in which participants only advance a single demand.

Alignment Protests
Alignment protests concern fundamental aspects of the prevailing political order-such as regime change or the status of a specified territory-and, therefore, imply that the protesters take sides and align themselves with either the government or the rebels.Protests in which participants oppose the prevailing political order and articulate support for a rebel group are examples of anti-government protests.Examples include protests in which participants call for regime change, a change of political system, secession, or denounce the government.An example of an anti-government protest occurred in Cotabato City in the Philippines in 1986, when tens of thousands attended a three-day prayer rally to show support for the Moro Islamic Liberation Front (Barter 2014, 177).Another example was the protest in Sanaa mentioned in the introduction.In contrast, pro-government protests are protests in which participants oppose a change to the established political order and show support for the government.An example of a pro-government protest took place in Addis Ababa in Ethiopia in 2021, when civilians demonstrated their support for the government's fight against the Tigray People's Liberation Front (France24 2021).

Intervention Protests
Intervention protests concern political or military efforts by secondary or third parties to regulate the civil war's main incompatibility, for instance, through the use of armed force or economic sanctions.A key difference between intervention and alignment protests is that intervention protests target foreign actors and do not involve protesters taking sides on the overall incompatibility.Protesters can either oppose the foreign actor's involvement in anti-intervention protests, or endorse the foreign actor's involvement in prointervention protests.Examples of demands articulated in intervention protests are an end to external military support for the government or rebel forces, calls for UN intervention, or an end to economic sanctions.For example, in April 2021, protesters in the Democratic Republic of the Congo demonstrated to demand MON-USCO's departure (Reuters 2021).The opposite occurred in Indian-controlled Kashmir in August 2016, when separatist supporters marched in Srinagar to call upon the UN to intervene (Euronews 2016).

Peace-Related Protests
Peace-related protests concern manifest or proposed efforts by the main warring parties to find a negotiated solution to the civil war, such as initiating peace negotiations or signing a ceasefire or peace agreement.Even when peace-related protests emerge from partisan factions, they differ from alignment protests because civilians oppose or support particular conflict resolution efforts rather than calling for or opposing regime change or secession.One subtype of peace-related protest is pro-peace protests in which participants advocate for one or several armed actors to initiate or deepen efforts to negotiate an end to the war.Such was the case in Liberia, where the Women in Peace Network organized marches and sit-ins to push the warring parties to the bargaining table without putting their weight behind a particular side (Dudouet 2021, 7).Civilians can also advocate against a peace process, peace agreement, or provision in an agreed agreement in anti-peace protests.An anti-peace protest took place in Bangui in the Central African Republic in 2019, when the Ë Zîngo Bîanî opposition coalition marched to oppose the Khartoum Agreement for excluding opposition groups (The New Humanitarian 2019). 4

Reform Protests
Reform protests concern an armed actor's wartime governance and behavior toward civilians, such as violence against civilians, service provision, or imposition of particular wartime rules.These protests differ from alignment protests in that they do not concern the political system as a whole, but seek to change particular policies (Arjona 2015, 183-4;Dudouet 2021, 10).Pro-reform protests refer to protests in which civilians articulate demands on an armed actor to reform their wartime governance or change its behavior, but refrain from making claims pertaining to the prevailing political order.One example of a pro-reform protest took place during Ansar Dine's occupation of Timbuktu in Mali, when dozens protested against the jihadists' imposition of sharia law (Svensson and Finnbogason 2021, 581).The opposite of pro-reform protests are anti-reform protests in which participants oppose a change to an armed actor's wartime governance or behavior.As reform protests often concern issues for which there is broad civilian consensus, anti-reform protest is probably rare and related to prior pro-reform protest.An example of an anti-reform protest took place in militia-controlled Benghazi in Libya in 2012, when civilians gathered to oppose an earlier proposal to abandon city-wide sharia law (The Washington Post 2012).Although anti-reform protests may be rare, ignoring such protests may lead scholars to overlook that civilians are sometimes divided about wartime governance issues, a common feature in civil wars with deep identity cleavages.

