Crisis managers but not conflict resolvers: Mediating ethnic intrastate conflict in Africa

Abstract

Instability and conflict within African countries are on the rise. What are the best means for third parties to promote short-term crisis management and long-term conflict resolution in these situations? Often, these two tasks are at odds with one another, and certain approaches to intervention may be more or less effective. This study grapples with these issues by focusing on one particularly difficult set of cases—violent crises that are rooted in ethnic divisions and are part of protracted conflicts in Africa during the post-Cold War era—and one approach to intervention—mediation. We also view mediation as a multidimensional strategic process, and we test a series of hypotheses linking specific mediation styles to various crisis outcomes. The data and analyses reported in this study grew out of a new project named Mediating Intrastate Crises that is focused on uncovering the dynamics of successful mediation efforts during crisis situations at the intrastate level, which are important but understudied phenomena. Our findings indicate that mediators are highly effective at managing crises in the short term, particularly when they adopt a more intrusive approach. However, they have insignificant effects on long-term conflict resolution, showing little ability to stem the tide of recurrent violence.

Keywords

Africa civil crisis ethnic conflict mediation

Introduction

Ethnic violence across many regions of Africa presents the international community with wrenching dilemmas and difficult decisions. Instability is on the rise in the region (Hewitt, 2012), and many African countries have experienced serious outbreaks of violence in the recent past—Southern Sudan, Ethiopia, Somalia, the Democratic Republic of Congo (DRC), Rwanda and Burundi to name just a few. Episodic violence also has high recurrence rates, indicative of the intractability of many of these conflicts.

Often, the conceptions of short- and long-term peace in these situations are at odds with one another. Immediate humanitarian concerns or heightened tensions between the disputants demand a focus on more short-term outcomes. In this case, it is valuable and necessary for the parties to make basic concessions and forge agreements that curb hostilities and de-escalate the situation. However, the ultimate goal is long-term peace, and outcomes that simply provide short-term relief from strain may be insufficient to achieve that larger goal.

What are the best means for promoting short-term crisis management and long-term conflict resolution? Are crisis management and conflict resolution fundamentally different tasks, for which certain approaches to intervention may be more or less effective? This study grapples with these issues by focusing on one particularly difficult set of cases—violent crises that are rooted in ethnic divisions and are part of protracted conflicts in Africa during the post-Cold War era—and one approach to terminating these crises—mediation.

In practical terms, it is important for policymakers to be aware of any differences in stylistic approaches to mediation and in short- vs long-term tasks, and especially of any tradeoffs that exist among these various options. Furthermore, there should be a strong imperative for regional and international communities to help resolve the particular set of cases under analysis in this study.

The data and analyses reported in this study grew out of a new project named Mediating Intrastate Crises (MISC) that is focused on uncovering the dynamics of successful mediation efforts during crisis situations at the intrastate level (see Eralp et al., 2012; Wilkenfeld et al., 2010). The underlying argument examined in this study is that appropriate mediation strategies in intrastate crises can lead to the achievement of formal negotiated agreements—a short-term outcome that we refer to as crisis management—and post-crisis tension reduction among parties—a long-term outcome that we refer to as conflict resolution. Our findings indicate that mediators tend to be successful at managing crises in the short term, particularly when they adopt a more intrusive approach. However, they show little ability to stem recurrent violence in the long term.

Intrastate crisis

While the literature on international politics is rich in both theoretical and empirical studies of interstate crises (Beardsley, 2010; Brecher and Wilkenfeld, 2000), the concept of crisis has rarely been applied systematically at the intrastate level. At this level, crises occur between the state and one or more organized non-state actors.

We have adapted the International Crisis Behavior (ICB) project definition of international (interstate) military-security crisis (Brecher and Wilkenfeld, 2000) for the intrastate crisis context. An intrastate crisis has the following necessary and sufficient conditions: (1) an increase in intensity and/or change in type of disruptive interactions between a state government and one or more opposition organizations, with a higher-than-normal likelihood of violent hostilities that, in turn, (2) destabilizes their relationship, challenges the structure and threatens basic values of the state. An intrastate crisis is triggered when the state actor perceives a threat to one or more basic values, an awareness of finite time for response to the value threat, and a heightened probability of involvement in military hostilities.

The key reason why the subject of crisis is important for conflict studies is that crisis events have significant downstream effects on the costs and benefits of war, serving as potential incubators of high-level hostilities (Lebow, 1981; Slantchev, 2005). Crisis bargaining is almost always marked by reciprocal escalation (James, 2004) owing to audience considerations, military buildups, and at the intrastate level, armed ethnic opposition groups with lower opportunity costs for rebellion. These various factors also increase the probability of war (Fearon and Laitin, 2003; Lai, 2004; Slantchev, 2005; Tarar, 2006).

A crisis-based approach to understanding mediation offers value-added over the existing literature in several ways. First, crises are not necessarily violent.¹ The dominant frame of analysis of internal strife is the armed conflict or civil war, which uses a threshold of violence—typically operationalized in terms of number of battle-deaths—to define fluctuations in the intensity of conflict (Fearon and Laitin, 2003; Gleditsch et al., 2002; Sambanis, 2004). In contrast, a crisis-based data frame allows for research into the prevention of war, and not just its resolution. Mediation efforts may have very different effects if they are conducted when fighting is at a low level, rather than when violence is intense. For instance, in Burundi in 1993, MISC records a crisis triggered by a coup attempt by Tutsi soldiers, whereas the Uppsala Conflict Data Program/Peace Research Institute Oslo (UCDP/PRIO; Harbom and Wallensteen, 2010) and hence the mediation datasets that are built on it—Civil War Mediation (DeRouen et al., 2011) and Managing Low-Intensity Intrastate Conflict (Melander et al., 2009)—record no conflict activity. Intuitively, a coup attempt by soldiers from one of two competing ethnic groups represented a major increase in conflict intensity even if it did not produce a sufficient number of deaths to breach the UCDP/PRIO conflict threshold. This episode, in fact, prompted an Organization of African Unity mediation effort. The MISC approach allows us to highlight a vital period during which conflict prevention, including via mediation, is of particularly high importance.

Second, crises are episodes of particularly high intensity; unlike the usual violence frame, they are not necessarily continuous periods of fighting (although they can be). For example, in Burundi in the period 1991–1992, UCDP/PRIO lists a single, continuous armed conflict from 26 November 1991 to the end of December 1992. Our data, in contrast, identifies two separate events in this period in which rebel Palipehutu forces emerged, fought the government briefly, and retreated. These two events represent clear spikes followed by lulls in the intensity of conflict. Thus, even when they do involve violence, the episodic nature of crisis captures fluctuations in conflict intensity that signal important events in the course of a conflict.

MISC is also concerned with perception as the defining aspect of crisis, which distinguishes it from the Early Conflict Prevention in Ethnic Crises (ECPEC) dataset (Öberg et al., 2009). ECPEC defines a crisis on the basis of the stated goals/threats and actions of ethnic challengers, and it views escalation to violence as one of several possible outcomes of crisis rather than a dynamic dimension of crisis. Like ICB before it, MISC maintains that statements and actions from opponents only cause a crisis when the state perceives them as threatening, and the state ceases to be in crisis only when the three perceptions discussed earlier—threat, time constraint and probability of increased violence—subside. Escalation to violence does not necessarily erase these perceptions, and in fact rarely does so. Unlike ECPEC, then, MISC is able to focus on events of particular importance to the state and to analyze crises across different levels of violence. MISC thus bridges an important gap in the data, between mediation in violent episodes (Regan et al., 2009) and that in crises short of violence (Öberg et al., 2009).

