Making disorder more manageable: The short-term effectiveness of local mediation in Darfur

Abstract

Much of the quantitative conflict resolution literature focuses on mediation between states or on efforts to end a civil war through a comprehensive peace agreement that brings peace to the entire country. This article instead analyses the effectiveness of mediation between a wide range of armed actors on a subnational level. Utilizing unique data on Darfur covering the January 2008–August 2009 period, this article finds that mediation efforts following armed clashes in a given area significantly prolong local lulls in fighting in this area. This finding remains robust when controlling for the presence of a peacekeeping base, battle-deaths and the type of armed actors engaged in armed clashes. In addition, the finding remains robust when accounting for the non-random assignment of mediation efforts through matching similar observations in the dataset. Finally, anecdotal evidence from sites of armed conflict beyond Darfur suggest that the findings from this study might also hold in other armed conflicts.

Keywords

conflict resolution Darfur local mediation local peace peacekeeping United Nations

Introduction

Following the failure of the Darfur Peace Agreement (DPA) in 2006, the United Nations (UN) and the African Union (AU) were much criticized between 2007 and 2009 for their inability to build on the DPA and resolve the conflict in Darfur, as well as for their inability to bring the government of Sudan (GoS) and the major Darfurian rebel groups together for peace talks. However, within this time period many different smaller, less publicized initiatives to curb armed violence in Darfur took place. Broadly speaking, two approaches to deal with armed conflict can be recognized. The first approach aims to conclude an elite-level bargain between a government and a rebel party, potentially paving the way for a peacekeeping mission that ensures continued commitment to peace of the conflict parties. The second approach aims to mitigate violence and build peace on the local level, either though grass-roots initiatives led by local people or with the support of state-level or international mediators supporting the peacemaking effort. It was this second type of mediation that was conducted throughout Darfur in the wake of the failed DPA in order to manage armed violence.

Much of the quantitative conflict resolution literature focuses on mediation efforts between states or on efforts to end a civil war through a comprehensive peace agreement that brings peace to the entire country (Wallensteen & Svensson, 2014; Duursma, 2014). This article, instead, focuses on mediation in conflicts that are ‘local’ and take place below the surface of the highly publicized peace processes that have the potential to end in a Nobel peace prize ceremony in Oslo. National-level mediation efforts have been criticized for being out of touch with realities of a conflict on the ground (Autesserre, 2017, 2010). However, these critiques of mediation efforts aimed at resolving national-level conflicts ignore the fact that while international mediation and peacekeeping indeed often fail, a wealth of scholarly studies show that these type of international conflict management efforts at least greatly increase the likelihood of conflict termination and durable peace (Bercovitch & Gartner, 2006; Fortna, 2008).

On the other hand, armed violence commonly does not end when fighting between government forces and rebel groups stops. This means that violence between other pairs of actors than the government forces and rebels persists even if a civil war has technically ended. For instance, Debos uses the term ‘inter-war’ to describe those periods in Chad where there was no large-scale fighting but low violence between a wide array of armed groups which was the best Chadians could hope for (Debos, 2016).

Subnational mediation efforts are often portrayed as being an effective mechanism to deal with armed violence on the local level between other types of pairs of armed actors than government forces and rebels (Autesserre, 2017; O’Bryan, Rendtorff-Smith & Donati, 2017). Yet, beyond participant observations and several case studies, there is little evidence to support the effectiveness of local mediation. As noted by Autesserre in a review essay, all of the authors studying local peace efforts ‘come from anthropological or interpretive traditions, making it more difficult for the findings to travel to mainstream positivist research in political science and international relations. Quantitative and experimental inquiries could help to overcome these limitations by evaluating the strength and exportability of the qualitative findings’ (Autesserre, 2014: 495).

This article aims to systematically examine the effectiveness of local mediation efforts. Specifically, this article is concerned with what local mediation efforts can accomplish in sites of armed violence in which there is a breakdown in central political authority and where many different armed actors operate. I conduct a quantitative analysis of local mediation efforts, either with or without international support, in Darfur between January 2008 and August 2009. Darfur is a suitable case not only because UN data on Darfur are available for this period, but also because there is a high degree of spatial variation in levels of violence in Darfur between January 2008 and August 2009, which makes it possible to examine whether this variation is driven by mediation efforts on the subnational level. Moreover, while Darfur experienced a civil war in the studied time period, a diverse set of different types of armed actors, including militias and rebel groups, engaged in armed clashes. De Waal refers to the armed violence in Darfur as a ‘Hobbesian’ situation of generalized insecurity even during periods of low-intensity fighting between the armed actors (de Waal, 2015). Finally, and crucially, no country-level mediation efforts took place in Darfur between January 2008 and August 2009, only some failed efforts to initiate peace negotiations (Svensson & Wallensteen, 2010). This makes it easier to isolate the effect of subnational mediation in Darfur.

While most of the previous research focuses on how mediation efforts can bring peace to an entire country, this article will consider how subnational mediation efforts can create local lulls in armed fighting, contributing to making disorder more manageable. While this article contributes to our understanding of subnational mediation, the evidence presented here should be seen as preliminary evidence that can lay the basis for future research. The data on Darfur only pertain to a specific area and the data only cover around one and a half years. Yet, crucially, this article lays the basis for the start of a research agenda that draws on systematic, quantitative research methods to study subnational mediation processes and their effects on battle intensity. The next section reviews how the quantitative conflict resolution literature is state-focused and explains why subnational mediation is effective to manage armed fighting.

Theoretical framework

Local mediation can be defined as a conflict management activity, which does not involve the use of physical force or invoke the authority of law, used to change, affect or influence the perceptions or behaviour of conflict parties that are engaged in a conflict at the subnational level (Bercovitch, 1992: 7; Gorur & Vellturo, 2017: 7). Of course, organized armed violence always takes place somewhere at a local level (Aas Rustad et al., 2011), but ‘conflict at the subnational level’ in this definition of local mediation refers to the conflict issues the parties fight about rather than where armed violence takes place. A conflict issue is a disputed point or question, the subject or controversy of a conflict (Diehl, 1992: 333). A provincial governor or even international peacekeeping staff mediating between groups of farmers and herders fighting each other over control of a piece of land would thus constitute cases of local mediation based on the definition of local mediation used in this article.

However, while local conflict is often between conflict parties that are solely embroiled in conflict concerning local conflict issues, such as between herders and farmers, this does not have to be the case. Parties involved in a national-level conflict can also be involved in a local mediation process, as long as this mediation effort pertains to resolving a local conflict issue rather than a national one. For example, the Free Syrian Army (FSA) fights Syrian government forces for control over Syria, but it has been involved in various local mediation processes in which a local faction of the rebel group negotiated over local conflict issues (Turkmani et al., 2014).

