5

“FROM A THOUSAND EYES TO A THOUSAND VOTES”

Transitioning into Politics

The previous chapter demonstrated that the organization rebels build directly shapes the organization rebels can become. However, organizational transformation is only half of the rebel-to-party battle. After surviving the end of war, the demobilization process, and the internal dynamics of party building, rebel successor parties still have a monstrous task ahead: convince people that they are worth voting for. Thus, while chapter 4 explained how militant organizations become parties in form, the final question I address is whether they can become parties in function as well. In other words, if it looks like a party and it’s registered as a party, can it win like a party?

Halfway through the war, the ERP leader Joaquín Villalobos was asked whether the organization was in a position to mobilize voters with the same success it had in mobilizing recruits. He responded optimistically, arguing that the people who supported the FMLN during the war—providing information, supplies, and sanctuary—would show up in equal numbers if the rebellion transitioned to electoral politics. “Desde mil ojos a mil de votas” (From a thousand eyes to a thousand votes), he said.1 The question I ask here is, did Villalobos get it right? Does logistical support during the war translate into inevitable voter mobilization when the struggle moves from the battlefield to the ballot?

Wartime organization clearly matters for a rebel group’s capacity to forge a party organization on the heels of conflict. But when the time comes to mobilize the thousand eyes on which they relied during war into the votes they need on election day, are those same structures up to the task? When it comes to explaining rebels’ electoral prospects, proto-party structures have benefits that transcend the organization’s boundaries. Organizational subdivisions dedicated to governance, social service provision, and political messaging not only imbue the rebels with skill sets needed to create a party organization, but also establish the type of political linkages needed for the successor party to act (and win) like one.

Chapter 4 focused on explaining transformation: the structural changes needed to forge a party organization.2 But transformation does not capture the stakes of building integrated rebel successor parties. Transition is the shift toward operating in a new environment.3 Organizational transformations and environmental transitions are related, yet they capture different processes, entail different challenges, and have different observable indicators.4 As I note in previous work, this distinction gives rise to a conceptual hitch in the literature.5 Although scholars occasionally allude to different processes, the conceptual line demarcating transition and transformation is never directly engaged.6 More often than not, the concepts and their indicators are used interchangeably, which undermines clarity, concept validity, and the scope of questions we can even ask.7 This chapter pivots from the internal transformation to examine how rebels make the outward transition into the electoral arena.

Parsing Transformation and Transition

To motivate the distinction in rebel-to-party outcomes, it helps to revisit the stakes.8 In scholarship and policy making alike, rebel-to-party provisions are justified by a single argument: giving former belligerents access to legal avenues for dissent will prevent them from using violent ones. In so doing, postconflict integration reduces the likelihood of conflict recurrence and paves a smoother path for stability, development, and democratization.9 The stakes involve some of the most central dynamics to our discipline. If indeed these benefits derive from the formation and inclusion of rebel successor parties, then we need to know both how these parties form and how they function.

Drawing on the organizational change literature, we can distinguish rebels’ internal, organizational transformations from their external, environmental transitions.10 Organizational transformation and environmental transition are related but distinct outcomes, which come with distinct challenges. In the post-war context, it is possible to have one without the other. For example, a militant group may reorganize in a way that mirrors a party organization’s structure, but simply remain a local political actor without participating in electoral politics (e.g., the Zapatistas). A transformation without transition will likely have different implications for both the group and the political system than a transformation accompanied by a successful move into legal politics. Conversely, registering as a party without the organizational changes required to successful campaign, mobilize, and govern is considerably less likely to result in the long-term stability and democratization that rebel-to-party scholars tout as the benefits of postwar electoral inclusion. Many groups survive beyond the war’s end, but some get much further in the electoral arena than others.

While the outcomes are related, parsing transformation and transition is crucial for two reasons. On the one hand, transformation does not happen in a vacuum; a changing organization both affects and responds to its environment—especially when the environment is changing as well. Beyond overcoming internal dynamics, rebel successor parties must operate under new rules, with new competition, by performing new tasks, in service of new goals (which themselves feed back into how they navigate reorganizing). On the other hand, the transition to a new environment is actor dependent: some organizations will be poised to confront these new challenges and perform these new tasks better than others. As such, the process of organizational transformation is considerably less salient if it doesn’t translate into competent functioning.

This distinction is mirrored in the parties literature, where the analogue is party formation versus party consolidation.11 Distinguishing transition from transformation enables future analyses to more directly speak to the parties literature, allowing scholars to examine the interaction between the functional move into politics and the organizational changes that promote (or inhibit) it.

While a distinct logic of transition and transformation is not explicated in the literature, most scholars define rebel-to-party outcomes in terms of electoral participation, which aligns with a transition into the political environment.12 But this tack raises an important question: How much participation is enough for a transition to count? If the stakes outlined earlier guide conceptual evaluation—as they should—then the trade-offs of different approaches are highly consequential.13 Existing approaches have laid an important theoretical and methodological foundation—paving the way for rich analyses and prodigious data-collection efforts. However, they exhibit a few unresolved conceptual issues.

First, scholars often demarcate successful rebel-to-party transition according to minimalist benchmarks of participation, such as party registration or ballot appearance.14 While these indicators have the benefit of parsimony and observability across cases, neither fully captures political integration or the full scope of Sartori’s criteria for qualifying as a party in the first place. Yet, the problem goes beyond concept validity. Beyond inadequately capturing what Robert Adcock and David Collier call the “background concept,” minimalist indicators introduce problematic heterogeneity into the set of positive cases (particularly if the outcome is dichotomized, which it most often is).15 On the flip side, overly strict criteria for successful transition—for example, achieving blackmail or veto power in government—needlessly excludes parties that nonetheless have access to government and can meaningfully contribute to policy making.16

The second conceptual issue concerns how or whether scholars define rebel-to-party failure. Although definitions of transition are fairly common, codified definitions of failure are considerably less so. Yet, defining failure is crucial for any comparative analysis.17 After all, if we want to understand why some rebel successor parties succeed in the electoral arena while others fail, how we define the others is going to affect the answers we get. One tack—increasingly common among those using multipurpose datasets to conduct large-N analyses—involves borrowing the dataset’s (broad) inclusion criteria to serve as the de facto scope conditions for the transition variable.18 Researchers score transitions (however conceived) as a 1, then fill in 0s for all remaining observations in the dataset. Thus, for Benjamin Acosta and Aila Matanock, the implied definition of “failure” is “any militant group that did not participate in electoral politics.” This approach, however, sometimes results in comparing successes to nonviable contenders: groups lacking the intent or even the legal opportunity to transition. For example, organizations like the Legion of Doom or the Animal Liberation Front are likely systematically different from those that tried and succeeded at party formation and those that tried and failed.19 Consequently, this conceptual decision may produce empirical selection effects.

The third and related issue concerns the role of disarmament. While some distinguish “disarmament” from “party formation” to explore their relationship as a salient line of inquiry, one sect of the literature identifies “disarmament” as a necessary condition for both transformation and transition.20 As such, any group that has failed to disarm entirely is deemed a failed transition, irrespective of its electoral outcomes. The logic is straightforward: hybrid militant-party organizations can use violence to influence or circumvent the political process. The “transition to peaceful politics” is, thus, incomplete.21 To be sure, this hypothesis is compelling and important. However, translating this hypothesis into a conceptual rule prevents us from actually testing whether and to what extent it holds. Embedding disarmament as a defining feature of transition gives rise to three problems.

For one, this approach conflates an organizational outcome with a functional and environmental one. Whether the group demobilizes its armed wing is distinct from its electoral standing.22 If an organization runs in elections and places candidates in office through those elections, it does not behoove us as analysts to say “that doesn’t count” because something about the organization is off-kilter.23 More broadly, engaging in activities in addition to the pursuit of goals through policy channels does not inherently compromise the organization’s status as a party unless these other channels are supplanting the political process. Hezbollah, for example, owns and operates restaurants, cafés, and a museum in addition to its political party and its armed wing. Do those activities make it less of a party? During the war, did those activities make it less of a rebel group?

Conceptualizing successful transition in terms of disarmament also forces scholars to conflate political party formation with cessation of violence, which are two empirically (and conceptually) distinct outcomes.24 This conflation not only distorts the mechanisms at work but also limits the scope of questions we can ask. If we are interested in rebel-to-party transitions because we want to know whether incorporating rebel successor parties into the political process affects the recurrence of violence or civil war, then whether the party chooses to retain an armed wing is likely an important variable to consider in addition to party formation. If instead we are interested in what prompts rebel groups to disarm—and becoming a political party is just one of many possible routes to disarming—then party formation and disarmament are distinct concepts on different sides of the equation. Disaggregating party transition from disarmament opens up many lines of inquiry that are closed off when the former is defined in terms of the latter.

Moreover, the disarmament criterion is not applied equally across all actors in the conflict. After all, many office-holding parties maintain control over paramilitary organizations—that is, extrajudicial armed wings that function in service of the party—yet these parties are not deemed “incomplete.” For example, during the first half of the Salvadoran Civil War, the ruling party relied heavily on the paramilitary group ORDEN to exercise control over rural areas.25 While the government’s use of ORDEN was criticized, it was not deemed a nonparty because it had an armed wing. As a result, this conceptual requirement creates a double standard for militant groups on the one hand and incumbent parties on the other without justifying how or whether these two actors truly represent differences in kind.

Conceptualizing Rebel-to-Party Transition

Broadly, rebel-to-party transition occurs when a militant organization qualifies as a legal, functioning political party. As such, the definition must parse groups that are functioning as parties from those that are not. At a minimum, the benchmarks for successful transition must align with the benchmarks of how scholars define political parties. On this point, the rebel-to-party scholarship is virtually unanimous. Nearly every analysis that provides an explicit definition of party relies on Sartori’s definition:26 “any political group identified by an official label that presents at elections, free or unfree, and is capable of placing through elections candidates for public office.”27 However, for analyses to test what we want to test, positive instances of transition must also capture the outcome undergirding the stakes: the political integration of rebel successor parties. This is not just an exercise in mincing words. If integration into the political system is the true source of rebel-to-party benefits, then the stakes of adequately capturing this outcome are pivotal to both analysis and security.

Drawing on Zaks 2022, I propose a conceptual framework that captures the full scope of political integration and lays the foundation for a viable measurement strategy. Keeping in mind Sartori’s instruction that concepts are not just theoretical components but “data containers,” conceptual frameworks are only valuable insofar as they inform inclusion criteria (who we should sort) and measurement (how we should sort them).28 In accordance with principles of conceptualization explicated by Giovanni Sartori, Robert Adcock and David Collier, and Gary Goertz, I define the scope conditions on “failed transition” and the criteria for two further conceptual categories that capture different levels of political integration.29

FAILED TRANSITION

Since a core goal of conceptualization is to facilitate meaningful comparisons, the first conceptual category explicitly defines failed transitions or nontransitions.30 Rebel-to-party failure occurs when a militant group engaged in a political conflict against a target government had the desire and legal opportunity to enter electoral politics but did not achieve even minimal benchmarks of party status.31 As such, it must not register or appear on a ballot—even for a single election. In the empirical section that follows, I use this definition to specify key conditions that guide data collection and measurement.

NOMINAL PARTICIPATION

The next conceptual distinction captures minimal benchmarks for party status, which I call nominal participation. Groups in this category formally register as parties and sometimes appear on a ballot, but fail to win any seats. Thus, they transition into the electoral environment but fall short of political integration (functioning in a governing environment). The distinction is especially salient when groups run in only one or two elections before dropping out. Following the literature, I agree that taking logistical steps toward integration sets nominal participants apart from those that never even register. Yet, the factors associated with running are likely different from those associated with winning, and their implications for postconflict dynamics may be different as well. Accounting for the heterogeneity in rebel-to-party outcomes requires conceptual differentiation at consequential junctures.

SEATED PARTICIPATION

The final conceptual category distinguishes groups that have won at least some seats in postconflict elections. Seated participation captures a more comprehensive scope of functions that accompany political integration, maps more thoroughly to Sartori’s definition of parties, and facilitates testing the mechanism by which rebel integration facilitates peace. Specifically, if peace hinges on the group’s access to legal channels for dissent, our conceptualization should differentiate among groups with qualitatively different levels of access (those who pursue office versus those who take office). Though winning seats is not a requirement of party status, it is a salient conceptual category for theorizing the causes, implications, and mechanisms of rebel-to-party transition.

What Affects Transition?

Since transition is fundamentally about performing new functions in a new environment, explaining it means accounting for what rebel successor parties must do and the opportunities and constraints they face in the process. Rebels’ transition into electoral politics is shaped by their capacity to perform preelection functions (campaigning, party rallies, fundraising), their ability to mobilize voters on election day, and the institutional context in which they operate.

Scholars diverge on the source of rebels’ capacity for political mobilization. Some contend that the extent to which voter mobilization has its roots in war is largely a function of whether ex-combatants get out the vote for their former commanders.32 Others argue that “reliance on civilian aid” in territorial strongholds forces militant organizations to establish a social contract, which can then be leveraged into election-day support.33 Finally, for others still, rebels’ capacity for campaigning and voter mobilization originates before the war began. According to Manning and Smith, groups that began as political parties before the war broke out can leverage their historical political experience into campaign success.34 In sum, political mobilization is currently attributed to in-group loyalty, territorial control, and prewar experience.

Each of these arguments identifies a constituency (i.e., a set of potential voters) and establishes a link between the constituents and the new party (former membership, a wartime social contract, or territorial administration). To be sure, these conditions are critical for the party to function. However, the relations they describe are insufficient alone to account for mobilization, and the proposed mechanisms by which militant histories are connected to electoral successes overlook potential variation in their efficacy. The problem, I argue, is that we are not accounting for systematic differences in the types of relations or links that rebels forge with the communities they encounter. As such, these arguments are built on the assumption that one type of linkage (e.g., wartime support) can be easily translated into another (e.g., postwar votes). This assumption, however, does not find support in theory or on the ground.35 Rather, in the same way that only certain structures are useful for party transformation, only certain types of relations are conducive to political mobilization. After all, we would not expect groups to be able to easily leverage coercive or extractive relationships into electoral support—which is largely why we find rebel-to-party transition puzzling in the first place.

The parties literature is instructive here. Scholars have developed a rich typology of the linkages political parties forge with their constituents because they understand that mobilization is contingent on the nature and trade-offs of different linkage strategies.36 In the context of explaining rebel successor parties, however, it becomes clear that both literatures are incomplete. The parties literature presumes the relations between citizens and parties have always (and only) been peaceful; the postconflict literature does not specify the nature of relations at all.37

Proto-party structures bridge this theoretical gap by accounting for the variation in wartime relations. Broadly, organizations with functioning proto-party structures are less bounded than their combat-centric counterparts. Their dayto-day operations require interacting with the local population on dimensions beyond security and extraction, thereby thinning the border between who’s in and who’s out and creating opportunities for more diverse linkages. Specifically, governance and social service provision structures as well as affiliations with cooperatives and professional associations forge political links with local populations during wartime. Unlike relations characterized by violence and extraction, these relations characterized by political, ideological, material, and educational exchange are more easily (and intuitively) convertible into electoral support. The book’s main hypothesis follows accordingly:

H1: Groups that build more extensive proto-party structures during war will be more successful in electoral politics than those that do not.

Zooming out, this explanation once again highlights the utility of taking an explicitly organizational approach—even when we move beyond the organization’s boundaries. Disaggregating the types of roles and subdivisions present in militant organizations allows us to more precisely characterize the types of relations those structures forge with people outside or on the margins of the organization. In contrast, if we model rebel organizations only in terms of their capacity for producing violence and the logistical divisions needed to support the armed apparatus, then our characterization of relations between rebels and local populations will be limited. As Mampilly argues, reducing complex militant organizations “to their most gasp-inducing components” obstructs the “broader set of interactions that [rebel groups] constantly engage in with local communities.”38

The reality of electoral politics is that no one running for office comes with a built-in constituency, much less a loyal and mobilized one. Even popular, incumbent politicians in functioning democracies spend inordinate amounts of time and money on constituent outreach, campaigning, voter-registration efforts, and get-out-the-vote initiatives. And these candidates are not operating in a new system or facing an electorate that has never before seen a ballot. In the context of postconflict elections, mobilizing voters on election day requires more than just physical presence and political experience.

Finally, while organizations matter, so too does the institutional environment in which they operate. As rebel successor parties gear up to compete in elections, they face new rules of engagement, as it were. Specifically, the electoral institutions present will shape the nature and severity of obstacles to entering the political market. Since proportional representation (PR) systems allocate seats in proportion to the percentage of votes cast for each party, these institutions are more conducive to the inclusion of multiple parties, facilitate the emergence of smaller parties, and lower barriers to entry.39 All else equal, I anticipate that rebel successor parties are more likely to experience initial levels of electoral success under PR systems. Crucially, however, while electoral institutions are likely to affect entry at early stages of transition, I do not expect them to have an effect on longevity. In other words, these institutions allow rebel successor parties to get their foot in the door, but they do little to determine whether they can stay. The second hypothesis follows:

H2: Groups that emerge in proportional representation systems will be more likely to gain seats in immediate postconflict elections.

Insurgent Structures and Outcomes (ISO) Dataset

The end of the Cold War brought with it a boon in electoral opportunities for former rebels.40 More and more, civil conflicts were ending through negotiated settlements rather than one-sided victory. With those settlements came provisions for electoral participation and other power-sharing arrangements. Scholars have seized on the increasing frequency of rebel-to-party opportunities to either create or expand cross-national datasets with variables suited to testing the causes and consequences of transition.41 These data-collection efforts and the analyses based on them have laid a crucial analytic foundation in a high-stakes field. As I have noted in previous work, however, existing rebel-to-party datasets exhibit inconsistencies in both the universe of cases deemed relevant for comparative analysis and the coding of the outcome variable—even when they define transition the same way.42 The overinclusion of nonviable contenders (e.g., the Animal Liberation Front) and the omission of groups that ultimately transitioned (e.g., Hezbollah in Lebanon) create samples that suffer from selection bias, which quickly finds its way into the results.43

I have constructed the Insurgent Structures and Outcomes (ISO) Dataset to build on these foundations while overcoming the pitfalls. The dataset collects empirically and conceptually informed data on rebels’ wartime organizational structures and a nuanced scope of rebel-to-party outcomes. ISO seeks to identify the universe of potential and successful rebel-to-party transitions. I compiled the sample of cases by first collecting the full set of observations across existing data-sets and omitting those that violated the inclusion criteria outlined later in this chapter. Then, drawing on primary sources, secondary sources, and consultation with country or group experts, I collected data on the wartime organizational characteristics of the remaining groups. The resulting ISO dataset is composed of seventy-eight rebel groups that operated between 1974 and 2018.

Keeping in mind Sartori’s instruction that concepts are “data containers” as much as they are theoretical components, I draw on the conceptual framework laid out earlier to inform the inclusion criteria (who we should sort) and measurement (how we should sort them).

Inclusion Criteria and Scope Conditions

In light of the established selection bias present in previous analyses, the inclusion criteria for this dataset warrant special attention. Although there is room for expansion—since rebel organizational structures have analytic utility beyond understanding the outcome of this study—the scope of the ISO Dataset is currently limited to the set of potential transitions. The reasoning for this is purely pragmatic: collecting data on the internal structures of clandestine organizations is difficult and time intensive. As such, the current inclusion criteria of the ISO data align with the minimum conditions for rebel-to-party failure.

In a departure from some approaches, which rely on civil war termination or peace agreements to define the universe of cases, ISO compiles observations based on characteristics of the group rather than characteristics of the conflict.44 As a phenomenon, rebel-to-party transition is not bound by the standard operational definition of civil war, nor is it limited to conflicts that have ended in peace agreements.45 Thus, limiting the scope of observations to cases that satisfy tangentially related inclusion criteria risks omitting viable or even important cases from the analysis. Since the analytic objective is to use information about the group during conflict to predict the group’s political trajectory, I derived inclusion criteria with two goals in mind: including the full scope of positive cases and defining the boundaries of an appropriate comparison set.

The inclusion criteria are twofold: First, the group must have made political claims on a government—either calls for expanded rights for a marginalized group, democratic reforms, revolution, or territorial secession.46 This criterion does double duty. To count among potential transitions militant organizations must have expressed the desire to govern in one capacity or another. As such, it omits groups with objectives that are either too broad (e.g., establishing a regional caliphate), too narrow (e.g., animal liberation), or too unrealistic (e.g., anarchists) that party formation around those goals is unlikely or untenable.47 Moreover, becoming a party requires a target government in which to participate. Specifying where groups would become parties is crucial to assessing the legality of and obstacles to transition. Including only groups that make political claims against a target government filters out groups with transnational objectives such as al-Qaeda or Boko Haram (both of which have appeared as failures in rebel-to-party analyses).48 For these groups, a within-state political solution is either unlikely or undefined. To assess whether the outcome is even legal, we must be able to identify the specific government into which a group would transition. Unless a transnational group expresses interest in transitioning within a single state, it should not be compared to groups that do.

The second criterion is much more straightforward: rebels must have a legal path to party formation. This path can take a few forms. One route is through explicit electoral participation provisions codified in a negotiated settlement. As Matanock observes, nearly half of “conflict-ending settlements” since 1990 contain provisions allowing ex-rebels to participate in elections.49 Unfortunately, while provisions for electoral participation are a convenient scope condition, not all rebel successor parties emerge through this path. In other cases, state institutions may simply allow for party formation, and militant organizations proceed accordingly. Hezbollah in Lebanon, for example, transitioned into politics without any explicit provision enabling it to do so. Legality as a condition may seem obvious, but some existing rebel-to-party datasets include among the failures groups that operate in states in which party formation is prohibited.50 To truly fail, the group must have attempted transition in a place where party formation was possible. After all, one cannot treat rebel-to-party status as a variable if the outcome cannot vary.

Before moving onto coding, I address an implicit counterpoint. The inclusion criterion of Acosta’s Revolutionary and Militant Organizations Dataset (REVMOD) and Matanock’s Militant Group Electoral Participation (MGEP) dataset is all militant groups, defined as “nongovernmental entities using extra-legal violence to achieve political aims.”51 According to Matanock, broad inclusion criteria are important because “any of these groups may participate in elections.”52 Of course, the accidental omission of viable contenders is not ideal. However, their inclusion is not without costs. The trade-off takes the following form: collect data that (1) compare groups that succeeded to a narrower sample of groups whose desire for power-sharing was clear, yet unrequited, or (2) compare groups that succeeded to a much wider sample of groups that tried but failed and groups that never tried in the first place.

Ultimately, even an incomplete universe based on conceptually informed sampling is preferable for a few reasons. First, overly broad inclusion criteria will almost inevitably include organizations that are unlikely, unwilling, or unable to attempt transition. When the zeroes are made up of two systematically different populations, any analysis based on that sample is liable to suffer from selection bias.53 Second, because the scope conditions on ISO are explicit, the range of groups to which inferences apply is similarly well defined. Thus, even if one finds a transition outside ISO’s current scope, the condition that needs alteration will also be clear.

Variables, Measurement, and Data Collection

ISO contains a host of variables relevant to conflict and postconflict dynamics, but here I focus on outlining the variables relevant to my theory and the alternative explanations I test from the literature. For each variable, I discuss the measurement strategy, approach to data collection, and potential limitations or challenges.

REBEL-TO-PARTY TRANSITION

The core benefit of the conceptual framework is that it is theoretically bounded, yet operationally flexible. This discussion draws heavily on the measurement strategy I lay out in my 2024 article. I propose two measurement strategies for the outcome—one four-stage measure and one dichotomous measure—not only to test whether the theory is robust (or sensitive) to different coding schemes, but also to assess whether proto-party structures (and other variables) affect some steps in the transition process more than others. To briefly situate this discussion in the literature, all previous rebel-to-party analyses employ a dichotomous measure of transition coded according to different benchmarks of electoral participation. The coding schemes and dataset summaries are given in table 5.1.

TABLE 5.1. Overview of existing measures of rebel-to-party (RtP) transition1

I propose a disaggregated measure of transition to reduce the heterogeneity evident in binary measures of the outcome. The four-stage measure synthesizes conceptual and empirical insights to justify adding nuance on two dimensions. Drawing on the conceptual framework, I first distinguish nominal participation (party registration or ballot appearance) from seated participation (groups that garner enough votes to take office). Then, upon examining postconflict elections, the data reveal that while many rebel successor parties become persistent seat holders in their respective governments, a significant number lack that same staying power. In this second group, successor parties win some seats in the first post-conflict election, but ultimately fail to pass muster in the long term. By the second or third election, their vote share drops below the threshold to take office and never recovers. To capture this empirical trend, I disaggregate seat winners into short-term participants and persistent contenders.54 Placing these “one-hit wonders” in a distinct category enables scholars to test whether they exhibit systematic differences from groups that never make it beyond the ballot as well as groups that exhibit electoral persistence.55 The measurement strategy is enumerated here:

• (0) Failures: groups with the opportunity to transition and desire for political participation, but that fail to register.

• (1) Nominal Participants: groups that registered as parties and/or appeared on a ballot but failed to win seats.

• (2) Short-Term Participants: groups that won seats, yet dropped out within three election cycles.56

• (3) Persistent Contenders: groups that won seats in three or more elections.

Due in part to the scarcity of cases and in part to the desire for parsimony, I also propose a dichotomous measure of rebel-to-party transition. Moreover, as table 5.1 illustrates, all previous analyses employ a dichotomized measure as well, thereby establishing a demand for a clean demarcation between “success” and “failure.” Given that any binary measure of a complex variable will exhibit heterogeneity within categories, the challenge is to identify a cut point at which the heterogeneity is least problematic. The measure I propose departs from previous approaches, yet aligns with the underlying conceptual motivations in the literature: capturing rebels’ integration into legal politics. While this approach departs from the rebel-to-party literature, Adrienne LeBas contends that holding seats after elections is a reliable indicator to capture both party status and political integration.57 As such, I propose grouping nominal participants and nonparticipants together as failures and coding all seat winners as successes:

• (0) Failures: nonparticipants and nominal participants in the electoral system

• (1) Successes: rebel successor parties that won seats in at least one election

Collecting data on party registration and electoral outcomes was relatively straightforward. Much of the data were available from election databases, news sources, and databases of registered political parties worldwide. I also cross-checked my coding against all other rebel-to-party datasets.

Nevertheless, this approach is not without limitations. The most pressing issue with this measure (and the underlying conceptual framework) is its election-centric view of political parties. However, parties routinely perform important off-the-ballot functions like rallies, local organizing, and advocacy, which this framework does not capture. As a result, rebel groups that become active political organizations outside of the electoral arena are liable to fall between the conceptual cracks.

I argue, however, that the trade-off is worthwhile for a couple of reasons. First, holding seats after elections is a reliable way to capture both party status and political integration.58 More important, capturing only groups that function in a governing capacity (or fail to) allows us to more directly test the mechanism underlying the stakes: that access to legal channels for dissent will prevent groups from seeking violent ones. While off-the-ballot parties perform important tasks, they do not as reliably have access to these channels as those integrated in government. Second, from both a practical and conceptual standpoint, functioning as a party in government is more easily standardized and operationalized in a way that implies similar outcomes across cases. Although it is an imperfect proxy, data collection is considerably more straightforward, reliable, and comparable using election-centric indicators.

PROTO-PARTY STRUCTURES

The book’s core argument is that rebel groups with proto-party structures will have a decisive advantage as they both transform into political parties and transition into electoral politics. Chapter 3 identified three broad domains that correspond to the key roles of party organizations: governance, social services, and political messaging. To operationalize this concept I identify five distinct types of subdivisions, each of which falls under one of the categories: administration and governance, health services, educational provisions, affiliated community groups (e.g., a youth organization or a women’s organization), and political messaging.

• Governance and Administration: This variable is coded 1 if I find evidence of routine governance in either controlled or contested territory, and 0 if I did not. Indicators of administrative capacity include resolving disputes; allocating resources (e.g., food, money, or basic goods); or building and managing public infrastructure (e.g., roads, electricity, waste removal).

• Health Services: This variable is coded 1 if the organization set up medical services accessible to local populations, and 0 otherwise. The extent of health provisions ranges from providing basic medical treatment to sick or injured civilians (as in the case of the National Council for the Defense of Democracy-Forces for the Defense of Democracy [CNDD-FDD] in Burundi) to constructing full-service hospitals (as in the case of Hezbollah in Lebanon).59

• Community Groups: This variable is coded 1 if the movement had an explicit youth, women’s, and/or student organization during wartime, and 0 if they did not. Community groups capture both embeddedness among local populations and political education of those either outside or at the fringes of the movement. Because I want to capture long-term embeddedness, the community groups had to exist at least one year before the end of the war and they had to either have their own name or evidence of repeated meetings.60

• Educational Provisions: This variable is coded 1 if the rebel group provided basic educational services to local populations such as literacy programs or an equivalent to primary schooling, and 0 otherwise. To be sure, most, if not all, educational services included some political component, but for this variable to get a 1, I required evidence that a curriculum existed outside of the organization’s message. Where I found evidence only of “political education,” this variable is coded 0, and directed education about the movement counts instead toward political messaging.

• Political Messaging: This variable is coded 1 if the group provides a regular source of news or propaganda. I do not elevate one medium over another, so I take any of the following as an indicator of a political-messaging wing: dissemination of written work (e.g., newspapers, flyers, pamphlets), creation of a radio or television station, or holding regular political education groups with local populations.

Using these indicators, I created a composite index of proto-party structures. The index is additive and unweighted since no theoretical reasons point to the relative importance of one domain over the other.61 The proto-party structures variable thus captures the extent to which a given militant organization has diversified into salient political domains. Thus, a higher value on proto-party structures indicates greater political diversification. This variable will be used to test the book’s core hypothesis:

H1: A greater diversity of proto-party structures increases the likelihood that groups will achieve more advanced levels of rebel-to-party success.

POLICING AND SECURITY

I also collected data on whether rebel groups provided policing and security services to local populations during the conflict. This variable is also dichotomous and is coded a 1 if I found evidence that the group took over policing activities in an area. Examples of security services include civilians calling on the organization in the event of a robbery or harassment or rebels pointedly protecting civilians from attacks (either from the government or from other rebel groups). Policing captures the enforcement side of administration, but itself does not translate into a specific party feature. As such, I do not include it as part of the proto-party structure index.

I include this variable in the dataset because the goal is to have as thorough of an account of rebel organizational structures as possible. I have no theoretical reason to believe that policing has any specific effect on rebel-to-party transformations. I did, however, rerun the models with policing added as an additional control variable to test whether any diversification is good for rebel-to-party transition. The coefficient estimate is consistently small (typically around 0.1) and the standard error hovers around 0.7, suggesting that not just any diversification, but rather, proto-party diversification is the driver of rebel-to-party success.

Data Collection

The organization-level data derives from a variety of primary and secondary sources. These sources include firsthand accounts; archived communiques, pamphlets, and newspapers from within the rebel organization; state and NGO reports from conflict zones; scholarly works from individual conflicts; and media coverage of the war. I consulted a minimum of two different types of sources per case to minimize source-centric biases. As I discuss in the next section, accounts from rebel organizations, governments, and media outlets are each likely to exhibit different types of bias for different reasons.

Challenges, Limitations, and Trade-offs

While the outcome variable is clearly a contested concept, the most pressing empirical challenge is collecting systematic data on rebel groups’ wartime organizational structures. The overarching challenge—as others have noted—is the difficulty of finding reliable data on clandestine organizations.62 Potential problems in data collection stem from systematic sources of bias as well as incidental misrepresentation.

The systematic sources of bias are multifold and go in both directions. On the one hand, rebel groups have incentives to overrepresent noncombat activities—especially social service provisions—as a means of boosting their legitimacy.63 And office-seeking groups may be even more prone to this tendency. On the other hand, government and some NGO accounts of rebel groups are likely to downplay or erase any political activities that may bolster the perception of rebels at home or abroad. A similar bias is likely evident in scholarship and media, albeit for different reasons. Due in part to the importance of documenting patterns of violence, and in part to the combat-centric organizational framework that shapes the literature, scholarship and news reports on rebel groups are liable to overdocument the actions and structure of the military to the exclusion of other divisions of the organization.64

Beyond the sources of systematic bias, data collection is also subject to incidental sources of error. Thinking back to the Salvadoran case, for example, the picture one paints of the FMLN’s organization will look very different depending on whether the war is documented from Chalatenango, San Salvador, or Morazán.65 While the FMLN exhibited more pronounced geographical distinctions in its local structures than other groups, this phenomenon is not unique either.

To be sure, the potential for missing data is high. Yet, the risk of associated bias is mitigated in two ways. First, while rebel groups are likely reticent about discussing ongoing military operations, they tend to be open—if not eager—to advertise the work of their political wings and their capacity to provide social services. Moreover, when it comes time to transition into politics, spotlighting their social and political experience during the war is a strategically beneficial way of mobilizing voters. However, their tendency to overrepresent proto-party structures is actually a double-edged sword. While it may lead to some biased accounts in documenting their wartime organization, the incentives to advertise the scope of their politically salient structures make it likely that absence of evidence for a given subdivision is actually evidence of absence.

Additional Variables

In addition to the proto-party index, I include a series of control variables capturing other salient organizational factors and alternative explanations: size, fragmentation, external sponsorship, rebel-to-party provisions, and postwar electoral institutions. I omit systemic correlates of rebel-to-party transition that do not contribute to our explanation of the phenomenon, such as the Cold War ending and whether the transformation took place on the African continent.66 These variables predict civil war termination, which is correlated with rebel-to-party transformation, but do not explain it.

SIZE

Larger organizations are expected to exhibit greater diversification and possess more slack resources, both of which feed into an organization’s resilience and capacity to adapt.67 Moreover, large organizations during war potentially contribute to larger bases of political support at the ballot.68 To the extent possible, the variable captures the estimated size of the organization at the end of the war (i.e., leading up to transition). Size estimates came from a variety of primary sources, secondary literature, and the Non-State Actor Data.69 Where estimates diverged by more than a thousand members, I took the mean across estimates. All values are logged for the analysis.

H2: Larger organizations are more likely to survive the end of war and transition into electoral politics.

FRACTURING

Staniland demonstrates that wartime resilience and adaptability are at least in part a function of organizational cohesion.70 I draw on this argument to test whether the same features matter for resilience off the battlefield. I use organizational fracturing as an inverse proxy for cohesion. Fracturing is a count variable starting at 0 if the group exhibits no evidence of infighting. The variable is coded 0.5 if I find evidence of intraorganizational tensions, but no splits, and the count increases by 1 for every time the organization split over its lifetime. Drawing on the literature, I expect that high rates of fracture will detract from an organization’s capacity to transition into a party by contributing to ideological and logistical gridlock as the organization negotiates their new tasks and their new role.

H3: Fracturing is negatively associated with transition.

ELECTORAL INSTITUTIONS

Another factor potentially influencing rebels’ postconflict opportunity to launch a successful party are the country’s electoral institutions. I include a binary control variable indicating whether the country’s legislature is elected via proportional representation versus single-member districts. The rules governing postconflict elections may create an exogenous barrier to entry for new parties trying to break into electoral politics. Specifically, since single-member districts are more likely to foster two-party systems, former rebels otherwise well poised to succeed as a political party may face an institutional obstacle unrelated to their organizational advantages.71 The variable PR System is coded 1 if at least half of the seats in the national legislature are allocated via proportional representation, and 0 otherwise. I anticipate that PR systems will make breaking into politics more likely, but will have little impact on longevity of the party.

H4: Proportional representation systems facilitate early stages of transitioning (winning seats in the first postconflict elections), but do not have an effect on the longevity of rebel successor parties.

EXTERNAL SPONSORSHIP

According to Acosta’s analysis, one of the most significant predictors of rebel-to-party transition is the extent to which the militant group has secured external sponsorship.72 External sponsorship is a count variable documenting the rebels’ number of state patrons. For the groups present in my data and his, I draw on REVMOD for the rebel group’s number of sponsors; otherwise I adopt the same coding rule: the count increases for every country that provided the militants with funding, weapons, training, or sanctuary.73 For a variety of reasons, I do not believe external sponsorship will predict transition, but I test the hypothesis from the literature. The first alternative hypothesis I test follows:

A1: External sponsorship is positively associated with rebel-to-party success.

REBEL-TO-PARTY PROVISIONS

While scholarship is mixed on the effects of party transition clauses, I control for whether the conflict ended with a negotiated settlement that included explicit rebel-to-party provisions. Söderberg Kovacs and Hatz, for example, argue that “rebel-to-party provisions are likely to function as one mechanism among many that serve as a guarantee for the political and organizational survival of the rebel group.”74 Rebel-to-party provisions is a dichotomous variable coded 1 if the group signed an agreement with an explicit transition clause according to the UCDP Peace Agreement Dataset, and 0 otherwise. Similarly, the theoretical framework guiding this analysis leaves me with no priors on the efficacy of rebel-to-party provisions. I expect that in previous analyses, this variable captured the opportunity to transition, rather than a mechanism facilitating it.

A2: Rebel-to-party provisions are positively associated with rebel-to-party success.

Empirical Analysis

This section examines whether rebels’ wartime organizational structures predict electoral success. To test how proto-party structures perform in explaining rebel groups’ transition into politics, I conduct a variety of statistical analyses on the ISO data. I turn to large-N analyses here to evaluate the extent to which the explanatory variables of the theory are observable across cases and to assess whether the theory travels beyond the Salvadoran context. After all, if proto-party structures were specific to the FMLN or even just the region, that fact would place an important scope condition on inference and call into question the contributions of this study.

First, I estimate a series of ordinal logistic regressions on the fully disaggre-gated outcome variable. Model 1 tests how organizational traits—proto-party structures, size, and fracturing—affect the likelihood and extent of transformation. Model 2 controls for the financial aspect of opportunity by estimating the impact of third-party sponsors. Model 3 controls for the institutional aspects of opportunity, adding in variables for rebel-to-party provisions in the settlement and whether the electoral system uses proportional representation. Finally, Model 4 presents a fully saturated model. Table 5.2 presents the results of the regression and table 5.3 report’s a group’s predicted probabilities of successful integration as its number of proto-party structures increases.

In addition to testing the model on the four-stage outcome variable, I also run the fully saturated model on three alternative codings of rebel-to-party transition, representing different levels of rebel successor party integration. Specifically, I recode transition as a series of dichotomous variables at three different cut points for success: nominal participation, seated participation, and long-term integration. Since the outcome is now dichotomous, I use logistic regression to estimate the model.75

TABLE 5.2. Ordinal models of rebel-to-party transition

TABLE 5.3. Predicted probabilities of transition for X proto-party structures

These analyses serve two main functions. First, by rerunning the analyses on a binary measure of nominal success, I can test whether the results of prior studies hold on their own turf, or whether they are driven more by inclusion criteria and omission of key organizational traits (as anticipated). Second, testing my hypotheses across a variety of cut points enables me to parse whether certain explanatory factors are more important for some stages of transition than others. The results of the analyses are depicted in figure 5.1a–d.

Results

The results from the main analyses on the four-stage outcome provide strong evidence in favor of H1: Rebel groups with more proto-party structures consistently experience higher levels of integration as their successor parties enter politics. The estimated size of the organization also consistently achieves significance, thus corroborating H2. Obviously, the mechanism for this result is not elucidated in a statistical model, though it is likely some combination of greater diversification and a larger base of potential voters.76 Furthermore, as H3 predicts, higher rates of organizational fracturing are associated with lower levels of political integration. Though the fracturing variable loses significance as other variables are added, this outcome is likely a result of inflated standard errors penalizing the admittedly small sample.

FIGURE 5.1A. Coefficient effect plots at different cut points for rebel-to-party success. Here, the cut point used is the full ordinal DV.

FIGURE 5.1B. Coefficient effect plots at different cut points for rebel-to-party success. Here, the cut point used is nominal participation.

Turning to the control variables that represent rebels’ opportunity to transition, the results align with my expectations. Higher levels of rebel-to-party transition appear to be more likely under proportional representation systems than in single-member districts, which supports H4. The two alternative hypotheses that I test from the literature do not fare well. Regarding hypothesis A1, external sponsorship never achieves significance in any of the models. This result further substantiates the intuition I lay out in my 2024 article: statistically robust estimates of the effect of external sponsorship were likely driven by artifacts in the data used to estimate the model. Specifically, by including hundreds of small, less-viable contenders for rebel-to-party transition, this variable was likely sorting viable political aspirants from irrelevant failures. As such, third-party sponsorship was associated with transition in the dataset, but, in reality, it does little to parse groups that tried and failed from groups that tried and succeeded.

FIGURES 5.1C. Coefficient effect plots at different cut points for rebel-to-party success. Here, the cut point used is a single electoral victory.

Furthermore, rebel-to-party provisions do not reliably predict transition success, thereby calling A2 into question. My hunch is that built-in provisions are likely good at parsing groups that had the opportunity to transition from those that did not, but since this dataset omits groups that lacked the opportunity to move into electoral politics, that feature is not captured.

FIGURES 5.1D. Coefficient effect plots at different cut points for rebel-to-party success. Here, the cut point used is persistent contenders.

Running the saturated model on a variety of binary coding schemes proves useful for a few reasons. First, the results parse the point at which certain variables matter. Second, they reveal how sensitive rebel-to-party models are to how we operationalize the outcome.

Figure 5.1b reveals that slicing the dependent variable at a nominal conception of success causes almost every variable to lose significance. Indeed, this model specification is the only one in which proto-party structures fall out of significance. This result is unsurprising. Given the increasing rate at which rebel-to-party provisions are included in negotiated settlements, party registration is often an incidental result of merely implementing the negotiated settlement.77 The factors that facilitate the actual organizational transformation, transition into electoral politics, and the party’s long-term success are unlikely to predict settlement implementation.

Moreover, despite using the same coding for rebel-to-party success as previous analyses, expectations from the literature are not supported. The implications for the rebel-to-party literature are substantial. The results suggest that using “party registration” or “electoral participation” to demarcate transition is creating a heterogeneous success group comprising subpopulations that are systematically different on both the outcome and explanatory variables. The null results suggest that selection effects in dataset composition likely drove the results of prior models using this outcome.78

The results shift drastically when the outcome variable is coded to reflect a higher benchmark for success. Figure 5.1c depicts the coefficient estimates for a logit model in which rebel-to-party success is coded 1 if the group placed any candidates in office in at least one election, and 0 otherwise. At the point that winning seats is considered a defining feature of successful transition, proto-party structures become key to parsing success from failure. Additionally, PR systems become significant in this model, which corroborates the intuition that institutions affect groups’ initial capacity to gain representation.

Finally, the last model, depicted in Figure 5.1d, codes a transition as successful only if the party occupies a lasting position in national or regional politics. The results of this model continue to align with the predictions of my theoretical framework, yet they differ from the previous model. Namely, electoral institutions are no longer a significant predictor of success. This result follows logically if we consider how electoral institutions should matter. Specifically, we should expect the institutional arrangement to affect whether a party can break into politics in the first place, but whether the party lasts once it is there is more a function of the party’s political aptitude.

Proto-Party Structures and Municipal Election Results in El Salvador

The results just described provide compelling evidence that proto-party structures built during war do a reasonable job of predicting rebel-to-party success. The hitch, of course, is that these models are based on aggregate data, which is less than ideal after I spent the preceding chapters demonstrating what we risk losing when we treat organizations exclusively at the aggregate level. The question these models leave unanswered is whether the organizations that build proto-party structures exhibit some other systematic difference that predicts rebel-to-party success, or whether these structures are really doing the heavy analytic lifting.

To ensure that proto-party structures are not merely capturing some correlated, aggregate trait of the groups, I dive into the FMLN case to test whether these dynamics are reflected at the local level. As demonstrated in chapters 3 and 4, proto-party structures were common throughout the FMLN, but not uniformly so. What makes the FMLN such an ideal case for testing rebel-to-party theories is that the suborganizational variation is geographically constrained. While the FPL, PRTC, PCS, and RN built robust political subdivisions during the conflict, the ERP’s wartime structures functioned largely in service of its combat operations. Getting comprehensive and granular data about the presence (and absence) of particular structures built during the war would be a nearly insurmountable task for any country—even small ones. As such, I use the consistent presence of the respective OP-Ms as a proxy to represent whether proto-party subdivisions were present in a given municipality.

Drawing on hundreds of archival documents from the rebels, the US government, NGOs, and the Salvadoran government as well as secondary sources (scholarship and news reports), I create a municipal-level dataset accounting for different OP-M presence during the conflict, which I then merge with municipal-level electoral results from El Salvador’s 1994 legislative elections. The dataset includes six binary variables: one for each of the five politico-military organizations, and a sixth denoting evidence of FMLN presence, but without clear indication of which OP-M occupied the territory (which occurred in only six instances).79 In total, I find evidence of a strong FMLN presence in sixty-six municipalities (out of 262).

On average, the FMLN took 16 percent of the vote in a given municipality. In light of the new geopolitical data of FMLN presence, I slice the data in two ways and run analyses on each. First, I assess the average vote share in municipalities with a strong FMLN presence for each OP-M. Summary statistics are reported in table 5.4. I also collapse the proto-party OP-Ms into a single group to compare to the ERP. To test whether OP-M presence—and thus, a wartime legacy of proto-party structures—is associated with greater postconflict vote share, I run a series of two-tailed t-tests on the new data.

In line with my theory, I expect to find lower vote shares for the FMLN in ERP-dominant municipalities. Crucially, this is a difficult test for two reasons—one substantive, one methodological. First, as Leigh Binford notes, ERP-occupied areas (particularly in Morazán) comprised the strongest and most autonomous rearguard during the war.80 Scholars have explicitly argued that strong bureaucracy and territorial control during war should translate into postconflict gains for rebel successor parties.81 Additionally, Morazán saw a disproportionate amount of active fighting. As a result, the security provided by the ERP as well as the localized demobilization of former combatants may well act as a countervailing force in the FMLN’s favor. Second, this test is difficult simply because of the relatively small number of cases.

TABLE 5.4. Summary of vote share in OP-M-dominant municipalities

Notwithstanding the analytic challenges, the results are clear. The municipalities occupied by OP-Ms that prioritized building diverse political and administrative subdivisions were considerably more likely to vote for the FMLN in the postconflict elections. To continue the ERP-FPL comparison, the average difference in FMLN vote share between ERP-dominant municipalities and FPL-dominant municipalities is nearly 10 percent. The results are nearly identical when the other OP-Ms are collapsed into a single group and compared to the ERP, though these estimates are significant at a higher level of confidence.82 The results are illustrated in figures 5.2 and 5.3. For reference, I have also plotted the average vote share for the FMLN for every municipality and the average vote share for the FMLN in the municipalities occupied by any OP-M. The FMLN’s vote share in the ERP zones is statistically indistinguishable from its countrywide average.

From a Thousand Proto-Party Structures to a Thousand Votes

The FMLN’s overall showing in the 1994 elections was respectable for a multi-party system, but it was by no means breathtaking. However, a more complex story emerges when we disaggregate votes by municipality and compare them to the FMLN’s local organizational structures. The results of the municipal-level analyses cast doubt on the optimistic quote that opened the chapter. Unfortunately for both Villalobos and the theories that propose the same, rebels need more than wartime collaborators—a thousand eyes—to guarantee postwar votes. To be sure, the ERP had a robust territorial stronghold, it exhibited a strong bureaucracy, and it rarely committed violence against local populations. Yet, these other wartime traits were insufficient to win it the votes it needed from the population on which it had come to rely. Rather, to win like a party, rebels actually have to look like one first.

FIGURE 5.2. Vote share disaggregated by OP-M Note: Each bar represents the FMLN’s vote share (1) across the country as a whole (2) aggregated in all FMLN Zones (3) in ERP-controlled zones, (4) in FPL-controlled zones, etc. ERP = People’s Revolutionary Army, FPL = Popular Liberation Forces, RN = National Resistance, PRTC = Central-American Workers Revolutionary Party, PCS = Communist Party of El Salvador.

Broadly, this chapter makes three core contributions. First, parsing transition from transformation lays the conceptual foundation to create more tailored theories and to more incisively sort the rebel-to-party literature. In light of this more explicit framework, one can more easily identify which scholars are focused on one versus the other and which factors are meant to explain which outcome. Moreover, the distinction paves the way for future explorations of the relationship between the two outcomes.

Crucially, this is not a matter of semantics. In Colombia’s 2018 peace agreement with the Colombian Revolutionary Armed Forces (FARC), for example, one of the terms of the settlement affords the new FARC party a minimum quota of seats for the first two election cycles—irrespective of its vote share. The party’s initial vote share would not have otherwise been sufficient to place it in office were it not for the quota. Since its initial integration into politics is not contingent on successful party formation or performance, we can now ask whether transition into politics begets transformation. Will the organization be more likely to consolidate as a party given that it has to perform like one? In other words, can forced transition beget transformation? And will party consolidation help ensure its electoral longevity moving forward?

FIGURE 5.3. Comparing vote share under the ERP to proto-party OP-Ms

The second contribution is the introduction of the new dataset, which is specifically tailored in both its coding and composition to test theories of rebel-to-party transition. While others have conducted statistical analyses of transition in the past, previous attempts have expanded datasets that were originally curated for other purposes. As a result, ISO represents the first dataset that is constructed explicitly with rebel-to-party analyses in mind. The inclusion criteria are derived from the nuanced conceptual framework, which helps close an oft-ignored gap between composition, coding, and conceptualization.

Third, and most important for the book, the results of the analyses in this chapter provide strong evidence supporting the theory and its generalizability. While the previous chapters focused on demonstrating the mechanism by which wartime structures affect party formation, this chapter takes the critical step of demonstrating that these structures also affect party performance. As a result, not only does the new outcome variable capture political integration, but also the analyses demonstrate that proto-party structures built during war strongly improve the prospects for integration when the fighting stops.