The Heterogeneous Consequences of Civilian Protest
A key proposition in the emerging literature is that nonviolent action increases the political costs of inaction and can, hence, pressure armed actors to change their policies and behavior (Avant et al. 2019, 17;Dudouet 2021, 3;Nilsson et al. 2020, 227; Paffenholz  (León 2017, 4), at least to the extent that they activate different mechanisms.A comprehensive examination of the potential consequences of civilian protest on civil war dynamics is beyond the scope of this study.Instead, I theorize the adverse consequences of civilian protests that seemingly oppose peaceful conflict resolution for three well-studied and consequential outcomes: one-sided violence (OSV) against civilians, violence involving peacekeepers, and conflict resolution.
A first potential consequence is that alignment protests increase the likelihood of OSV against civilians.Alignment protests can spur OSV by putting pressure on armed actors to act more forcefully to defeat the enemy, which may prompt military leaders to disregard indiscipline within their ranks and discount international humanitarian law (Jo 2015, 52-3).Moreover, alignment protests may provoke other armed actors to view civilians as essential support structures and, therefore, as strategic and legitimate targets of violence (León 2017, 7-8).This mechanism may, therefore, push armed actors to exert revenge against groups or communities that have participated in hostile alignment protests.By revealing civilian loyalties, alignment protest may also foment tensions that trigger local armed actors like militias to target civilians to deter threats against their constituents (Kalyvas 2006, 178-81;León 2017, 7-8).Finally, alignment protests often involve the use of harsh language or hate speech that securitizes enemy supporters as a threat.Such language can dehumanize perceived enemy supporters and, hence, make attacks on civilians more palpable to an armed actor's constituency (Valentino, Huth, and Balch-Lindsay 2004, 381).Since alignment protests concern both the government and the rebel side and can activate both rallying and repression mechanisms, alignment protests should be associated with both government and rebelperpetrated OSV. 5 Hence, I expect that: Hypothesis 1: Alignment protests increase the likelihood of one-sided violence against civilians. 6 second potential consequence is that antiintervention protests increase the likelihood of violence involving peacekeepers.Anti-intervention protests often depict intervening third-parties like UN missions as biased rather than taking a neutral role in the conflict.Additionally, anti-intervention protesters often propagate the belief that third parties are acting in their own interest and should, therefore, be seen as invaders rather than impartial interveners.By popularizing such discourses, anti-intervention protests can promote the idea that peacekeepers are legitimate targets of violence (Salverda 2013, 710), hence making both armed actors and civilians more likely to employ violence against peacekeepers.Moreover, anti-intervention protests can deepen mistrust in and resentment toward peacekeepers, which may discourage dialogue and information-sharing between peacekeepers and civilians (Nomikos 2021, 197).Because intelligence is essential for conducting peacekeeping activities (Bove andRuggeri 2019, 1634), such lack of information and dialogue can make peacekeepers more prone to escalate tense situations and more vulnerable to surprise attacks by armed actors.Thus, I expect that: Hypothesis 2: Anti-intervention protests increase the likelihood of violence involving peacekeepers. 7 third potential consequence is that anti-peace protests generate obstacles for successful conflict resolution.Several case studies demonstrate that pro-peace protests can contribute to conflict resolution by opening up opportunities for peace and dialogue (Abbs 2021, 12) and imposing costs on the conflict parties (Dudouet 2021, 3;Nilsson et al. 2020, 227).Unfortunately, these mechanisms go both ways.Anti-peace protests can signal to the conflict parties that core constituents are opposed to a negotiated settlement and that signing or implementing a peace agreement will, therefore, threaten their future hold on political power (Lilja 2011, 316).More disruptive protests may also impose direct costs on the conflict parties that cripple efforts to make peace (Paffenholz 2014, 88).At the societal level, anti-peace protests can close opportunities for peace and dialogue between citizens by dividing pro-peace movements and emboldening potential spoilers to act (Abbs 2021, 12), thereby making it more difficult for pro-peace movements to organize or for the conflict parties to implement peace agreement provisions (Paffenholz 2014, 83).Finally, as efforts by the conflict parties to appease anti-peace protests may involve downplaying or distancing themselves from peace agreements, such protests can exacerbate credible commitment problems (Walter 2002).Taken together, I expect that: Hypothesis 3: Anti-peace protests decrease the likelihood of successful conflict resolution.8 (Balcells 2017, 156-7).