Local mediation efforts have received relatively little scholarly attention. Indeed, the quantitative conflict resolution literature is extremely state-focused. Many of the hypotheses tested in the mediation literature have been tested based on data on international crises (Beardsley et al., 2006). Furthermore, numerous quantitative studies have addressed mediation within civil war countries, yet these studies almost always focus on mediation between a government and a rebel party (for an overview, see Wallensteen & Svensson, 2014; Duursma, 2014). While the quantitative conflict resolution literature has mainly focused on mediation between governments and rebels, armed violence often also takes place between different rebel parties, between communal groups, between different groups within the government security apparatus, or between armed militias and other armed groups. A notable exception in the literature is a recent study by Karakus & Svensson (2017) in which they examine the durability of ceasefires, including local ceasefires, concluded by various types of armed actors in Syria.

The lack of focus on the micro-level dimensions of mediation between different types of armed actors in the quantitative conflict literature contrasts with the qualitative research that has emerged in this regard (for an overview, see Mac Ginty & Richmond, 2013; Autesserre, 2014). The rapidly expanding body of literature on subnational mediation is referred to as the ‘local turn’ within peace studies (Mac Ginty & Richmond, 2013). It is argued in most studies on subnational mediation that it is effective (Autesserre, 2014). However, what is currently still missing within this body of literature is a quantitative assessment of the effectiveness of local peace efforts.

In this article, I test the argument that mediation on the local level makes future armed clashes less likely, thus prolonging local lulls in armed fighting. The focus is purposively on the short-term effect of local mediation as this makes it possible to examine whether local mediation is an effective conflict management tool when armed fighting is ongoing and a country-level peace process is not possible. An observation on Darfur made by Flint illustrates the merit of a short-term focus: ‘Mediators tend to assume that the conflicting parties want to establish a stable “normality” based on a written agreement, but in Darfur today there is so little confidence in any form of normality that the parties are seeking short-term advantage and expecting continuing turmoil. Realistically, the challenge is not to create stable normality, but to make the disorder manageable’ (Flint, 2010: 45–46). So how can local mediation efforts contribute to making disorder more manageable?

I put forward three arguments for why mediation on the local level reduces the likelihood of future armed violence. First, mediators can help facilitate information exchange. The underlying logic of this argument is that conflict parties take up arms when there is disagreement about the balance of power, as this prevents leaders from bargaining to a settlement that would avoid the costs of armed fighting (Fearon, 1995; Reiter, 2009). There is generally relatively little communication between local conflict parties when armed fighting is ongoing (Gorur & Vellturo, 2017). Mediators can establish contact between conflict parties by acting as a go-between (Kelman, 2005; Duursma, 2017c). The subsequent exchange of information between the conflict parties helps them to reach a settlement that reflects the balance of power (Fearon, 1995; Reiter, 2009). With regard to local conflicts, some scholarly attention has been given to how ‘insider mediators’ establish contacts between the conflict parties and facilitate information exchange. Insider mediators are mediators who are not members of the conflict parties, but who nevertheless ‘through their localized networks can make connections, establish contact, and facilitate dialogue between the belligerents’ (Karakus & Svensson, 2017: 6). The mayor of a given town mediating between two tribes living in close proximity to this town would be an example of an ‘insider’ mediation effort.

Second, a third party facilitating information exchange not only helps the conflict parties reach agreement over the balance of power, but also allows the conflict parties to engage in problem-solving and resolve the underlying conflict issues. Mediators can act as a repository of trust between the conflict parties and provide a relatively neutral space for peace talks (Kelman, 2005; Duursma, 2017b). This allows the local conflict parties to resolve their differences through negotiations. Blattman, Hartman & Blair (2014) find robust evidence that alternative dispute resolution training, which includes training in mediation and problem-solving skills, for customary leaders and ordinary citizens in Liberia led to a higher resolution of land disputes and a reduction in levels of violence. A telling example of how the UN has facilitated problem-solving is a peace conference organized by civil affairs officers within the United Nations Mission in Sudan (UNMIS) in 2011. The goal of this conference was to make peace between two rival tribes in south Kordofan. UNMIS arranged the transport for representatives of the conflict parties to attend the peace conference. At the conference, the civil affairs officers helped the tribes to agree upon three thematic issues that needed to be discussed: homicide and blood money compensation payment; peaceful co-existence and development of relations; and joint services for consolidating the reconciliation. Three thematic groups of 20 participants (ten from each side) were subsequently formed. Each thematic group discussed how to address issues of concern and signed a document detailing what they had agreed upon. Later in the peace conference, all conference participants from both tribes endorsed these agreements (O’Bryan, Rendtorff-Smith & Donati, 2017).

Third, and lastly, mediators can contribute to making armed clashes less likely through pressuring the local conflict parties towards concluding a ceasefire. While the exchange of information and facilitation of problem-solving both help the conflict parties find a settlement within the bargaining range, third-party pressure increases the cost of not making peace for the conflict parties, thus widening the bargaining range. There is robust evidence that pressuring conflict parties in international crises makes the conclusion of a peace agreement more likely (Beardsley et al., 2006). Mediators can also put pressure on local conflict parties. Bradbury et al. (2006: 88) observe, for example, that in Darfur, customary mediation by mediators known as ajaweed often involves significant pressure being put on local conflict parties: ‘The Ajaweed are, by traditional practice, elderly people who are versed in communal customs and customary laws. They are not neutral in the Western sense of the term. Their practice is to exert pressure on the party resisting a settlement, until they accept the recommendations the Ajaweed have settled on.’

In short, mediators in local conflicts can help the conflict parties find an agreement that is in line with the balance of power, resolve the conflict issues, or provide incentives to stop fighting. Hence, I expect mediation to prolong local lulls in armed fighting.

Hypothesis 1: Mediation following an armed clash between local conflict parties prolongs the time until an armed clash between these conflict parties reoccurs.

The argument on local mediation outlined above relates to mediation in local conflicts in general, but there are two specific reasons for why the involvement of UN peacekeeping staff in local mediation efforts further increases the prospect of mediation success. First, the involvement of UN peacekeeping staff in local peace processes helps the conflict parties to overcome commitment problems that would otherwise make peaceful forms of resolution difficult to pursue (Hultman, Kathman & Shannon, 2014; Ruggeri, Dorussen & Gizelis, 2017). Local conflict parties find it hard to commit to peace since they cannot be sure the enemy will commit to peace. The involvement of a third party, especially a third party with monitoring capacity like a peacekeeping mission, can raise the confidence among the local conflict parties that the local ceasefire is implemented in good faith (Wall & Druckman, 2003; Ruggeri, Dorussen & Gizelis, 2017). Karakus & Svensson (2017) call into question the added value of international involvement when it comes to mediating local conflicts, pointing out that local ceasefires derived without any international involvement have been more likely to succeed in Syria. Yet, no peacekeepers are deployed in Syria and the added value of international staff involved in local peace process is likely to be different when UN peacekeepers are on the ground. Indeed, research on the effectiveness of peacekeeping suggest that UN peacekeepers can play a positive role in prolonging local peace. Ruggeri et al. find that local conflict episodes last for shorter periods when peacekeepers are deployed to conflict-prone locations inside a country, even with comparatively modest deployment. They partly explain this finding by pointing out that peacekeepers can mediate between local conflict parties and, in doing so, can assure compliance with any concluded peace deal through monitoring or even enforcing these local peace deals (Ruggeri, Dorussen & Gizelis, 2017: 167–168).

