State violence and participation in transitional justice: Evidence from Colombia

Abstract

Can the legacy of state violence undermine participation in transitional justice services designed to consolidate peace after conflict? We argue that, in the aftermath of armed fighting, state-perpetrated violence leads to reduced uptake of government reconciliation policies. We leverage spatial and temporal variation in levels of civilian victimization by perpetrator and find that, in contrast to violence committed by non-state groups, violence carried out by state forces against civilians is associated with lower levels of enlistment in Colombia’s state-run victims’ registry. We replicate this relationship using survey data and present evidence linking victimization by the state to lower levels of trust in the government overall. Together, our analyses demonstrate that disaggregating the identity of armed actors can provide significant theoretical and empirical advances in our understanding of peacemaking and post-conflict reconstruction. In the Colombian case, the legacy of state violence leads to the systematic exclusion of certain types of victims from transitional justice and undermines trust in the institutions responsible for building durable pathways to peace. Consequently, our findings have implications for the design of transitional justice policies, the study of the legacies of conflict on political and social outcomes, and processes of post-conflict peacebuilding.

Keywords

political violence reparations transitional justice trust

Participation in transitional justice and reconciliation policies can require civilians to interact with state institutions that failed to protect them or, in many cases, perpetrated violence. This simple fact creates tension for individual civilians and for whole societies. Victims may be reticent to engage with the state, even when doing so could yield material and symbolic benefits, and the state may deliberately exclude those it victimized in the past. For either or both reasons, reconciliation and reparation may remain incomplete, potentially increasing the likelihood of continued or recurrent violence.

We examine this tension empirically and test whether, all else equal, victimization by the state¹ leads to differential uptake of victim registration relative to victimization by non-state armed groups in the context of Colombia’s five-decade-long civil war. We then examine what these findings imply for our understanding of peacemaking, transitional justice, and post-conflict reconstruction. While many studies have shown that exposure to violence can increase prosociality and community group participation (Bauer et al., 2016), we examine a context where political engagement requires interacting directly with state institutions and can be subject to state consent. We hypothesize that this dynamic leads to systematically lower victim registration rates among individuals and communities that experienced higher levels of violence at the hands of state forces despite equal eligibility.

We advance the existing literature both theoretically and empirically. By disaggregating victims and their experiences by perpetrator, we highlight a pathway to create more nuanced theory that reflects the complexities of wartime and its aftermath. In considering differences by perpetrator, we also develop a better understanding of the scope of transitional justice’s impacts and ability to address inequities after violence. In particular, we suggest that when the state holds responsibility for victimization, certain individuals are more likely (whether voluntarily or involuntarily) to be excluded from transitional justice. This suggests that the legacy of violence is heterogeneous and that the potential benefits of transitional justice may accrue to only a subset of those affected by conflict.

Empirically, we leverage administrative panel data in conjunction with individual-level survey data to test our claims. This multifaceted approach allows us to capture intertemporal dynamics within and across geographic regions. It also facilitates interpersonal comparisons between victims of different armed groups and between victims and non-victims. As our main dependent variable, we consider whether victims are part of the Single Registry of Victims (Registro Único de Víctimas, hereafter RUV) held by the Victims’ Unit (Unidad para la atención y reparación integral a las víctimas). RUV inclusion is determined via a bureaucratic procedure, and those on the registry are eligible for material reparations – a transitional justice policy first implemented by the government in 2011. This outcome captures transitional justice participation – a behavior rarely studied but consequential for phenomena of broader interest to political scientists and policymakers alike, including reconciliation, democratic consolidation, and peacebuilding. The particular type of participation – inclusion in the RUV – could be considered a hard test of our proposed theory, as registration promises individual, material benefits and minimizes potential deterrent effects of other transitional justice policies (such as coming face-to-face with one’s direct perpetrator).

We report several key findings. From the analysis of the administrative data, we find that observed overall violence in a given year and municipality is strongly and positively correlated with the number of victims included in the RUV who declare the victimizing event occurred in said year and municipality. However, when we separate out and control for violence by guerrillas and paramilitaries, state-led violence is not associated with any increase in the number of registrants declaring victimization occurring in a given municipality-year. We replicate this finding at the individual level using survey data from the Latin American Public Opinion Project’s Americas Barometer (LAPOP, 2018), where respondents who were victimized are asked to identify one or more perpetrator. All else equal, individuals who self-report victimization by non-state actors exhibit greater likelihood of registering and receiving reparations compared to individuals who report state victimization (whose registration and reparation rates are statistically indistinguishable from non-victims).

We then use these data to explore whether a similar pattern can be observed between victimization by the state and trust in state-run institutions. We find that not only are state victims less likely to participate in state-run transitional justice, they also have lower trust in the government and its institutions more broadly, consistent with findings in other contexts (De Juan & Pierskalla, 2016; Wang, 2021). These trends are consequential, as distrust may undermine processes of peacebuilding and could ultimately lead to conflict recurrence. While this relationship could be a cause or consequence of reduced inclusion in transitional justice, the fact that state victimization undermines trust in state institutions and can lead to exclusion from transitional justice suggests that a segment of conflict victims may have persistent grievances with the state. Moreover, it suggests that the current deployment of transitional justice does not assuage these grievances but instead may deepen them. These findings have implications for how we conceive of conflict’s aftermath and design policies to address the legacies of human rights violations.

Over decades of civil conflict, Colombia has amassed one of the world’s largest populations of war victims. This level of victimization has necessitated transitional justice services to advocate for the rights of victims and to advance reconciliation. With over 9 million victims registered to date,² this policy constitutes the most broadly utilized transitional justice policy in existence internationally. Considering this scale, our results are troubling. Our findings suggest that the construction of transitional justice programs as a state-led enterprise can lead to systematic inequalities in the provision of justice. Exclusion from transitional justice can also have downstream consequences. For example, the receipt of compensation has been linked to changes in political engagement (Voytas, 2021). More broadly, this research highlights an ongoing need to disaggregate and deconstruct ‘victimhood’ when considering the legacies of violence and pathways to enduring peace. Though we test our expectations in the Colombian case, our findings may also describe other cases of civil war and repression where state institutions hold responsibility for civilian harm.

Perpetrator identity, participation, and trust in the state

In brief, we expect the identity of perpetrators of violence to shape how victims engage with peacebuilding and post-conflict institutions. We anticipate that state-led victimization will reduce trust in state institutions and lead to lower levels of participation in transitional justice.