The set of cases under analysis in this study provides an especially severe set of problems that make both peaceful crisis management and post-crisis conflict resolution difficult to achieve. Eralp et al. (2012) elucidate five factors that make violent, ethnic, African, intrastate crises so challenging for mediators: heightened levels of threat and stress, acute information problems, more severe insecurity and lower levels of trust, problems of commitment, and risky behavior. These conditions stem from a potent mixture of time-pressure, heightened stakes and hostility, autocratic or unstable states with low levels of capacity and weak institutions, disadvantaged opposition organizations that tend to articulate zero-sum goals, actors with loss-averse frames and incentives to bluff, a prevalence of spoilers that try to thwart peace processes, and a backdrop of protracted ethnic tension (Birnir, 2007; Colaresi and Thompson, 2005; DeRouen and Goldfinch, 2005; Fearon, 1995; Gurr, 1990; Kahneman and Tversky, 1979; Lebow, 1981; Mishali-Ram, 2006; Russett and Oneal, 2001; Stedman, 1997). Investigating whether or not mediation can succeed under this adverse set of circumstances will provide crucial insight into what type of influence mediators can expect to have on intrastate crises and conflict writ large.

Mediation

Third parties face a daunting task in trying to mediate intrastate crises. Nevertheless, we argue that mediators can be effective when they tailor their intervention approach to overcoming these obstacles. We also argue that certain styles or approaches to mediation will be more successful than others.

Bercovitch et al. (1991: 8) define mediation as “a process of conflict management where disputants seek the assistance of, or accept an offer of help from, an individual, group, state, or organization to settle their conflict or resolve their differences without resorting to physical force or invoking the authority of the law”. For mediation to occur or continue, the disputants must both accept the third party’s involvement.

Mediation occurs with greater frequency in crisis events than during periods of lower conflict intensity (Dixon, 1996). Yet crisis mediation has received scant attention from scholars, particularly in the intrastate context. Only a handful of systematic analyses exist (see Beardsley, 2008; Beardsley et al., 2006; Dixon, 1996; Morgan, 1994; Quinn et al., 2006; Wilkenfeld et al., 2003, 2005), and only one at the intrastate level (see Öberg et al., 2009).

Prior studies of mediation in civil and ethnic wars tend to examine mediation as a singular process. However, Regan and Aydin (2006) contend that the tendency to analyze intrastate mediation incidents in an aggregate fashion is one reason why empirical work has generally produced inconclusive evidence on the effectiveness of this form of intervention. This is unfortunate, given that “the policy community conceives interventions as a more complex undertaking involving a range of alternative strategies, coordination, and sequences” (Regan and Aydin, 2006: 737). We also view mediation as a multidimensional strategic process, and we develop a series of hypotheses linking specific mediation styles to various crisis outcomes.

We adopt what is probably the most common conceptualization of mediation style in the international relations literature, a three-tiered scheme first proposed by Touval and Zartman (1985) and further refined by Bercovitch and associates (see especially Bercovitch and Houston, 1996)—facilitation (i.e. communication), formulation (i.e. procedural) and manipulation (i.e. directive). Each of these styles entails a different level of substantive involvement and dictates specific parameters of mediator behavior. The styles are also often cumulative in nature, with higher forms typically being accompanied by lower forms.

Facilitation encourages communication but is the only style that does not contribute substantively to the negotiation process. The African Union’s provision of formal communication channels between Angola and the Front for the Liberation of the Enclave of Cabinda–Renewed in the summer of 2006 led to the signing of a peace agreement ending the Cabinda separatist conflict.

Formulation entails promoting or coordinating solutions in a non-coercive manner. France, Algeria and Burkina Faso intervened as a team of mediators during the crisis between Niger and Tuareg rebel groups in 1994. By making substantive proposals that protected the territorial integrity of Niger while promoting the development of areas populated by Tuaregs, their formulative mediation led to a comprehensive peace agreement in 1995.

Manipulation involves the mediator using pressure and leverage to direct the parties toward outcomes (Beardsley et al., 2006; Touval and Zartman, 1985). The Central African Republic effectively ended a 1994 crisis between the Chadian government and the southern rebel organization the Committee of National Revival for Peace and Democracy by pressuring both sides to come to a peace agreement and by establishing a monitoring committee made up of international and regional powers.

Crisis outcomes: formal agreement and post-crisis tension reduction

We explore the impact of mediation, particularly the styles reviewed above, on the achievement of formal negotiated agreements in the short term, and on the reduction of tensions between crisis actors in the long term. By examining both the short-term outcomes that terminate escalation periods (i.e. crises) and the longer-term, post-escalation (and in some cases, post-war) stability of relations between parties, we are hoping to arrive at a more extensive understanding of the effect of mediation on conflict and crisis dynamics.

Compromise is an inseparable component of peaceful crisis management and conflict resolution. Concessions help disputing parties move further toward or into their shared bargaining space. Formal negotiated agreement is a type of compromise-based outcome of particular interest. Formal agreements are more binding and contain higher costs for defaulting than informal, tacit or unilateral compromise outcomes. Creating and implementing formal agreements require at least a moderate amount of resources, so disputants will try to avoid the sunk costs brought on by signing an agreement that they project will collapse. Formal agreements also contain punishments for noncompliance, and contractual norms and audience costs legitimize retaliation should the agreement be violated (Fortna, 2003a). Owing to this cost structure, formal agreements are more likely to be reached when their provisions lie within the parties’ overlapping range of preferable outcomes. This array of factors means that formal agreements are typically more sustainable relative to other non-violent outcomes.

Even though achieving formal agreement at crisis termination is important, the best-case scenario would involve a sustained reduction in tensions between the parties following the crisis. However, there are significant threats to durable peace in the set of cases under analysis.

In the immediate aftermath of intrastate crisis events, parties often continue to experience higher-than-normal stress as they emerge from periods of intense hostility. This stress is likely to devolve into a resumption of crisis-level tensions when one or more actors: (a) sign an agreement whose terms are barely above their reservation points; (b) have devious objectives and use the break in tensions to re-arm; (c) experience a change in capabilities after crisis termination; or (d) reassess how the distribution of capabilities between themselves and their opponent(s) matches with the distribution of benefits provided by an agreement (Powell, 1999; Richmond, 1998; Werner, 1999). In each of these cases, if the actor is unsatisfied with the terms of the crisis outcome and believes that its utility will be maximized by returning to the battlefield or brinkmanship, it is likely that peace agreements will break down.

Conversely, crisis agreements are more durable when they lie well above each side’s reservation point and when the parties receive benefits in accordance with their distribution of capabilities. Peace is also more stable when each side credibly commits to ensuring each other’s security (Fortna, 2003a; Hartzell et al., 2001).

Mediation and crisis outcomes

Studies of interstate crises have shown a substantial positive relationship between mediation and the achievement of formal agreement (see Beardsley et al., 2006; Quinn et al., 2006; Wilkenfeld et al., 2005). Mediation has also been linked to durable peace in civil wars (see Kydd, 2003; Kirschner and Von Stein, 2009; Regan and Aydin, 2006; Walter, 1997). We expect similar results for intrastate crises. Mediators help disputing parties overcome the difficulties associated with uncovering mutually acceptable solutions, making concessions, accepting the costs associated with signing formal agreements, and finding lasting solutions to their problems.