Second, UN peacekeeping staff can compensate for possible biases of local mediators in those local conflicts in which the UN mediates jointly with local third parties. Elfversson (2019) demonstrates that conflict resolution is unlikely when biased government actors mediate in a local conflict because the conflict parties cannot trust the government’s willingness to uphold any agreement reached. The involvement of UN peacekeeping staff can help the conflict parties commit to an agreement in spite of the government seeking to promote the interest of only one party. In addition to the UN having the monitoring capacity to assess whether the agreement is implemented in good faith, a core principle of UN peacekeeping is that peacekeepers act impartially. In other words, while biased local-level or national-level mediators may sometimes be necessary to push local conflict parties towards peace, UN peacekeeping staff can help mitigate the fears of the conflict parties about any of these local-level or national-level mediators favouring the other side.

In short, the involvement of UN peacekeeping staff in mediation processes in local conflicts helps the parties to overcome commitment problems. This leads to the following testable implication.

Hypothesis 2: Mediation efforts with the involvement of an international third party in local conflicts are more effective than mediation efforts without the involvement of UN peacekeeping staff.

Research design

In order to test the impact of mediation on the subnational level, I draw on event data that have been compiled by the Joint Mission Analysis Centre (JMAC) between 3 January 2008 and 31 August 2009, in real time, to support the day-to-day operations of the United Nations–African Union Hybrid Operation in Darfur (UNAMID). The data have been provided by the African Union High-Level Panel on Darfur (AUPD, 2009) with assistance from Alex de Waal who acted as an adviser to the AUPD (de Waal et al., 2014). The JMAC data are uniquely detailed (Duursma, 2017a). A major advantage of these data is that not only observations on armed clashes are included, but also data on mediation efforts. This allows for an assessment of mediation efforts in Darfur by looking at temporal and spatial variation of armed clashes across Darfur’s different localities.

Another advantage of the JMAC dataset is that it contains non-public data. Hence, unlike UN staff that work on reports that will be published in the public realm, JMAC analysts do not have to worry about whether their reporting will paint the UN activities in a certain light. This means that the JMAC data are relatively unlikely to underreport or overreport certain UN activities or activities by actors outside the UN. Since political considerations are not taken into account by JMAC analysts when processing information, there is no reason to suspect that only successful cases are reported.

There is also little reason to assume that the JMAC data overrepresent the role of the UN, while underestimating local third-party efforts. Supporting the mediation of local conflict has been an important part of UNAMID’s mandate. The need to address local conflict in Darfur was highlighted as early as July 2005, when the conflict parties signed the Declaration of Principles, which stated that the ‘promotion of reconciliation, the restoration of the traditional and time honoured peaceful coexistence among the communities of Darfur’ are essential to restore peace in Darfur and called for a ‘Darfur–Darfur dialogue’ (African Union, 2005: 2). When UNAMID was deployed in January 2008, UNAMID Civil Affairs started to try to support this Darfur–Darfur dialogue aimed at resolving conflict between various state armed groups, which is why JMAC has also collected data on mediation efforts not conducted by the UN. Indeed, a civil affairs officer reflected that ‘[i]t is crucial for us to know what local peacemaking efforts are conducted in order to see where we can be of added value’ (author interview in New York, 19 July 2017). Nevertheless, mediation between different government units and between government and militias is arguably less likely to be picked up by JMAC analysts because these conflicts are commonly resolved internally behind closed doors (de Waal, 2015). As a robustness test, the empirical analysis will therefore drop these types of armed clashes in order to see whether this changes the results.

I have constructed a dataset, on the basis of the JMAC data, in which the unit of analysis is the occurrence of an armed clash in one of Darfur’s 63 localities. I deem the locality an appropriate level of analysis because armed actors often compete over control of territory within the confines of these administrative borders (Gorur & Vellturo, 2017). Whether armed clashes would reoccur in a given grid cell or a particular village would not capture this dynamic.

This article focuses on the short-term effectiveness of mediation on the local level. I therefore look at the duration of lulls in armed fighting. It should be noted that this is different from looking at peace, as peace requires the belligerents to wilfully no longer engage in violence with one another for a clearly defined period of time. Rather than selecting an arbitrary time frame to assess the duration of lulls in armed fighting, I measure the number of days between the date when an armed clash occurred between a set of armed actors and the date when an armed clash reoccurred between the same armed actors. An armed clash is thus considered to have reoccurred if an armed clash between the same conflict parties takes place again in a given locality. Accordingly, by necessity, this article focuses on armed clashes of which the conflict parties are known, as it is impossible to identify whether an armed clash reoccurs or not if the parties are unknown. If an armed clash has not reoccurred by 31 August 2009 (the end date of the JMAC data), then the number of days from the date of the armed clash until 31 August 2009 is coded. In order to model the duration of ‘lulls in arms fighting’ in number of days, I employ hazard analyses. Using hazard models to study conflict reocurrence makes it possible to right censor those armed clashes that had not yet reoccurred by 31 August 2009. The right censoring of observations is a real advantage considering the relative short time frame of the dataset. Cox duration models are used because this procedure is semi-parametric and thus less dependent on any parametric assumption.

An example of a mediation effort, included in the JMAC data, in which both a local mediator and UNAMID were involved in Darfur is the effort to terminate armed fighting in the vicinity of Ladob in Gereida locality in mid-May 2008: ‘UNAMID forces and civilian component have been mediating the conflict that opposes the Birgit to the Zaghawa and their respective affiliated movements, the SLA FW and the SLA MM. The wali of South Darfur has also addressed the issue’ (JMAC observation 462). An example of a local third-party effort is the following mediation effort on 20 November 2008 in response to fighting between Masalit and Arab tribesmen: ‘The Wali of west Darfur mediated between the Masalit and Arab tribesmen following an armed clash on 17 Nov. 08’ (JMAC observation 1608).