CASE INTRODUCTION AND DATA
The Ivorian civil war constitutes an ideal case to explore patterns and consequences of civilian protest.First, protest constituted a prominent mode of political contention in the Ivorian civil war: Côte d'Ivoire witnessed more conflict-related protests than any other African country in 1992-2013 (Vüllers and Krtsch 2020, 9).Thus, focusing on Côte d'Ivoire allows me to examine a civil war in which civilian protest was a key feature.Second, much of the existing literature on nonviolent civilian action in civil war focuses on Colombia.Hence, studying Côte d'Ivoire allows me to interrogate patterns and consequences of wartime civilian protest outside this paradigmatic case.Third, multiple armed actors, including pro-government militias, UN peacekeepers, and a French stabilization mission, were involved in the Ivorian civil war, which enables me to look at civilian protests beyond the main governmentrebel dyad.Finally, extensive reporting on the war made constructing a fine-grained dataset on wartime civilian protest based on newspaper reports feasible.

Identifying Protest Events
The dataset covers the period from 19 September 2002 to 27 April 2011, when the "Battle for Abidjan" and President Gbagbo's arrest ended the civil war.Most variables and coding decisions draw upon the Nonviolent Action in Violent Contexts (NVAVC) dataset, which the authors state can serve "as a pilot for other regions" (Chenoweth, Hendrix, and Hunter 2019, 297).I created the dataset by hand-coding more than 7,5009 English and French newspaper articles identified in (1) the Factiva database, (2) Ivorian archives, (3) existing datasets including protest events,10 and (4) relevant secondary sources.This combined sampling approach helped alleviate under-reporting in hard-to-accessareas: some 31% of all events in rebel-held areas were only identified through archival research and secondary sources, compared to 12% of all events in government-held areas.The dataset includes 412 protest events targeting either the government, FN, French forces, UN peacekeepers, or pro-government militias.All observations include information on protest demands.The dataset records and classifies the primary type of protest demand proclaimed against each individual target, which means that protest events involving demands on multiple targets are coded for several stated demands, whereas protest events involving multiple demands on the same target only include the most salient demand.The Code Book outlines all inclusion criteria, operationalization, and ethical considerations.11Before proceeding to the analysis, I present an overview of general patterns in the data.

Protests Opposing Peaceful Conflict Resolution Were Frequent and Large
Pro-reform protests constituted the most common protest type, followed by anti-intervention and antigovernment protests (Table 2).Demands that opposed peaceful conflict resolution by signaling support for the belligerents, denouncing peacekeepers, or challenging the peace process were frequent and featured in a full 59% of all events.On average, these protests were much larger; 38% of alignment, anti-intervention, and anti-peace protests gathered more than one thousand participants, whereas only 12% of the pro-intervention, pro-peace, and reform protests did so.Thus, in Côte d'Ivoire, focusing only on civilian protests to resist violence and war would lead us to overlook frequent and large protest events.

Armed Actors Sometimes Organized Protests
The FN was involved in organizing 19% of the events in rebel-held areas, whereas civil society groups with strong ties to the government, notably the ultranationalist Jeunes Patriotes, were involved in 31% of events in government-held areas.Protests with armed actor involvement often featured more militant demands, including explicit disapproval of the peace process and calls to escalate military operations.The FN and government's implication in orchestrating civilian protest suggests that armed actors themselves viewed nonviolent action as a useful tool for furthering their political goals, a dynamic that would be missed if we discarded certain protests based on stated demands.