Few studies directly measure engagement with such services. Instead, research often focuses on attitudes toward transitional justice and finds that victims generally support reparative policies (Nussio, Rettberg & Ugarriza, 2015; Gibson, 2002; Nalepa, 2010; Laplante & Theidon, 2007; David & Choi, 2009). However, research also suggests that heterogeneous experiences predict distinct transitional justice preferences (Pham, Weinstein & Longman, 2004). For example, Hall et al. (2018) show that while exposure to violence heightens retributive desires, contemporary contact with perpetrators increases support for restorative justice. In Spain, Aguilar, Balcells & Cebolla-Boado (2011) suggest that ideology, familial victimization, and regional context predict attitudes toward transitional justice. Samii (2013) similarly finds that partisanship shapes transitional justice preferences in postwar Burundi. In Colombia, Tellez (2019) and Branton et al. (2019) find that exposure to conflict influences support for the peace process and concessions to armed actors, while Garbiras-Diaz, Garcia-Sanchez & Matanock (2020) and Kreiman & Masullo (2020) suggest that support is contingent on elite cues and the faction responsible for victimization, respectively.

Another related body of work examines political participation after violence but rarely considers transitional justice participation. Many studies document heightened political engagement and prosociality after violence (Blattman, 2009; Bauer et al., 2016; Bateson, 2012). However, others find negative consequences of violence on political behavior (Zkuhov & Talibova, 2018; Balcells, 2012). The mixed findings suggest a complex relationship between violence and political engagement. By emphasizing perpetrator identities, we join a growing number of scholars that take a more nuanced approach, examining the heterogeneous effects of violence exposure. This work has found that the relationship between violence and political behavior depends on gender (Hadzic & Tavits, 2019), kinship networks (Dorff, 2017), and armed groups’ strategies (Gallego, 2018), to name a few.

We build on these literatures in three ways. First, we extend the notion that differences in victimization matter for subsequent political attitudes and behavior and argue that the identity of the perpetrator is an important dimension of conflict experiences. Second, we expand the repertoires of political behavior to consider participation in transitional justice, a highly consequential form of engagement in conflict and post-conflict environments. Third, we complement data on reported attitudes with behavioral engagement.

When violence was perpetrated by non-state groups, we expect transitional justice participation to be relatively straightforward and to offer important benefits. By engaging in transitional justice, victims can receive recognition of their suffering. After violence, many victims face severe economic hardship, and the material benefit from some transitional justice policies is sorely needed (e.g. material reparations). When individuals were victimized by the state, however, the dynamics of transitional justice become more complicated. Participation might require victims to interact with representatives of the same entity responsible for the abuses, and in some cases, to denounce their perpetrator, an uncomfortable and potentially perilous task. Engaging with state entities might be perceived as especially dangerous when parts of the state remain capable of targeting citizens with future violence (Rozenas & Zhukov, 2019). State entities could also be reluctant to validate victim accounts of violence by state forces because verification implicitly acknowledges and recognizes state guilt. Consequently, those victimized by state forces may be unable to participate fully in transitional justice. For these reasons, we advance our first hypothesis:

Hypothesis 1: Victimization by state forces will be associated with lower levels of transitional justice engagement compared to victimization by non-state groups.

This pattern may in part be caused by, and also contribute to, lowered trust in state institutions. As De Juan & Pierskalla (2016) point out, it is especially in fragile and conflict-affected contexts where trust in state institutions is of importance. Belief in the government’s ability and willingness to stand by bargains made during negotiations can influence support for the peace process and reduce the risk of repeat violence (Sacks & Larizza, 2012; Hutchison & Johnson, 2011).

Experiencing violence is traumatic and can leave a lasting impact on those affected. Thus, post-violence attitudes are unlikely to shift easily. Distrust is likely front of mind when one considers engaging with an institution affiliated with perpetrating past harm. Among those victimized by the state, we expect potent memories of violence to transcend the specific violent episode and to continue to define one’s relationship with the state more broadly, even when the specific state entities or representatives differ (Wang, 2021).

Existing literature on the relationship between wartime violence and trust links conflict exposure to lower levels of generalized trust (De Luca & Verpoorten, 2015; Kijewski & Freitag, 2018; Rohner, Thoenig & Zilibotti, 2013). Across a range of contexts, citizens also lose faith in the actors and institutions perpetrating or failing to prevent violence (Gates & Justesen, 2013; Ishiyama et al., 2018). Kreutz & Nussio (2019) suggest that in Colombia, the government’s policy of extradition to the United States eroded trust among former combatants and Kaplan & Nussio (2018) highlight how government distrust can fuel recidivism.

Our analysis examines the context of conflict victims. Overall, we expect that victims of state-perpetrated violence will have lower trust in the state. When violence is perpetrated by non-state entities, such as guerrilla groups, we do not expect victims to react to state institutions and policies in the same way. Though trust might be damaged because state entities failed to protect innocent civilians, views of the government will be less negative when the state did not directly carry out violence or when violence was predominantly carried out by non-state groups. In this setting, individuals may turn to political systems to obtain personal benefits, such as material ones that accrue from reparations policies.

Trust in state institutions is important in its own right, but also illuminates some of the mechanisms which beget unequal access to transitional justice and other forms of civic engagement. Scholars highlight how certain groups are less likely to claim their rights from the state (Kruks-Wisner, 2018; Abrego, 2008) and may be skeptical of government assistance (Ewick & Silbey, 1998). The same logic can be extended to victim populations in post-conflict contexts who differentially claim reparative benefits. As suggested by the literature on trust and civic participation generally (Levi & Stoker, 2000) and on compliance with regulation (e.g. taxation) (Fjeldstad, 2004), citizens’ decisions to comply and participate often rest on the assumption that the government is trustworthy. In post-conflict settings distrust can be rampant and even the financial incentives associated with some transitional justice policies may not be enough to overcome the barriers to entry.

At the same time, trust can be impacted by differential transitional justice experiences. Scholars have noted how citizens’ experiences engaging with public policies shape subsequent attitudes toward the government (Pierson, 1993; Mettler, 2005). Interactions with the state inform an individual’s evaluation of government trustworthiness (Soss, 1999; Weaver & Lerman, 2010). In post-conflict Sierra Leone, Wong (2016) found that service delivery increased trust only when it mirrored citizens’ needs. If impacts on trust depend on government performance during policy implementation, experiences with transitional justice could further lower trust among state victims. This could happen, for example, if state victims appeal for inclusion in transitional justice but are denied (although in recent years approximately 93% of victims’ registrations have been approved (Rivas, 2016)). Because state-run transitional justice policies operate through government employees, this presents an opportunity for strategic exclusion of those accusing the state of wrongdoing.

Together, these expectations lead us to our second hypothesis, which could be a cause or a consequence of the patterns described in Hypothesis 1, and which we examine using individual-level survey data in Colombia:

Hypothesis 2: Victimization by state forces will be associated with lower levels of trust in the state.

Regardless of the pathway, the observed systematic under-representation of a particular subset of victims from justice and their elevated distrust constitutes an issue that is at once normatively important, theoretically informative, and indicative of an under-recognized failure of public policy in fragile contexts. The present article focuses on the high-level patterns between identities of perpetrators, trust, and inclusion in transitional justice, though we elaborate on possible mechanisms in the empirics section. Further research will be required to provide causal weight to either proposed pathway and inform potential policy responses.

It is worth emphasizing that citizens may be victimized by multiple perpetrators or unable to identify a responsible group at all.³ Still, we expect that any experiences of state-led victimization will negatively impact

engagement with transitional justice and reduce trust in the state and its institutions. Put differently, ceteris paribus, victimization by state forces undermines participation in state-run reparative policies. Our empirical strategy adopts this perspective, allowing us to consider the marginal association between experiences of state violence and registration as a victim accounting for victimization by non-state groups.

Conflict and victimization in Colombia

We expect the logic outlined above to extend to a wide range of conflict and post-conflict settings internationally where state forces carried out violence. In these cases, we believe that similar dynamics may create barriers and preclude certain victims from accessing transitional justice. These obstacles can be particularly prohibitive for those abused by the state, who must encounter their perpetrator’s affiliates (when transitional justice is state-run) as a prerequisite to participation.

Reparations policies have been administered nearly 200 times during conflict or in countries transitioning to peace or democracy (Olsen, Payne & Reiter, 2010; Loyle & Binningsbo, 2018) and have become an increasingly common provision in peace agreements (see lower panel in Figure 1). Beyond compensation, the results we find and mechanisms we examine are relevant for other commonly implemented state-run transitional justice policies such as truth commissions and trials. These initiatives are premised on victim participation or testimony and involve state cooperation – stipulations that can be complicated for any victim and for victims of state forces in particular.

We also expect the findings to apply to many post-violence cases where reparations have been administered. However, we note several important dimensions of the Colombian case. First, Colombia’s policy is unique in its size and scope, which could perhaps contribute to the idea that registering is futile (few victims, regardless of perpetrator, have been paid) but could alternatively increase the policy’s visibility, making registration more common. Though its size is atypical, Colombia’s reparations program consequently has a large number of stakeholders, making understanding its workings all the more important. Second, the RUV was established prior to Colombia’s peace deal, meaning that victims could register before or after its signing. While Loyle & Binningsbo (2018) have documented that transitional justice during conflict is in fact quite common globally, participation while conflict is ongoing may differ from participation in policies implemented during transitions. Future work will examine how levels of violence affect uptake of during- and post-conflict justice.

Despite these caveats, contemporary Colombia provides an ideal environment to test our expectations. During the conflict, both state and non-state entities carried out violence, allowing us to compare the associations between each type and total registration while holding other variables constant. Doing so allows us to account for many relevant omitted variables that might drive the observed relationships. Additionally, the duration and intensity of Colombia’s conflict make the fear of recurrence particularly salient. Thus, understanding the impacts of policies designed to build peace and prevent repeat violence is critical. In this section, we further describe the Colombian case and justify our dependent variable: inclusion as a victim in the RUV held by the state-run Victims’ Unit.

Prior to the 2016 peace deal, Colombia had been engaged in armed conflict for decades. The conflict has involved multiple groups: paramilitary forces, guerrilla groups, crime syndicates, and the government. Dynamics of violence changed considerably throughout the conflict and can generally be categorized into four distinct time periods (Ch et al., 2018). In the late 1980s to mid-1990s, violence was largely perpetrated by the FARC. During this time, no victim registry existed. By the time our data on victim registration begin in 2001, paramilitaries were increasingly powerful and had consolidated under the umbrella of United Self-Defense Forces of Colombia (AUC). The share of victimization by paramilitary groups soon decreased as paramilitaries agreed to a ceasefire and demobilized in 2003–05. From late 2006 to 2010, the Colombian military and police increasingly established a presence throughout the country, while the FARC weakened. In 2012, the FARC and government began peace talks.

The factors that make Colombia an ideal testing ground for our hypotheses also introduce some theoretical and empirical complexity. First, the documented relationships between the paramilitaries and several arms of the Colombian government likely shape how victims assess the state and its services – particularly those related to the conflict. However, there are many reasons why experiences of victimization by paramilitaries may be meaningfully different from victimization by state forces themselves. For instance, states often sanction paramilitaries to give themselves plausible deniability in instances of violence. Moreover, the explicit connection between high-level political actors and the paramilitaries was relatively less well known until the onset of the ‘Parapolitics’ scandal. After this scandal, the state underwent a highly publicized campaign of arrests and prosecutions of those involved. Thus, while this relationship is certainly important, we do not have strong expectations regarding the extent to which victims will treat paramilitaries as state or non-state actors in our framework. Empirically, we code paramilitary and state violence as separate categories and let the data speak for themselves.

Second, while we examine state-perpetrated violence, it is worth mentioning that we are specifically interested in violence carried out by armed entities. While some state entities were responsible for human rights violations in the Colombian case, others, such as the Defensoría del Pueblo, the Procuraduría General de la Nación, and individual judges, sought to sanction and curb state repression (Grupo de Memoria Histórica, 2013).

Indeed, in 1997, the government established the Single Registry of Displaced Populations for victims of forced disappearance (Rivas, 2016). In 2011, the Victims’ Law mandated an expanded database, the Single Registry of Victims in Colombia, which includes victims of forced disappearance and sequestration, torture or inhumane treatment, sexual violence, forced recruitment, forced displacement, assassination, and other violence that produced injuries (see Figure A1 and Table A19 in the Online appendix for a current breakdown of victim type). In cases of assassination and forced displacement, family members (spouses/partners, children, and parents) of direct victims are also eligible to register. When the criteria expanded, victims of newly eligible crimes could register for events dating back to 1985; we examine the year of reported event occurrence. Victims can contact their closest Victims’ Unit office (there are over 200) and request to join the RUV virtually, by mail, in-person, or over the phone. To enlist, victims must present the required documentation (identification and two witness declarations) and provide information about the timing of victimization. The Victims’ Unit then verifies a victim’s application by cross-referencing her account with databases within the National Information Network for Care and Reparation for Victims.⁴ Victims are eligible regardless of the identity of their perpetrator, but registration is voluntary, requiring victims to take the actions listed above. Once an application is filed, according to the 2011 Victims’ Law:

In the event that the victim mentions the name or names of the potential perpetrator of the damage he claims to have suffered […] this name or names shall in no case be included in the administrative act by which the registration is granted or denied. (Ley de Víctimas y Restitución de Tierras, Law 1448)

Once the case has been reviewed, the state issues a mailed administrative notice of approval or rejection within 120 days.

After victims initiate the reparation process, they receive a personalized note from the government confirming inclusion in the RUV and expressing its commitment to reparation. For most, this is the sole response received, as only 11.1% of victims had been paid as of April 2020 (Unidad para las Victimas, 2020).⁵ We are concerned primarily with an antecedent to compensation: inclusion in the RUV.

Empirical strategy

We argue that victims of state violence will have reduced trust in the government and be less likely to engage in transitional justice relative to those victimized by non-state actors. We test these hypotheses across multiple sources of data and find support in each. We begin by examining the association with recorded incidents of victimization and registration as a victim at the municipality-year level from 2001 to 2017. We then replicate these findings and provide initial evidence for the trust mechanism using individual survey responses from two rounds of the LAPOP Americas Barometer (LAPOP, 2018).

Municipality panel analysis

The two main data sources for the municipality-year analysis are (1) El Centro de Investigación y Educación Popular’s (CINEP) violent events database⁶ and (2) administrative data on victim registrations obtained from Colombia’s Victims’ Unit. Trends for these data are presented in Figure 2. Based on regional and national press coverage, the CINEP data catalogs location, number of victims, type of human rights violation, and groups identified as being involved in the event. Our main independent variables are the total number of victims in a given municipality-year that are associated with each of our three categories of armed actors: guerrillas, paramilitaries, and the state.⁷ We conduct a similar exercise for the many guerrilla and paramilitary groups and aggregate the total number of victims in events attributed to each category of armed actor to the municipality-year level. Consequently, some events involve multiple armed groups. For instances involving state actors, 74% are only attributed to state actors. Such cases often involve instances of human rights violations such as arbitrary detention, extrajudicial killing, sexual assault, or torture.⁸ Many of these cases include the misrepresentation and extrajudicial execution of civilians as guerrilla members occurring during the False Positives scandal (Acemoglu et al., 2020). We also account for victims in cases where the perpetrator is unidentified and control for these separately. Our dependent variable is constructed by aggregating the total number of registered individuals who report their victimization occurred in each municipality-year. We adopt this strategy for two reasons. First, we can use the reported date of victimization itself to merge and then compare victim registration and CINEP violence data at the municipality and year of event level. Second, on a practical level, our dataset does not include other temporal information, such as the

date a victim declared to the Victims’ Unit, which is unknown to us. Because individuals who have been displaced may not register in the location where victimization occurred, and because the coverage of non-displacement registration varies over time, we test our hypotheses using the total number of registrations and excluding individuals whose main grievance is displacement. A breakdown of injury recorded in the registry can be found in Figure A1 in the Online appendix.

The total number of registered victims often exceeds the number of observed victims from the CINEP data in part due to imperfect coverage by CINEP but also because the victim registry can include individuals’ family members indirectly affected by the conflict. Nevertheless, we see in Figure 3 that the total number of observed victims per 10,000pop is highly correlated with the level of registered victims. Our goal is to examine whether this relationship varies depending on who carried out violence.

We first estimate a fixed-effects specification following Equation 1. We test whether victimization $V$ by armed group $\in {G, P, S, U}$ or guerrilla, paramilitary,

state, or unidentified perpetrator respectively, is associated with levels of victim registration for events occurring in a municipality (

m

) in year (

t

) controlling for time-invariant municipality-specific characteristics (

δ_{m}

), and year-specific fixed-effects (

γ_{t}

). We also control for the

number of victims where no perpetrator was identified, which we denote as ‘unidentified perpetrator’, or U-P. The central challenge in identifying the causal effect of violence, at the individual or administrative level, is that violence is a strategic outcome and therefore inherently selective. We include municipality fixed effects to remove time-invariant geographical factors which may influence levels of violence and registration. We also include year fixed-effects to remove year-on-year variation in overall levels of violence. We implicitly assume that the remaining systematic variation in victim registration can be captured by observed levels of victimization by armed groups – including the state. However, our findings are robust to the inclusion of additional time-varying controls and to the exclusion of reports where the main grievance was displacement. We discuss these and other robustness checks in more detail below and in the Online appendices.

{Registration}_{m t} = β_{G} V_{m t}^{G} + β_{P} V_{m t}^{P} + β_{S} V_{m t}^{S} + β_{S} V_{m t}^{U} + δ_{m} + γ_{t} + ε_{m t}

To examine Hypothesis 2, we test the null hypothesis that the coefficient on the number of government victims ( $β_{S}$ ) is the same as the coefficient on the number of guerrilla victims ( $β_{G}$ ). Our theoretical framework predicts that $β_{S}$ will be substantively and significantly less than $β_{G}$ . Figure 4 and Table I present our main findings. Each panel of Figure 4 presents the coefficients from Column (7) of the table and data points adjusted for each other variable. Because victim registration and victim counts are not normally distributed, we present findings using several data transformations and the results are similar throughout. For each specification, we calculate heteroskedasticity consistent robust standard errors clustered at the municipality level.

Across all specifications, violence attributable to guerrillas or to paramilitaries is positively associated with the total number of victims who eventually register with the state and declare the victimizing event occurred in a given year. In contrast, the estimated coefficient for state-led violence is often insignificant at conventional levels and can be positive or negative depending on the specification. We use Chi-squared tests to compare $β_{S}$ and $β_{G}$ . The relationship between victimization by the state and RUV year of occurrence is significantly less than the relationship between victimization by guerrilla groups and inclusion at the .1% level across all specifications. This finding provides strong support for Hypothesis 2.

In Online appendix Tables A1–A4, we present robustness checks to this main result, examining whether the removal of year-specific effects, different lag structures (to account for historical trends in victimization), matching on treatment histories (Imai, Kim & Wang, 2018), and time-varying municipality characteristics change our inference. Altogether, the results are consistent with the outcome of our primary specifications. These findings provide strong evidence in support of the idea that the nature of victimization shapes engagement in transitional justice policies and that the identity of the perpetrator in some way determines the size of that effect.

Individual-level analysis

To test this association at the individual level, we use survey data from the 2014 and 2016 rounds of LAPOP’s

Table I.

Two-way fixed effects estimates of victimization on registration

	Total registration				Excluding displacement
	Raw (1)	Per cap. (2)	$log (y)$ (3)	arcsinh( $y$ ) (4)	Raw (5)	Per cap. (6)	$log (y)$ (7)	arcsinh( $y$ ) (8)
State victims	$-$ 8.684	$-$ 3.484	0.044	0.016	1.729 $^{* *}$	$-$ 0.163	0.056	0.046
	(7.035)	(7.244)	(0.085)	(0.070)	(0.748)	(0.617)	(0.078)	(0.074)
Guerrilla victims	27.147 $^{* * *}$	33.541 $^{* * *}$	0.629 $^{* * *}$	0.494 $^{* * *}$	5.543 $^{* * *}$	3.451 $^{* * *}$	0.638 $^{* * *}$	0.569 $^{* * *}$
	(7.005)	(9.808)	(0.094)	(0.077)	(0.891)	(1.080)	(0.074)	(0.067)
Paramilitary victims	26.321 $^{* * *}$	43.319 $^{* * *}$	1.020 $^{* * *}$	0.824 $^{* * *}$	1.978 $^{* * *}$	3.027 $^{* * *}$	0.688 $^{* * *}$	0.617 $^{* * *}$
	(6.237)	(10.000)	(0.115)	(0.097)	(0.567)	(0.825)	(0.082)	(0.078)
U-P victims	17.052 $^{* * *}$	15.202	0.775 $^{* * *}$	0.651 $^{* * *}$	7.931 $^{* * *}$	2.464	0.796 $^{* * *}$	0.748 $^{* * *}$
	(5.710)	(10.803)	(0.172)	(0.152)	(1.865)	(1.528)	(0.121)	(0.115)
$χ^{2} : (β_{S} = β_{G})$	149.67	232.66	46.96	39.27	94.46	131.23	75.28	62.26
$P r (> χ^{2})$	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000
Municipio FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Clusters	1,172	1,172	1,172	1,172	1,172	1,172	1,172	1,172
Observations	16,890	16,030	16,030	16,030	16,890	16,030	16,030	16,030
R-squared	0.085	0.168	0.060	0.048	0.312	0.083	0.068	0.056

Heteroskedasticity consistent robust standard errors clustered at the municipality level in parentheses. Columns (1) through (4) present results where the dependent variable is a transformation of the total number of registered victims per municipality-year. Columns (5) and (6) exclude victims registering as displaced. Columns (1) and (5) present estimates from raw count figures; in Columns (2) and (6) all key variables are per-capitized; in (3) and (7) variables are per-capitized and then logged; finally (4) and (8) present results using the inverse hyperbolic sine transformation of our key variables. The row denoted $χ^{2} (β_{S} = β_{G})$ presents the $χ^{2}$ of a linear hypothesis test of whether the coefficient on the state victims $(β_{S})$ is the same as the coefficient on the variable measuring victims of guerrilla forces ( $β_{G}$ ). Estimation includes both municipality and year fixed effects.

Significance levels: *** .1%; ** 1%; * 5%.

Americas Barometer in Colombia. For each round we have basic respondent demographic characteristics, trust in state institutions, and several questions related to experiences of victimization. For the 2016 round, we also have information about victim registration and whether the respondent has received reparations.

We first estimate a regression specification analogous to Equation (1) where individual-level victim registration status is regressed on dummy variables indicating whether the respondent self-reports as a victim of a particular group⁹ as well as basic demographic information (sex, age, a quadratic age term, ethnicity, years of education, employment status, and marital status) using the 2016 data. Column (2) of Table II reports this analysis. Individuals who self-report as victims of the guerrillas or victims of the paramilitaries are more likely than non-victims to register as a victim conditional on demographic characteristics. In contrast, victimization by the state is statistically indistinguishable from non-victims in terms of registration rates. A similar pattern holds for whether individuals received reparations, suggesting that lower registration among those victimized by the state implies foregoing potential income. These data also allow us to examine one of our proposed mechanisms, that state victims have lower trust and decreased participation in forms of engagement which are predicated on that trust, such as registration as a victim.

We test whether victimization by different actors is associated with levels of trust in state institutions (compared to non-victims) using ordinary least squares (OLS) regressions controlling for demographic factors and including primary sampling unit (PSU)-Wave fixed-effects and standard errors clustered at the PSU-Wave level.¹⁰ Figure 5 plots estimated coefficients of guerrilla, paramilitary, and state victimization on trust in the judiciary, armed forces, congress, police, etc., as well as an average index of each of these

Table II.

Individual victimization, registration, and reparations

	Trust index	Register as victim	Receive reparations
	(1)	(2)	(3)
Victim of:
State	$- {0.507}^{*}$	$0.026$	$- 0.047$
	$(0.242)$	$(0.067)$	$(0.037)$
Guerrilla	$- 0.106$	${0.173}^{* * *}$	$0.032$
	$(0.090)$	$(0.028)$	$(0.017)$
Paramilitary	$0.005$	${0.159}^{* * *}$	${0.085}^{*}$
	$(0.114)$	$(0.040)$	$(0.033)$
Other	$0.051$	${0.079}^{*}$	$0.022$
	$(0.112)$	$(0.035)$	$(0.024)$
Demographics:
Female	$- 0.010$	$0.019$	${0.026}^{*}$
	$(0.072)$	$(0.014)$	$(0.011)$
Age	$- 0.024$	$- 0.000$	$0.002$
	$(0.015)$	$(0.003)$	$(0.002)$
Age²	${0.000}^{*}$	$0.000$	$- 0.000$
	$(0.000)$	$(0.000)$	$(0.000)$
Eth: Indigenous	$0.045$	$- 0.001$	$0.026$
	$(0.281)$	$(0.066)$	$(0.056)$
Eth: Mestizo	$- 0.041$	$- 0.012$	$- 0.014$
	$(0.162)$	$(0.043)$	$(0.036)$
Eth: Mulatto	$- 0.119$	$- 0.009$	$0.021$
	$(0.206)$	$(0.053)$	$(0.048)$
Eth: Other	$- 0.270$	$- 0.028$	$- 0.007$
	$(0.236)$	$(0.059)$	$(0.041)$
Eth: White	$- 0.051$	$- 0.023$	$0.008$
	$(0.172)$	$(0.043)$	$(0.036)$
Years educ.	$- {0.026}^{*}$	$- {0.007}^{* *}$	$- 0.002$
	$(0.011)$	$(0.002)$	$(0.002)$
Unemployed	$- 0.037$	$0.006$	$0.004$
	$(0.102)$	$(0.023)$	$(0.017)$
Married	$- 0.028$	$0.008$	$0.018$
	$(0.078)$	$(0.017)$	$(0.012)$
FE	YES	YES	YES
R $^{2}$	$0.293$	$0.384$	$0.234$
No. obs.	$1470$	$1471$	$1471$
N Clusters	$251$	$251$	$251$

Heteroskedasticity consistent robust standard errors clustered at the PSU-Survey Round level in parentheses. Estimates based on LAPOP’s 2016 survey round (registration and reparation questions were not asked in earlier round) and include PSU fixed effects.

Significance levels: *** .1%; ** 1%; * 5%.

seven-point Likert-type items. The data include pooled observations from the 2014 and 2016 survey rounds. Overall, victimization by guerrillas is not systematically associated with trust in state institutions. Similarly, while paramilitary victimization is associated with lower trust in the police, there is no systematic relationship. In contrast, individuals who have been victimized by the state report lower levels of trust in the judiciary, armed forces, congress, police, the president, and elections. These coefficients are significant at the 5% level for the judiciary, the police, and elections. Those reporting state victimization also score lower on an index averaging across trust in all institutions. A full tabular presentation of these results may be found in Table A5 (Online appendix).

We want to highlight that these results are derived from a small sample. Our analysis is necessarily limited to those surveyed by LAPOP, and while that restricts the precision and confidence of our estimates, they nonetheless provide evidence in line with both our theoretical expectations and municipality-level analyses. These results are also associational. While we control for a set of pre-treatment, observable characteristics, there are likely many unobserved qualities that make individuals more or less likely to be victimized by particular groups. Though we cannot directly examine the effects of these characteristics, we can use coefficient stability approaches to assess the degree to which selection on unobserved characteristics might bias our results and limit our ability to make causal claims using the approach presented in Oster (2019). More details on this test can be found in the Online appendix, but we find that to reduce our observed effect to zero, we would need to be able to account for more than twice the variation in our trust index compared to our controlled regression, and selection on unobservable characteristics would need to be twice as strong as selection based on observed variables and operate in the opposite direction. This analysis suggests that this relationship is not particularly sensitive to selection on unobservables.

Potential pathways

The purpose of this article is to shed light on high-level disparities in transitional justice uptake among potential stakeholders. We described two possible pathways behind this phenomenon. The first is a demand-side mechanism wherein victim experiences of violence committed by the state lead to negative perceptions of the state and politics and, therefore, lower participation in state-run transitional justice. The second is from the supply side: the state may – deliberately or otherwise – de-prioritize the registration of individuals who declare to be victims of the state. While we cannot adjudicate between these two mechanisms empirically, this section expands on these ideas for the purposes of guiding future research.

Selection into reporting of state-led victimization

Victims may choose to not register as a victim due to a

lack of trust in the state and an unwillingness to engage with it. This decision could be a rebuke of the policy (Cronin-Furman & Krystalli, 2020; Sandoval Rojas, 2015). It may also be generated by fear, which is a conclusion reached by the Red Cross in Colombia. They write that underreporting arises because those affected ‘have fear of speaking, a lack of resources, or are unaware of the options for recourse’ (Comite Internacional de la Cruz Roja, 2017). It is easy to imagine how speaking out would be particularly tolling for those who were victimized by the state, the very entity to which they must report their abuses. Remaining silent in this context can be a way for victims to protect themselves, even if silence means exclusion from potential material benefits (Riano-Alcalá & Baines, 2011). Additionally, if experiences of violence erode trust in the state more generally, it may be that victims do not believe the state will uphold its end of the bargain. Thus, from their perspective, registration may constitute a ‘no-win’ scenario in which they must make themselves more visible to the state with little expectation of actual compensation.

Selective approval of RUV applications

As discussed earlier, the 2011 Victims’ Law prohibits exclusion based on a perpetrator’s identity. Still, an alternative explanation is that state victimization is reported but is excluded from the RUV through the Victims’ Unit approval process. This could generate both the lower inclusion rates we document and lower state trust. Though it is difficult to document deliberate exclusion of state victims, it is worth noting that the Victims’ Unit has received hundreds of thousands of tutelas, or legal measures taken by citizens when their rights are violated or threatened (El Tiempo, 2016). Many of these are initiated from cases rejected by the Victims’ Unit. Disappointment in the experience is common, even when individuals are included in the RUV (Pham et al., 2016; Cronin-Furman & Krystalli, 2020). The Victims’ Unit has been criticized by public officials who suggest that it fails to assist all victims (El Tiempo, 2019). Though we cannot and do not wish to assert that the state is deliberately excluding victims, we note that this is an additional possible cause for lower enlistment among state victims. We also note that in recent years approximately 93% of victims’ registrations have been approved (Rivas, 2016). Either or both of the two possible explanations reviewed are consistent with the empirical analysis.

Registration capacity

An additional supply-side mechanism could be due to limits on the capacity to register victims that are correlated with incidences of state violence. For instance, if lower state capacity is correlated with both higher incidences of civilian victimization and lower registration capacity, then we would expect to see lower victim registration when controlling for instances of state-led violence. However, in our robustness checks, we control for municipality-level government spending, a weak proxy for state capacity, and find no systematic correlation with registration rates.

Conclusion

Across administrative and survey data we find that the consequences of victimization during civil war vary by perpetrator. In contrast to victimization by non-state actors, state violence does not lead to a significant rise in transitional justice engagement. Though previous work has documented overall increases or decreases in political engagement after violence, we join a growing number of scholars suggesting that the effects of violence are context dependent. Our results suggest that understanding the legacies of conflict requires a more nuanced look at the processes of civilian victimization. In particular, victimization by state forces during conflict has implications for state legitimacy and consolidation of peace after conflict. Given that victims of state-led violence access governmental benefits at lower rates than other victims, they may continue to distrust and feel aggrieved by the state, posing challenges in post-conflict settings prone to conflict reversion.

This is not to say that victimization by non-state actors is less destructive than state-perpetrated violence. Unfortunately, violence perpetrated by state and non-state forces has negatively affected millions of Colombians. Even once victims manage to be included in the RUV, the vast majority await state assistance. Those who do receive it may find the amount unable to meaningfully reduce their economic vulnerability. We emphasize the hardships of accessing transitional justice, but note that further difficulties may arise when it comes to the implementation of transitional justice policies themselves.

It is difficult to say whether the patterns we find will travel beyond the Colombian case and beyond the specific context of transitional justice engagement. However, the heterogeneity we detect could more broadly shape engagement in peacebuilding and post-conflict institutions if the patterns we revealed generalize to other contexts where the state perpetrates violence (heightening the psychological burden of participating) and where political engagement requires interacting with governmental representatives (bringing victims in proximity with the perpetrating institution). An example of participation that fits these criteria is turning out to vote, and indeed, scholars find lower electoral participation after state repression in Russia and Ukraine (Zkuhov & Talibova, 2018). Other types of mobilization may operate differently: recent work links state violence to involvement in civil defense forces (Schubiger, 2021). Taken together, if experiences of state violence simultaneously reduce transitional justice engagement and drive citizens to take up arms, the implications for conflict reversion and peace are more dire.

We have shown that experiences of state-sponsored violence reduce state trust after violence and transitional justice engagement. These patterns may arise because victims link state violence to the entities overseeing reparations. We thus expect our results will be more likely to apply when this link is evident. Perceived and actual connections between violence and transitional justice may be greatest when violence is ongoing and where settlements are negotiated, with perpetrators retaining sizable political influence in peacetime. Transitional justice engagement and political participation may be more appealing when victims can be sure that perpetrators have fully left power and demobilized. Research examining cases of clear victory or other clean breaks from past regimes will help to explore this possibility.

Our analysis has implications for the study and design of transitional justice policies. Increasingly, researchers are evaluating how these policies shape individuals’ political behavior and attitudes (Cilliers, Dube & Siddiqi, 2016; Balcells, Palanza & Voytas, 2022). In addition to considering these important dimensions of transitional justice, we should emphasize the process of participating in transitional justice policies from a victim’s perspective. Our results suggest that those for whom reconciliation is often most needed (e.g. between victims of the state and the state itself) are the least likely to actually engage in transitional justice. Consequently, the ideal transitional justice regime may differ according to who perpetrated violence. When state entities are perpetrators, it may be advantageous to have international or non-state groups involved to a greater extent in the process so that all victims can benefit from transitional justice measures without having to confront and engage with an entity linked to their victimization. Alternatively, states and other organizations should explicitly cater to victims of state violence, recognizing that distrust likely exists. Involving victims like these in the design of transitional justice policies may help to signal that the state is trustworthy and intent on helping victims. More generally, analysis of the uptake of transitional justice policies is a critical antecedent to understanding the practical use and consequences of transitional justice programs.

Though we believe our disaggregation is a step in the right direction, much more remains to be done. We examine one dimension of heterogeneity: perpetrator identity. But victims in Colombia and elsewhere differ along many additional dimensions, including gender, socio-economic status, and victimization experience (Rettberg, 2015). They have different knowledge of or encounters with other state entities and pro-victim initiatives that could help mitigate negative perceptions of the state. These victim-level characteristics may also generate differential rates of inclusion in transitional justice policies. By considering perpetrator identities, we hope to encourage future research to disaggregate victim populations and examine inequalities in the reach of policies meant to aid victims.

Footnotes

Replication data

The dataset, codebook, and code for the empirical analysis in this article, along with the Online appendix, can be found at http://www.prio.org/jpr/datasets.

Acknowledgments

We thank numerous anonymous reviewers, Tomás Flores, Oliver Kaplan, Jake Shapiro, and audiences at MPSA, ESOC, and the Princeton Latin American Politics Workshop for valuable feedback.

ORCID iD

Elsa Voytas

Notes

References

Abrego

Leisy

(2008) Legitimacy, social identity, and the mobilization of law: The effects of Assembly Bill 540 on undocumented students in California. Law & Society Review 33(3): 709–734.

Acemoglu

Daron

Fergusson

Leopoldo

Robinson

James

Romero

Dario

Vargas

Juan F

(2020) The perils of high-powered incentives: Evidence from Colombia’s false positives. American Economic Journal: Economic Policy 12(3): 1–43.

Aguilar

Paloma

Balcells

Laia

Cebolla-Boado

Hector

(2011) Determinants of attitudes toward transitional justice: An empirical analysis of the Spanish case. Comparative Political Studies 44(10): 1397–1430.

Balcells

Laia

(2012) The consequences of victimization on political identities: Evidence from Spain. Politics & Society 40(3): 311–347.

Balcells

Laia

Palanza

Valeria

Voytas

Elsa

(2022) Do transitional justice museums persuade visitors? Evidence from a field experiment. Journal of Politics 84(1): 496–510.

Bateson

Regina

(2012) Crime victimization and political participation. American Political Science Review 106(3): 570–587.

Bauer

Michal

Blattman

Christopher

Chytilová

Julie

Henrich

Joseph

Miguel

Edward

Mitts

Tamar

(2016) Can war foster cooperation? Journal of Economic Perspectives 30(3): 249–274.

Blattman

Christopher

(2009) From violence to voting: War and political participation in Uganda. American Political Science Review 103(2): 231–247.

Branton

Regina

; Jacqueline Demeritt, Amalia Pulido & James Meernik (2019) Violence, voting & peace: Explaining public support for the peace referendum in Colombia. Electoral Studies 61(4): 102067.

10.

Rafael

Shapiro

Jacob

Steele

Abbey

Vargas

Juan

(2018) Endogenous taxation in ongoing internal conflict: The case of Colombia. American Political Science Review 112(4): 996–1015.

11.

Cilliers

Jacobus

Dube

Oeindrila

Siddiqi

Bilal

(2016) Reconciling after civil conflict increases social capital but decreases individual well-being. Science 352(6287): 787–794.

12.

Comite Internacional de la Cruz Roja (2017) En Colombia hay mas de 20.000 victimas de violencia sexual cometida en el marco del conflicto armado [In Colombia, there are more than 20,000 victims of sexual violence committed during the course of the armed conflict].

13.

Cronin-Furman

Kate

Krystalli

Roxani

(2020) The things they carry: Victims’ documentation of forced disappearance in Colombia and Sri Lanka. European Journal of International Relations 27(1): 79–101.

14.

David

Roman

Choi

Susanne

(2009) Getting even or getting equal? Retributive desires and transitional justice. Political Psychology 30(2): 161–192.

15.

De Juan

Alexander

Pierskalla

Jan Henryk

(2016) Civil war violence and political trust: Microlevel evidence from Nepal. Conflict Management and Peace Science 33(1): 67–88.

16.

De Luca

Giacomo

Verpoorten

Marijke

(2015) Civil war, social capital and resilience in Uganda. Oxford Economic Papers 67(3): 661–686.

17.

Dorff

Cassy

(2017) Violence, kinship networks, and political resilience: Evidence from Mexico. Journal of Peace Research 54(4): 558–573.

18.

Tiempo

(2016) Las mas de 570.000 tutelas que asfixian a la unidad de victimas [The more than 570.000 tutelas that suffocate the Victims’ Unit].

19.

Tiempo

(2019) En 8 anos solo se ha reparado el 13 por ciento de las victimas [In 8 years, only 13 percent of victims have been received reparations].

20.

Ewick

Patricia

Silbey

Susan

(1998) The Common Place of Law: Stories from Everyday Life. Chicage, IL: University of Chicago Press.

21.

Fjeldstad

Odd-Helge

(2004) What’s trust got to do with it? Non-payment of service charges in local authorities in South Africa. Journal of Modern African Studies 42(4): 539–562.

22.

Gallego

Jorge

(2018) Civil conflict and voting behavior: Evidence from Colombia. Conflict Management and Peace Science 35(6): 601–621.

23.

Garbiras-Diaz

Natalia

Garcia-Sanchez

Miguel

Matanock

Aila

(2020) Using political cues for attitude formation in post-conflict contexts. ESOC Working Paper (19).

24.

Gates

Scott

Justesen

Mogens K

(2013) Political trust, shocks, and accountability: Quasi-experimental evidence from a rebel attack. Journal of Conflict Resolution 64(9): 1693–1723.

25.

Gibson

James

(2002) Truth, justice, and reconciliation: Judging the fairness of amnesty in South Africa. American Journal of Political Science 46(3): 540–556.

26.

Grupo de Memoria Histórica (2013) Â¡Basta ya!: Colombia: Memorias de guerra y dignidad [English translation].

27.

Hadzic

Dino

Tavits

Margit

(2019) The gendered effects of violence on political engagement. Journal of Politics 81(2): 676–680.

28.

Hall

Jonathan

Kovras

Iosif

Stefanovic

Djordje

Loizides

Neophytos

(2018) Exposure to violence and attitudes towards transitional justice. Political Psychology 39(2): 219–233.

29.

Hutchison

Marc L

Johnson

Kristin

(2011) Capacity to trust? Institutional capacity, conflict, and political trust in Africa, 2000–2005. Journal of Peace Research 48(6): 737–752.