H1. Mediated crises are more likely to end in formal negotiated agreements than are unmediated crises.

H2. A reduction in tensions between actors is more likely to follow mediated than unmediated crises.

Even in its most basic form, mediation assists parties in overcoming the myriad information problems that influence crisis bargaining. Facilitative mediators help to open or maintain lines of communication between disputants. Facilitators can also structure the negotiation so that key issues are given sufficient consideration. Information sharing through a facilitative mediation process helps the parties to clarify individual positions, judge each other’s commitment, and make more accurate estimates of their relative power distribution and shared zone of agreement. Facilitators can also help rebuild trust between parties (Fisher, 1972), often eroded by insecurity and ethnic hostility. Increased trust spurs increased commitment. Trust may be especially important for maintaining peace in the long run when workable short-term compromises fall close to one or more parties’ reservation points. Finally, hands-off mediators allow parties to invest their own resources and reputations into an agreement, providing disincentives for reneging at a later point.

Formulative mediators should be more successful than simple facilitators at achieving formal agreements as they highlight and coordinate the parties’ foci on mutually acceptable concessions. If zones of agreement do not exist or at least one party views the issues in zero-sum terms, formulators can try to re-define the issues to alter the substance of the bargaining game. Formulation tactics of suggestion and persuasion are perhaps most valuable during bargaining stalemates, when parties tend to view each other’s proposals as invalid or to equate concessions with capitulation (Schelling, 1960). While previous research at the interstate level has shown that formulative mediators are less effective at ensuring post-crisis tension reduction (Beardsley et al., 2006; Quinn et al., 2006; Wilkenfeld et al., 2005), we expect that mediators that work to achieve mutually preferable solutions will be more necessary in the intrastate context owing to the parties’ tendencies to adopt hardline, zero-sum positions.

Manipulative mediators have demonstrated the greatest success at achieving agreements (Beardsley et al., 2006; Bercovitch and Houston, 1996; Morgan, 1994; Wilkenfeld et al., 2005), including in the intrastate African context (Rothchild, 1997; Zartman, 1989). Manipulators have also been shown to effectively promote long-term peace (Kirschner and Von Stein, 2009; Walter, 1997). Manipulative mediators typically have greater means by which to independently supply and apply information than facilitators (Fey and Ramsay, 2010), and unlike formulators they can aggressively push parties to adopt proposals (Bercovitch, 1997; Kydd, 2003). These tactics should give manipulators an advantage in terms of revealing and exploiting zones of agreement in order to achieve specific outcomes.

The linchpin of successful manipulation, however, is the mediator’s use of leverage, bolstered by its status and resources. This is a tool that can be used by the mediator to “ripen” a situation for resolution (Rubin, 1991). Manipulators can apply or threaten negative sanctions (i.e. “sticks”) to raise the costs or decrease the benefits of continued conflict and induce a “mutually hurting stalemate”. In other cases, manipulative mediators can offer a “way out” of the stalemate (Zartman and Touval, 1996) by applying positive leverage (i.e. “carrots”) to increase the benefits or decrease the costs of cooperation. Carrots include agreeing to pay some of the costs of formal agreements or providing side payments to the parties. Carrots help to stem risky behavior by shifting the parties’ foci from losses to gains (Levy, 1996). By altering the parties’ incentives for fighting and/or cooperation, manipulators can expand, and perhaps even create, zones of agreement (Hopmann, 1996; Wilkenfeld et al., 2003).

Peace has a greater chance of enduring in the post-crisis setting if commitments are subject to enforcement. Manipulative mediators often act as agreement guarantors, either explicitly via the terms of agreements or threats to punish non-compliance, or implicitly through commitments to assist with implementation processes (Fortna, 2004; Touval and Zartman, 1985; Walter, 1997). Agreement guarantee and enforcement mechanisms decrease the costs of maintaining the peace or raise the costs of breaking the peace.

This discussion of the expected effects of mediator styles on formal negotiated agreements and post-crisis tension reduction leads to the following hypotheses:

H3a. Manipulative mediation is more likely to bring about formal agreement than all other mediation styles.

H3b. Of the other two styles, formulative mediation is more likely to bring about formal agreement than facilitation.

H4a. Manipulative mediation is more likely to bring about post-crisis tension reduction between actors than all other mediation styles.

H4b. Of the other two styles, formulative mediation is more likely to bring about post-crisis tension reduction between actors than facilitation.

In some cases, manipulative tactics that do not address security concerns may be insufficient. Vulnerability contributes to a situation where parties are preoccupied with relative gains and fear that their opponent will renege on the terms of security arrangements. As a result, they are unwilling to sign more binding agreements and have strong inclinations to hold out for better future bargains. Even if they do sign agreements, insecurity often remains in the aftermath of high intensity conflict periods—such as violent crises—owing to “the fears, memories, and sunk costs associated with high levels of casualties” (Hartzell et al., 2001: 190). Residual insecurity increases the probability of accidental violence or “trigger-happy” responses to perceived agreement violations.

In this context, manipulative mediators that can generate security guarantees for the parties should be more effective at securing agreements and long-term peace (Doyle and Sambanis, 2006; Walter, 1997; Werner and Yuen, 2005). Security guarantees can take the form of oversight committees for military-related provisions of agreements, or actual peacekeepers (Fortna, 2003a; Smith and Stam, 2003).

Security guarantees serve several purposes. First and foremost, they raise the costs of non-agreement and defection from accords that are signed. Peacekeepers and security monitors help to enforce commitment by acting as referees, assigning blame to parties that violate security arrangements. Sometimes they also act as punishers, threatening to target violators with negative sanctions (Fortna, 2003a). Second, security guarantees decrease the costs and increase the benefits of agreement and commitment by shifting the onus of security responsibilities from the parties themselves to the mediators. Finally, security guarantees can help the parties feel safer in a general sense (Regan and Aydin, 2006) and less fixated on relative gains, which should increase opportunities for compromise.

H5. Manipulative mediators will be most effective at securing formal agreement when they provide or arrange for security guarantees.

H6. Manipulative mediators will be most effective at bringing about post-crisis tension reduction between actors when they provide or arrange for security guarantees.

Methodology

Data

This study extends initial findings by Eralp et al. (2012) using the MISC dataset. MISC is designed to uncover the dynamics of successful crisis mediation at the intrastate level. Although the MISC project aims to be global in scope, it was necessary to begin by collecting data on a specific subset of cases. As alluded to earlier, this “first-cut” subset has been conceptualized as follows:²

1990–2005, that is, post-Cold War;

Africa;

armed conflict according to the Uppsala Conflict Data Program (UCDP, 2010);³

opposition group included in the Minorities at Risk (MAR) dataset (Minorities at Risk Project 2009);

opposition group represented by an organization that qualifies for inclusion in the Minorities at Risk Organizational Behavior dataset (Asal et al., 2008).

This subset covers 116 crises. Table 1 lists the intrastate crises covered in this study, grouped into 24 protracted conflicts.

Table 1.