The Timing and Location of Events Varied by Protest Type
Different protest types followed differential temporal and spatial patterns, consistent with what we should expect of protests concerning different issues.Figure 1 reports the number of protest events by quarter-year and types of demands.The graph provides several insights.First, different protest types followed divergent time trends.Although the total number of events (top panel) was fairly consistent over time, the other panels indicate that the issue at stake changed over time.Second, some protest types-notably alignment and intervention protests-were more common in periods with active fighting.Third, peace-related protests were temporally associated with the peace process, especially the Linas-Marcoussis Agreement process in 2003.Finally, reform protests constituted the most common type of protest during the "no war, no peace" period, when civilians challenged the belligerent's abuse of civilians and lack of service provision, and peaked in 2010, when preparations for the presidential election triggered protests concerning voter registration.Although an almost equal share of protest  events occurred in rebel-and government-held areas (44% vs. 56%), different protest types also followed distinct spatial patterns (Figure 2).

THE ADVERSE CONSEQUENCES OF CIVILIAN PROTEST
I test H1 and H2 quantitatively using two datasets created from the database: a time series dataset measured at the conflict-week (N = 449) and a panel dataset measured at the district-week (N = 6,286). 12To avoid capturing protest repression, all models use a 1-week temporal lag for the independent variable.I test H3 through a short case study of the 2003 failed peace process. 13Given the study's observational setup, the empirical analyses aim to provide a plausibility probe rather than a conclusive test of the hypotheses.