I control for several possible confounding variables that might influence both the reoccurrence of armed clashes and the initiation of mediation efforts. First of all, I control for the presence of a peacekeeping base in the locality. Previous research shows that peacekeepers are most likely to be deployed on the front line (Ruggeri, Dorussen & Gizelis, 2016), and it is plausible to surmise that UN civil affairs officers are more likely to engage in subnational mediation in close proximity to peacekeeping bases. In addition, I control for the number of battle-deaths in a given armed clash. Mediation could be more likely in more severe armed clashes, while previous research on the impact of peacekeeping on the local level has found that conflict intensity is a significant predictor of future armed fighting (Costalli, 2013). I also control for structural factors related to the locality that may influence both the reoccurrence of armed clashes and the initiation of mediation efforts. A binary variable that indicates the presence of mountains (Jebel Moon, Jabel Marra, Jebel Umkardos, Jebel Adola and the Midoub mountains) in the locality is included in the models. Expert studies on Darfur indicate that much of the fighting takes place in the mountains (e.g. Flint & de Waal, 2008), but these areas are also harder to reach for potential mediators. The opposite is true for urban centres. Fighting has mainly taken place outside of the areas in which Darfur’s major towns are located (Flint & de Waal, 2008), whereas mediation in these areas should be highly likely if fighting takes place. The models therefore include a binary variable that indicates the presence of a major town with more than 75,000 inhabitants (Nyala, Al Fasher, Ad Daein and Al Geneina) in the locality where the armed clash occurs. A binary variable that measures whether the armed clash took place during the rainy season is also included because both mediation and armed violence are less likely to occur during rainy season in Darfur as result of roads being impassable. Finally, I control for the types of armed actors that clashed.

Findings and discussion

The JMAC dataset includes 199 armed clashes (of which the identity of the conflict parties can be established) in Darfur between 1 January 2008 and 8 August 2009. Mediation efforts were initiated following 44 of these armed clashes. This constitutes around 22% of the armed clashes. Of the total of 199 armed clashes in Darfur between January 2008 and August 2009, 69 reoccurred within this period of time. Several armed clashes reoccurred as soon as the next day. For instance, on 17 February 2008, fighting broke out at Durum, Hamra and Lamena villages between the SLA AW and the SLA FW over control of the area (JMAC observation 93). The next day, on 18 February 2008, fighting restarted in the same area between the rival rebel groups (JMAC observation 97). Of all the armed clashes that eventually reoccurred, the armed clash between government forces and the Janjaweed faction led by Muhammad Hamdan Hamiti in El Fasher locality on 2 March 2008 took the longest to recur, namely after 441 days on 17 May 2009 (JMAC observations 130 and 2908). Only five out of the 44 armed clashes followed by a mediation effort reoccurred, as opposed to 64 out of the 155 armed clashes not followed by a mediation effort. This descriptive already provides strong support for the short-term effectiveness of local mediation.

In order to assess the duration of local lulls in armed fighting (as opposed to solely the recurrence of armed clashes) and control for possible cofounding variables, Table I shows the relationship between mediation efforts and armed clash recurrence through Cox regressions on armed clash recurrence in days. The hazard coefficients of the Cox regressions are reported. A negative coefficient in the models in Table I means that as the independent variable increases (e.g. a mediation effort is initiated), the time to the reoccurrence of an armed clash increases (e.g. it takes longer for the armed clash to reoccur). Model 1 thus shows that armed clashes in response

Table I.

Hazard estimates on the durability of local lulls in armed fighting in Darfur, January 2008–August 2009

Variables	Model 1	Model 2	Model 3	Model 4
Mediation	–1.680**		–1.910**
	(0.485)		(0.603)
Only local involvement		–1.577*		–1.249
		(0.768)		(1.026)
UN involvement		–1.738**		–2.125**
		(0.603)		(0.728)
Peacekeeping base in locality	0.0135	0.0132	–0.714	–0.708
	(0.316)	(0.316)	(0.448)	(0.448)
Battle-deaths	0.00204	0.00198	0.0514*	0.0520*
	(0.00508)	(0.00512)	(0.0209)	(0.0210)
Major town in locality	0.303	0.301	1.037	1.031
	(0.463)	(0.464)	(0.876)	(0.877)
Mountains in locality	0.452	0.454	0.316	0.314
	(0.410)	(0.410)	(0.664)	(0.665)
Rainy season	0.301	0.294	0.521	0.493
	(0.273)	(0.276)	(0.566)	(0.569)
Government versus rebels	Reference category	Reference category	Reference category	Reference category
Government versus government	–1.223*	–1.226*	omitted	omitted
	(0.555)	(0.555)
Government versus militia	–2.141*	–2.147*	omitted	omitted
	(1.022)	(1.022)
Rebels versus rebels	–0.512	–0.517	0.798	0.757
	(0.416)	(0.417)	(0.757)	(0.761)
Rebels versus militia	–0.323	–0.325	omitted	omitted
	(0.375)	(0.375)
Militia versus militia	0.116	0.103	0.0354	0.0168
	(0.378)	(0.386)	(0.440)	(0.441)
Likelihood ratio chi-square	34.69**	34.71**	27.90**	28.35**
Observations	199	199	111	111

* significant at 5%; **significant at 1%.

to which a mediation effort takes place are on average followed by longer lulls in armed fighting.

The positive impact of local mediation on the duration of local lulls in armed fighting in Darfur in the studied period is in line with Alex de Waal’s observation in a briefing in September 2009 that the major reason for the decline in intertribal fighting is a robust action by the South Darfur state authorities who mediated between the rival tribes (de Waal, 2009: 2). UNAMID supported these tribal reconciliation efforts. For instance, the JMAC data include numerous observations on the Civil Affairs Division (CAD) meeting with representatives of the Habaniya and the Fallata during March and April 2009 (JMAC observations 2741, 2774, 2797 and 2823).

The impact of mediation efforts on the reoccurrence of armed clashes is shown in a Kaplan-Meier graph in Figure 1. The mean of the numbers of days until an armed clash reoccurs, of those armed clashes that eventually reoccur, is 53 days. Figure 1 indeed shows

Figure 1.

Kaplan-Meier estimates on the recurrence of armed clashes in Darfur

that the survival functions steeply drop in the first two months following an armed clash that is not followed by a mediation effort. Around 56% of the armed clashes that are not followed by a mediation effort do not reoccur, whereas this figure is 89% (only 5 out of 44 reoccurred) for armed clashes that are followed by a mediation effort.

To test the impact of international involvement in local mediation efforts, Model 2 looks at the relative effectiveness of mediation efforts conducted solely by locals or efforts in which UNAMID is involved as well. UNAMID had a role in 29 of the 44 mediation efforts that followed an armed clash. It follows from Model 2 that both mediation efforts solely conducted by local third parties and local mediation efforts in which UNAMID has been involved significantly prolong local lulls in armed fighting, though the effect of solely local mediation is significant at the 5% level. Hence, Model 2 does not provide support for Hypothesis 2.