MISC cases 1990–2005

Protracted conflict	Intrastate crises	Actor-level crises	Crisis dates^a
Angola: Cabinda	6	13	1991–2006
Angola: National Union for the Total Independence of Angola	6	12	1989–2002
Burundi: Hutus–Tutsis II	10	23	1991–2006
Chad: Southerners II	6	12	1992–1998
Djibouti: Afars II	4	8	1991–2000
DRC (Zaire): Union for Democracy and Social Progress (UDPS)	11^b	26^b	1991–2005
DRC (Zaire): Union of Federalists and Independent Republicans (UFERI)	4^b	9^b	1991–1995
DRC (Zaire): Tutsis	4	8	1996–2004
DRC: ex-Zairean armed forces	1	2	2004
Ethiopia: Tigreans	1^c	3^c	1988–1991^c
Ethiopia: Eritreans	1^c	2^c	1990–1991^c
Ethiopia: Oromiya	6^c	12^c	1989–2008^c
Ethiopia: Afars I	2^c	4^c	1989–1997^c
Ethiopia: Amhara	4	8	1990–2004
Ethiopia: Ogaden	6	13	1994–2007
Mali: Azawad	7	16	1990–2001
Niger: Tuareg	3	7	1991–1997
Republic of the Congo: Lari–M’Boshi–Nibolek/Bembe	10	23	1993–2005
Rwanda: Hutus–Tutsis I	7	14	1990–2004
Senegal: Casamance	7	14	1990–2003
Sierra Leone: Temne–Mende	5	12	1991–2000
Somalia: Ogaden	1	4	1988–1991
Sudan: Southerners I	4	9	1989–2003
Sudan: Darfur	4	10	2003–2005
Non-protracted conflict	2^d	4^d	1993–1994
Total	116	259

The dates listed in this table range from the trigger of the first crisis to the termination of the last crisis in the protracted conflict that fit this criterion.

Three intrastate crises and six actor-level crises are shared by the UDPS and UFERI protracted conflicts in DRC (Zaire).

One intrastate crisis and one actor-level crisis are shared by the Tigreans, Eritreans, Oromiya and Afars I protracted conflicts in Ethiopia. This intrastate crisis was triggered by Tigrean rebels in 1988, while Afar and Oromiya rebels entered the crisis in 1989 and Eritreans entered in 1990.

Two intrastate crises—the Tutsi army coup attempt in Burundi in 1993 and the Mbanza–Ngungu army mutiny in Zaire in 1994—and four actor-level crises are not part of any protracted conflict. These crises were included because of their close relation to the Hutus–Tutsis II and UDPS protracted conflicts, respectively.

Overall, the mediation rate for violent, African, ethnic, intrastate crises is 60% for the period 1990–2005, compared with 55% for violent interstate crises for the entire international system during the same period. Intrastate crises terminate in formal agreement 34% of the time and recur at an alarmingly high rate of 74%. Compared with the population as a whole, mediated crises terminate in formal agreement at a somewhat higher rate of 47%, while the crisis recurrence rate is virtually the same (75%). The main objective of this study is to address how different approaches to mediation affect the moderate rates of agreement and high rates of recurrence among intrastate crises.

Dependent variables

We examine the impact of mediation on two types of crisis outcomes, both measured dichotmously. Form of agreement is an indicator of the short-term, termination-point outcome of a crisis, or what we term a crisis management outcome. A formal agreement may be a treaty, armistice or cease-fire between the state and its non-state opponent(s). The second outcome variable measures post-crisis tensions between opponents. We refer to this as a conflict resolution outcome since it indicates the long-term aspects of crises on the overall conflict between parties. It is measured in terms of the presence or absence of a subsequent crisis between the opponents within five years.

Independent variables

We utilize several variables to assess mediation. The first is a dichotomous measure that simply indicates the existence of a mediation effort.

To test our hypotheses about mediation style, we use a variable that identifies the highest style of mediation that occurs during a crisis. “Highest” is defined according to the intensity of the substantive contribution that a mediator makes to the negotiation process. The lowest level is facilitation, followed by formulation, and then manipulation. The cumulative nature of highest mediation style has some implications for our empirical results, particularly in the case of manipulation. Four dummy variables identify the three styles and no mediation.

Finally, we include measures of whether or not manipulative mediators generated security guarantees during the crisis. Security guarantees include such things as peacekeepers or teams tasked with monitoring compliance with ceasefires. Mediators can provide these guarantees during the crisis or arrange for them to occur after crisis termination. Dummy variables indicating manipulative mediation with and without security guarantees are included as substitutes for manipulation as the highest form variable in several models.

Control variables

We also analyze the impact of a set of eight control variables that reflects the strategic environment in which the intrastate crisis takes place. Characteristics of both the individual actors and their interactions are likely to affect crisis outcomes. Table 2 presents summary information and descriptive statistics for all variables.

Table 2.

MISC Africa—descriptive statistics (total number of observations = 116)

Crisis feature	Level of measurement	Source	Statistics
Formal agreement	Dichotomous	MISC	N = 39 (34% of crises)
Tension reduction	Dichotomous	MISC	N = 28 (26% of crises)
Mediation	Dichotomous	MISC	N = 70 (60% of crises)
Highest mediation style	Nominal	MISC
Facilitation			N = 9 (8% of crises; 14% of mediated crises)
Formulation			N = 11 (10% of crises; 17% of mediated crises)
Manipulation			N = 45 (41% of crises; 69% of mediated crises)
Mediator security guarantees	Dichotomous	MISC	N = 35 (32% of crises; 55% of mediated crises)
Mediator goals	Dichotomous	MISC
Crisis management			N = 13 (11% of crises; 19% of mediated crises)
Conflict resolution			N = 56 (49% of crises; 81% of mediated crises)
Severity of violence	Ordinal	MISC	Minimum = 1 (minor clashes), maximum = 3 (full-scale war); mode = 1 (minor clashes)
Group discrimination	Scale, index (political discrimination + economic discrimination)	MAR	Minimum = 0, maximum = 8; mean = 4.66; Standard deviation = 2.61
Group power	Scale, index (lootable resources + external support + troop support)	MAR & MISC	Minimum = 0, maximum = 5; mean = 3.03; standard deviation = 1.59
Regime type (last crisis year)	Scale	Polity	Minimum = −8; maximum = 8; mean = −1.03; standard deviation = 4.19
Group grievances	Ordinal	MAR & MISC	Minimum = 1 (anti-discrimination), maximum = 4 (regime overthrow); mode = 3 (independence)
Crisis duration (in months)	Scale	MISC	Minimum = 0, maximum = 93; mean = 13.85; standard deviation = 17.32
Gravity of threat	Ordinal	MISC	Minimum = 0 (limited military), maximum = 4 (regime); mode = 3 (territorial)

We control for characteristics of both the state and the opposition group represented by the non-state actor. State characteristics include regime type and the gravity of threat that it perceives. Opposition group characteristics include its articulated grievances, the level of discrimination that it experiences, and its power capabilities. Generally speaking, a state that is more authoritarian and more discriminatory in its treatment of the opposition group is less willing to respond to the group’s interests or credibly commit to ensuring the group’s security. In turn, the opposition group will perceive a lower likelihood of altering the status quo through conventional means. Furthermore, a state that perceives a higher level of threat or that is facing an opposition group articulating more radical goals is less willing to concede to the group’s demands. Radical goals also indicate an equal-but-opposite unwillingness to compromise on the part of the opposition group. Finally, the opposition group’s power is an indicator of its bargaining strength.