Alignment Protests Correlate with Violence against Civilians
To test whether alignment protests are associated with a higher likelihood of OSV (H1), I use data from the UCDP Georeferenced Event Dataset (GED) to capture conflict-weeks and district-weeks in which armed actors deliberately killed civilians (Davies, Pettersson, and Öberg 2022;Sundberg and Melander 2013).I use a dichotomous measure of the dependent variable because the main difference across weeks and districtweeks is OSV occurrence rather than severity.OSV occurred in 13% of the weeks and 1.2% of the districtweeks, whereas alignment protests occurred in 12% of the weeks and 1.6% of the district-weeks.Since the dependent variable is dichotomous and positive values are rare, I use logistic regression and correct for biases arising in rare events models (Kosmidis and Firth 2009).Models 1-3 use the week as the unit of analysis.Model 1 probes the bivariate relationship between a 1-week temporal lag of alignment protest occurrence and OSV.Model 2 adds a number of possible confounders, and controls for a 1-week and 2-week temporal lag of OSV; weeks with negotiations using the African Peace Processes dataset (Duursma and Gamez 2022); weeks and district-weeks with battles using the UCDP-GED; and the number of thousands of peacekeepers deployed in a specific week or district-week using the Geocoded Peacekeeping Operations dataset (Cil et al. 2020).Other time-invariant confounders are held relatively constant by virtue of the focus on a single country.Model 3 disaggregates the independent variable into pro-government and anti-government protests to uncover differential effects across subtypes.Models 4-6 use the district-week as the unit of analysis, and include district-level fixed effects to account for time-invariant, unobserved heterogeneity across districts that could confound the relationship, such as ethnopolitical polarization (Balcells 2017).
The results are shown in Table 3. Models 1-2 indicate that alignment protest occurrence was associated with a higher likelihood of OSV the week after.The substantial association is considerable: based on model 2, the predicted probability of OSV following weeks with alignment protests is about twice as high as following weeks without alignment protests (21% compared with 12%).14Model 3 further suggests that this association is primarily due to pro-government protest; while positive, the coefficient for antigovernment protest is not statistically significant.The results from the district-week analysis provide similar results.Model 5 demonstrates that alignment protest occurrence in a particular district was associated with a three times higher predicted probability of OSV the week after in that district (3.8% compared with 1.2% based on model 5).Finally, model 6 suggests that this association was due to pro-government protests also at the district-week level.These results remain stable across several robustness checks (see Section A3 of the Supplementary Material).Moreover, I find that anti-peace protests were also associated with a higher likelihood of OSV, whereas other protest types were not consistently associated with OSV (Table A3 in the Supplementary Material).
Qualitative evidence lends additional support for H1.First, dehumanizing language about enemy supporters often accompanied alignment protests.The use of hate speech during the Ivorian crisis is well documented (HRW 2006).Hate speech was particularly common during pro-government protests (HRW 2011, 41-4), which may account for why pro-government but not anti-government protest was associated with a higher OSV risk.The UNSG's special adviser explicitly linked hate speech to violence, warning that it "could exacerbate already worrisome and widespread violations of human rights" (UN News 2004).Likewise, progovernment protests during the 2011 post-electoral crisis often saw Jeunes Patriotes leaders depict Ivorians of immigrant decent as "vermin," and called upon their supporters to "enroll in the army to liberate Côte d'Ivoire from these bandits" (HRW 2011, 42).Second, several reports link alignment protests to subsequent OSV.For instance, the "surge of patriotic fervour" that accompanied a pro-government rally on 2 October 2002 was associated with violent attacks against immigrants throughout the city in subsequent weeks. 15Likewise, at a pro-government rally on 18 March 2004, Jeunes Patriotes leader Goudé labeled anti-government protesters "the enemies of peace" and threatened to prevent a scheduled anti-government protest. 16When the Note: All models correct for rare events bias using mean bias-reducing adjusted scores.Robust standard errors are reported in parentheses.*p < 0.1, **p < 0.05, ***p < 0.01.