30.

Ishiyama

John

Higareda

Felipe Carlos Betancourt

Pulido

Amalia

Almaraz

Bernardo

(2018) What are the effects of large-scale violence on social and institutional trust? Using the civil war literature to understand the case of Mexico, 2006–2012. Civil Wars 20(1): 1–23.

31.

Kaplan

Oliver

Nussio

Enzo

(2018) Explaining recidivism of ex-combatants in Colombia. Journal of Conflict Resolution 62(1): 64–93.

32.

Kijewski

Sara

Freitag

Markus

(2018) Civil war and the formation of social trust in Kosovo: Posttraumatic growth or war-related distress? Journal of Conflict Resolution 62(4): 717–742.

33.

Kreiman

Guillermo

Masullo

Juan

(2020) Who shot the bullets? Exposure to violence and attitudes toward peace: Evidence from the 2016 Colombian referendum. Latin American Politics and Society 62(4): 24–49.

34.

Kreutz

Joakim

Nussio

Enzo

(2019) Destroying trust in government: Effects of a broken pact among Colombian ex-combatants. International Studies Quarterly 63(4): 1175–1188.

35.

Kruks-Wisner

Gabrielle

(2018) Claiming the State: Active Citizenship and Social Welfare in Rural India. Cambridge: Cambridge University Press.

36.

Laplante

Lisa

Theidon

Kimberly

(2007) Truth with consequences: Justice and reparations in post-truth commission Peru. Human Rights Quarterly 29(1): 228–250.

37.

LAPOP (2018) The Americas Barometer by the Latin American Public Opinion Project (LAPOP).

38.

Levi

Margaret

Stoker

Laura

(2000) Political trust and trustworthiness. Annual Review of Political Science 3(1): 475–507.

39.

Loyle

Cyanne

Binningsbo

Helga Malmin

(2018) Justice during armed conflict: A new dataset on government and rebel strategies. Journal of Conflict Resolution 62(2): 442–466.

40.

Mettler

Suzanne

(2005) Soldiers to Citizens: The G.I. Bill and the Making of the Greatest Generation by Suzanne Mettler. Oxford: Oxford University Press.

41.

Nalepa

Monika

(2010) Skeletons in the Closet: Transitional Justice in Post-Communist Europe. Cambridge: Cambridge University Press.

42.

Nussio

Enzo

Rettberg

Angelika

Ugarriza

Juan E

(2015) Victims, nonvictims and their opinions on transitional justice: Findings from the Colombian case. International Journal of Transitional Justice 9(2): 336–354.

43.

Olsen

Tricia D

Payne

Leigh

Reiter

Andrew

(2010) Transitional Justice in Balance: Comparing Processes, Weighing Efficacy. Washington, DC: United States Institute of Peace.

44.

Oster

Emily

(2019) Unobservable selection and coefficient stability: Theory and evidence. Journal of Business & Economic Statistics 37(2): 187–204.

45.

Pham

Phuong

Weinstein

Harvey

Longman

Timothy

(2004) Trauma and PTSD symptoms in Rwanda: Implications for attitudes toward justice and reconciliation. JAMA 292(5): 602–612.

46.

Pham

Phuong Ngoc

Vinck

Patrick

Marchesi

Bridget

Johnson

Doug

Dixon

Peter

Sikkink

Kathryn

(2016) Evaluating transitional justice: The role of multilevel mixed methods datasets and the Colombia reparation program for war victims. Transitional Justice Review 1(4).

47.

Pierson

Paul

(1993) When effect becomes cause: Policy feedback and political change. World Politics 45(4): 595–628.

48.

Rettberg

Angelika

(2015) Victims of the Colombian armed conflict: The birth of a political actor. In: Bagley

Bruce

Rosen

Jonathan

(eds) Colombia’s Political Economy at the Outset of the Twenty-First Century. Blue Ridge Summit, PA: Lexington, 111–139.

49.

Riano-Alcalá

Pilar

Baines

Erin

(2011) The archive in the witness: Documentation in settings of chronic insecurity. International Journal of Transitional Justice 5(3): 412–433.

50.

Rivas

Jairo

(2016) Official victims’ registries: A tool for the recognition of human rights violations. Journal of Human Rights Practice 8 (1): 116–127.

51.

Rohner

Dominic

Thoenig

Mathias

Zilibotti

Fabrizio

(2013) Seeds of distrust: Conflict in Uganda. Journal of Economic Growth 18(3): 217–252.

52.

Rozenas

Arturas

Zhukov

Yuri

(2019) Mass repression and political loyalty: Evidence from Stalin’s ‘terror by hunger’. American Political Science Review 113(2): 569–583.

53.

Sacks

Audrey

Larizza

Marco

(2012) Why quality matters: Rebuilding trustworthy local government in post-conflict Sierra Leone. Washington, DC: World Bank.

54.

Samii

Cyrus

(2013) Who wants to forgive and forget? Transitional justice preference in postwar Burundi. Journal of Peace Research 50(2): 219–233.

55.

Sandoval

Rojas

, Carolina

Nathalia

(2015) Movilizarse ante la Corte: Trayectoria y efectos de tres episodios de movilizacion legal constitucional de feministas, indigenas y victimas de crimenes de Estado en Colombia [Social mobilizations in times of the constitution: Feminists, indigenous people and victims of crimes of state before the Colombian Constitutional Court]. Bogota: Universidad de los Andes.

56.

Schubiger

Livia Isabella

(2021) State violence and wartime civilian agency: Evidence from Peru. Journal of Politics 83(4): 1383–1398.

57.

Soss

Joe

(1999) Lessons of welfare: Policy design, political learning, and political action. American Political Science Review 93(2): 363–380.

58.

Tellez

Juan Fernando

(2019) Worlds apart: Conflict exposure and preferences for peace. Journal of Conflict Resolution 63(4): 1053–1076.

59.

Unidad para las Victimas (2020) Nueve años de la mano de las víctimas del conflicto, trabajando por su reparación integral [Nine years hand in hand with the victims of the conflict, working for their comprehensive reparation].

60.

Voytas

Elsa

(2021) What can reparations do? The political consequences of compensation. Working paper.

61.

Wang

Yuhua

(2021) The political legacy of violence during China’s cultural revolution. British Journal of Political Science 51(2): 463–487.

62.

Weaver

Vesla

Lerman

Amy

(2010) Political consequences of the carceral state. American Political Science Review 104(4): 817–833.

63.

Wong

Pui-Hang

(2016) How can political trust be built after civil wars? Evidence from post-conflict Sierra Leone. Journal of Peace Research 53(6): 772–785.

64.

Zkuhov

Yuri

Talibova

Roya

(2018) Stalin’s terror and the long-term political effects of mass repression. Journal of Peace Research 55(2): 267–283.