We also control for two characteristics of the actors’ interactions: crisis duration and severity of violence. On the one hand, longer and more violent crises will increase the costs of conflict, as well as the potential for exhaustion. On the other hand, these two factors may also harden positions and increase commitment problems by decreasing trust, increasing insecurity or increasing perceptions of stalemate.

Finally, we control for mediator goals. If crisis management is the principal goal, we expect that the mediator will be less concerned with achieving a lasting peace. In contrast, loftier conflict resolution goals may inhibit the ability of the mediator to de-escalate the crisis in the short-term.

Models

Several sets of binary logit models are used to test the hypotheses, with one set dedicated to each of the two crisis outcome measures described above. Three variations within each set assess the impact of mediation on crisis outcomes. The first and second variations examine the impacts of mediation as a whole and the highest style of mediation, respectively. The final variation looks at the differential effects of manipulation with and without security guarantees.

Recent studies argue that the observed relationships between long-term conflict resolution and both mediation and the achievement of agreement are mis-stated without considering selection effects (see Beardsley, 2010; Bercovitch and Gartner, 2006; Gartner, 2010; Svensson, 2005; Werner and Yuen, 2005). The idea is that mediation and less durable agreements are selected in the most difficult conflicts, thereby deflating any positive causal effects that these factors may have on durable peace. To assess potential endogeneity, we employed seemingly unrelated, recursive, bivariate probit models (Beardsley, 2008; Gartner, 2010; Svensson, 2005). Mediation incidence and form of agreement appeared as endogenous regressors in separate selection equations, and the outcome of interest was post-crisis tensions.⁴ In each case, p was statistically insignificant, indicating a low likelihood of unobserved factors affecting both processes. In this situation, it is acceptable and practical to treat the two processes as independent and to estimate the equations separately (Greene, 2003). Thus, our logit models of post-crisis tensions generally provide unbiased estimates, and the bivariate probit models go unreported.

Since most of the crises in the MISC dataset occur within protracted conflicts, we also expect there to be some clustering in the data. We attempted to account for clustering, but applicable statistical methods can introduce significant bias into the data analysis when sample sizes are as small as ours (Maas and Hox, 2004).⁵ The results were inconclusive and go unreported. We feel confident in treating the individual crises in the MISC dataset as distinct and self-contained events. First, crises typically involve a single or smaller subset of all issues at stake in a protracted conflict (Brecher and Wilkenfeld, 2000), and in some cases new issues arise between the parties that cause subsequent crises. Second, the same set of actors is not always involved in each crisis. Third parties also change. Patrons on both sides fluctuate over time, altering the balance of power and subsequently the probability of different crisis outcomes. Mediators also change: multiple interventions are often occurring, and the primary mediation team and/or leaders of that team can differ from crisis to crisis.⁶ Finally, time periods between crises are sometimes extended (e.g. 10 years elapsed between the second and third crises in the protracted conflict between Ethiopia and Amharas), adding to the plausibility of treating them as distinct events.

Results

Table 3 presents the results of our analyses for form of agreement and post-crisis tensions. For ease of interpretation, the results are discussed by calculating the predicted probability of observing a “1” on the dependent variable, given a specific value of an independent variable and holding all other variables constant. The predicted probabilities associated with all statistically significant mediation variables are displayed in summary in Table 4.

Table 3.

Logit models—effects of mediation on form of agreement and post-crisis tensions

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8
	Form of agreement				Post-crisis tensions
Mediation	2.67*** (0.74)	—	—	—	−1.00 (0.79)	—	—	—
Facilitation	—	2.41** (1.04)	2.56** (1.08)	1.24 (1.26)	—	−0.97 (1.20)	−0.97 (1.20)	−0.22 (1.34)
Formulation	—	2.23** (0.93)	1.95** (1.01)	0.63 (1.17)	—	−0.97 (1.13)	−0.65 (1.15)	0.10 (1.23)
Manipulation	—	2.87*** (0.80)	—	—	—	−0.88 (0.87)	—	—
Manipulation, no security guarantees	—	—	1.32 (1.09)	—	—	—	−0.75 (1.06)	—
Manipulation, security guarantees	—	—	3.68*** (0.92)	2.36** (1.08)	—	—	−1.11 (0.98)	−0.36 (1.01)
No mediation	—	—	—	−1.32 (1.09)	—	—	—	0.75 (1.06)
Severity of violence	−0.07 (0.41)	−0.17 (0.44)	−0.16 (0.48)	−0.16 (0.48)	2.11*** (0.64)	2.03*** (0.65)	2.13*** (0.67)	2.13*** (0.67)
Group discrimination	0.12 (0.10)	0.08 (0.10)	0.02 (0.11)	0.02 (0.11)	−0.08 (0.11)	−0.10 (0.11)	−0.08 (0.11)	−0.08 (0.11)
Group power	−0.30 (0.21)	−0.27 (0.22)	−0.38 (0.23)	−0.38 (0.23)	−0.67*** (0.26)	−0.64** (0.26)	−0.67** (0.27)	−0.67** (0.27)
Regime type, last year	−0.10* (0.06)	−0.10* (0.06)	−0.05 (0.06)	−0.05 (0.06)	0.22*** (0.07)	0.21*** (0.07)	0.20*** (0.07)	0.20*** (0.07)
Mediation goals	−0.80 (0.71)	−0.74 (0.72)	−0.86 (0.75)	−0.86 (0.75)	0.70 (0.87)	0.64 (0.88)	0.61 (0.90)	0.61 (0.90)
Group grievances	−0.26 (0.25)	−0.27 (0.26)	−0.55* (0.30)	−0.55* (0.30)	0.37 (0.27)	0.36 (0.27)	0.44 (0.30)	0.44 (0.30)
Crisis duration	0.01 (0.01)	0.01 (0.01)	0.02 (0.02)	0.02 (0.02)	0.00 (0.02)	0.00 (0.02)	0.00 (0.02)	0.00 (0.02)
Gravity of threat	0.04 (0.22)	0.03 (0.23)	0.05 (0.23)	0.05 (0.23)	−0.57** (0.25)	−0.56** (0.26)	−0.58** (0.26)	−0.58** (0.26)
Form of agreement					0.88 (0.62)	0.81 (0.62)	0.99 (0.67)	0.99 (0.67)
Constant	−1.61** (0.84)	−1.36 (0.84)	−0.35 (0.96)	0.97 (1.28)	−2.01** (0.96)	−1.90** (0.97)	−2.25** (1.08)	−3.00** (1.48)
N	109	106	105	105	101	99	98	98
Likelihood ratio chi-square	24.99***	24.38***	29.62***	29.62***	22.61**	21.97**	22.48**	22.48**

p ≤ 0.1; ** p ≤ 0.05; *** p ≤ 0.01.

Table 4.

Predicted probabilities—effects of mediation on likelihood of formal agreement

	Model 1	Model 2	Model 3	Model 4
No mediation	0.0853^a	0.0884^a	0.0729^a	—
Mediation	0.5412	—	—	—
Facilitation	—	0.4809	0.4865	—
Formulation	—	0.4433	0.3531	—
Manipulation	—	0.5883	—	—
Manipulation, no security guarantees	—	—	—	0.2381^a
Manipulation, security guarantees	—	—	0.7234	0.7319

Baseline predicted probability for the model and reference category in question, calculated by holding all other binary variables at their modal values and all ordinal and continuous variables at their mean values. All predicted probabilities are computed based on the logit estimates in Table 2. Predicted probabilities are only presented for statistically significant relationships.

Formal agreement

Table 3 shows that all of our statistical models concerning the achievement of formal agreement are highly significant (p ≤ 0.01). Mediation is by far the strongest determinant of whether or not a crisis terminates in formal agreement. Hypothesis 1 is confirmed. Mediated crises are 46% more likely to terminate in formal agreement than unmediated crises (model 1).

We also find considerable evidence to support Hypothesis 3a. All three styles of mediation have a positive and statistically significant effect on the likelihood of formal agreement, with manipulation showing the largest impact of the three (model 2). Crises in which mediators intervene and use manipulative tactics are 50% more likely to end in formal agreement than those that go unmediated. The influence of formulation and facilitation on formal agreement is less powerful than that of manipulation but nevertheless sizeable. The relative effects of these two styles are roughly equal, contrary to Hypothesis 3b.

Finally, we also see strong evidence to support Hypothesis 5. When manipulative mediators provide or arrange for security guarantees, they are 49% more likely to achieve formal agreement (model 4). In fact, manipulators appear to have little success when such guarantees are not part of their efforts (model 3).

Tension reduction

In contrast to its profound effect on formal agreement, Table 3 shows that mediation has no statistically significant effect on post-crisis tensions between parties (model 5). The same holds for the three individual styles of mediation (model 6), as well as mediator-generated security guarantees (models 7 and 8). Thus, we find no evidence to support Hypotheses 2, 4a, b and 6.

We are confident that these non-findings are not attributable to model misspecification or the relatively moderate number of mediated cases that terminated in tension reduction (n = 17). First, the likelihood ratio chi-square test statistics for all post-crisis tensions models indicate sufficient goodness-of-fit (p ≤ 0.05). Second, reduced models were attempted with just the mediation variables in one case and just the control variables that attain statistical significance in another case, and the results were the same as in the full model. Third, there is no difference between mediated cases and the population as a whole in terms of the percentage of crises that led to tension reduction. Instead, the results simply show that non-mediation factors consistently have a more powerful influence on long-term peace.

Crises involving more authoritarian regimes and high-level threats from stronger non-state actors are more likely to recur. These results are consistent with our expectations. Higher levels of violence between crisis actors have a strong effect in reducing the probability of crisis recurrence. This may indicate that exhaustion increases the chances of long-term peace.

Discussion

Our findings regarding the effectiveness of mediation are mixed. There is strong evidence pointing to a major role for mediation in managing violent intrastate ethnic crises in post-Cold War Africa. However, effective crisis management through mediation does not appear to lead to long-term resolution of these conflicts.

We find that not all mediation styles are equally capable of enhancing crisis management. Manipulative mediation is the most effective, achieving formal agreements at far higher rates than less intrusive styles. By using leverage to alter incentive structures, manipulators are most adept at shifting bargaining positions and altering risky behavior to create space for and to sometimes force agreements.

More often than not, however, manipulative mediators need to provide or arrange security guarantees between parties to maximize their probability of achieving agreements. Pressure and sanctions are not always enough to induce cooperation. Security guarantees help to generate commitment to peace in the short term by managing the considerable vulnerability problem experienced by parties in the particular set of cases under analysis.

Relative to manipulative mediation, formulation and facilitation have less powerful but nevertheless significant and positive effects on formal agreement. By suggesting and coordinating mutually acceptable solutions, formulators help to overcome bargaining stalemates and shortsightedness, and to persuade ethnic parties to reconsider zero-sum positions. Unexpectedly, mediators that focus solely on facilitation are equally effective as formulators. In the face of violent, time-pressured situations set against a backdrop of protracted ethnic hostility, sometimes all that is needed is a mediator that can forge basic communication channels, bridge information gaps or help the parties to begin to understand and trust one another. Since the styles of mediation are cumulative to some degree, formulation and facilitation may also be key components of an overall manipulative strategy.

Nevertheless, the direct effects of mediation on long-term peace are negligible (for evidence of this same trend, see also Bercovitch and Gartner, 2006; Richmond, 1998; Werner and Yuen, 2005). We posit two interrelated explanations for this finding: the inability of mediators to resolve long-term commitment issues among the parties and sample features.⁷ We also address the relevance of our non-findings to the relationship between security guarantees and long-term peace for studies of civil conflict in general.

Mediators seem to be able to induce parties to commit to the costs of setting up agreements and compliance checks in the first place, but these commitments are not very credible if they secure only short-term peace. The failure of manipulative mediators to bring about long-term peace is especially significant, since these mediators are most successful at achieving short-term crisis management and conduct almost 90% of post-crisis mediation efforts aimed at promoting long-term conflict resolution.

Mediators face inherent trade-offs between the time-intensive process of conflict resolution and the time-pressured process of de-escalating violent crisis situations. Manipulators attempt to resolve the underlying issues at stake in a large majority of the crises in which they are involved, but the intensity of these situations may also force them to use their leverage in an expedient manner.⁸ In doing so, they can deflate bargaining points without ensuring that the distributions of capabilities and benefits match one another, that the outcomes lie comfortably within the zone of agreement between the parties or that the parties at least perceive the outcomes as fair and reasonable (Beardsley et al., 2006; Favretto, 2009; Werner and Yuen, 2005). In this scenario, the parties’ commitments to the outcomes are precarious: post-crisis re-evaluations or changes in capabilities have a high probability of leading to re-escalation. Initially stronger parties are especially likely to be dissatisfied with mediators that expediently manipulate incentives or alter bargaining spaces. Manipulator moves that shrink power differentials can also embolden initially weaker parties (Aggestam, 2002; Richmond, 1998).

Our findings regarding mediation and long-term peace are also partially a function of our sample. While ICB data indicate that about half of all post-Cold War crises at the interstate level occur within the context of protracted conflicts and roughly one-third contain an ethnic element, virtually all of the African intrastate crises in this study are part of protracted ethnic conflicts. Commitment problems are enhanced in this context owing to extended hostile interactions and issues that are multitudinous, interconnected and often perceived as indivisible. Individual components of the conflict are difficult to resolve without resolving the whole (Colaresi and Thompson, 2005). Parties are also less assuaged by security guarantees (Doyle and Sambanis, 2000) and less invested in divide-the-pie arrangements such as power-sharing (Mishali-Ram, 2006; Favretto, 2009), a problem that plagues African conflicts (Spears, 2000). Resolution of such conflicts tends to be protracted and piecemeal (Zartman, 1989). Less intrusive styles of mediation, while successful in interstate crises, may be less equipped to deal with the severe commitment problems that typify our sample (Bercovitch and Gartner, 2006; Rubin, 1980; Walter, 1997).

On the basis of evidence from prior studies, we strongly expected to find that security guarantees effectively deter a return to the battlefield, even if all other tactics fail. Guarantors attempt to act as “boundaries” between opposing sides in order to decrease their probabilities of winning future battles (Smith and Stam, 2003). However, in our sample, they fail to have a sufficiently adverse effect on such calculations over the long term. To some degree, this finding is also attributable to data-related issues. Many of the cases in our sample involve actors that perceive severe harm and exhibit coherency issues such as factionalism, weak leadership and undisciplined troops. Such conditions increase the likelihood of accidental or spiraling escalation and dampen the ability of security guarantees to perform their function. The long-term success of peacekeeping missions may also vary by the strength and type of mission (Doyle and Sambanis, 2000; Fortna, 2003b; Smith and Stam, 2003).

Nevertheless, our findings raise larger questions about how studies tend to assess long-term peace and the value of security guarantees in the intrastate context. While we are concerned with the resumption of violent crises, most prior studies show that security guarantees effectively prevent a resumption of civil war (Doyle and Sambanis, 2006; Fortna, 2003b; Walter, 1997). However, a lack of war does not equate to peace. Enduring hostilities between groups often remain serious issues even if they do not escalate to the level of war again (e.g. the Republic of Congo experienced five additional, residual violent crises following its civil war). Furthermore, there are a number of long-running, violent protracted conflicts that never reached the level of civil war but still pose significant problems for their societies (e.g. Angola–Cabinda, DRC–UDPS, Niger–Tuareg, etc.). Thus, even if security guarantees do help to prevent additional civil wars, we find that such guarantees are not necessarily successful at deterring civil violence—including serious bouts thereof—in the long term, at least in Africa.

Conclusion

The results of this study indicate that mediation and its component styles have a profoundly positive influence on intrastate crisis outcomes in the short term but little effect over the long term. Even security guarantees—previously found to be a highly reliable mechanism for maintaining peace—fail to prevent crisis recurrence. These findings are confined to a particular problem-heavy set of intrastate crises—those that are violent, ethnic-related and located in post-Cold War Africa.

Although the data on this set of intrastate crises do not show an impact of mediation on long-term conflict resolution, the achievement of agreement in the short term has intrinsic value during a violent intrastate crisis. De-escalation of an intensely stressful, pressurized and hurting situation is an immediate goal for the parties and mediators. Sometimes it means the difference between stemming serious humanitarian tragedies or allowing them to continue. In crises where such pressing issues combine with high degrees of issue indivisibility—frequent in number in the set of cases under analysis—mediators can be effective by setting aside the larger issues at stake for the time being and using their leverage instead to lessen the humanitarian toll (Favretto, 2009). De-escalation also allows the parties breathing room to negotiate the broader issues under less threatening and less pressurized conditions, encouraging deliberation and optimization rather than shortcuts and satisficing. Furthermore, the agreements that terminate these crises often serve as the basis for comprehensive solutions further down the road.

The results of this study indicate that the dominant, negative theme in scholarship on intrastate conflict and mediation is inadequate. Mediators do often fail as long-term conflict resolvers, but they can be highly effective as short-term crisis managers. Practitioners of mediation should also take note of this finding, as lofty goals of permanent conflict termination during all interventions may neither be realistic nor the best use of their talents and resources. Crisis management is a worthwhile and important undertaking, and mediation may be better viewed as a short-term way to deal with highly intense situations rather than the primary means of bringing about long-term solutions. Future studies should further explore the powerful, independent effects that higher levels of violence have on reducing tensions between crisis actors. We also advise a deeper investigation of the post-crisis environment and how the international community can tailor its efforts to the realities thereof in order to more effectively support the process of long-term resolution.

Footnotes

Authors’ note

An earlier version of this paper was presented at the annual meeting of the International Studies Association, San Diego, California, 1–4 April 2012.

Funding

Funding for this project was provided by the Folke Bernadotte Academy under its Program in Conflict Prevention.

Notes

References

Aggestam

(2002) Mediating asymmetrical conflict. Mediterranean Politics 7(1): 69–91.

Asal

Pate

Wilkenfeld

(2008) Minorities at Risk Organizational Behavior Data and Codebook Version 9/2008, http://www.cidcm.umd.edu/mar/data.asp

Beardsley

(2008) Agreement without peace? International mediation and time inconsistency problems. American Journal of Political Science 52(4): 723–740.

Beardsley

(2010) Pain, pressure and political cover: Explaining mediation incidence. Journal of Peace Research 47(4): 395–406.

Beardsley

Quinn

Biswas

Wilkenfeld

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

Bercovitch

(1997) Mediation in international conflict: An overview of theory, a review of practice. In: Zartman

Rasmussen

(eds) Peacemaking in International Conflict: Methods and Techniques. Washington, DC: United States Institute of Peace Press, pp. 125–153.

Bercovitch

Gartner

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

Bercovitch

Houston

(1996) The study of international mediation: Theoretical issues and empirical evidence. In: Bercovitch

(ed.) Resolving International Conflicts: The Theory and Practice of Mediation. Boulder, CO: Lynne Rienner, pp. 11–35.

Bercovitch

Anagnoson

Wille

(1991) Some conceptual issues and empirical trends in the study of successful mediation in international relations. Journal of Peace Research 28(1): 7–17.

10.

Birnir

(2007) Divergence in diversity? The dissimilar effects of cleavages on electoral politics in new democracies. American Journal of Political Science 51(3): 602–619.

11.

Brecher

Wilkenfeld

(2000) A Study of Crisis, 2nd edn. Ann Arbor, MI: University of Michigan Press.

12.

Colaresi

Thompson

(2005) Alliances, arms buildups and recurrent conflict: Testing a steps-to-war model. The Journal of Politics 67(2): 345–364.

13.

DeRouen

Jr Goldfinch

(2005) Putting the numbers to work: implications for violence prevention. Journal of Peace Research 42(1): 27–45.

14.

DeRouen

Jr Bercovitch

Pospieszna

(2011) Introducing the Civil Wars Mediation (CWM) dataset. Journal of Peace Research 48(5): 663–672.

15.

Dixon

(1996) Third-party techniques for preventing conflict escalation and promoting peaceful settlement. International Organization 50(4): 653–681.

16.

Doyle

Sambanis

(2000) International peacebuilding: A theoretical and quantitative analysis. American Political Science Review 94(4): 779–801.

17.

Doyle

Sambanis

(2006) Making War and Building Peace: United Nations Peace Operations. Princeton, NJ: Princeton University Press.

18.

Eralp

Quinn

Wilkenfeld

(2012) Delivering peace: Options for mediators in African intrastate crises. In: Hewitt

Wilkenfeld

Gurr

, with Heldt B (eds) Peace and Conflict 2012. Boulder, CO: Paradigm, pp. 85–98.

19.

Favretto

(2009) Should peacemakers take sides? Major power mediation, coercion, and bias. American Political Science Review 103(2): 248–263.

20.

Fearon

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

21.

Fearon

Laitin

(2003) Ethnicity, insurgency, and civil war. American Political Science Review 97(1): 75–90.

22.

Fey

Ramsay

(2010) When is shuttle diplomacy worth the commute?: Information sharing through mediation. World Politics 62(4): 529–560.

23.

Fisher

(1972) Third party consultation: A method for the study and resolution of conflict. Journal of Conflict Resolution 16(1): 67–94.

24.

Fortna

(2003a) Scraps of paper? Agreements and the durability of peace. International Organization 57(2): 337–372.

25.

Fortna

(2003b) Inside and out: Peacekeeping and the duration of peace after civil and interstate wars. International Studies Review 5(4): 97–114.

26.

Fortna

(2004) Peace Time: Cease-Fire Agreements and the Durability of Peace. Princeton, NJ: Princeton University Press.

27.

Gartner

(2010) Signs of trouble: Regional organization mediation and civil war agreement durability. Paper presented at the Annual Meeting of the Folke Bernadotte Academy Working Group on Conflict Prevention, New York.

28.

Gleditsch

Wallensteen

Eriksson

Sollenberg

Strand

(2002) Armed conflict 1946–2001: A new dataset. Journal of Peace Research 39(5): 615–637.

29.

Greene

(2003) Econometric Analysis, Fifth Edition. Upper Saddle River, NJ: Prentice Hall.

30.

Gurr

(1990) Ethnic warfare and the changing priorities of global security. Mediterranean Quarterly 1(1): 82–98.

31.

Harbom

Wallensteen

(2010) Armed conflicts, 1946–2009. Journal of Peace Research 47(4): 501–509.

32.

Hartzell

Hoddie

Rothchild

(2001) Stabilizing the peace after civil war: An investigation of some key variables. International Organization 55(1): 183–208.

33.

Hewitt

(2012) The peace and conflict instability ledger: Ranking states on future risks. In: Hewitt

Wilkenfeld

Gurr

, with Heldt B (eds) Peace and Conflict 2012. Boulder, CO: Paradigm, pp. 4–17.

34.

Hopmann

(1996) The Negotiation Process and the Resolution of International Conflicts. Columbia, SC: University of South Carolina Press.

35.

James

(2004) Systemism: Social mechanisms, and scientific progress: A case study of the International Crisis Behavior Project. Philosophy of the Social Sciences 34(3): 352–370.

36.

Kahneman

Tversky

(1979) Prospect theory: An analysis of decision under risk. Econometrica 47(2): 263–292.

37.

Kirschner

Stein

(2009) The pieces of peacemaking: Understanding implementation of civil war settlements. Civil Wars 11(3): 279–301.

38.

Kydd

(2003) Which side are you on?: Bias, credibility, and mediation. American Journal of Political Science 47(4): 597–611.

39.

Lai

(2004) The effects of different types of military mobilization on the outcome of international crises. Journal of Conflict Resolution 48(2): 211–229.

40.

Lebow

(1981) Between Peace and War: The Nature of International Crisis. Baltimore, MD: Johns Hopkins University Press.

41.

Levy

(1996) Loss aversion, framing, and bargaining: The implications of prospect theory for international conflict. International Political Science Review 17(2): 179–195.

42.

Maas

CJM

Hox

(2004) Robustness issues in multilevel regression analysis. Statistica Neerlandica 58(2): 127–137.

43.

Melander

Möller

Öberg

(2009) Managing intrasate low-intensity armed conflict 1993–2004: A new dataset. International Interactions 35(1): 58–85.

44.

Minorities at Risk Project (2009) Minorities at Risk Dataset. College Park, MD: Center for International Development and Conflict Management, http://www.cidcm.umd.edu/mar/ (accessed 16 November 2010).

45.

Mishali-Ram

(2006) Ethnic diversity, issues, and international crisis dynamics, 1918–2002. Journal of Peace Research 43(5): 583–600.

46.

Morgan

(1994) Untying the Knot of War: A Bargaining Theory of International Crises. Ann Arbor, MI: University of Michigan Press.

47.

Öberg

Moller

Wallensteen

(2009) Early conflict prevention in ethnic crises, 1990–1998: A new dataset. Conflict Management and Peace Science 26(1): 67–91.

48.

Powell

(1999) In the Shadow of Power. Princeton, NJ: Princeton University Press.

49.

Quinn

Wilkenfeld

Smarick

Asal

(2006) Power play: Mediation in symmetric and asymmetric international crises. International Interactions 32(4): 441–470.

50.

Regan

Aydin

(2006) Diplomacy and other forms of intervention in civil wars. Journal of Conflict Resolution 50(5): 736–756.

51.

Regan

Frank

Aydin

(2009) Diplomatic interventions and civil war: A new dataset. Journal of Peace Research 46(1): 135–146.

52.

Richmond

(1998) Devious objectives and the disputants’ view of international mediation: A theoretical framework. Journal of Peace Research 35(6): 707–722.

53.

Roodman

(2009) Estimating fully observed recursive mixed-process models with cmp. Center for Global Development, Working Paper no. 168.

54.

Rothchild

(1997) Managing Ethnic Conflict in Africa: Pressures and Incentives for Cooperation. Washington, DC: Brookings Institution Press.

55.

Rubin

(1980) Experimental research on third-party intervention in conflict: Toward some generalizations. Psychological Bulletin 87(2): 379–391.

56.

Rubin

(1991) The timing of ripeness and the ripeness of timing. In: Kriesberg

Thorson

(eds) Timing the De-escalation of International Conflicts. Syracuse, NY: Syracuse University Press, pp. 237–246.

57.

Russett

Oneal

(2001) Triangulating Peace: Democracy, Interdependence, and International Organizations. New York: Norton.

58.

Sambanis

(2004) What is civil war? Conceptual and empirical complexities of an operational definition. Journal of Conflict Resolution 48(6): 814–858.

59.

Schelling

(1960) The Strategy of Conflict. Cambridge, MA: President and Fellows of Harvard College.

60.

Slantchev

(2005) Military coercion in interstate crises. American Political Science Review 99(4): 533–547.

61.

Smith

Stam

(2003) Mediation and peacekeeping in a random walk model of civil and interstate war. International Studies Review 5(4): 115–135.

62.

Spears

(2000) Understanding inclusive peace agreements in Africa: The problems of sharing power. Third World Quarterly 21(1): 105–118.

63.

Stedman

(1997) Spoiler problems in peace processes. International Security 22(2): 5–53.

64.

Svensson

(2005) Do mediators go where they are needed the most? Selective mediators in internal armed conflict. Paper presented at the Annual Meeting of the American Political Science Association, Washington, DC.

65.

Svensson

(2011) Crowded with conciliators: Exploring multiparty mediation in civil wars. Peace and Policy, 16.

66.

Tarar

(2006) Diversionary incentives and the bargaining approach to war. International Studies Quarterly 50(1): 169–188.

67.

Touval

Zartman

(1985) Introduction: Mediation in theory. In: Touval

Zartman

(eds) International Mediation in Theory and Practice. Boulder, CO: Westview Press, pp. 7–17.

68.

Walter

(1997) The critical barrier to civil war settlement. International Organization 51(3): 335– 564.

69.

Werner

(1999) The precarious nature of peace: Resolving the issues, enforcing the settlement, and renegotiating. American Journal of Political Science 43(3): 912–933.

70.

Werner

Yuen

(2005) Making and keeping peace. International Organization 59(2): 261–292.

71.

Wilkenfeld

Young

Asal

Quinn

(2003) Mediating international crises: Cross-national and experimental perspectives. Journal of Conflict Resolution 47(3): 279–301.

72.

Wilkenfeld

Young

Quinn

Asal

(2005) Mediating International Crises. London: Routledge.

73.

Wilkenfeld

Brecher

Asal

Quinn

Eralp

McLauchlin

(2010) Mediating intrastate crises: Africa database. Center for International Development and Conflict Management, Working Paper.

74.

Zartman

(1989) Ripe for Resolution: Conflict and Intervention in Africa, updated edn. New York: Oxford University Press.

75.

Zartman

Touval

(1996) International mediation in the Post-Cold War era. In Crocker

Hampson

Aall

(eds) Managing Global Chaos: Sources of and Responses to International Conflict. Washington, DC: United States Institute of Peace Press, pp. 445–461.