anti-government march went ahead on 25 March, the Ivorian army used heavy force, killing 120 people and wounding 274.17Similar dynamics unfolded during the 2010-11 election crisis, when pro-Gbagbo rallies were often followed by a "marked increase" in the number of attacks against opposition supporters (HRW 2011, 42).Thus, I find both quantitative and qualitative evidence in support of H1.

Anti-Intervention Protests Correlate with Violence Involving Peacekeepers
To test whether anti-intervention protests are associated with a higher likelihood of violence involving peacekeepers (H2), I draw on data from the Peacemakers at Risk (PAR) dataset, which records the location and timing of violent incidents involving French and UN peacekeepers (Lindberg Bromley 2018).Since the PAR dataset only extends to the end of 2009, this reduces the number of conflict-weeks from 449 to 380 and the number of district-weeks from 6,286 to 5,320.Peacekeeper violence occurred in 5% of the weeks and 0.6% of the district-weeks, whereas antiintervention protests occurred in 10% of the weeks and 1.3% of the district-weeks.I use a similar setup as in testing H1.Model 7 probes the bivariate relationship between a 1-week temporal lag of anti-intervention protest occurrence and peacekeeper violence.Model 8 adds the same potential confounders used above, including a 1-week and 2-week temporal lag of peacekeeper violence. 18odels 9 and 10 use the district-week as the unit of analysis, and include district-level fixed effects. 19able 4 reports the results.Models 7 and 8 demonstrate that anti-intervention protest occurrence was associated with a higher likelihood of peacekeeper violence the following week.Again, the association is substantial: based on model 8, the predicted probability of peacekeeper violence following weeks with antiintervention protests is almost three times as high as following weeks without anti-intervention protests (12% compared with 5%).The results from the district-week analysis provide similar results.Model 10 shows that anti-intervention protest occurrence in a particular district was associated with an almost three times higher predicted probability of peacekeeper violence the week after in that district (1.8% compared with 0.7%).These results remain consistent across several robustness checks (see Section A4 of the Supplementary Material).No other protest type has a consistent statistically significant association with peacekeeper violence (Table A7 in the Supplementary Material), suggesting that anti-intervention protests may have unique downstream effects on civil war dynamics.
Several pieces of qualitative evidence illustrate a potential link between anti-intervention protests and peacekeeper violence.First, the most common antiintervention protest target was also the most common actor involved in the violence.Some 70% of all antiintervention protests targeted French involvement, and 85% of all peacekeeper violence events involved French forces.Second, anti-intervention protests often featured a rhetoric that stressed resisting foreign involvement as a "war of second independence" (Piccolino 2011, 2).This "virulent anti-United Nations propaganda" cast interveners as partial and "resulted in the extensive looting and destruction of assets and property of United Nations," as well as repeated attacks on French troops (UNSC 2006, 2).The Jeunes Patriotes, who mobilized 30% of all anti-intervention protests, were infamous for their "serious inflammatory statements" against intervention forces (UNSC 2006, 2).Jeunes Patriotes leader Goudé, for example, told a large anti-French demonstration that they would "send them back where they came from," whereas an FPI leader threatened "systematic abduction of white expatriates" if the French did not disarm the rebels. 20Third, anti-intervention protests often involved more confrontational nonviolent tactics that sustained low-intensity violence against peacekeepers, including rock-throwing and property occupations.Indeed, some 34% of anti-intervention protests turned violent, compared with 25% of the events without anti-intervention demands.Thus, I find both quantitative and qualitative evidence in support of H2.