However, a limitation of the estimates in Model 1 and Model 2 is that the non-random assignment of mediation is insufficiently taken into account. It could be that those armed clashes that are followed by a mediation effort have very different characteristics. Before turning to the statistical analysis that deals with this problem, a reflection on the likely direction of potential selection bias is in order. Expert reports on local conflicts suggest that local conflicts that are most likely to be mediated are also most prone to reoccurrence (Gorur & Vellturo, 2017; O’Bryan, Rendtorff-Smith & Donati, 2017). Mediation seems to occur more frequently following armed clashes between tightly organized groups that have a history of armed fighting, because mediators are already familiar with this conflict and they know how destructive the conflict can be. At the same time, conflicts of this type are also more likely to reoccur than armed clashes between conflict parties that are loosely organized and have no conflict history. An unexpected clash between farmers and herders because of cattle destroying crops would indeed seem less likely to reoccur than an armed clash between two tribal militias with a longstanding history of fighting over a particular piece of land. Finally, of those clashes that are followed by mediation, clashes followed by the involvement of UN Civil Affairs officers seem especially prone to reoccur. The UN often acts as a ‘mediator of last resort’ in situations ‘where there is nobody else to do the work and there is a major long-term risk of escalation’ (Brockmeier & Rotmann, 2016: 17). One UN Civil Affairs officer working in Darfur acknowledged that given the limited resources of the UN, he and his colleagues were forced to prioritize managing conflicts that are most prone to escalation (author interview in Khartoum, 10 December 2014). Telling in this regard is that the average number

Table II.

Results of matching: treatment = mediation effort

	L1 distance before matching	L1 distance after matching
Peacekeeping base in locality	0.31	0.00
Battle-deaths	0.16	0.08
Major town in locality	0.15	0.00
Mountains in locality	0.09	0.00
Rainy season	0.02	0.00
Government versus government	0.12	0.00
Government versus militias	0.05	0.00
Rebels versus rebels	0.06	0.00
Rebels versus militia	0.15	0.00
Militia versus militia	0.33	0.00
Global L1 distance	0.57	0.16
Matched cases		31
Unmatched cases		13

of deaths of armed clashes that are followed by a mediation effort is 13.5, while this figure is 6.2 for armed clashes that are not followed by a mediation effort; and conflict intensity has been found to be a significant predictor of conflict recurrence in previous spatially disaggregated research (Costalli, 2013).¹ Overall, this means that if there are selection effects, these probably run in the opposite direction of Hypotheses 1 and 2.

Nevertheless, a matching design is used to deal with the methodological problem of the non-random assignment of mediation (see Rubin, 1979). More specifically, I use the coarsened exact matching (CEM) method introduced by Blackwell et al. to coarsen the independent variables and recode them so that similar values are grouped together. The exact matching algorithm is then used to identify the matches. Next, the coarsened values are abandoned and the actual values of the independent variables in the matched data can be used to estimate the causal effect of mediation (Blackwell et al., 2009). The CEM method thus essentially follows the logic of a most similar design. Table II summarizes the results of the application of the CEM procedure.

In addition to the unidimensional measures of L1 for each variable separately, Table II reports the Global L1 distance. The latter is an index that measures the global balance between the two subsamples: armed clashes that are followed by a mediation effort and armed clashes that are not followed by a mediation effort. The characteristics of the two subsamples would be completely the same if the L1 statistic has a value of 0, whereas an L1 value of 1 would mean that the characteristics of the two subsamples are completely different. The Global L1 distance is 0.57 before matching, which means that only around 43% of the empirical distributions of the two subsamples overlap. The value 0.57 serves as a baseline reference for the unmatched data. As follows from Table II, the CEM matching procedure reduces the imbalance in the data, as the Global L1 distance is 0.16 after matching.

Successful matching requires simultaneously achieving an acceptable level of imbalance in the data and a sufficiently large matched data subset. However, minimizing imbalance is at the expense of the size of the matched data subset and thus might lead to an unacceptably high level of variance, whereas maximizing the size of the subset of matched data will be at the expense of a reduction of the imbalance of the data and thus might lead to biased estimations. Many matching procedures either determine the matched sample size and then minimize imbalance or determine the imbalance and subsequently maximize the sample size. A distinctive feature of the CEM matching procedure used here is that it jointly maximizes balance and sample size (Blackwell et al., 2009; see also King, Lucas & Nielsen, 2017). As noted above, the CEM matching procedure reduces the imbalance to a Global L1 distance of 0.16, but the matched sample size is also sufficiently large. Table II shows that 31 cases of mediation following an armed clash could be matched, while 13 could not be matched. With 31 treatment observations among a total of 44 observations with the treatment in the dataset, the matching procedure thus tests for significance on 70.1% of the treated observations. The total number of matched observations is 111 out of a total of 199 observations in the dataset. In short, the CEM matching procedure has reduced the imbalance, while ensuring a sufficiently large matched subsample.

Since the imbalance is not completely removed, I subsequently employ a duration model to estimate the causal effect of mediation based on the matched data. As follows from Model 3 and Model 4 in Table I, the analysis based on the matched data does not alter the main finding on the impact of mediation. The significant and negative correlation between mediation and the reoccurrence of armed clashes remains, even when taking into account the non-random assignment of mediation efforts. This provides additional support for Hypothesis 1. However, while the impact of local mediation efforts with international involvement remains significant, the impact of solely local mediation turns statistically insignificant. Yet, it should be noted that the relatively small number of observations on which this analysis is based (as a result of matching the data) means that one should be careful to not draw any definite conclusions.

Nevertheless, the descriptions of the mediation efforts in which UNAMID was involved suggest that UNAMID’s contribution was supportive in most of these local mediation efforts, arguably even crucial at the same time. For instance, in late June, armed fighting broke out between the Tarjam and Bani Halba in the area of Bulbul (JMAC observation 786). Upon UNAMID advice, the Wali of South Darfur went to talk to the conflict parties to calm down the tensions. UNAMID provided air transportation for these visits to take place (JMAC observation 800). UNAMID and local actors have in several instances also acted in parallel in Darfur. For instance, an armed clash took place between the SLA FW and the SLA MM in Ladob village in early May (JMAC observation 431). The JMAC notes that following this armed clash, ‘UNAMID forces and civilian component have been mediating the conflict between the SLA FW and the SLA MM. The Wali of South Darfur has also addressed the issue’ (JMAC observation 462). A subsequent report notes the conclusion of an agreement on 30 June and another report in July notes ‘[o]n 3 July 2008 UNAMID DJSR met with representatives of SLA/MM and SLA/FW regarding the commitment of the two factions to the agreement of 30 June 2008 to cease hostilities in Shaeiria Locality. Mr. Anyidoho, DSJR emphasized UNAMID’s support to the agreement at all stages. For the agreement to be sustainable, the two factions recommended to have strong and transparent implementation mechanism in cooperation between UNAMID, SLA/MM, and SLA/FW’ (JMAC observation 801). Finally, UNAMID has also supported local peace processes through both mediating and following up with patrols. An example from the JMAC data in this regard is that UNAMID civil affairs officers were involved in mediation between two local Gimir tribes in Katayla locality in late 2009 and that UNAMID patrols subsequently tried to prolong this peace. For instance, the JMAC report that ‘[n]o clashes have been reported between the Gimir North (who had migrated from Kulbus, West Darfur, to Antikina during the droughts of 1984) and the Gimir South (the landowners and leaders of the Native Administration in Katayla) since 12 December. Despite the lull, the situation remains tense, as a militia of about 300 men arrived from El Geneina and Kulbus via Mukjar towards Rehed Al Birdi locality (150 km southwest of Nyala) to fight alongside the Gimir North. Local community leaders have tried to persuade the group to return and not to further complicate the already tense situation. A UNAMID patrol has been tasked to meet their leaders in an attempt to convince them to withdraw’ (JMAC observation 1783).

Finally, three more robustness checks are conducted. First, in order to test whether the data under study meet the proportional hazards assumption, I employ a non-proportionality test using Schoenfeld residuals (Box-Steffensmeier, Reiter & Zorn, 2003). With a chi2 score that is statistically significant, this test shows that the proportional hazards assumption is violated. In other words, the effects of the independent variables included in the model may be greater or smaller, or even change signs, depending on the amount of time that has elapsed for a given observation. This means that a single hazard ratio describing the impact of the variables used in the models in Table I might not be appropriate. Box-Steffensmeier, Reiter & Zorn (2003) show that the proportional hazards assumption can be relaxed by including interaction terms of time and the independent variables. The underlying idea of this approach is that non-proportionality is very similar to other instances in which the effects of independent variables change depending on the value of some other variable. Time can thus be treated as a covariate that has a conditional effect on other variables. Consequently, I replicate the models, but I interact all the variables with a variable that measures the time since the last armed clash between a given dyad in a given locality. Including these interaction terms allows a variable’s effect to increase or decrease over the duration under study. The models also include the square and cube of the time variables in order to further account for temporal dependence (Box-Steffensmeier, Reiter & Zorn, 2003). The interaction terms of time and the mediation variables in the models are not statistically significant. This suggest that using a single hazard ratio to describe the impact of these variables is appropriate.

Second, the models in Table II are replicated using a logit model with armed clash reoccurrence used as a dependent variable. This does not alter the sign of the coefficient of the mediation variables and neither does it change the statistical significance.

Third, in order to examine whether the findings are driven by an overall decline in armed violence in Darfur from February onwards, I estimate the impact of mediation in Darfur based on a subsample of the data. For this analysis, all armed clashes between March and August 2009 are dropped. Crucially, I right censor those armed clashes that had not yet reoccurred by 28 February 2009 rather than 31 August 2009 and I do not take any of the armed clashes after 28 February 2009 into account to code the conflict reoccurrence variable. This robustness check does not change the main finding that local lulls in armed fighting after armed clashes followed by a mediation effort tend to be significantly longer. In short, this article shows that local mediation is effective – and that peacekeeping staff, especially civil affairs officers, contribute to these mediation efforts.

Beyond Darfur

Local mediation is not unique to Darfur. This final section briefly considers whether the findings above would also hold in sites of armed conflict other than Darfur. Autesserre convincingly shows that most of the international mediation efforts in the DRC, as well as those conducted on the subnational level, have failed (Autesserre, 2010). Yet, Autesserre uses a much more ambitious benchmark for mediation success than used in this article. While the analyses above focus on short-term violence, Autesserre uses a situation-specific definition of success in which she considers a mediation effort to be successful when ‘a large majority of the implementers (international and local peacebuilders) and intended beneficiaries (including local elites and ordinary citizens) perceive it as having promoted peace in the area of intervention’ (Autesserre, 2017: 117). While mediation efforts in the DRC might indeed have been unsuccessful according to this definition of success, there is some anecdotal evidence that suggest that subnational mediation efforts in the DRC have made a positive difference in the short term on various occasions. A telling example in this regard is that with tens of thousands of people being displaced by armed violence, the head of the peacekeeping mission in the DRC, Namanga Ngongi, mediated between the agriculturalist Lendu and pastoralist Hema ethnic groups in Ituri in March 2003, resulting in a ceasefire on 18 March (Guéhenno, 2015: 127). Jean-Marie Guéhenno, the Under-Secretary General for UN Peacekeeping Operations at the time, asserts that the efforts led by Ngongi ‘would make an enormous difference; without him, many more people would have died in Ituri’ (Guéhenno, 2015: 127).

The positive short-term impact of local mediation has also been reported in relation to the war in Syria. Hassan Hassan noted in early 2014 that ‘[a]lthough rebels with radical ideologies currently seem to dominate the scene, most Syrians have more everyday demands – dignity and freedom. This group, which is bent on the dismissal of [President] Assad but not on the wholesale dismantling of Syrian state institutions, has proved willing to strike deals in certain conditions’ (Hassan, 2014). Indeed, by early 2014, several ceasefires had been concluded between government troops and various local factions of the Free Syria Army (FSA) based in several suburbs of Damascus, including Barzeh, Moadamiya, Bibilla, Bait Sahem and Dumayr (Turkmani et al., 2014: 25–30; Hassan, 22 January 2014). These local ceasefires were quite successful at managing some of the worst armed violence in Damascus. This, in turn, prompted UN Special Envoy for Syria, Staffan de Mistura, to shift gear in late 2014 and pursue a local approach to manage armed violence in Syria .

Mediation on the subnational level has also been moderately successful – at least in the short term – in some instances in South Sudan. For example, community leaders in South Sudan’s Bor-Pibor area engaged in local mediation in May 2017 in responses to a series of clashes between Dinka Bor and Murle tribal militias in this area. The local mediation effort culminated in the signing of a ceasefire agreement in Juba, which stipulated that the two communities would immediately cease hostilities and create an atmosphere conducive to a comprehensive and inclusive dialogue. This ceasefire succeeded in preventing any further large-scale violence between these communities in the subsequent months (Sudan Tribune, 2017a,b). United Nations Mission in South Sudan (UNMISS) staff provided support for this mediation effort. In short, mediation on a subnational level has not only been successful in short-term violence reduction in Darfur, but also in other sites of armed conflict like the DRC, Syria and South Sudan.

Conclusion

Some have argued that the conventional model of a peace process in which two sides of equal legal standing and with roughly similar military capability engage in square-table negotiations is ill suited to make disorder more manageable in places in which central authority has broken down and in which many different armed actors operate (for example, see de Waal, 2015). This article does not argue that international mediation efforts aimed at bringing peace to an entire country are ineffective – as previous research clearly shows these efforts are effective – but this article rather shows that local mediation can be an effective supplementary tool to manage armed clashes between a wide range of different armed actors.

