Rebel group legitimacy, ideology and durable peace

Abstract

Purpose

This study aims to explore the role of rebel group legitimacy and ideology in durable peace (DP) following peace agreements to end civil wars. It builds on earlier research showing that justice and civil society involvement are critical in achieving DP. This study adds the impacts of rebel group activities and support on DP. Activities include service delivery and mobilization. Support is gauged with outcomes of presidential and parliamentary elections held following peace agreements.

Design/methodology/approach

Five data sets were used to measure the key variables: DP, inclusive commissions (IC), legitimacy symmetry (electoral outcomes), service delivery and ideological mobilization. A measure of rebel group integration in the political system was also constructed. Impacts of the integration, legitimacy and ideology variables were assessed with a hierarchical regression model (HRM). This study begins with a base model drawn from earlier research showing the key predictors were procedural justice (PJ) and IC. The authors ask about the extent to which the rebel group variables contribute additional variance to the prediction of DP.

Findings

The main contributors to the prediction of DP were PJ, IC and integration in the political system. None of the legitimacy or mobilization variables added significant variance to the prediction. Only one of the mobilization variables, forced recruitment, was significant. The decision to integrate into the political system following the agreement did not mediate the relationship between PJ in the negotiation process and DP. Results of a factor analysis showed that DP, PJ, IC and integration formed a cluster with strong loadings on the first factor.

Research limitations/implications

The negative results for the legitimacy and mobilization variables may not be the last word on rebel group influences. Lack of support for the key hypotheses spurs attempts to discover other sources that contribute to the survival of rebel group actors in the political system and, in turn, to DP.

Practical implications

The issues raised by this study contribute to debates about ways to attain peaceful relations among competing groups following a civil war. It appears that attention to factors inside and around the negotiation process (PJ, ICs and conversion) may be more important than rebel group activities outside of these processes. The results call attention, in particular, to the important role played by political integration. From a policy perspective, it would be useful to develop levers for encouraging rebel groups to emerge as political actors in the post-agreement environment.

Originality/value

Developing measures of the symmetry of rebel group legitimacy and integration in the context of a comparative case study are the primary original contributions of this study. Furthermore, the mode of analysis (HRM) is novel in this literature. This approach builds on and extends the earlier research on factors influencing DP.

Citation

Druckman, D., Vuković, S. and Verbeek, N. (2024), "Rebel group legitimacy, ideology and durable peace", International Journal of Conflict Management, Vol. 35 No. 1, pp. 215-241. https://doi.org/10.1108/IJCMA-01-2023-0015

Introduction

In this study, we focus on factors that influence durable societal peace (DP). Recent research has provided insights into the roles played by justice during the negotiation process (Druckman and Wagner, 2019) and civil society participation around that process (Cuhadar and Druckman, 2023). We found that adherence to principles of procedural justice (PJ) within the negotiation sets in motion a process that culminates in long-term peace. We also discovered that a particular kind of civil society activity, referred to as inclusive commissions (IC), increased the chances that peace will occur during an eight-year period following the agreement. We probe further in this study by introducing other variables that capture activities outside the negotiation process and are hypothesized to influence DP. These consist of a cluster of rebel group performances and activities that occurred during the pre- and post-agreement phases of a peace process. They include rebel group electoral performance, pre-agreement service delivery and ideological mobilization. This sequence of analyses is construed as adding complexity to the prediction of DP: starting with justice inside the negotiation, we moved to civil society activities around the process and then, in this study, to activities of one of the conflict actors. The increased complexity is evaluated with a hierarchical regression model (HRM).

Continuity and complexity are two themes that guide our research on DP. Each study builds on the previous work. The next study in the sequence adds layers of complexity to our understanding of factors that influence DP [1]. The sections to follow evince these themes. We begin with a discussion of DP, followed by how PJ and IC were defined in the earlier studies. We then move to the conceptual underpinnings of our new probe. Importantly, we discuss the role played by rebel groups in the peace process, drawing on a large number of earlier studies. Then, we craft sections on the role played by conversion to political parties, the concept of legitimacy in a rebel group context, and activities involved in ideological mobilization. Next, we describe how the variables were defined, including the data sets used to measure them. The HRM model is described, followed by the results and discussion. Before concluding, we discuss several innovative contributions made by the study along with policy implications. In that discussion, we also raise cautions about some limitations.

Defining durable peace

Participation by both conflicting parties is a core governance component that contributes to DP (Cousens et al., 2001; Matanock, 2017). Earlier research has identified four components of DP: the extent of reconciliation, improvement of security and governing institutions and economic development (Wagner and Druckman, 2017).

Reconciliation is an immediate consequence of a peace agreement. It consists of activities that seek to redress past offences to transform adversarial relations. Peace processes encourage parties to come to terms with past offences through different means, including efforts to address crimes and establish truth commissions (Hayner, 2011). Peace agreements also contain provisions for reception and reintegration, improving community relations and resettlement of refugees (Balcells and Justino, 2014). The growing field of transitional justice focuses on these processes as mechanisms for making progress towards societal peace and reconciliation. Whether they are effective remains an empirical question raised by Thoms et al. (2010) and Millar (2011). These authors call attention to culturally contingent effects and backsliding due to the provocative arousal of refreshing wounds suffered during the war.

Institutional changes reflect longer-term consequences. Democracies and capitalist economics, “depend on public institutions to uphold rules, to maintain order, to resolve disputes impartially, and to regulate behavior incompatible with the preservation of market democracy itself” (Paris, 2004, 205). Three types of institution-building challenges in a peace process are typical in the areas of security, governance and economic institutions. Security reform consists of changes in military and police organizations, demobilization and disarmament processes, a decrease in violent crime and holding of the negotiated cease-fire (Diehl and Druckman, 2010). Governance reform includes power-sharing arrangements, establishment of the rule of law and legal reform, electoral reform and human rights protection (Cousens et al., 2001). The economic component of institution building includes progress towards economic stability or growth, including measures of income equality, changing living standards, support programs from lending institutions and, generally, the pursuit of conditions and policies for sustainable economic growth (O’Reilly, 2014). Without institutional change, reconciliation is likely to be threatened. Together, these components of DP provide a multidimensional concept that captures transitions in a conflict and the way a peace process gathers momentum or dissolves over time following an agreement.

A second measure of durability used in the Druckman and Wagner’s (2019) study is referred to as stable agreements. It was measured over a five-year period and consisted of a three-step scale: agreement remains in force through this period; agreement remains in force with some conflict events occurring; and the agreement was abrogated during the five-year period. This variable was used as a DV to supplement the DP analyses.

This is the third study that focuses on DP. The previous work documented the impacts of justice and civil society variables on DP. Building on these findings, we extend the range of factors that can influence peace. But we also have the advantage of continuity by retaining a statistical model that enables the cumulation of findings. We ask whether the additional variables strengthen the prediction of durability. In the next section, we discuss the previous findings, referred to as the base model. The sections to follow provide the conceptual underpinnings for the new variables measured in this study.

Procedural justice and civil society

Justice during the peace negotiation process and inclusive civil society commissions were shown to improve the prospects for DP in the post-agreement period. The Druckman and Wagner’s (2019) study pinpointed the important role played by PJ, consisting of four parts: fair treatment, transparency, representation or voice and voluntary decision-making. These parts have in common a focus on building or hindering the development of relationships between negotiating teams (Lind and Tyler, 1988). The Druckman–Wagner findings showed that those aspects of PJ referred to by Hollander-Blumoff (2017) as social conduct, namely, courtesy (fair treatment) and respect (transparency) from one negotiating delegation to another, were particularly strong predictors of DP. A double statistical mediation modelling also showed that PJ initiates a process leading to distributive justice (fair agreements) in the outcome, stable agreements and DP. We refer to PJ as part of a base model for evaluating the effects of other variables on DP.

The role played by civil society groups around the negotiating table was also shown to influence DP. Using an HRM, Cuhadar and Druckman (2023) found that the key predictor of DP was ICs, one of seven civil society modalities. The HRM results showed that the ICs added significant variation to predictability based on PJ. Thus, we learned that DP is influenced by processes that occur both inside and outside the negotiation. Continuing in this vein, we ask whether another type of variable, referred to as rebel group activities, improves the predictability to the model based on PJ and ICs. By adding these variables to the model, we increase our probe into the complexity of intra-societal conflicts and attempts to resolve them through peace negotiations. Taken together, PJ and IC are considered as the base model from which we build further components of variation.

Why study rebel groups?

We extend the DP model in the direction of rebel group activities. This focus moves this stream of research from what happens inside and around the negotiation to the broader conflict, which may be considered as the context of peace negotiations and implementation. It also shifts our attention from dyadic (regime and rebels) to monadic (the rebel group side) analyses. It can be argued that, as agents of social change, rebel group demands are pivotal in the escalation or de-escalation of the conflicts. As such, they remain an indispensable actor for any conflict management activity, including peace-making processes that aim to yield a negotiated settlement. Notwithstanding the political risks of including rebel groups into a peace process, the choice of excluding them from such efforts carries even greater risks of perpetuating their spoiling effects and undermining prospects of reaching a peaceful end to conflict (Stedman, 1997). In fact, rebel groups that get a seat at the table are more likely to commit to the implementation of what has been agreed, even in the face of potential spoiling effects of those excluded from the process (Nilsson, 2008). This is also the case for the government.

Rebel group activities are also critical during the post-agreement transitions from war to peace. To overcome the security dilemma faced by rebel groups in the post-agreement phase, the implementation of the peace agreements is most often complemented with specific provisions that offer security guarantees, monitoring activities and implementation assistance by third parties (Walter, 1999; Fortna, 2003; Westendorf, 2015). The most challenging task in this phase is to achieve a meaningful demobilization and disarmament of rebel groups as a way of eliminating the choice to relapse into violence (Spear, 2002; Schulhofer-Wohl and Sambanis, 2010). At the same time, the strength of the peace agreements is often linked to the provisions that manage to adequately address the underlying grievances that instigated the rebellion and provide power-sharing mechanisms that would grant access to decision-making roles to former rebel groups (Sisk, 1996; Hampson, 1996; Hartzell and Hoddie, 2015).

The emerging scholarship on rebel-to-party transition indicates that electoral participation provisions in negotiated settlements provide a creative way of mitigating the commitment problems, as they offer a unique anti-spoiling incentive for former rebel groups and an entry point for external actors to effectively monitor the implementation of the peace agreement (Matanock, 2018). This would explain why some rebel groups opt to convert into a political party at the onset of the implementation phase and that might affect the durability of peace achieved. Indeed, a majority of the cases in our data set, discussed below, consisted of rebel conversion immediately following the agreement. Consequently, their pre-agreement efforts at mobilizing supporters and providing services would seem to contribute to their viability as political actors in the post-conflict environment, which, in turn, influences long-term reconciliation and institutional change as key aspects of DP. In this study, we ask first about the relationship between rebel group transitions and DP. We then probe for explanations of this relationship by analysing rebel group legitimacy and ideological mobilization, which capture both the demand (electoral performance) and supply sides of rebel group activities. Both sides of activities are evaluated at the same level of analysis, referred to as meso or communal. It is located between the micro (PJ in the peace negotiations) and macro (durable societal peace) levels in the sequential model explored by Druckman and Wagner (2019). The scope of the study consists of connecting processes at these levels of analysis by asking how group activities influence such societal processes as changes in security and governance institutions. This kind of cross-level framing is at the heart of the study of roles played by social-psychological factors in international relations (Druckman, 1994; Druckman, 2003).

Integration into the political system

The first question is what happens to the rebel groups just after the peace agreement. Some groups become political parties and compete in elections. Others become integrated into the political system without establishing identities as parties. And still others retain their identities as rebels and remain outside the system. A second question is whether rebel-to-party transitions or other forms of integration in the system promote DP. The primary focus in the post-civil war literature has been on rebel-to-party transformation, usually in the form of provisions in peace agreements.

As Zaks (2023) noted, rebel groups may outright fail to transition into political parties despite their expressed desire for transition and concurrent legal opportunity to achieve such status. Similarly, she noted that the attempt to transition might result in only “nominal participation”, where rebel groups “register as parties and sometimes appear on a ballot, but fail to win any seats” (2023, p. 6). A second possible outcome is “seated participation”, where rebel groups manage to win at least some seats in post-conflict elections. This outcome presents a “more comprehensive scope of functions that accompany political integration […] and facilitates testing the mechanism by which rebel integration promotes peace” (2023, p. 6). Thus, rebel-to-party transition is only one form of post-agreement integration into the political system.

When rebel groups take part in elections, they have an enticing opportunity to seize power and decision-making authority through a process that grants them much-needed societal legitimacy (Söderberg Kovacs and Hatz, 2016). Pre-conditions for a successful rebel to party transformation include pre-conflict political dynamics, the conflict characteristics and the manner in which the conflict ended (Manning and Smith, 2016; Lyons, 2016). Post-settlement effects of rebel group inclusion in governmental politics have been shown to impact on the likelihood of post-settlement peace, measured in terms of battlefield deaths in the period following the agreement (Marshall and Ishiyama, 2016).

In the post-settlement environment rebel-to-party transition is neither automatic nor unhampered. In practice, these provisions are frequently met with strong domestic and international resistance and criticism, as they are believed to grant amnesty and impunity to perpetrators of violence during conflict (Tellez, 2019). Yet, bringing former rebel groups into “normal politics” represents an indispensable step in achieving long-term stability (Marshall and Ishiyama, 2016; Matanock, 2018; Manning et al., 2022; Daly, 2021). Manning et al. found that in the post-Cold War era between 1990 and 2021, when rebel-to-party provisions became a prominent feature of peace settlements, there were only “13 instances in which a party had returned to conflict prior to a scheduled election, out of 392 elections” (Manning et al., 2022, p. 6). Thus, for peace-brokers and policymakers, the dilemma of allowing participation rests on their ability to articulate provisions that withstand short-term resistance and criticism for the sake of long-term objectives. As Ishiyama and Batta argue, “durable peace settlements require the active involvement and cooperative engagement of these political groups” (2011, p. 6). As such, we expect that rebel to party transition, where former rebels continue to pursue their political goals through elections, will have a direct impact on DP.

The literature reviewed suggests that there are several forms of post-agreement transition taken by rebel agreements: rebel to party transition, incorporation in the political system and remaining outside the political system. These decisions are also shown to affect the duration of peace as measured by battle deaths following the agreement. These earlier findings suggest a first hypothesis:

H1.

Durable peace is stronger when rebel groups transition to political parties or are integrated into the political process following the peace agreement.

In our study, we take into account the various forms of post-agreement decisions. We also use a broader and deeper measure of DP that includes reconciliation and institutional change. But, we take our study a step further by considering mediating processes. The Druckman and Wagner (2019) double mediation analysis shows a sequence that travels from PJ in the process to DP in society. Building on this finding and on H1, we suggest that PJ during the negotiation process predicts rebel party transition or integration in the political process. Taking this prediction a step further, we also suggest that these transitions mediate the relationship between PJ and DP. This is a time sequence where PJ in the negotiation process proceeds the transition outcome, which, in turn, results in a more DP. These predictions take the form of two hypotheses.

H1a.

Higher levels of PJ during the negotiation process make rebel group political integration more likely.

H1b.

Rebel group political integration explains the relationship between PJ and DP.

Legitimacy

Scholarship on rebel group legitimacy is an under-theorized area of study. A growing interest in the conceptual underpinnings of rebel groups’ legitimacy faces several challenges associated with the task of transposing a concept that is linked with state institutions to non-state actors. From a normative standpoint, the prevailing approach treats states as legitimate, whereas rebel groups are primarily defined as pathological (Della Porta, 2004; Podder, 2014). Similar to state institutions, the politics of legitimacy remains central to the activities of rebel groups with long-term goals of survival, irrespective of whether these goals are personal, ideological, political or socio-economic (Schlichte and Schneckener, 2015; Jo, 2015). As is the case with many other social actors, rebel groups face continued domestic and international pressure to rationalize and vindicate their methods and goals; their strategies and tactics are contingent on their ability to attract and maintain wider support for their actions. As Schlichte and Schneckener argue, legitimacy remains the key variable that explains rebel groups long-term political success, which is exemplified by their “ability to gain political power and maintain such a position over a significant period of time”, implying a transfer of “(military) control into some sort of institutionalized power” (2015, p. 410). In fact, the transition from winning on the battlefield to winning in the post-conflict political landscape presupposes, at a minimum, continued popular support for the rebel group (Kalyvas, 2006). An attempt is made in this study to explore the relationship between legitimacy and DP.

One aspect of legitimacy explored in this study is service provision. Popular support for rebel governance may depend on the provision of basic services such as access to water, electricity, means of communication, education, justice, public health services, welfare services, infrastructural projects, policing and religious services (Grynkewich, 2008; Wagstaff and Jung, 2020; Asal et al., 2022). As Podder points out, “service provision by armed groups generates output or performance legitimacy among the decision to domestic constituency” (2017, p. 690). Similarly, Mampilly notes:

It is only by replicating some of the functions and forms of the nation-state […] that will allow an insurgent organization to derive attitudinal support for its political authority and achieve some form of legitimacy (2011, p. 8).

Service delivery is the first aspect of legitimacy that we explore in this study.

Another aspect is electoral performance. In post-conflict societies, elections represent one of the most fundamental processes in building political institutions and political legitimacy (Reilly, 2002; Sisk, 2013). As Höglund, Jarstad and Kovacs underscore, “ideally, post-war elections may serve to install a legitimate, democratic government, may contribute to the consolidation of a peace agreement, and may promote reconciliation” (2009, p. 534). Yet, there are several obstacles to achieving such legitimacy, including the persistence of incentive structures that promote reliance on violence for achieving political goals (Paris, 2004; Steenkamp, 2005), weak or non-existent electoral participation provisions that increase uncertainty about compliance and cause commitment problems for parties to accept the results (Matanock, 2017), as well as a rushed or poorly timed run to the ballots (Paris, 2010). Notwithstanding these impediments, post-war elections represent an indispensable mechanism for effective state-building of war-torn societies, as they provide a baseline for inclusion, representation, accountability and a fair distribution of power. As such, elections remain a “critical mechanism to legitimize new leaders and institutions following state failure and civil war” (Lyons, 2004, p. 273). Electoral performance is the second aspect of legitimacy that we explore in this study [2].

Service provision

The provision of public goods by rebel groups is linked to their ambition to contest, erode and, if possible, depose the authority of the state. By providing basic services, rebel groups project the ambition to present themselves as a viable alternative to the incumbent regime, capable of performing all activities that can legitimize their authority in the eyes of populations that are subject to their rule (Stewart, 2018). In other words, “the provision of social services can help improve nonstate actor’s own legitimacy while also helping it to challenge the legitimacy of the state” (Asal et al., 2022, p. 841). Moreover, a rebel group’s legitimacy is constructed against the backdrop of historical experiences of civilians living under weak rule of law and bad governance by previous state authorities, and this effect is strengthened when the rebel groups provide fundamental services, curb corruption and improve the rule of law (Revkin, 2021).

The costs of service provision can be justified by utilitarian motives, as rebel groups may use service provision to make the population reliant on them. This, in turn, provides unique opportunities for the rebel groups to mobilize new recruits into their ranks, who are incentivized by additional privileges that come from joining the rebellion (Berman et al., 2011; Kostelnik and Skarbek, 2013). At the same time, the provision of services that carry a symbolic imprint may facilitate the process of identifying the most devoted and dependable recruits (Berman and Laitin, 2008; Berman, 2011). Thus, in addition to utilitarian motives, rebel groups may justify their service provision as a direct manifestation and materialization of their ideological convictions (Davis and Robinson, 2012; Keister and Slantchev, 2014).

Rebel group’s war-time service provision may have an impact on a range of social aspects in the post-conflict phase. It is directly correlated with post-conflict democratization processes (Huang, 2016), post-conflict rule of law (Loyle, 2020) as well as post-conflict approaches in formulating appropriate health-care systems (Ghobarah et al., 2003). In cases where rebel governance included electoral processes, these practices facilitated and simplified the transition to post-conflict elections (Cunningham et al., 2021). Moreover, state-building practices face less resistance and are more streamlined in cases where structures of rebel governance can be co-opted (Loyle et al., 2023). The discussion in this section leads to the following hypothesis:

H2.

The more services provided by rebel groups, the higher the durable peace.

In their study of 27 countries that experienced civil war, which comprised of 42 rebel groups that participated in elections between 1996 and 2013, Ishiyama and Wiedmeier found that:

[n]ot only is rebel governance not related to the measure of electoral success, but the sign of the relationship is negative. Thus, it makes little difference whether the rebel group had developed a “counter state” in explaining the political success of the rebel party later (2020, p. 57).

Yet, the relationship between electoral performance, regarded as the demand side of legitimacy, and service provision as a supply side feature remains largely unexplored (Florea and Malejacq, 2018; Loyle et al., 2022; Teiner, 2022). This question is prompted by Shesterinina’s (2022) argument about the way that legitimacy is shaped by service provision during conflict and manifested through electoral support enjoyed in the post-agreement phase. This argument also has implications for the link between electoral performance and DP.

Electoral performance

Daly’s (2021) study of post-conflict electoral outcomes shows that in the period from 1970 to 2015, on average, incumbent successor parties (i.e. governments) have gained 41.8% of votes compared to rebel successor parties that on average reached 25.6% of votes. These results reaffirm the basic assumption that incumbent successor parties enjoy significant structural advantages through their control of the state apparatus and their expertise in governing (Daly, 2019). Manning and Smith (2016) study of rebel groups’ electoral performance in the post–Cold War era between 1990 and 2016, when rebel-to-party provisions became a prominent feature in peace settlements, found significant variation in outcomes: in 28% of elections, former rebel groups won at least one seat but no more than 10% of the votes, in 15% of elections these parties won between 10 and 30% of the vote, in 14% of elections the results ranged between 30 and 60% of the vote and in 9% these parties won more than 60% of the vote. Daly’s findings also suggest a strong path-dependency between the first and second elections in the post-settlement phase, with a correlation of 0.7, indicating that “the first postwar elections set the political trajectory of post-conflict countries” (2021, p. 843).

The electoral performance of former rebel groups, viewed from the standpoint of legitimacy, may be considered in relation to the governing regime. Their relative legitimacy may be more or less symmetrical. It is more symmetrical when there is a balance of public support for the two sides, as each is able to claim authority over a specific polity and govern it accordingly. This balance may consist of an alteration of power across different electoral cycles, as well as a distribution of power that sees one side dominating the central (national) level and the other a significant portion of local (municipal) level polities. Both sides assume governing roles and project their legitimacy through the performance and delivery of basic governing duties. It is less symmetrical when election results reaffirm the political dominance of the regime or when the rebels make significant gains in popularity, overtaking the initial advantages of governing the population.

Participation by all conflicting parties is a core governance component that contributes to DP (Cousens et al., 2001; Matanock, 2017). For this measure, Druckman and Wagner (2019), electoral performance was treated as one of four parts of the governance component of DP. Less attention was paid to the implications for differences between the government and rebels in legitimacy. An attempt is made in this study to explore the relationship between electoral outcomes as indicators of post-agreement legitimacy and DP. This exploration extends the earlier work by probing more deeply into this aspect of governance.

The idea of symmetry was raised and explored by Druckman and Green (1995) in the context of a conflict between the Philippines Aquino administration and the National Democratic Front (NDF) insurgency. Using a performance-based indicator of legitimacy, they examined the way that changes in relative legitimacy between the regime and rebel groups influenced decisions to negotiate. Increases in legitimacy for both actors encouraged them to seek a peaceful resolution of the conflict. Interestingly, negotiation was also the choice when the NDF’s legitimacy increased while the regime’s legitimacy decreased. These results suggest that relative legitimacy is an important incentive for seeking a peaceful resolution of the conflict. Building on the earlier study, we devise a measure of relative legitimacy in the context of DP. We go beyond the short-term decisions to engage in negotiations by assessing long-term societal peace. We also go beyond an in-depth case study by assembling a multiple-case data set and performing statistical analyses. The discussion above leads to the following hypothesis:

H3.

The more similar shares of the votes obtained by the successor parties of the rebel group and the incumbent regime, the higher the durable peace.

Ideological appeals and durable peace

A question arises about how to define ideology in the context of this study on DP. One approach is to rely on categories that depict the groups’ political doctrines, such as left-wing, right-wing, anti-government, ethnonationalist, Islamist and/or other religious types (Polo and Gleditsch, 2016; Wood and Thomas, 2017; Sarwari, 2021; Asal et al., 2021). These categories are useful for capturing political incompatibility with the incumbent government’s official policies. They do not help in understanding how rebel groups mobilize support and recruit new members with messages that resonate with critical target audiences.

The way groups use ideological appeals to recruit and mobilize support complements the way material resources are used in delivering services. These appeals may include invoking underlying grievances against the incumbent regime, highlighting the oppressive nature of the government’s policies, using claims that underscore the pervasiveness and injustice of extractive practices and depriving populations of resources and opportunities in a given system (Weinstein, 2005; Humphreys and Weinstein, 2008; Thaler, 2022; Soules, 2023). However, it may also be the case that these appeals are insufficient to attract the hoped-for support. If so, rebel groups may rely on more assertive and forceful practices of recruitment (Eck, 2014; Sawyer and Andrews, 2020).

The discussion above raises questions about the relationship between ideological mobilization and DP. Ideological claims by rebels increase conflict duration (Basedau et al., 2022). Nilsson and Svensson (2021) suggested that civil wars where actors embrace ideologies with a transnational character – such as Islamism and Marxism – tend to be the most intractable type of armed conflict. More importantly, according to their findings, conflicts over an Islamist ideology display a higher likelihood of recurrence. Yet, according to Deitch (2022), while religious conflicts have a propensity to last longer than non-religious ones, religion seems to lack a significant impact on conflict intensity or on conflict reoccurrence. Such seemingly competing findings offer a useful baseline from which we aim to explore the link between ideological mobilization and DP. Ideological mobilization that expounds anti-governmental sentiments may produce more committed members willing to make sacrifices for the group. Highly ideological groups remain very uncompromising and absolutist in their demands (Pruitt, 2006). As a consequence, this research proposes the following hypothesis:

H4.

The stronger the ideological appeals or anti-government rhetoric used by rebel groups, the weaker the DP.

Measurement of variables

Measures of DP and its components were drawn from the 50 cases compiled by Druckman and Wagner (2019). The DP variable consists of 16 components within the following four DP areas:

• reconciliation (four components);

• security institutions (five components);

• governance institutions (four components); and

• economic stability (three components).

Using data from the Peace Accords Matrix (Kroc Institute for International Peace Studies, 2013) and Uppsala Conflict Data Programme (UCDP, 2013), among other sources, the coders assessed the cases for each of the 16 components according to the following four-point scale: 4 (very successful); 3 (mostly successful); 2 (moderately successful); or 1 (failure). After coding each component, an average score for each DP area was calculated. These averages were summed to create a single overall DP score ranging from 0–16.

In addition, we used the PJ measure from that data set. For this measure, Druckman and Wagner (2019) evaluated primary and secondary accounts of negotiations to identify instances in which four PJ elements played a role in the negotiations:

• transparency;

• fair representation;

• fair treatment, fair play; and

• voluntary agreement.

If any of these PJ variables played a partially or fully “satisfied” role in the negotiation process, the variable was assessed for whether it had a “highly significant”, “important” or “marginal” influence on the process. Satisfied variables were those instances in which the justice component had a positive influence on the talks. Highly significant instances of PJ were assigned “3”, important instances of PJ were assigned “2” and marginal instances of PJ were assigned “1”. For each of the four PJ variables, the authors added the relevant scores and developed an average for the PJ variable. They added these averages to develop a PJ-sat index.

The IC variable was drawn from Cuhadar and Druckman (2023). This dummy variable measures the presence or absence of IC, considered to be commissions with formal status to prepare or conduct a peace process which explicitly includes civil society. The representation of all of society in the peace process-usually in the form of IC-contributes significantly to addressing past grievances from civil wars and charting a path towards peace. It is one of the six civil society modalities included in the Cuhadar–Druckman analyses. It was the strongest predictor of DP in that study.

A measure of rebel group post-agreement transition party was constructed by Soules (2023). The scale ranges from 0 to 32 years. Zero scores capture transition as an outcome of the agreement. It includes both transitions to a political party and integration in the political system. Twenty-two of the 39 cases received a zero score. Of these, 17 transitioned to a political party, usually by a power sharing formula in the agreement, and competed in the election cycles. Five cases also transitioned by showing some forms of participation in the government (Mali, Niger, PNG, Bangladesh and Yugoslavia). The other 17 remained anti-establishment rebel groups before disbanding, from one year in the case of Rwanda following the 1993 agreement to 32 years in the case of RENAMO following the 1984 Mozambique agreement.

A second data set was constructed. These data consisted of judgements about the symmetry or asymmetry of legitimacy between the government and rebel groups. The judgements were based on voting outcomes: it was measured by how widely the votes received by each of the two parties diverged (voting spread), taking into account both presidential and parliamentary elections where they occurred during the eight-year period following the peace agreement. The public vote distribution over the two election cycles was averaged and used to code the legitimacy symmetry variable as shown below in Table 1. The voting spread (government vs rebel groups) was divided into four categories referred to as legitimacy asymmetry, coded as 1 (more than 30% electoral spread that favours the government); moderate asymmetry, coded as 2 (between 20% and 30% electoral spread that favours the government); moderate legitimacy symmetry, coded as 3 (less than 20% spread that favours the government); and legitimacy symmetry, coded as 4 (less than 10% favouring the government, including a more popular rebel group). In general, the observation period of eight years after the peace agreement included two election cycles. Sometimes, however, there was only one election cycle, meaning that it received more weighting. Special weight was given to the one observed election cycle when there was only one election in the observation period and the next election was clearly outside the observation period (>1 year). If a second election was only just outside the observation period of eight years, for example, a few months, this election was considered.

The historical electoral data came from various sources (AED, 2012; Nohlen, 2005; Nohlen et al., 2001a; Nohlen et al., 2001b). We focused attention on the nationwide elections (head of state and parliamentary elections), irrespective of the exact constitution of the electoral system of the country in question. As a rule, the results of both elections were closely linked, meaning that clear winners in one election are usually also winners in the other election or close winners/losers in one election have close results in the other election. In the exceptional cases where two elections (presidential/parliamentary) within one election cycle produce significantly different results for competitors, we calculated an average spread across the two elections. The coding system is summarized in Table 1, with case examples in each box.

Two extended examples serve to illustrate the way the coding was performed. The first is the peace agreement in Mozambique in 1992 between FRELIMO and RENAMO. This case was assigned the category “Legitimacy Symmetry”, coded as 4. In the presidential elections, FRELIMO achieved 53.3% of the vote while RENAMO ended up with 33.3%, resulting in a voting spread of 20%. However, in the parliamentary elections, FRELIMO achieved an electoral result of 44.3%, while RENAMO received 37.8% of the voters, resulting in a voting spread of 6.5%. This means that in the first election cycle, the average difference in voting outcomes between the two groups across the presidential and parliamentary elections was 13.25%. Thus, the legitimacy symmetry score is 3.

In the 1999 Mozambican elections, FRELIMO won 52.3% of the votes in the presidential elections, whereas RENAMO won 47.7%, resulting in a voting spread of 4.6%. In the parliamentary elections, FRELIMO received 48.5% of the vote, whereas RENAMO obtained 38.8%, resulting in a voting spread of 9.7%. On average, the voting spread was 7.15%. The average voting spread over both election cycles was thus 10.2%. According to our coding, legitimacy asymmetry occurs when the average voting spread is below or around 10%, so this example was coded as legitimacy symmetry.

The second is the 1989 peace agreement in Angola between the MPLA, then the government, and UNITA. In the first presidential elections after the peace agreement in 1992, MPLA won 47.57% in the presidential elections and UNITA received 40.07%, a voting spread of 7.5% in favour of the government. In the congressional elections, MPLA won 53.74% and UNITA 34.10%, a voting spread of 19.64% in favour of the government. On average, these voting outcomes fall into the category of “moderate legitimacy symmetry”, coded as 3; the voting spread is less than 20%. In 1993, however, Angola had already fallen back into civil war, meaning that there was no second election cycle. Accordingly, the first election cycle was decisive for the assessment of legitimacy symmetries.

A third data set was constructed to measure the service delivery of the rebel groups. This was a composite measure that used variables from Albert’s (2022) Rebel Quasi-State Institutions. It is not unusual for armed rebels fighting to take control of a state to create state-like governing institutions. The Rebel Quasi-State Institutions data set covers 235 rebel groups and codes annually for 25 institutions during the entire existence of the group. Service provision is measured along three dimensions: political, economic and social. The composite measure consists of seven variables that capture various services in the areas of health, education and justice. The variables are summed without weighting, resulting in a scale ranging from 0 to 7. Coding details are described in Appendix 1.

The fourth set of variables comes from the rebel appeals and incentives data set assembled by Soules (2023). This data set rates and aggregates various aspects of rebel groups’ ideology with an emphasis on types of appeals. These measures can be considered as proxies for the mobilization of recruits to rebel groups. It includes information on 232 militant groups for the period 1989–2011. Coding details are described in Appendix 2. Summary statistics for the variables are shown in Table 2, including the coding ranges, for example, 1–4 for electoral symmetry, as shown also in Table 1 above.

A challenge in using multiple data sets is to match them to assure that each case has scores on each of the variables: DP, PJ, IC, political party conversion, legitimacy symmetry, service delivery and ideology. In total, 11 of the 50 cases from the Druckman and Wagner (2019) data set were excluded due to incomplete data. Thus, the final data set used for these analyses varied between 33 (SD) and 39 (all the other variables except IC with 38) cases. The HRM analyses reported below are based on an N of 33 only when SD is included in the analysis. To ensure that this reduced number of cases (39, 38 or 33) represented the complete number, we performed sensitivity analyses that compared correlations among DP, reconciliation and PJ for the 50 and 39 case data sets. Only minor changes in the correlations occurred with the reduced data set. All of the correlations were non-significant at about the same p level.

A summary of the key characteristics for each variable is shown in Table 3. Time in play refers to the period during which the variable is measured. Location refers to when the focal process or activity occurs during the period. Level of analysis is whether the variable is measured at a micro (small group), meso (community) or macro (societal) level. The column on the data set provides the references from which the data were drawn. The HRM entry is whether the variable was used as a predictor or DV. We now turn to a discussion of the model, followed by the results.

The model

The statistical analyses are performed with HRM. The framing of our questions in terms of sequential predictors of DP moves us from ordinary least squares to HRM. Like stepwise regression, hierarchical regression is a sequential process involving the entry of predictor variables into the analysis in steps. Unlike stepwise regression, the order of variable entry into the analysis is based on conceptual or principled rationale. In our analyses, we prioritize the results obtained in the earlier study. A key advantage of this mode of analysis is robustness with regard to assumptions. As noted by Woltman et al. (2012, p. 56), “(i)n addition to HLM’s ability to assess cross-level data relationships and accurately disentangle the effects of between- and within-group variance, it is also a preferred method for nested data because it requires fewer assumptions to be met than other statistical methods (Raudenbush and Bryk, 2002) […]. (e)ffect size estimates and standard errors remain undistorted and the potentially meaningful variance overlooked using disaggregation or aggregation is retained”. It is an especially flexible technique for analysing the kind of cross-variable, multi-settings problem presented by our research questions.

Focusing attention on the negotiation process, the key predictor of DP was PJ (Druckman and Wagner, 2019). We then moved outside the negotiation process by examining the effects on DP of civil society inclusion (Cuhadar and Druckman, 2023). Thus, the justice-based DP model prioritizes PJ activities inside the negotiation process. We added to this model civil society inclusion variables and asked about the extent to which they enhance the predictions made by the justice model. Our focus in these analyses has been on the added variance explained (R-squared) by including new variables. We found that civil society inclusion, notably the IC modality, added significant explanatory variance (12% at p < 0.007) to predictions based on PJ. We turn now to the question of whether the added complexity of rebel group activities improves the prediction of DP beyond that explained by PJ and IC.

The HRM process used in these analyses consists of sequential additions of the new variables. These variables are entered in the order in which they were discussed in the earlier sections. Firstly, we evaluate the variance contributed by PJ. Secondly, we evaluate the added variance contributed by the IC variable. We refer to these variables as the base model. The rebel group variables are entered next. Since these variables do not correlate with each other, with DP or with PJ and IC, the order of entry is irrelevant. Thirdly, we add the political integration variable. Then, we turn to the legitimacy variables of electoral outcomes and service delivery. Next, we insert the aggregate ideology variable. This process adds complexity at each step in the process. In this study, the complexity consists of the four types of rebel group activities before and after the peace negotiations. It will reveal the extent to which added complexity improves DP predictability.

Results

A series of HRM analyses were performed. Each analysis is an attempt to predict DP [3]. The first model was run with PJ and IC (base model), adding political integration as a predictor of DP. Supporting H1, the results show that integration increases the predictability of the base model. This variable adds 10% to the R-square, which is significant at the 0.02 level (see Table 4). The key predictors are PJ and political integration. Furthermore, we compare the DP means for cases where political integration occurred (n = 22) and when it did not occur at the time of the agreement (n = 17) (p < 0.02 by independent samples t-test). Similarly, for the reconciliation measure, the respective means were 1.53 and 0.53 (p < 0.002 by independent samples t-test). The political integration variable was also divided into three parts: conversion (1); under 10 years as a rebel group (2); and over 10 years as a rebel group (3). An ANOVA shows a significant difference among these categories (p < 0.02). A Tukey multiple comparison test indicates that this result is accounted for by the difference between 1 (political party) and 3 (over 10 years) (p < 0.012). These results suggest that rebel group integration at the time of or shortly after the peace agreement increases DP and reconciliation. This effect is particularly strong for the comparison of political parties and rebel groups with long tenure.

Table 5 adds the legitimacy and mobilization variables to the prediction equation. The results show that political integration remains a strong predictor of DP. However, none of the other three variables (electoral performance, service delivery or ideological mobilization) adds significant variance to the prediction. Thus, H2 and H3 are not supported. Nor is H4 on mobilization supported. Ideological appeals do not predict DP. Further analyses of the mobilization variable are presented below.

The next analysis addresses H1a. It reverses the role of rebel group integration. Here, we consider this variable as a DV. The results are shown in Table 6. Supporting H1a, PJ is a strong predictor of DP. None of the other variables (IC, legitimacy, SD or mobilization) add significant variance to the prediction. Taken together with the results reported in Table 4, the findings show that integration predicts DP and is only predicted by PJ. Integration into the political system follows high levels of PJ during the negotiation process and facilitates societal peace over the long term.

This interpretation suggests the possibility that political integration explains the effects of PJ on DP, as stated in H1b. A single statistical mediation model was evaluated following the procedures suggested by Hayes et al. (2019, Model 4). The analysis was performed with both the scaled and categorical measures of political integration. The results indicate that political integration is not a significant mediator; nor did either measure of integration explain the relationship between PJ and DP. This result does not support H1b. Thus, political integration and PJ are relatively independent influences on DP.

Probing further, we examined the seven components of the ideological mobilization variable. Only the forced recruitment variable adds significant variance to the base model. The final HRM analysis examines the variance added by this component of ideology. The results are shown in Table 7. Forced recruitment adds significant variation to the prediction of DP [4].

In summary, the three rebel group legitimacy variables do not increase the predictive power of the base model. Only one component of ideology provides a significant increase in predictability.

The pattern of results presented above suggests a factor structure for the variables. The loadings for each of the three extracted factors are shown in Table 8. The first factor explained 32% of the total explained variance. As expected, substantial loadings on this factor were obtained for the cluster of DP, PJ, IC, reconciliation and political integration. Integration and mobilization had strong loadings in opposite directions on the second factor, whereas the electoral and SD variables loaded on the third factor. This is an interesting factor structure. Firstly, it places DP in a cluster with its key predictors: IC, PJ, reconciliation and political integration. The HRM analyses show that each of these variables predicts DP. Secondly, it shows that the three variables that do not predict DP, mobilization, SD, and electoral performance, load together on the third factor. As we noted earlier in the section on legitimacy, electoral performance and SD, the two parts of the legitimacy concept, load on the third factor in the same direction. This supports our decision to conceive these variables as part of the same concept (see Footnote 2). These results highlight two clusters, one consists of predictors of DP, which includes negotiation processes at or around the table and an outcome. The other consists of the rebel group variables that do not predict DP [5].

Discussion

The guiding themes of continuity and complexity are evident in this research. Continuity is longitudinal in the sense of building sequentially on previous results. The findings obtained from these analyses provide the groundwork for our next studies. Complexity is a layered concept in the sense of adding bricks at each of several stages of construction. As more complexity is built into each study, we move closer to a holistic framework for understanding DP. Each of these themes is highlighted in our discussion of the findings in this section.

We began this study with a general question: Do activities by rebel groups before and after a peace agreement to end civil wars influence DP? Analyses performed with several large N case data sets revealed that few of the rebel group activities variables had significant effects on DP or stable agreements. Regarding continuity, this study is the third in a sequence of investigations on DP and related variables. The first study by Druckman and Wagner (2019) showed that PJ was a strong predictor of DP. The second study by Cuhadar and Druckman (2023) showed that IC added significant variance to the prediction of DP. This study showed that transition to political parties or integration in the political system also adds significant variance to the prediction of DP based on PJ and IC. It appears that what happens during and around the negotiation process, including the outcome, is more important than what rebel groups do outside that process. Integration into the political system can be construed as part of the negotiation outcome. Thus, we are only adding one additional layer of complexity to the base model that includes only PJ and IC. That layer reinforces our early conclusions that what happens in and around the peace negotiations matters for long-term peace. But it is also the case that certain rebel group activities matter. Interestingly, it is a political transition decision made by rebel groups at the time of the agreement rather than their legitimizing and mobilization activities before and after the talks [6].

Of particular interest is the finding that a decision to integrate into the political system predicts DP and is part of a factor structure that includes DP, reconciliation, PJ and IC. However, the statistical mediation analyses show that political integration does not explain the relationship between PJ and DP. Thus, PJ during the process and political integration as an outcome are relatively independent routes to DP. The former may be regarded as setting in motion the relational dynamics that strengthen long-term peace. These dynamics are especially important for the reconciliation aspect of DP. The latter may be regarded as the structural changes that are needed for the peaceful functioning of the political system and other institutions. These changes are important for institutional stability, which is another part of the DP index. The political integration variable also loads significantly on the second factor, which includes mobilization. However, they load in different directions. Transition into the political system is more likely to occur when mobilization efforts are unsuccessful. Ideological mobilization may discourage rebel groups from becoming parties or otherwise integrating into the political system.

The findings suggest that war-time service delivery does not have a significant impact on DP and the stability of agreements. Accordingly, while war-time service delivery may help rebel groups achieve their interests at the negotiation table, they have no impact in facilitating the peace to follow; nor do they contribute to the endurance of the achieved agreements. Service delivery is primarily a manifestation of a rebel group’s short-term utilitarian goals, and is used to attain societal support, promote discipline and sanction defection. Such efforts to govern during wartime do not translate into post-conflict reconciliatory behaviour. Similarly, large support in the post-agreement phase, mirrored in electoral symmetry in the first two electoral cycles, has no discernible impact on the durability of peace or on the stability of agreements. This symmetry may generate a type of political stalemate where former and continuing rebel groups have no incentive to change the status quo that grants them sufficient access to power. While favourable outcomes such as conversion to political parties do promote DP, attempts made by former or continuing rebel groups to enhance their legitimacy do not. Rather, the groups may view electoral symmetry as a satisficing device to liven their campaigns and retain their ideological identities.

Similarly, most of the mobilization variables did not impact on DP. Rather, they may reinforce the group’s political identity in attempts to strengthen their claims on the government and larger society. Although ideological appeals may be instrumental in shaping the war-time identities of rebel groups, the varied types of verbal appeals largely remain abstract constructs that do not impact a rebel group’s credible commitment to long-term peace. In fact, appeals that invoke societal grievances are primarily an instrument of war-time recruitment, which in the post-agreement phase may represent an impediment for long-term reconciliation. On the contrary, forced recruitment deviates from this trend by having a significant impact on DP. The tactic of forced recruitment is not a manifestation of weakness but rather of a rebel group’s strength and willingness to deter defection and induce compliance (Sawyer and Andrews, 2020). As such, the ability to project authority among the recruits and enforce compliance through coercive means may translate into more resolute behaviour of former rebel groups in the post-agreement phase. They may be seen to maintain commitment in implementing peace agreements and participating in a DP.

Yet there may be another way of explaining the largely negative results for rebel group activities. We referred earlier to our analyses as monadic in the sense of focusing on one of the conflict parties. We take government activities out of the predictive framework [7]. Measurements of rebel group activities are used to predict a societal variable. No doubt, these activities are only a part of the unfolding of events over the eight years following the agreement. A question of interest is whether including government activities as additional predictors would strengthen the prediction of DP. These activities could be treated separately or combined with rebel group activities in the form of a dyadic index. From a statistical perspective, this could be done with multi-level modelling techniques (Hox et al., 2017). This is a challenge for further research on DP following civil wars.

Innovations, limitations and policy relevance

This study contributes several innovations but also calls attention to limitations that open space for further research. As we noted earlier, focusing on the DP impacts of rebel groups adds to the earlier research on dyadic (government and rebels) analyses. We have learned, however, that rebel group activities do not add significantly to predictions of DP based on negotiation process variables. But, we also learned that conversion to political parties as an outcome does add significant variance to the prediction. Another innovative contribution that extends earlier research is the original measure of legitimacy (a) symmetry. Using data from two post-agreement elections, we were able to gauge the increasing or decreasing popularity of rebel groups. Although the measure did not predict DP in these analyses, it does extend earlier case-study research in the direction of comparative research. It may be used in other studies that compare trajectories of the course of conflict between competing groups. A third contribution consists of the effort to match the scores for cases on five data sets. Only 11 cases were dropped from the Druckman and Wagner (2019) study. The HRM provided an opportunity for evaluating the increased complexity provided by adding rebel group activities to the process variables of PJ and IC (civil society participation).

With regard to limitations, the study raises the need for other indicators of peace that would complement DP. The distinction between negative and positive peace is germane. With respect to the former, the number of years without violence would be useful (Marshall and Ishiyama, 2016). Regarding the latter, various indicators of quality of life, including community building, would extend the reach of our analyses. Similarly, other indicators of rebel group activities may improve predictions of peace. For example, the fragmentation of rebel groups – those that convert and those that do not convert to parties – during the eight-year post-agreement period could provide useful information. As noted above, developing dyadic party measures could help to improve predictability. More generally, various ways of conceptualizing conflict complexity deserve more attention.

From a policy perspective, these results help with decisions concerning where to invest resources in attempts to quell civil wars. For advisors to rebel groups, the findings suggest that they should help to steer these groups towards becoming parties or otherwise integrating their activities within the political system and then support their participation by making the former rebels fixtures in the political system. For those who advise governments, we suggest that they promote PJ and civil society participation in the negotiation but also develop measures that foster political stability (multi-party institutions) and long-term peace by removing incentives to incite violent challenges to a democratic political system.

Our findings also suggest policy restraint from overstating the importance of a rebel group’s perceived war-time legitimacy. Consequently, we suggest that rebel groups, including the most coercive ones that practised forced recruitment, be mainstreamed into a peace process that yields a DP. Negotiated settlements should include provisions that facilitate rebel to party transformations and offer incentives for them to remain parties over the long term. War-time behaviour should not be treated as a qualifying or disqualifying factor in efforts to achieve agreements that endure over time.

Conclusion

This study fills a lacuna in the research literature on rebel groups. Drawing on this broad literature, we evaluated a set of factors hypothesized to influence DP. Emphasizing the importance of continuity in research, the study builds on findings from recent research on justice and civil society inclusion in peace negotiations. Analyses were conducted on five data sets, including measures of DP, party integration, rebel group service delivery, electoral performance (legitimacy symmetry) and ideological mobilization. The results showed that processes inside and around the negotiations were better predictors of DP than rebel group activities. DP at a macro-level depends more on relational variables than on strategies for building loyalty to rebel groups. But it is also the case that political integration makes a difference. Interestingly, the negotiation variables (PJ and IC) and political integration were relatively independent predictors of DP. Integration did not emanate from the process in the form of mediating variables. This finding should spark further research on rebel groups. Other research suggestions are discussed along with implications for government policies and rebel group strategies for creating communities of intra-societal peace.

Table 1.

Measuring electoral performance

Electoral performance	Voting spread	Assigned value
Electoral asymmetry	The voting spread between the government and the rebel group [or ideologically similar parties] is more than 30% in favour of the government. Case examples Guatemala, 1996 Indonesia, 2005 Liberia, 2003	1
Tending towards asymmetry	The voting spread between the government and the rebel group is more than 20%. Case examples Niger, 1995 Rwanda, 1993 Sudan, 2005	2
Tending towards symmetry	The voting spread between the government and the rebel group is less than 20%. Case examples Nicaragua, 1988 Mozambique 1992 DR Congo, 2002	3
Electoral symmetry	The voting spread between the ruling party and the rebel group is less than 10%. Case examples Mozambique, 1999 United Kingdom, 1998 Cote d’Ivoire (Ivory Coast), 2003	4

Source: Authors’ own work

Table 2.

Descriptive statistics

	DP	PJ	IC	Reconciliation	RG party conversion after peace agreement	Service delivery	Electoral symmetry (1–4)	Ideology (0–4)
Mean	4.33	4.12	0.447	1	5	2	2	2.94
Maximum	12	9	1	3	32	5	4	4
Minimum	0	0	0	0	0	0	1	1
Variance	11.07	5.87	0.26	1.13	68.7	2.53	1.10	1.26
Frequency	39	39	38	39	39	33	39	39

Source: Authors’ own work

Table 3.

Variables by characteristics

Variable/characteristic	Time in play	Location in sequence	Level of analysis	Data set	HRM entry
DP	Eight years post-agreement	After the agreement occurs	Maco/societal	Druckman and Wagner (2019)	DV
SA	Five years post-agreement	Implementation period	Meso/rebel and government	Druckman and Wagner (2019)	DV
PJ	During the negotiation	Part of the negotiation process	Micro/negotiation conversations	Druckman and Wagner (2019)	Base predictor
IC	During the negotiation and implementation	Around the negotiation process	Meso/community groups	Cuhadar and Druckman (2023)	Base predictor
Party conversion	At the time of the agreement	Outcome of the negotiation	Meso/rebel group decision	Soules (2023)	Predictor and DV
SD	Before and during the negotiation	Rebel group activities outside the negotiation	Meso/rebel group activities	Albert (2022)	Predictor
Elections	During the eight-year post-agreement	When rebels become parties	Macro/societal	African Elections Database (AED, 2012) (example)	Predictor
Mobilization	Before and during the negotiation	Rebel group activities outside the negotiation	Meso/rebel group activities	Soules (2023)	Predictor

Source: Authors’ own work

Table 4.

Political integration and DP

Model	R	R-square	Adjusted R-square	Std. error of the estimate	Change statistics
					R-square change	F change	df1	df2	Sig. F change
1	0.471a	0.222	0.200	2.996	0.222	10.253	1	36	0.003
2	0.534b	0.285	0.245	2.911	0.064	3.124	1	35	0.086
3	0.623c	0.388	0.334	2.733	0.103	5.719	1	34	0.022

Notes:

^aPredictors: (constant), PJ; ^bPredictors: (constant), PJ, IC; ^cPredictors: (constant), PJ, IC, political integration

Source: Authors’ own work

Table 5.

Political integration, legitimacy and DP

Model	R	R-square	Adjusted R-square	Std. error of the estimate	Change statistics
					R-square change	F-change	df1	df2	Sig. F change
1	0.478a	0.228	0.203	2.761	0.228	9.163	1	31	0.005
2	0.524b	0.275	0.227	2.720	0.047	1.940	1	30	0.174
3	0.627c	0.393	0.331	2.530	0.118	5.659	1	29	0.024
4	0.640d	0.410	0.325	2.540	0.016	0.777	1	28	0.386
5	0.646e	0.417	0.310	2.570	0.008	0.357	1	27	0.555
6	0.656f	0.430	0.299	2.589	0.013	0.589	1	26	0.450

Notes:

^aPredictors: (constant), PJ; ^bPredictors: (constant), PJ, IC; ^cPredictors: (constant), PJ, IC, political integration; ^dPredictors: (constant), PJ, IC, political integration, electoral symmetry; ^ePredictors: (constant), PJ, IC, political integration, electoral symmetry, service delivery; ^fPredictors: (constant), PJ, IC, political integration electoral symmetry, service delivery, mobilization

Source: Authors’ own work

Table 6.

Predictors of political integration

Model	R	R-square	Adjusted R-square	Std. error of the estimate	Change statistics
					R-square change	F change	df1	df2	Sig. F change
1	0.405a	0.164	0.137	7.831	0.164	6.086	1	31	0.019
2	0.460b	0.212	0.159	7.731	0.048	1.811	1	30	0.189
3	0.509c	0.259	0.183	7.622	0.048	1.861	1	29	0.183
4	0.521d	0.272	0.168	7.692	0.012	0.477	1	28	0.496
5	0.528e	0.279	0.145	7.795	0.007	0.265	1	27	0.611

Notes:

^aPredictors: (constant), PJ; ^bPredictors: (constant), PJ, IC; ^cPredictors: (constant), PJ, IC, electoral symmetry; ^dPredictors: (constant), PJ, IC. electoral symmetry, service delivery; ^ePredictors: (constant), PJ, IC. Electoral symmetry, service delivery, mobilization

Source: Authors’ own work

Table 7.

Forced recruitment and DP

Model	R	R-square	Adjusted R-square	Std. error of the estimate	Change statistics
					R-square change	F change	df1	df2	Sig. F change
1	0.471a	0.222	0.200	2.996	0.222	10.253	1	36	0.003
2	0.534b	0.285	0.245	2.911	0.064	3.124	1	35	0.086
3	0.640c	0.410	0.358	2.684	0.125	7.175	1	34	0.011

Notes:

^aPredictors: (constant), PJ; ^bPredictors: (constant), PJ, IC; ^cPredictors: (constant), PJ, IC, forced recruitment

Source: Authors’ own work

Table 8.

Factor loadings

	Factor 1	Factor 2	Factor 3
DP	0.817	−0.200	0.374
Mobil	−0.185	−0.637	−0.416
IC	0.666	0.442	−0.219
PJ	0.737	0.054	−0.271
SD	−0.362	0.190	0.517
Elections	−0.159	0.376	0.560
Recon	0.802	0.107	0.242
Party conversion	−0.499	0.584	0.015
SA	0.407	−0.471	0.559
Recruit	−0.081	0.820	−0.090
Extraction method: principal component analysis

Note:

^aThree components extracted

Source: Authors’ own work

Notes

1. The layering idea may be captured by the metaphor of peeling an onion. In this study, the onion is pealed from the inside (negotiation process) out (rebel group activities in the larger conflict).

2. Supporting the decision to consider service delivery and electoral performance as two parts of the same concept are the results of a factor analysis to be discussed in the section on Results below. These variables will be shown to load in the same direction on the same factor.

3. The control variables in the Druckman and Wagner’s (2019) study were contextual (conflict environment, news coverage and peacekeeping operations) and case features (type of agreement, mediator presence, length of negotiation, region and decade in which agreement was done). None of these variables was a significant predictor of DP.

4. Additional HRM analyses were conducted with the measure of stable agreements (SA). Only the political integration variable added significant variance (17%) to the prediction of SA. This result suggests that the agreement holds better for rebel groups that integrate into the political system after the agreement.

5. Mobilization and political integration load on the second factor in opposite directions. This suggests that the decision to integrate is more likely to occur when mobilization is weak.

6. Further research is needed on separating the effects on DP of political party transition from integration into the political system without party transition.

7. Although the measure of legitimacy (a)symmetry may be considered dyadic, our focus is on the relative performance of rebel groups vis-à-vis the government. Thus, our interpretation of these data is framed in terms of rebel group outcomes.