Anti-Peace Protests Can Hinder Conflict Resolution
To test the proposition that anti-peace protests decrease the likelihood of successful conflict resolution (H3), I draw on process-tracing evidence from the secondary sources underlying the database and examine how anti-peace protests undercut the Linas-Marcoussis Agreement (LMA).The LMA constitutes a suitable case because many anti-peace protests targeted this agreement.As described below, anti-peace protests undermined the LMA in three different ways.First, anti-peace protests forced key decision-makers to back-track on their commitments to mitigate political costs.Second, anti-peace protest popularized the sentiment that provisions like the FN's control of key ministries were deeply unfair and part of a French-led plot (Piccolino 2011, 14).Third, attempts to appease protesters often aggravated credible commitment problems between President Gbagbo and the rebels.The extended analysis in Section A6 of the Supplementary Material discusses alternative explanations.
The LMA, signed in Paris on 23 January 2003, was a comprehensive power-sharing agreement brokered by the French government.The agreement stipulated the creation of a reconciliation government headed by President Gbagbo, disarmament of all forces, and organization of elections (Mitchell 2012, 178-9).Additionally, the agreement allocated control of the interior and defense ministries to the rebels (Mehler 2009, 466).The LMA led to a temporary reduction in fighting, but infuriated government hardliners that believed that the President had conceded too much (Mitchell 2012, 179).
Protests against the LMA and French custodianship marked the immediate post-agreement weeks (Figure 3).The first anti-peace protest erupted when President Gbagbo announced a reconciliation government on 25 January, when hardliners in the Jeunes Patriotes demonstrated in Abidjan demanding that President Gbagbo renege on the deal. 21The next day, tens of thousands of people joined the protesters outside the French embassy. 22Protest leaders explicitly rejected the power-sharing formula and demanded that the "rebels take no part in the new government" 23 and accused "Paris of forcing Gbagbo to yield to rebel demands." 24Dissent spread further on 27 January, when protests in Abidjan, Gagnoa, San Pedro, and Soubré gathered tens of thousands of participants. 25 The first protest wave culminated on 1 February, when some one hundred thousand government supporters attended an anti-peace march through Abidjan.According to one journalist, the march showed "that Ivorians had rejected en masse the power-sharing deal." 26Another seven civilian protests triggered by the LMA occurred the week after, exacerbating government-rebel tensions.
The first protest wave signaled to President Gbagbo that key constituents vehemently opposed the LMA, a concerning development with elections on the horizon (Mitchell 2012, 179).Even worse, after a broad segment of society opposed the LMA (FEWER 2003), hardliners such as the Ivorian army leadership and the FIGURE 3. Civilian Protests after the Linas-Marcoussis Agreement Note: Each dot represents a protest event.
President's own wife publicly rejected the agreement (ICG 2003, 31).Army leaders viewed the LMA as an unacceptable threat to their power (Mehler 2009, 466).For example, in conjunction with an anti-peace protest on 26 January, the army issued a communique saying that "parts of the peace accord were 'humiliating' to the army and security forces."27With dissent rising, observers noted that President Gbagbo had lost support from the security forces, leading politicians, traditional leaders, and the Jeunes Patriotes over the peace deal. 28Although it is difficult to ascertain whether political elites were swayed by the protests or would have opposed the LMA nevertheless, it is noteworthy that elites only spoke out against the agreement after the protests revealed widespread popular discontent.
With pressure mounting, President Gbagbo started back-tracking on the agreement.In a 27 January speech to protesters, he downplayed the peace deal as "a set of propositions," 29 and said that the process could not move ahead without the people's approval. 30On 2 February 2003, a presidential adviser told the international press that the peace agreement had "shocked the Ivorian people" and said that "the renegotiation of the Marcoussis accord is inevitable." 31Referring to the anti-peace march the day before, he lamented that "having rebels in strategic government positions seems intolerable for Ivorians." 32When President Gbagbo eventually called upon his supporters to give the accord a chance in a long-awaited speech on 7 February, he also assured his followers that the military would not disarm and insisted that giving the rebels control over crucial ministries was out of the question. 33According to peace monitors, Gbagbo's 7 February speech "was greatly influenced by the voice of the street" and directly aimed at appeasing "the menacing street demands" (FEWER 2003).
President Gbagbo's efforts to calm opposition to the LMA raised MPCI leader Soro's suspicion that the government was acting in bad faith and had no intention to implement the deal.Public backpedaling aimed at soothing protesters provoked further unwillingness on the rebels' part to implement the accord (Mitchell 2012, 180).The day after the presidency called for renegotiating the LMA, MPCI-leader Soro demanded that the government implement the deal or face military action. 34MJP-leader Gaspard reacted to President Gbagbo's 7 February speech in a similar way and imposed that the President "has to apply what was decided," 35 whereas Soro urged the international community to pressure Gbagbo to implement the settlement, or else the rebels would march on Abidjan. 36  On 10 February, once again citing the President's lack of commitment, the rebels boycotted an ECOWAS summit on implementing the LMA.Explaining the boycott, Soro reiterated that the interior and defense portfolios were "not negotiable for us" and hardened his stance on the Ivorian army disarming. 37 The LMA never picked up any momentum after its first tumultuous weeks.Disagreement continued to undermine the reconciliation government and further implementation defied consecutive mediation attempts.In parallel, protests against the LMA and French intervention continued in government-held areas, as protesters decried amnesties for the rebels 38 and inhibited the installation of rebel ministers in Abidjan. 39Simultaneously, large anti-government protests took off in rebel-held areas denouncing President Gbagbo's sabotage of the reconciliation government. 40 As civilians, notably the Jeunes Patriotes, set out to "physically paralyze" the peace process (Piccolino 2011, 14), President Gbagbo appeared convinced that implementing the LMA was political suicide and continued to cripple the agreement (FEWER 2003).When the rebels recognized that the government had no intention to honor its commitments (Mitchell 2012, 179), they walked off the reconciliation government, leaving the LMA to wither and die.