The findings in this article suggest that local mediation efforts are effective at preventing any more armed clashes. In addition, tentative evidence seems to suggest that local mediation efforts in which internationals are involved are particularly effective. This article thus shows that peace operations not only deter actors from using armed violence, but also contribute towards the management of armed violence through mediation efforts.

Three remaining questions and possible avenues for future research stand out. First, previous research has found that the conflict in Darfur is difficult to resolve because many different stakeholders operate at different levels, including local, national, regional and international levels. The many interlinkages between these different levels add to the intractable nature of the Darfur conflict (Brosché & Duursma, 2017). One fruitful avenue for future research would be to examine how the connections of local conflict parties influence their propensity to make peace or not (Duursma, 2019).

Second, future research should examine possible negative effects of local mediation efforts. For example, non-state armed groups in the Central African Republic (CAR) have signed non-aggression pacts, while subsequently ganging up on parties that were not a signatory of the concluded pact and thus redirecting the violence elsewhere (Lombard, 2016).

Third, and lastly, an important question for future research is how local peace spells translate to long-term stability. This question mirrors research on the stability of agreements aimed at ending government–rebel violence on the country level (e.g. Quinn et al., 2013). Local peace spells could create the conditions for the durable termination of the local conflict, but local conflict parties could also resume fighting after having prepared for another bout of armed fighting. However, answering this question will require data covering a longer period and ideally also data from other peacekeeping operations. The UN could benefit from making more of its data available for researchers to facilitate research on the effectiveness of local mediation. Understanding under what conditions local mediation works could ultimately guide the work of local mediation efforts by UN staff.

Nevertheless, anecdotal evidence suggest that at least some of the local mediation efforts taken into account in this article have proven durable. For example, a situation report drafted by UNAMID Civil Affairs on 13 April 2019 notes that the ‘GCSS [Governance and Community Stabilization Section] as part of a UNAMID integrated team conducted a field mission to Tanakoro village (24 km south of Kass town) to assess the intercommunal related issues. The head of Sheiks from Rezeigat tribe informed that the security situation is stable and last intercommunal conflict in the area was in 2008 between Rezeigat and Tarjam tribes over land and pasture issues when a peace agreement was reached through mediation of UNAMID, the government, and native administration leaders’ (UNAMID GCSS, Daily Report 4 April 2019). Future research should examine the extent to which local mediation efforts are durable.

Furthermore, even if future evidence suggests that many local peace lulls eventually break down, this would not mean that local mediation efforts are not worthwhile. The modal outcome of national-level mediation efforts is also failure (Duursma & Svensson, 2019), but we also know that it only takes one successful attempt to accomplish a serious reduction in armed violence (Wallensteen & Svensson, 2014). The same might be true for local mediation efforts. Local mediation efforts are probably neither a placebo nor a panacea, but even one local mediation effort that proves durable – as the successful mediation effort between the Rezeigat and Tarjam tribes in the Tanakoro area illustrates – will have a positive effect on the everyday security of citizens. In addition, even relatively short local peace lulls might have a positive impact. Civilians are often targeted in the context of communal conflicts. Hence, a temporary lull in fighting in these type of conflicts provide citizens with some respite to gather or collect food, access humanitarian aid, or even flee the conflict area (Duursma, 2019). This relates to the point made by Flint (2010: 45–46) that mediation does not necessarily needs to lead a stable ‘normality’ based on a written agreement, but that it also can help to make disorder manageable.

In conclusion, local mediation efforts are effective at preventing any more armed clashes. Hence, this article shows that even when a stable ‘normality’ based on a written agreement is not within the realm of possibilities for a given civil war country, it might still be possible to make disorder more manageable through local mediation efforts.

Footnotes

Replication data

The dataset, codebook and do-files for the empirical analysis in this article, along with the Online appendix, can be found at .

Acknowledgements

I am grateful for very useful feedback from Alex de Waal, Govinda Clayton, Roger Mac Ginty, Roisin Read, Christine Cheng and Enzo Nussio. Any errors are my own.

Funding

The article is made possible through ESRC grant ES/ L007479/1 ‘Making Peacekeeping Data Work for the International Community’.

ORCID iD

Allard Duursma

Notes

References

Rustad

Siri

Aas

Camilla

Buhaug

Halvard

Falch

Åshild

Gates

Scott

(2011) All conflict is local: Modeling sub-national variation in civil conflict risk. Conflict Management and Peace Science 28: 15–40.

African Union (2005) Declaration of Principles for the Resolution of the Sudanese Conflict in Darfur.

African Union Panel on Darfur (2009) Mission to Sudan June 14–25 2009: Process, outcomes, observations (https://dl.tufts.edu/concern/pdfs/7m01bx989).

Autesserre

Séverine

(2010) The Trouble with the Congo: Local Violence and the Failure of International Peacebuilding. Cambridge: Cambridge University Press.

Autesserre

Séverine

(2014) Going micro: Emerging and future peacekeeping research. International Peacekeeping 21(4): 492–500.

Autesserre

Séverine

(2017) International peacebuilding and local success: Assumptions and effectiveness. International Studies Review 19(1): 114–132.

Beardsley

Kyle C

Quinn

David M

Biswas

Bidisha

Wilkenfeld

Jonathan

(2006) Mediation style and crisis outcomes. Journal of Conflict Resolution 50(1): 58–86.

Bercovitch

Jacob

(1992) The structure and diversity of mediation in international relations. In: Bercovitch

Jacob

Rubin

Jefery

(eds) Mediation in International Relations: Multiple Approaches to Conflict Management. Basingstoke: Macmillan, 1–29.

Bercovitch

Jacob

Gartner

Sigmund

(2006) Is there method in the madness of mediation? Some lessons for mediators from quantitative studies of mediation. International Interactions 32(4): 329–354.

10.

Blackwell

Matthew

Iacus

Stefano

King

Gary

Porro

Giuseppe

(2009) CEM: Coarsened exact matching in Stata. Stata Journal 9: 524–546.

11.

Blattman

Christopher

Hartman

Alexandra C

Blair

Robert A

(2014) How to promote order and property rights under weak rule of law? An experiment in changing dispute resolution behavior through community education. American Political Science Review 108(1): 100–120.

12.

Box-Steffensmeier

Janet M

Reiter

Dan

Zorn

Christopher

(2003) Nonproportional hazards and event history analysis in international relations. Journal of Conflict Resolution 47(1): 33–53.

13.

Bradbury

Mark

Ryle

John

Medley

Michael

Sansculotte-Greenidge

Kwesi

(2006) Local peace processes in Sudan: A baseline study. Research report . Nairobi: Rift Valley Institute.

14.

Brockmeier

Sarah

Rotmann

Philipp

(2016) Civil Affairs and Local Conflict Management in Peace Operations. Geneva: Geneva Centre for Security Sector Governance.

15.

Brosché

Johan

Duursma

Allard

(2017) Hurdles to peace: A level-of-analysis approach to resolving Sudan’s civil wars. Third World Quarterly 39(3): 560–576.

16.

Costalli

Stefano

(2013) Does peacekeeping work? A disaggregated analysis of deployment and violence reduction in the Bosnian war. British Journal of Political Science 44(2): 357–380.

17.

de Waal

Alex

(2009) Briefing: Darfur. September.

18.

de Waal

Alex

(2015) The Real Politics of the Horn of Africa: Money, War and the Business of Power. Cambridge: Polity.

19.

de Waal

Alex

Hazlett

Chad

Davenport

Christian

Kennedy

Joshua

(2014) The epidemiology of lethal violence in Darfur: Using micro-data to explore complex patterns of ongoing armed conflict. Social Science & Medicine 120: 368–377.

20.

Debos

Marielle

(2016) Living by the Gun in Chad: Combatants, Impunity and State Formation. London: Zed.

21.

Diehl

Paul F

(1992) What are they fighting for? The importance of issues in international conflict research. Journal of Peace Research 29(3): 333–344.

22.

Duursma

Allard

(2014) A current literature review of international mediation. International Journal of Conflict Management 25(1): 81–98.

23.

Duursma

Allard

(2017a) Counting deaths while keeping peace: An assessment of the JMAC’s field information and analysis capacity in Darfur. International Peacekeeping 24(5): 823–847.

24.

Duursma

Allard

(2017b) Partnering to make peace: The effectiveness of joint African and non-African mediation efforts. International Peacekeeping 24(4): 590–615.

25.

Duursma

Allard

(2017c) When to get out of the trench: Using smart pressure to resolve civil wars. Civil Wars 17(1): 43–61.

26.

Duursma

Allard

(2019) Local peace processes and the protection of civilians. IPI Global Observatory 27 September.

27.

Duursma

Allard

Svensson

Isak

(2019) Introducing an agenda-based measurement of mediation success: The divergent effects of the manipulation strategy in African civil wars. International Negotiation 24(2): 296–323.

28.

Elfversson

Emma

(2019) The political conditions for local peacemaking: A comparative study of communal conflict resolution in Kenya. Comparative Political Studies 52(13): 2061–2096.

29.

Fearon

James D

(1995) Rationalist explanations for war. International Organization 49(3): 379–414.

30.

Flint

Julie

(2010) Rhetoric and reality: The failure to resolve the Darfur conflict. HSBA Working paper 19. Geneva: Small Arms Survey.

31.

Flint

Julie

Waal

Alex de

(2008) Darfur: A New History of a Long War. London: Zed.

32.

Fortna

Virgina P

(2008) Does Peacekeeping Work? Shaping Belligerents’ Choices after Civil War. Princeton, NJ: Princeton University Press.

33.

Gorur

Aditi

Vellturo

Madeline

(2017) Local Conflict, Local Peacekeeping. Washington, DC: Stimson Center.

34.

Guéhenno

Jean-Marie

(2015) The Fog of Peace: A Memoir of International Peacekeeping in the 21st Century. Washington, DC: Brookings Institution Press.

35.

Hassan

(2014) Hope springs in Syria? How local cease-fires have brought some respite to Damascus. Foreign Affairs 22 January.

36.

Hultman

Lisa

Kathman

Jacob

Shannon

Megan

(2014) Beyond keeping peace: United Nations effectiveness in the midst of fighting. American Political Science Review 108(4): 737–753.

37.

Karakus

Dogukan C

Svensson

Isak

(2017) Between the bombs: Exploring partial ceasefires in the Syrian civil war, 2011–2017. Terrorism and Political Violence. https://doi.org/10.1080/09546553.2017.1393416

38.

Kelman

Herbert C

(2005) Building trust among enemies: The central challenge for international conflict resolution. International Journal of Intercultural Relations 29(6): 639–650.

39.

King

Gary

Lucas

Christopher

Nielsen

Richard A

(2017) The balance-sample size frontier in matching methods for causal inference. American Journal of Political Science 61(2): 473–489.

40.

Lombard

Louisa

(2016) State of Rebellion: Violence and Intervention in the Central African Republic. London: Zed.

41.

Mac Ginty

Roger

Richmond

Oliver

(2013) The local turn in peace building: A critical agenda for peace. Third World Quarterly 34(5): 763–783.

42.

O’Bryan

Tom

Rendtorff-Smith

Sara

Donati

Marco

(2017) The Role of United Nations Peacekeeping Operations in Addressing Local Conflicts: A Study of Practice. New York: United Nations Departments of Peacekeeping Operations and Field Support Policy, Evaluation and Training Division Policy and Best Practices Service.

43.

Quinn

David

Wilkenfeld

Jonathan

Eralp

Pelin

Asal

Victor

McLauchlin

Theodore

(2013) Crisis managers but not conflict resolvers: Mediating ethnic intrastate conflict in Africa. Conflict Management and Peace Science 30(4): 387–406.

44.

Reiter

Dan

(2009) How Wars End. Princeton, NJ: Princeton University Press.

45.

Rubin

Donald B

(1979) Using multivariate matched sampling and regression adjustment to control bias in observational studies. Journal of the American Statistical Association 74(36): 318–328.

46.

Ruggeri

Andrea

Dorussen

Han

Gizelis

Theodora-Ismene

(2016) On the frontline every day? Subnational deployment of United Nations peacekeepers. British Journal of Political Science 48(4): 1005–1025.

47.

Ruggeri

Andrea

Dorussen

Han

Gizelis

Theodora-Ismene

(2017) Winning the peace locally: UN peacekeeping and local conflict. International Organization 71(1): 163–185.

48.

Sudan Tribune (2017a) UN envoy warns over deteriorating security in S. Sudan. 11 May (http://www.sudantribune.com/spip.php?article62434).

49.

Sudan Tribune (2017b) S. Sudan: Bor and Murle communities sign cessation of hostilities accord. 24 May (https://www.sudantribune.com/spip.php?article62540).

50.

Svensson

Isak

Wallensteen

Peter

(2010) The Go-Between: Jan Eliasson and the Styles of Mediation. Washington, DC: United States Institute of Peace.

51.

Turkmani

Rim

Kaldor

Mary

Elhamwi

Wisam

Ayo

Joan

Hariri

Nael

(2014) Hungry for peace: Positives and pitfalls of local truces and ceasefires in Syria. London School of Economics, Department of International Development.

52.

Wall

James A

Druckman

Daniel

(2003) Mediation in peacekeeping missions. Journal of Conflict Resolution 47(5): 693–705.

53.

Wallensteen

Peter

Svensson

Isak

(2014) Talking peace: International mediation in armed conflicts. Journal of Peace Research 51(2): 315–327.