CONCLUSION
This study shows that just as civilians protest against violence and war, they also protest to ends at odds with peaceful conflict resolution, such as to signal support for armed actors, denounce efforts to make peace, and oppose peacekeepers.Drawing on a typology of wartime civilian protest types and novel data on the Ivorian civil war, I provide evidence that protests that opposed peaceful conflict resolution were associated with considerable increases in the risk of violence against civilians and violence involving peacekeepers, and played a critical role in undermining the 2003 peace process.
My aim is not to discount that civilians also act as forces of peace and restraint.My dataset records many instances in which civilians courageously used nonviolent tactics to call for peace, denounce violence, and request improvements to wartime governance.I show elsewhere that such civilian protests improved wartime governance (van Baalen 2021; van Baalen and Terpstra 2023).Supporting such civilian nonviolent action can sometimes provide an alternative to armed civilian protection (Jose and Medie 2015, 529;Masullo 2021, 906-7).Yet this study's key take-away is that treating civilians as agents of their own protection demands that we invest more in studying the full repertoire of civilian protests in civil war-including their role in fomenting violence.My study has three implications for research on civil war.
First, the study suggests that there is still much to uncover about the drivers of nonviolent civilian action in civil war.The typology, and the differential temporal and spatial patterns across protest types in Côte d'Ivoire, raises questions about when, where, and why civilians protest over particular issues in civil war.Under what conditions do civilians protest against peacekeepers?Why do some peace processes trigger anti-peace protests while others do not?And what degree of influence do governments and rebels exercise over alignment protests?Answering questions like these is important, as civilian protest location, timing, and frequency may help identify peace agreement weaknesses, peacekeeping shortcomings, and humanitarian relief priorities (see Vüllers and Krtsch 2020, 11).
Second, the study implies that we cannot fully understand the broader dynamics of civil wars that witness protests without close attention to civilian action.Civilian protest-and other forms of nonviolent civilian action (Avant et al. 2019)-is not only a reaction to civil war (Vüllers and Krtsch 2020), but also shapes civil war dynamics.Several studies show that civilian protests can bolster peace, limit violence, and improve wartime civilian welfare (Abbs 2021; Arjona 2016; Kaplan 2017;Nilsson et al. 2020).While more research is needed on the consequences of all civilian protest types, this study cautions that civilian protest may not only be a force for de-escalation and peace.Future research can continue to probe the adverse consequences of civilian protest with regard to patterns of violence, the onset and outcome of peace processes, and effectiveness of peacekeeping, as well as how different protest types influence political polarization.Additionally, future research could unpack how other protest characteristics or contextual factors, like partisan affiliation, protest tactics, or armed actors' territorial control, condition the relation between protest and civil war dynamics.
Third, the study suggests that permanently putting an end to civil wars that see frequent wartime civilian protest may demand greater investments in anchoring peace among civilians and demobilizing wartime protest networks (Nilsson and Svensson 2023).There are good reasons to believe that wartime protest brokers and social networks can leave lasting legacies, and both form a basis for strengthening peace and democracy as well as become breeding grounds for radical postwar politics.Future research can aid the above efforts by examining how protest demands influence whether civilian protests shape postwar attitudes and behaviors, and by interrogating the interventions that can contribute to civilian demobilization.

FIGURE 1 .
FIGURE 1. Number of Protest Events by Types of Demands and Quarter-Year

FIGURE 2 .
FIGURE 2. Location of Civilian Protest Events by Types of Demands

TABLE 1 .
A Typology of Wartime Civilian Protest Demands

TABLE 2 .
Breakdown of Protest Events by Types of Demands Numbers represent the number of events that included a specific demand.43 events included multiple demands on different targets, hence the proportions do not sum to 100%. Note:

TABLE 3 .
The Risk of Violence against Civilians Was Higher after Alignment Protests

TABLE 4 .
The Risk of Peacekeeper Violence Was Higher after Anti-Intervention Protests All models correct for rare events bias using mean bias-reducing adjusted scores.Robust standard errors are reported in parentheses.*p < 0.1, **p < 0.05, ***p < 0.01. Note: