Sage Journals: Discover world-class research

Abstract

The reintegration of former armed group affiliates into formal politics is a fundamental challenge for post-conflict societies. Arguably, this challenge is even more difficult in settings where ethnic discrimination is pervasive, since discrimination implies comparatively higher barriers to political reintegration for racialized ex-combatants. In this paper, we explore how ethnicity and armed group affiliation interact to shape electoral prospects. Specifically, we investigate whether participants in a conjoint experiment in Colombia discriminate between political candidates of European and African descent and whether politicians of different ethnic identities are differentially punished for ties to armed groups. We find that participants display a strong aversion toward politicians with armed group affiliations and that they discriminate against Afro-Colombian candidates. Furthermore, we demonstrate that biases are additive, in the sense that Caucasian and Afro-Colombian candidates are equally punished for associating with armed groups. Finally, we explore heterogeneous treatment effects to shed light on sources of ethnic discrimination and aversion toward former armed group affiliates.

Keywords

ethnic discrimination political reintegration voter behavior conjoint experiment

Introduction

On June 19, 2022, Colombia elected the first left-wing cabinet in the history of the country. However, the victory of President Gustavo Petro and his running mate, vice president Francia Marquez, was even more exceptional considering the identity traits of both candidates: Petro is a former member of a guerrilla group and Marquez became the first ever Afro-Colombian vice president. Historically, these groups have been scarcely represented in formal Colombian politics; the Afro-Colombian community has faced significant discrimination and affiliates of armed groups have been both excluded from,¹ and persecuted in,² the political arena.

Political exclusion may both trigger and entrench conflict (Wucherpfennig et al., 2012). Given that the opportunity to enter formal politics is a key concern for armed groups when negotiating peace terms, understanding voter aversion to former combatants and how such biases interact with ethnic discrimination can shed light on rebel incentives to demobilize as well as the prospect of successful post-conflict reintegration.

Using experimental survey data gathered just months before the 2022 Colombian presidential election, we analyze the impact of candidate ethnicity and (past) armed group affiliation on electability. The survey comprised two rounds with the same participants in Meta—a conflict-ridden department of Colombia—where the majority of residents are either white (Caucasian) or of mixed ethnic origin (Mestizo). In the first survey round, we gauge voter preferences through a conjoint experiment (Hainmueller and Hopkins 2015; Loizides et al. 2022), in which participants vote on candidates with randomly composed profiles in fictional elections (881 participants, 2579 choices), to tease out how candidates’ traits influence their electability. The experiment subtly manipulates candidates’ implied ethnicity by randomly varying their names as more Caucasian/European-sounding (Sebastían/Mariana) or more Afro-Colombian-sounding (Candelario/Petrona). Furthermore, we alter candidates’ armed group affiliation by randomly composing politicians who are rumored to have no, past, or present ties to an (unnamed) armed group. Since the conjoint experiment holds other relevant candidate traits constant by design,³ we can infer the causal impact of having an Afro-Colombian-sounding name (evaluated against having a Caucasian-sounding name) and armed group affiliation by estimating the average effect of the respective candidate trait on the likelihood of being chosen. The empirical analysis was pre-registered (https://aspredicted.org/NGJ_51R).

The results reveal that participants strongly punish politicians who have (had) ties with armed groups: past affiliation reduces the likelihood of being chosen by 14 percentage points, whereas present ties reduce the likelihood by 17.7 percentage points. Participants thus marginally distinguish candidates who have had ties with armed groups in the past from those who have current armed group affiliations.⁴ Furthermore, we find that Afro-Colombian candidates are chosen almost 6 percentage points less frequently than candidates with the exact same traits but with Caucasian-sounding names. Finally, we demonstrate that the effects of stigmatized candidate traits on electability are additive; the penalty for being associated with armed groups is of similar magnitude for Caucasian and Afro-Colombian candidates.

What can explain these results? After each fictional election, we ask participants to indicate whether they trust candidates to fulfill their campaign promises. We find significantly lower trust in candidates with armed group affiliations as well as in candidates with Afro-Colombian names. Accordingly, biases can, to some extent, be attributed to the fact that participants appear to trust both (former) armed group affiliates and Afro-Colombian candidates less. Finally, we investigate participant heterogeneity along two pre-registered dimensions: participants’ own ethnic identities and their support for (or opposition to) the peace agreement signed between the Colombian state and the FARC-EP in 2016. The first analysis reveals that the ethnic bias against Afro-Colombian candidates is strongest among participants who identify as Caucasian or “without ethnic identity”. Conversely, there is no evidence of an ethnic bias among participants who identify as Afro-Colombian, Indigenous, or report “Other ethnic identity”, although it should be noted that those latter tests are underpowered and, hence, the results should be interpreted with caution. The second analysis demonstrates that opponents of the peace accord are significantly less likely to vote for candidates who have had ties to armed groups in the past.

Our work links closely with an emergent strand of research using experiments to understand civilian preferences in post-conflict settings (see, e.g., Tellez 2019a; Kao and Revkin 2021), and especially studies focusing on preferences for ex-combatant reintegration (Bauer, Fiala and Levely 2018; Blair et al. 2021; Godefroidt and Langer 2021). We also relate to the significant scholarship concerned with empirically documenting discriminatory practices (see, e.g., Zschirnt and Ruedin 2016; Teele, Kalla and Rosenbluth 2018), a phenomenon that is inherently hard to measure. Methodologically, we build upon a burgeoning research paradigm using names to proxy for ethnicity (Doherty, Dowling, and Miller 2019; Thrasher et al. 2017; Portmann and Stojanović 2019; Portmann 2022), a practice that reduces the risk of social desirability bias (which otherwise mitigates the likelihood of detecting discrimination). The use of names to imply ethnicity and its application in a conjoint experiment jointly ensure participants “plausible deniability”, which increases the likelihood of exposing true preferences. By combining research on ex-combatant reintegration with the study of racial discrimination, the present paper provides the first experimental evidence of how ethnic identity shapes former combatants’ opportunities for political reintegration.⁵

The paper proceeds as follows. In the next section, we discuss the underlying theory and derive hypotheses. After that, we outline the study setting and describe the data. We then present the results. Finally, we discuss the findings and conclude.

Theory and Hypotheses

Social Identity and Differential Prospects for Political Reintegration

Social identity theory posits that individuals place themselves and others into different categories (Tajfel 1974). These categories, or social identities, create boundaries between “us” (the ingroup) and “them” (the outgroup) and provide cues (stereotypes) about individuals’ characteristics based solely on their group membership. Because members of the outgroup are typically ascribed more negative attributes, group identification tends to result in ingroup favoritism and/or outgroup discrimination (Everett, Faber, and Crockett 2015). This psychological tendency has been found to shape behavior across a range of domains, including moral decision-making (Chae et al. 2022; Agneman and Chevrot-Bianco 2023) and voting behavior (Portmann and Stojanović 2022).

Strong group identification can incite inter-group violence (Coser 1956), as group membership facilitates coordination and can be used to legitimize violent attacks (Bornstein 2003; Aspinall 2007). At the same time, conflicts tend to exacerbate group identification (Strömbom 2014), manifested, e.g., by an increase in group-based symbolism after conflict events (Martínez, Valentim and Dinas 2023). But what is the role of group identity in a society transitioning from war to peace? Specifically, how does social identity influence prospects for political reintegration of (past) armed group affiliates?

Protracted civil conflicts often breed animosity and mistrust toward rebels and their allies. Politicians and other authorities may intentionally exacerbate hatred as a strategic choice (Sierra and García 2019), e.g., in order to minimize civilian-enemy collaboration. In the aftermath of war, however, strategic incentives are typically reversed. For a stable and durable peace to materialize, former enemies need to be able to lay down arms and return to a normal life, which in turn requires popular acceptance of such a transition (Murillo Orejuela and Restrepo-Plaza 2021).

Since insurgencies often have political motives, the prospects for former combatants to enter formal politics are of first-order importance (Gilligan, Mvukiyehe, and Samii 2013). This objective is, however, challenging due to voters’ propensity to retrospectively punish ex-combatants for past transgressions (Podder 2012) as well as to consider their prospective political involvement a threat to post-conflict stability (Annan et al. 2011). Yet, attitudes vary both across and within contexts. Popular attitudes toward ex-combatant reintegration depend on factors such as conflict experiences and political views (Tellez 2019b), reintegration campaigns (Annan et al. 2011; Gilligan, Mvukiyehe, and Samii 2013), the general discourse (Blair et al. 2021), and the behavior of combatants during conflict (Godefroidt and Langer 2021). In addition, Bauer, Fiala and Levely (2018) emphasize structural conditions by explaining a lack of preference-based prejudice against former rebel fighters in Northern Uganda through the low levels of rebels’ agency (many of whom were abducted as children).

The temporal dimension also matters, although the speed of reconciliation varies significantly. In Colombia, for instance, the stigma against former members of the insurgent group the FARC-EP has decreased over time (Murillo Orejuela and Restrepo-Plaza 2021), but indicators of forgiveness and willingness to reintegrate remain low (López López et al. 2018). An illustrative case is the lack of success of The Commons (the political successor of the FARC-EP) after its foundation in 2016, attracting less than 0.4 percent of the votes in the 2018 general elections and even fewer votes in 2022 (Yordi 2022). Similarly, the inclusion of former right-wing paramilitaries in conventional Colombian institutions has sparked popular resistance, such as when President Petro appointed a former senior commander of the AUC (United Self-Defense Forces of Colombia) as a “peace manager”, responsible for peace negotiations (Taylor 2023). Taken together, previous research and aspects of the Colombian context⁶ suggest that voters may punish even weak and past links to armed groups. We propose the following interlinked hypotheses:

Hypothesis 1A

Participants punish politicians for rumored associations with armed groups.

Hypothesis 1B

Participants punish politicians with past and present ties to armed groups equally.

The expectation is that armed group affiliations significantly reduce electability on average. But armed group affiliates certainly experience varying barriers to political reintegration. Conflict is typically intertwined with other identity dimensions, such as ethnicity. For instance, while the civil war in Colombia was ideological rather than ethnic in nature, the conflict disproportionately affected peripheral regions characterized by weaker state presence and a higher concentration of Afro-descendant communities (Lambardi and Palacios 2022). The conflict thus further entrenched inter-ethnic inequalities,⁷ which can be traced to the systemic racism inherent to colonial institutions.

One reason for the lingering ethnic divide is that ethnic markers such as skin color allow for an external categorization that is both salient and impermeable (Huddy 2001). According to recent studies, it suffices to carry a name that implies group membership of a marginalized minority to face discrimination (see, e.g., Portmann and Stojanović 2022). In an investigation of discrimination in six Latin American countries, Colombia stood out as the country with the strongest association between skin color and perceived discrimination (Canache et al. 2014). Among other things, this tendency may influence descriptive representation in the political arena. Despite formal efforts to promote diversity in Colombian politics, e.g., through affirmative action guaranteeing two seats in the House of Representatives (Chavez-Dueñas, Adames, and Organista 2014), Afro-Colombians remain under-represented (Soendergaard 2014). A potential explanation for this pattern is ethnic prejudice among the electorate, or bottom-up discrimination, which implies a penalty to Afro-Colombian candidates solely due to their ethnic group membership. Our second hypothesis reads as follows:

Hypothesis 2

When deciding between identical politicians who differ only with respect to their implied ethnicity, participants tend to choose Caucasian over Afro-Colombian candidates.

Social identity is not unidimensional but rather constructed by a plethora of overlapping identity dimensions (Roccas and Brewer 2002; Kalin and Sambanis 2018). According to an emergent strand of research concerned with the intersectionality of biases, the signals of different group memberships may interact to shape discriminatory behavior (Lahey and Oxley 2018; Dahl and Krog 2018; Magni and Reynolds 2021). Related research suggests that marginalized groups are held to different moral standards. In a study of third-party punishment in a small-scale society, Bernhard, Fischbacher, and Fehr (2006) demonstrate that members of the outgroup are more harshly punished for moral transgressions than ingroup members. In a similar vein, Bauer et al. (2023) find that a stigmatized ethnic minority serves as a scapegoat and is punished even for moral transgressions for which they bear no blame.

Taken together, the experimental research (see also Schiller, Baumgartner and Knoch 2014; Yudkin et al. 2016) suggest that punishment is conditional upon group membership, a conjecture which is supported by research on judicial biases (Shayo and Zussman 2011; Alesinaand La Ferrara 2014). The key implication for the present study is that stigmas associated with armed group affiliation might be more sticky for Afro-Colombian candidates compared to Caucasian candidates. In other words, while Colombians of both African and European descent featured across different ranks of the major non-state armed factions (see a brief description in the Online Appendix Section A), the extent to which the public assigns them blame in the aftermath of war may vary.

At the same time, aspects such as forced recruitment, which should reduce perceived agency and, hence, ascribed blame (Bauer, Fiala and Levely 2018), may intersect with ethnic identity. For instance, forced recruitment was more widespread in Chocó (Uberti 2022), a predominantly Afro-Colombian department. If this statistical property reduces the perceived agency ascribed to Afro-Colombian candidates, this tendency might counteract the pre-supposed ethnic bias.

Yet, on balance, previous research point in the direction that stigmatized minorities are more likely to be punished for the very same moral transgressions. Accordingly, we propose a final hypothesis:

Hypothesis 3

Afro-Colombian political candidates are more likely to be electorally punished for ties to armed groups compared to their Caucasian counterparts.

Methodology

Sample

In order to investigate voter biases, we gather experimental data from the Colombian department Meta, located just south of the capital city, Bogotá. The population of Meta is largely White or of mixed ethnic origins (Mestizo); in our sample, 16.5 percent stated that they were White/Caucasian, 36.3 percent Mestizo, 6.7 percent Afro-Colombian, 6.9 percent indigenous, and 4.7 percent reported other ethnic identity. In addition, 21.7 percent reported “no ethnic identity” and 7.3 percent preferred not to state their ethnicity. The choice not to report ethnic identity may in some cases reflect a preference for a “Colombian identity”, which allegedly transcends ethnic boundaries.⁸

Meta was severely affected by the civil war that plagued the country for decades, with a strong presence of the major factions throughout the conflict. While left-wing guerrillas (FARC-EP) exercised considerable power, especially in the department’s western parts, right-wing paramilitary groups and other criminal syndicates were strong in the central areas. The resulting divide is evident in politics to this day. In the second round of the presidential elections in Colombia in 2022, the support for the left-wing candidate Gustavo Petro was strongest in the west and weakest in municipalities located in the center (Llaneras, Borja and Sevillano Pires 2022).

We sampled 16 of the 27 municipalities in Meta. The municipalities were purposefully selected to provide a sample of participants with diverse experiences with regard to the civil war. To this end, we sampled four municipalities that had been part of a demilitarized zone (DMZ) during the late 1990s and early 2000s and then featured strong presence by the FARC-EP and weak presence by the Colombian state; the eight municipalities closest to the DMZ (areas that were often contested by the armed factions);⁹ and the capital of Meta, Villavicencio, as well as the two most geographically proximate municipalities (areas where the Colombian state had a relatively strong presence).¹⁰

In each municipality capital, we randomly sampled eight urban and six rural areas (manzanas) from the universe of logistically reachable manzanas. We visited the areas on the ground instead of conducting online or telephone interviews to ensure a participant pool that reflected well the overall population of the sampled municipalities.¹¹ Between January 22 and February 4th, 2022, a team of 14 enumerators recruited participants following a door-to-door random-walk procedure in accordance with a number of predetermined rules (described in the Online Appendix Section B). The interviews were conducted in Spanish, and the enumerators recorded the answers on tablets. During the conjoint experiment, the enumerators would hand the tablets to participants so that they could read the candidate profiles themselves and make decisions in private.¹²

Surveying participants on sensitive matters in a post-conflict zone requires careful implementation to ensure the safety of both participants and enumerators. In the Online Appendix Section E, we discuss a number of measures that were taken and present information about the ethical approval, which was granted by a local institutional review board. In the next subsection, we describe the conjoint experiment and the key variables of interest.

Conjoint Experiment

We rely on conjoint experiments to identify the effect of candidate traits on electability. First, participants are introduced to the experiment through a short text stating that they will take part in fictitious elections and that they should vote as if they were real (Online Appendix Figure 4). Then, in three separate conjoint experiments, participants were presented with two different candidates and asked which candidate they would support if it was a real election. We randomly vary the traits (attributes) of would-be candidates along five dimensions (levels): their names (implies either an Afro-Colombian or Caucasian ethnic identity); their professions (business or union leaders); (rumored) armed group affiliation (none, past, or present); election promise (programmatic or clientelistic); and vote monitoring (a statement about whether candidates will monitor voter choices or respect the anonymity of ballots). In the Online Appendix Table C1, we present an example of a binary choice between candidates that a participant could have been presented with.

As per the pre-registration, the two attributes of main interest are candidates’ implied ethnicity and armed group affiliation, and how each of these factors influences electability. Nevertheless, the other features support the identification strategy as well. By informing participants about the candidates’ prior professions (and hence implicit socio-economic status) as well as political platforms, we mitigate the risk that such attributes, which may intersect with ethnicity and armed group affiliation, confound the analysis. Providing these additional pieces of information mitigates the risk that the factors of main interest (candidate ethnicity and armed group affiliation) are ascribed variation caused by omitted variables. Moreover, these additional variables ensure that the research question remains unknown to participants, thereby reducing the risk of social desirability bias (Gutiérrez-Romero and LeBas 2020). A paired conjoint design, as employed in the present study, has been shown to reveal real-world voting behavior remarkably well (Hainmueller, Hangartner, and Yamamoto 2015).

After each hypothetical election, participants are asked two additional questions: (1) whether they believe that not voting for a certain candidate might cause them problems (fear of retribution) and (2) whether they believe that candidates would fulfill their election promises (candidate trustworthiness). These variables were measured to enable the identification of participants’ underlying motives.

Participant-level Data

In addition to the conjoint experiment, we collected data on a number of respondent-level characteristics that enable treatment validation as well as investigations of heterogeneous treatment effects. In order to validate that names signaled ethnicity in the intended way, we collected additional data through a follow-up telephone survey (319 participants), where we asked participants to assign ethnic identities to hypothetical people with the names used in the conjoint experiments (Candelario; Petrona; Sebastián; Mariana). In the Online Appendix Section D, we demonstrate that Candelario and Petrona are assigned Afro-Colombian identities 82 percent and 68 percent of the time, respectively.¹³ In the follow-up survey, we also asked participants to declare their willingness to support an Afro-Colombian candidate in a hypothetical future presidential election. This variable enables us to investigate if presupposed ethnic discrimination is confined to participants who self-report as unwilling to support Afro-Colombians.

Finally, we gauge a number of participant traits that allow us to investigate treatment heterogeneity. First, in accordance with the pre-registration, we asked participants to self-report their ethnic identity. This variable allows us to shed light on the role of ingroup bias in shaping the hypothesized voter behavior. Second, in another pre-registered heterogeneity analysis, we investigate how participants’ views of the peace agreement shape their (assumed) biases against armed group affiliates and racialized candidates. Since peace agreement supporters displayed a stronger willingness to reconcile with ex-combatants (Tellez 2019a), the underlying hypothesis is that they should be more likely to vote for candidates with past armed group affiliations compared with participants who are negative to the peace agreement. In the specifications that investigate heterogeneous treatment effects, we focus on the electability of candidates with past armed group affiliation, since the purpose is to shed light on differences in the public’s willingness to support racialized and former armed group affiliate candidates. The operationalizations of all variables used in the empirical analysis are described in the Online Appendix Table F1.

Identification Strategy

Each of the 881 participants was asked to complete three (binary) elections and, hence, could consider six separate candidates, implying 5286 potential observations. However, 26 respondents participated in only two elections and 19 in only one election, yielding a total of 5158 observations. The dependent variables indicate whether Candidate X was (1) chosen; (2) feared; and (3) trusted, as a function of the independent variables (candidates’ traits and participants’ attributes). In accordance with the approach suggested in Hainmueller, Hopkins, and Yamamoto (2014), the data are analyzed through OLS regressions to infer statistical significance. We also rely on marginal means to graphically present how candidates’ electability depends on their traits. The regressions are all based on variants of the following equation:¹⁴

Y_{i j m} = β_{0 m} + \sum_{k = 1}^{K} β_{k m} \cdot {CandidateTrait}_{k} \cdot I (j \in S) + ε_{i j m}

Results

Baseline Results

The first hypothesis holds that participants should display aversion toward politicians with past and present ties to armed groups. According to the second hypothesis, we should further expect participants to discriminate against candidates with Afro-Colombian ethnic identities. To infer how candidates’ traits affect their vote shares, we leverage the fact that the conjoint experiment randomly composes candidate profiles, and compute the statistical links between traits (attribute levels) and a dummy variable indicating whether the candidate was selected (1) or not (0).

In Figure 1, we present the descriptive results (marginal means). In the graph’s bottom half, we display the likelihood that a candidate with a certain armed group history was selected. We find, as hypothesized, that participants display strong aversion toward politicians who are rumored to have an armed group affiliation, both past and present. In corresponding regression analyses presented in the Online Appendix Table G1, we show that the electoral penalty for a rumored armed group affiliation (compared to having no such affiliation) is about 16 percentage points (Column 1; p-value $< 0.001$ ). In Column 3, we disaggregate the measure of armed group affiliation into past and present, finding that present ties reduce the share of votes slightly more than past ties (−17.5 and −14.6 percentage points, respectively). In the Online Appendix Table 3 (Column 2), we show that this difference is only borderline significant, indicating that participants are almost as averse to candidates with past ties to armed groups as they are to candidates with present ties.

Figure 1.

Marginal means for different candidate traits.

Turning to the role of ethic identity, we confirm the pre-registered hypothesis that candidates with (implied) Afro-Colombian identities are less likely to attract votes, even though all other characteristics are held constant. The estimates in the red-shaded area depict the likelihood that candidates with Afro-Colombian names are chosen, whereas the pink-shaded area displays the estimates for candidates with Caucasian names. On average, participants are 6 percentage points less likely to select a candidate if they have an Afro-Colombian rather than a Caucasian-sounding name (Online Appendix Table G1 Columns 1 and 2; p-values $< 0.001$ ). Candidates’ names also imply gender, and notably female candidates are preferred over male candidates.¹⁵ Yet, candidates’ implied ethnicity is more consequential. While there is no statistical difference between participants’ preferences for “Mariana” or “Sebastián”, participants are significantly less likely to select candidates with the Afro-Colombian names “Petrona” and “Candelario” (Online Appendix Table G1 Columns 3 and 4; p-values $< 0.01$ ). In the Online Appendix Table 2, we repeat the baseline regression when including enumerator fixed effects—to ensure that no bias due to participant-enumerator matching influences the findings—and show that the main results are essentially unchanged in this robustness check.

In what follows, we consider how armed group affiliation and Afro-Colombian identity jointly determine politicians’ electability. The third hypothesis suggests that the penalty for armed group affiliation could be higher for Afro-Colombian candidates than for Caucasian candidates. To evaluate this assertion, we aggregate the identity traits into two categories: Afro-Colombian name (Candelario and Petrona) and armed group affiliation (past and present).

In Figure 2, we display the marginal means when these traits are interacted. As has already been established, the likelihood of being chosen is considerably lower if the candidate is Afro-Colombian and if s/he has an armed group affiliation. Accordingly, the least favored candidates are those that combine these two traits. However, there is no “Afro-Colombian penalty” for armed group affiliation, according to our data. We find that armed group affiliation reduces the likelihood that a voter selects an Afro-Colombian candidate by 14.9 percentage points, while it reduces the likelihood that a Caucasian candidate is selected by 17.1 percentage points (Online Appendix Table 4 Columns 1 and 2). Column 3 of the same table shows that the interaction term between Afro-Colombian identity and armed group affiliation is statistically insignificant. In sum, while participants display aversion toward both Afro-Colombian and armed group-affiliated politicians, there is no interaction between these traits. Rather, biases are found to be additive.

Figure 2.

Interactions between ethnic identity and armed group affiliation.

Next, we find support for the hypothesis that the impact of ethnically coded names is due to ethnic discrimination. In the follow-up survey, we asked participants to indicate their willingness to support a (hypothetical) Afro-Colombian presidential candidate. The majority of participants indicated that they would be willing to vote for an Afro-Colombian presidential candidate, but around 11 percent “definitely” or “probably” would not (Online Appendix Figure 5). In the Online Appendix Table 6, we show evidence of more pronounced discriminatory behavior among participants who self-report as unwilling to support an Afro-Colombian presidential candidate. For this group of participants, the effect of an Afro-Colombian name on the likelihood of being chosen is approximately three times as large compared with participants who self-report as willing to support Afro-Colombian candidates (Columns 1 and 2 of Online Appendix Table 6). Yet, even among participants who self-reportedly are open to voting for Afro-Colombians, we find evidence of systematic ethnic biases.

Finally, in the Online Appendix Table 5, we investigate whether Afro-Colombian candidates are more severely punished for other moral transgressions, namely clientelism and vote monitoring. We do not find statistically significant evidence of “Afro-Colombian penalties” in these domains either, although the signs of the interaction effects are negative. In the Online Appendix Table 10, we show that ethnic discrimination as well as electoral punishment for armed group associations is somewhat greater for candidates with a background as union leaders than for candidates with a background as business leaders.

What Explains Voter Biases?

Next, we ask why participants systematically decide against supporting Afro-Colombian and armed group-affiliated politicians. In line with social identity theory, we argue that ethnic identity and (rumored) armed group affiliation may activate an ingroup/outgroup logic among participants in our survey (who are largely White or Mestizo). Group membership, in turn, may provide cues as to whether candidates are trustworthy or dangerous. To investigate whether such inference played a role, we measured participants’ fear of retribution¹⁶ and their beliefs in candidates’ trustworthiness.¹⁷

In Figure 3, we display how the pre-registered mediators are influenced by the candidate traits of main interest. As should be expected, we observe that participants make no associations between candidates’ ethnicity and their capacity for retribution (left-hand graph of Figure 3; Column 1 of Online Appendix Table 7). However, participants are more likely to fear retribution from candidates who are/were affiliated with armed groups. Strikingly, participants believe that politicians with past ties to armed groups are just as likely to create problems for them as candidates with current ties to armed groups. These results offer a tentative explanation for why candidates with past ties to armed groups perform comparably to those with present ties; participants seem to attribute the same unfavorable qualities to candidates rumored to associate with armed groups regardless of whether the links are past or present.

Figure 3.

Trait effects on fear and trustworthiness.

In the right-hand graph of Figure 3 (Online Appendix Table 7 Column 3), we demonstrate lower levels of trust in both (implied) Afro-Colombian candidates and (rumored) armed group-tied candidates. The effect on trustworthiness is somewhat larger for candidates with present ties than for candidates with past ties. Since trustworthiness is a first-order concern for voters (Born, van Eck, and Johannesson 2018), this pattern may further explain the bias against Afro-Colombian and armed group-affiliated candidates. Indeed, we find that participants’ trust in candidates strongly predicts whether they will vote for them (Column 1 of Online Appendix Table 8). However, when we include fear or retribution and candidate trustworthiness as controls in the baseline specification, we find that the parameters for Afro-Colombian name and (rumored) armed group affiliation remain statistically significant (Online Appendix Table 8, Column 3). This implies that while fear of retribution and candidate trustworthiness partly explain voter biases, other mechanisms are also at play. Whether those other mechanisms are belief- or preference-based is beyond the scope of the present study, but this line of research constitutes a promising avenue for future work.

Heterogeneous Responses

In what follows, we turn to pre-registered heterogeneity analyses along two dimensions where we should expect variation in the propensity to select candidates based on their ethnic profiles and armed group affiliation. First, we investigate whether participants’ own ethnic identities condition their voting behavior. Second, we study how participants’ views on the peace agreement predict voting preferences. We are interested in whether these participant traits influence the willingness to reintegrate former armed group affiliates in politics and, hence, confine the analyses to candidates with no (reference group) or past armed group affiliations. Given that the experimental design pitted (implied) Afro-Colombian candidates against (implied) Caucasian candidates, social identity theory would predict more pronounced discriminatory behavior among participants of European descent. Conversely, we should expect less bias among Afro-Colombian participants and those of other minoritized ethnic groups.

In Table 1, we investigate the extent of voter discrimination against Afro-Colombian candidates and bias against former armed group-affiliated candidates, separately for participants of various (self-reported) ethnic identities.

Table 1.

Heterogeneous Effects: Participants' Ethnic Identities.

Dep. Var	Candidate Chosen
Sample	Indigenous	Afro-Col	Caucasian	Mestizo	No ethn	Other	No ans	Full
	N = 276	N = 286	N = 669	N = 1515	N = 885	N = 185	N = 315	N = 3816
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Afro-Col.^a	−0.00863 (0.0692)	−0.0458 (0.0556)	−0.0962 (0.0408)**	−0.0539 (0.0256)**	−0.0835 (0.0336)**	−0.0228 (0.0746)	−0.0470 (0.0555)
Past AG affil.^b	−0.0497 (0.0685)	−0.104 (0.0631)	−0.126 (0.0416)***	−0.171 (0.0276)***	−0.185 (0.0325)***	−0.0417 (0.0830)	−0.135 (0.0697)*
Minority^c								−0.0391 (0.0272)
Afro-Col. cand. × Minority								0.0465 (0.0418)
Past AG affil. × Minority								0.0965 (0.0445)**
Constant	0.561 (0.0426)***	0.566 (0.0370)***	0.616 (0.0250)***	0.586 (0.0159)***	0.629 (0.0194)***	0.575 (0.0553)***	0.577 (0.0350)***	0.606 (0.0110)***
Observations	276	286	669	1515	885	185	315	3816
R-squared	0.00202	0.0112	0.0198	0.0256	0.0343	0.00183	0.0149	0.0221
Dep. Var. Mean	0.543	0.514	0.538	0.513	0.539	0.551	0.521	0.528

Note: Table 1 displays regression estimates of the impact of candidates’ traits on the likelihood of a candidate being chosen, separately for participants who report their ethnic identity as indigenous (Column 1), Afro-Colombian (Column 2), Caucasian (Column 3), Mestizo (Column 4), and who do not report or report “other ethnicity” (Column 5). Reference categories: (a) Caucasian named candidate; (b) No AG affiliation; (c) Minority participant is a composite indicator for participants identifying as “Indigenous”, “Afro-Colombian”, and “Other ethnicity”, whose voting behavior is evaluated against “Caucasian”, “Mestizo”, and “No ethnicity” participants (note that we in this specification purposefully exclude the ambiguous category “No answer”). All columns report robust standard errors clustered at the participant level.

Standard errors clustered at the participant-level in parentheses *p < 0.10, **p < 0.05, ***p < 0.010.

Table 1 demonstrates negative coefficients on the Afro-Colombian dummy across all sub-samples. However, the effects are muted and statistically insignificant for participants who identify as Indigenous (Column 1), Afro-Colombian (Column 2), “Other” (Column 6), or who prefer not to state their ethnic identity (Column 7). In contrast, ethnic discrimination is significant among the other ethnic groups; Afro-Colombian candidates receive 9.6 pp. fewer votes among Caucasian participants (Column 3), 5.4 pp. fewer votes among participants reporting Mestizo ethnic identity (Column 4), and 8.4 pp. fewer votes among participants who report no ethnic identity (Column 5). The findings thus align with predictions of social identity theory, although it should be noted that the number of participants reporting Indigenous, Afro-Colombian, or “Other” ethnic identities is small and, therefore, that the statistical precision is low. Lastly, we also find that minority participants of the Indigenous, Afro-Colombian, and other ethnic communities display comparatively less aversion to candidates with past ties to armed groups, while the interaction effect with respect to candidate ethnicity falls below conventional significance levels (Column 8).

Next, we turn to an investigation of how peace agreement views predict voter behavior. We measure peace agreement views through two survey questions: one gauging voting intentions were Colombia to hold another referendum on the peace agreement¹⁸ and one measuring participants’ views of the FARC-EP’s compliance with the peace agreement (both questions are detailed in the Online Appendix Table F1). We dichtomize and aggregate these items into a count variable, Peace Agreement Views (PA Views), which indicates negative (1), neutral (2), or positive (3) views on the peace agreement.¹⁹

In Table 2, we display how candidate characteristics influence behavior in the different sub-samples. The results reported in Table 2 align with the pre-registered expectations: participants with negative views (Column 1) on the peace agreement display greater aversion toward politicians with past armed group affiliation (−27.2 pp.) compared to participants with neutral (−15.6 pp., Column 2) and positive (−9.6 pp., Column 3) views on the peace agreement. In Column 4, we demonstrate that peace agreement views significantly moderate the effect of past armed group affiliation on the likelihood of a candidate being chosen.

Table 2.

Heterogeneous Effects: Views on the Peace Agreement.

Dep. Var	Candidate Chosen
Sample	Negative	Neutral	Positive	Full
	N = 294	N = 1641	N = 1233	N = 3168
	(1)	(2)	(3)	(4)
Afro-Colombian^a	−0.00314 (0.0651)	−0.0828 (0.0252)***	−0.0331 (0.0276)	−0.0799 (0.0711)
Past AG affiliation^b	−0.272 (0.0597)***	−0.156 (0.0268)***	−0.0960 (0.0314)***	−0.323 (0.0710)***
PA views				−0.0294 (0.0185)
Afro-Colombian × PA views				0.0103 (0.0293)
Past AG affiliation × PA views				0.0782 (0.0301)***
Constant	0.596 (0.0424)***	0.610 (0.0155)***	0.562 (0.0170)***	0.657 (0.0449)***
Observations	294	1641	1233	3168
R-squared	0.0619	0.0243	0.00798	0.0206
Dep. Var. Mean	0.514	0.529	0.521	0.525

Note: Table 2 displays regression estimates of the impact of candidate traits on the likelihood of a candidate being chosen, separately for participants with positive (Column 1), neutral (Column 2), and negative views on the peace agreement (Column 3). Column 4 reports interaction effects, where peace agreement views is operationalized as a continuous variable. Reference categories: (a) Caucasian named candidate; (b) No AG affiliation. All columns report robust standard errors clustered at the participant level.

Standard errors clustered at the participant-level in parentheses *p < 0.10, **p < 0.05, ***p < 0.010.

The findings align with Tellez (2019a), who finds that supporters of the peace agreement are more willing to accept the political reintegration of members of the FARC-EP. In the Online Appendix Table 9, we go a step further and demonstrate that opponents of the peace agreement also are less likely to trust candidates with past armed group affiliation.²⁰

Taken together, the results of the heterogeneity analyses highlight significant variation in voter biases against Afro-Colombian and former armed group affiliated candidates. The investigation of how participants’ ethnic identities shape voter behavior aligns with social identity theory, while the study of how peace agreement views influence voter preferences indicates that ethnicity-based discrimination is not uniform. Indeed, if we study the interaction between participants’ ethnic identities and their peace agreement views, we find that participants reporting Caucasian, Mestizo, or no ethnic identity and who support the peace agreement (i.e., are coded as “positive” according to PA views)²¹ display smaller ethnic and past armed group biases compared to participants of the same ethnic identities that do not support the peace agreement (i.e., are coded as “neutral” or “negative”).²²

Discussion and Conclusion

Peace agreements often entail provisions for the reintegration of former rebels into conventional politics (Godefroidt and Langer 2021). However, low popular demand for ex-combatant politicians complicates such a transition. Given that rebels anticipate the public’s reluctance, they may worry that demobilization would result in marginalization, which, in turn, could prevent them from realizing political objectives. Consequently, a peace agreement that comprises political reintegration will appear less appealing if there are no prospects for future reconciliation with the general electorate.

Moreover, as the present study showcased, the attractiveness of a political solution may vary within armed groups, since members differ along dimensions (e.g., ethnicity) that shape their personal electoral prospects. We considered how (past) armed group affiliation and ethnic belonging, separately and jointly, shape political candidates’ electoral prospects. To this end, we investigated candidate electability through a conjoint experiment implemented in a conflict-ridden department of Colombia. The experiment unveiled a strong bias against candidates with (past) armed group affiliations and significant voter discrimination against candidates with Afro-Colombian names. The use of names to proxy for ethnicity allows us to circumvent issues of social desirability bias common to studies of discriminatory behavior. Yet, since a share of participants are unaware of the ethnic cue embedded in the names (resulting in treatment non-compliance), the results are arguably a lower-bound estimate of the true extent of discrimination against Afro-Colombian candidates.

We did not find evidence of a hypothesized interaction between Afro-Colombian identity and armed group affiliation, i.e., an ethnically differentiated punishment for ties to armed groups. Still, since former armed group affiliates from the Afro-Colombian communities face significantly weaker electoral prospects compared to former armed group affiliates of European descent, the desirability of a political solution may still be ethnically differentiated. We further showed that voter biases can partly be attributed to stereotyping, since participants displayed lower trust in armed group-affiliated candidates as well as in Afro-Colombian candidates. The magnitudes of biases, however, vary between groups of participants. In pre-registered heterogeneity analyses, we found that participants who identified as Indigenous, Afro-Colombian, or of “other ethnic origin” were more likely to support candidates with Afro-Colombian ethnicity and past armed group affiliation, as were participants who supported the peace accord.

Understanding voters’ aversion toward former combatants and how this bias interacts with ethnic discrimination can shed light on rebel incentives to demobilize as well as on the prospects of successful post-conflict political reintegration. Although Colombia recently elected a president with past ties to an armed group and a vice president from the Afro-Colombian community, our study demonstrates that voter biases remain widespread. While the present study did not find evidence of an “Afro-Colombian penalty” for being associated with armed groups, the lack of interaction does not imply that the public judges Afro-Colombian and Caucasian armed group affiliates in the same way. For instance, the expected ethnic bias in voting might have been balanced out by people ascribing different levels of agency to Caucasian and Afro-Colombian candidates, considering that Afro-Colombians were more likely to be forcibly recruited during the conflict. Future studies should attempt to unpack the causal pathways from ethnic belonging to reintegration prospects.

More generally, peace and conflict research should build on the present study to investigate other instances of discrimination and ethnic barriers after war. The contextual and (sometimes) subtle nature of discriminatory behavior underscores the need for scholars to consider a range of methodologies and settings when mapping the phenomenon and its implications for peace processes. Ultimately, applying social identity theory to the dynamics of reconciliation, in a similar manner as we have done to the dynamics of war, can pave the way for more context-sensitive and effective approaches to peace building.

Supplemental Material

Supplemental Material - How Ethnic Discrimination Shapes Political Reintegration After War: Insights From a Conjoint Experiment in Colombia

Supplemental Material for How Ethnic Discrimination Shapes Political Reintegration After War: Insights From a Conjoint Experiment in Colombia by Gustav Agneman and Lisa Strömbom in Journal of Conflict Resolution

Supplemental Material

Supplemental Material - How Ethnic Discrimination Shapes Political Reintegration After War: Insights From a Conjoint Experiment in Colombia

Footnotes

Acknowledgments

We are grateful for generous feedback provided by Priscyll Anctil Avoine, Maria del Mar Morales Burkle, Isabel Bramsen, Anne Lene Stein, the Comparative Politics Group at Lund University, and conference participants at the MPSA Conference 2023.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The research was funded by the Bank of Sweden Tercentenary Foundation within the umbrella project “Pushing the Boundaries of Peace Research: Reconceptualizing and Measuring Agonistic Peace” (RIK18-14416).

ORCID iD

Gustav Agneman

Data Availability Statement

Data and code necessary for the replication of the main findings are available at Zenodo (Agneman, 2023). The study was pre-registered at AsPredicted #85836 ().

Supplemental Material

Supplemental material for this article is available online.

Notes

References

Agneman

Gustav

Chevrot-Bianco

Esther

. 2023. “Market Participation and Moral Decision-Making: Experimental Evidence From Greenland.” The Economic Journal 133 (650): 537-581.

Agneman

Gustav

Strömbom

Lisa

Rettberg

Angelika

. 2024. Do Apologies Promote the Reintegration of Former Combatants? Lessons from a Video Experiment in Colombia. The Journal of Peace Research. Under review. 2024.

Alesina

Alberto

La Ferrara

Eliana

. 2014. “A Test of Racial Bias in Capital Sentencing.” American Economic Review 104 (11): 3397-3433.

Annan

Jeannie

Blattman

Christopher

Mazurana

Dyan

Carlson

Khristopher

. 2011. “Civil War, Reintegration, and Gender in Northern Uganda.” Journal of Conflict Resolution 55 (6): 877-908.

Aspinall

Edward

. 2007. “The Construction of Grievance.” Journal of Conflict Resolution 51 (6): 950-972.

Barbary

Olivier

. 2004. “Gente negra en Colombia: dinámicas sociopolíticas en Cali y el Pacífico,” in editor by Giraldo

F. U.

. Medellín: Editorial Lealon, 113-156.

Bauer

Michal

Jana

Cahlíková

Chytilová

Julie

Roland

Gérard

Želinskỳ

Tomáš

. 2023. “Shifting Punishment onto Minorities: Experimental Evidence of Scapegoating.” The Economic Journal 133 (652): 1626-1640.

Bauer

Michal

Nathan

Fiala

Levely

Ian

. 2018. “Trusting Former Rebels: An Experimental Approach to Understanding Reintegration after Civil War.” The Economic Journal 128 (613): 1786-1819.

Bernhard

Helen

Fischbacher

Urs

Fehr

Ernst

. 2006. “Parochial Altruism in Humans.” Nature 442 (7105): 912-915.

10.

Blair

Graeme

Littman

Rebecca

Nugent

Elizabeth R.

Wolfe

Rebecca

Bukar

Mohammed

Crisman

Benjamin

Anthony

Etim

Hazlett

Chad

Kim

Jiyoung

. 2021. “Trusted Authorities Can Change Minds and Shift Norms during Conflict.” Proceedings of the National Academy of Sciences 118 (42): e2105570118.

11.

Born

Andreas

van Eck

Pieter

Johannesson

Magnus

. 2018. “An Experimental Investigation of Election Promises.” Political Psychology 39 (3): 685-705.

12.

Bornstein

Gary

. 2003. “Intergroup Conflict: Individual, Group, and Collective Interests.” Personality and Social Psychology Review 7 (2): 129-145.

13.

Canache

Damarys

Hayes

Matthew

Mondak

Jeffery J.

Seligson

Mitchell A.

. 2014. “Determinants of Perceived Skin-Color Discrimination in Latin america.” The Journal of Politics 76 (2): 506-520.

14.

Cárdenas

Roosbelinda

. 2012. “Multicultural Politics for Afro-Colombians.” In Black Social Movements in Latin America, 113-133. Springer.

15.

Castillo

Sandra Soler

Abril

Neyla Graciela Pardo

. 2009. “Discourse and Racism in Colombia: Five Centuries of Invisibility and Exclusion.” In Racism and Discourse in Latin America, 131-170. New York: Lexington Books.

16.

Centro Nacional de Memoria Histórica . 2022. Bloque Mineros De Las Auc. Violencia Contrainsurgente, Economías Criminales y Depredación Sexual. informe n.° 12.

17.

Cepeda

Iván

. 2006. “Genocidio Político: El Caso De La Unión Patriótica En Colombia.” Revista Cetil 1 (2): 101-112.

18.

Chae

Jihwan

Kim

Kunil

Kim

Yuri

Lim

Gahyun

Kim

Daeeun

Kim

Hackjin

. 2022. “Ingroup Favoritism Overrides Fairness when Resources Are Limited.” Scientific Reports 12 (1): 1-11.

19.

Chavez-Dueñas

Nayeli Y.

Adames

Hector Y.

Organista

Kurt C.

. 2014. “Skin-color Prejudice and Within-Group Racial Discrimination: Historical and Current Impact on Latino/a Populations.” Hispanic Journal of Behavioral Sciences 36 (1): 3-26.

20.

Coser

Lewis A.

1956. The Functions of Social Conflict, Volume 9. New York.

21.

Dahl

Malte

Krog

Niels

. 2018. “Experimental Evidence of Discrimination in the Labour Market: Intersections Between Ethnicity, Gender, and Socio-Economic Status.” European Sociological Review 34 (4): 402-417.

22.

Doherty

David

Dowling

Conor M.

Miller

Michael G.

. 2019. “Do Local Party Chairs Think Women and Minority Candidates Can Win? Evidence from a Conjoint Experiment.” The Journal of Politics 81 (4): 1282-1297.

23.

Everett

Jim A. C.

Faber

Nadira S.

Crockett

Molly

. 2015. “Preferences and Beliefs in Ingroup Favoritism.” Frontiers in Behavioral Neuroscience 9: 15.

24.

Freire

Diaz-Bonilla

Orellana

S. S.

Lopez

J. S.

Carbonari

. 2018. Oficina Regional para América Latina y el Caribe Práctica Global Social, Urbana, Rural y de Resiliencia Práctica Global de Pobreza y Equidad The World Bank.

25.

Gilligan

Michael J.

Mvukiyehe

Eric N.

Samii

Cyrus

. 2013. “Reintegrating Rebels into Civilian Life: Quasi-Experimental Evidence From Burundi.” Journal of Conflict Resolution 57 (4): 598-626.

26.

Godefroidt

Amélie

Langer

Arnim

. 2021. “What Drives Attitudes Towards the Reintegration of Former Fighters? Insights from a Conjoint Experiment in Nigeria.” Journal of Peace Research 60 (3): 00223433221087070.

27.

Gray

Vanessa Joan

. 2008. “The New Research on Civil Wars: Does it Help Us Understand the Colombian Conflict?” Latin American Politics and Society 50 (3): 63-91.

28.

Gutiérrez-Romero

Roxana

LeBas

Adrienne

. 2020. “Does Electoral Violence Affect Vote Choice and Willingness to Vote? Conjoint Analysis of a Vignette Experiment.” Journal of Peace Research 57 (1): 77-92.

29.

Gutiérrez Sanín

Francisco

. 2008. “Telling the Difference: Guerrillas and Paramilitaries in the Colombian War.” Politics & Society 36 (1): 3-34.

30.

Hainmueller

Jens

Hangartner

Dominik

Yamamoto

Teppei

. 2015. “Validating Vignette and Conjoint Survey Experiments Against Real-World Behavior.” Proceedings of the National Academy of Sciences 112 (8): 2395-2400.

31.

Hainmueller

Jens

Hopkins

Daniel J.

. 2015. “The Hidden American Immigration Consensus: A Conjoint Analysis of Attitudes Toward Immigrants.” American Journal of Political Science 59 (3): 529-548.

32.

Hainmueller

Jens

Hopkins

Daniel J.

Yamamoto

Teppei

. 2014. “Causal Inference in Conjoint Analysis: Understanding Multidimensional Choices via Stated Preference Experiments.” Political Analysis 22 (1): 1-30.

33.

Hartlyn

Jonathan

. 1988. The Politics of Coalitional Rule in Colombia. New York: Cambridge University Press.

34.

Homburger

Julian R.

Moreno-Estrada

Andrés

Gignoux

Christopher R.

Nelson

Dominic

Sanchez

Elena

Ortiz-Tello

Patricia

Pons-Estel

Bernardo A.

Acevedo-Vasquez

Eduardo

Miranda

Pedro

Langefeld

Carl D.

Gravel

Simon

Alarcón-Riquelme

Marta E.

Bustamante

Carlos D.

. 2015. “Genomic Insights into the Ancestry and Demographic History of South America.” PLoS Genetics 11 (12): e1005602.

35.

Huddy

Leonie

. 2001. “From Social to Political Identity: A Critical Examination of Social Identity Theory.” Political Psychology 22 (1): 127-156.

36.

Hudson

Rex A.

2010. Colombia: A Country Study, Volume 550. United States: Government Printing Office.

37.

Ibanez

Ana Maria

. 2009. “Forced Displacement in Colombia: Magnitude and Causes.” The Economics of Peace and Security Journal 4 (1): 48-54.

38.

Kalin

Michael

Sambanis

Nicholas

. 2018. “How to Think about Social Identity.” Annual Review of Political Science 21: 239-257.

39.

Kao

Kristen

Revkin

Mara R.

. 2021. “Retribution or Reconciliation? Post-conflict Attitudes toward Enemy Collaborators.” American Journal of Political Science 67 (2): 358-373.

40.

La Semana (18 June 2022). Exguerrillero o Ingeniero? Colombia Elige a Su Nuevo Presidente Entre Gustavo Petro y Rodolfo Hernández, Quién es Quién? La Semana: https://www.semana.com/nacion/articulo/exguerrillero-o-ingeniero-colombia-elige-a-su-nuevo-presidente-entre-gustavo-petro-y-rodolfo-hernandez-quien-es-quien/202219/. Accessed 9 August 2022.

41.

Lahey

Joanna N.

Oxley

Douglas R.

. 2018. Discrimination at the Intersection of Age, Race, and Gender: Evidence from a Lab-In-The-Field Experiment. Technical report, National Bureau of Economic Research.

42.

Lambardi

German

Palacios

Paola

. 2022. “Land Use and the Incidence of Forced Displacement.” International Interactions 49: 1-30.

43.

Llaneras

Kiko

Borja

Andrino

Sevillano Pires

Luis

30 May. 2022. “Resultado de las Elecciones en Colombia Por Municipio.” El País. https://elpais.com/america-colombia/elecciones-presidenciales/2022-05-29/mapa-de-las-elecciones-en-colombia-con-los-resultados-en-cada-municipio.html. Accessed 10 August 2023.

44.

Loizides

Neophytos

Psaltis

Charis

Morgan-Jones

Edward

Sudulich

Laura

Popp

Raluca

Baykiz

Tekin

. 2022. “Citizens and Peace Mediations in Divided Societies: Identifying Zones of Agreement through a Conjoint Survey Experiment.” Journal of Conflict Resolution 66 (9): 1619-1649.

45.

López López

Wilson

Alvarado

Gonzalo Sandoval

Rodríguez

Sebastián

Ruiz

Camila

León

Juan David

Pineda-Marín

Claudia

Mullet

Etienne

. 2018. “Forgiving Former Perpetrators of Violence and Reintegrating Them into Colombian Civil Society: Noncombatant Citizens’ Positions.” Peace and Conflict: Journal of Peace Psychology 24 (2): 201.

46.

Magni

Gabriele

Reynolds

Andrew

. 2021. “Voter Preferences and the Political Underrepresentation of Minority Groups: Lesbian, Gay, and Transgender Candidates in Advanced Democracies.” The Journal of Politics 83 (4): 1199-1215.

47.

Martínez

Sergi

Valentim

Vicente

Dinas

Elias

. 2023. “Symbolic Conflict Resolution and Ingroup Favoritism.” Journal of Conflict Resolution 67 (10): 00220027231155323.

48.

Morrison

Judith A.

2015. “Behind the Numbers: Race and Ethnicity in Latin America.” Americas Quarterly 9 (3): 80.

49.

Múnera

Alfonso

. 2005. Fronteras Imaginadas. Bogota: Editorial Planeta Colombiana.

50.

Murillo

Mario A.

2003. Colombia and the United States: War, Unrest and Destabilization. New York: Seven Stories Press.

51.

Murillo Orejuela

Luis

Restrepo-Plaza

Lina

. 2021. “Reintegration and Forgiveness to Ex-Combatants in Colombia.” Development Studies Research 8 (1): 36-48.

52.

Nussio

Enzo

. 2011. “Learning from Shortcomings: The Demobilisation of Paramilitaries in Colombia.” Journal of Peacebuilding & Development 6 (2): 88-92.

53.

Phelan

Alexandra

. 2019. “Engaging Insurgency: The Impact of the 2016 Colombian Peace Agreement on Farc’s Political Participation.” Studies in Conflict & Terrorism 42 (9): 836-852.

54.

Podder

Sukanya

. 2012. “From Recruitment to Reintegration: Communities and Ex-Combatants in Post-conflict Liberia.” International Peacekeeping 19 (2): 186-202.

55.

Portmann

Lea

. 2022. “What Makes a Successful Candidate? Political Experience and Low-Information Cues in Elections.” The Journal of Politics 84 (4): 2049-2063

56.

Portmann

Lea

Stojanović

Nenad

. 2019. “Electoral Discrimination Against Immigrant-Origin Candidates.” Political Behavior 41 (1): 105-134.

57.

Portmann

Lea

Stojanović

Nenad

. 2022. “Are Immigrant-Origin Candidates Penalized Due to Ingroup Favoritism or Outgroup Hostility?” Comparative Political Studies 55 (1): 154-186.

58.

Roccas

Sonia

Brewer

Marilynn B.

. 2002. “Social Identity Complexity.” Personality and Social Psychology Review 6 (2): 88-106.

59.

Rodríguez-Garavito

César

Sierra

Tatiana Alfonso

Adarve

Isabel Cavelier

. 2008. “Racial Discrimination and Human Rights in Colombia.” Global Justice Series.

60.

Schiller

Bastian

Baumgartner

Thomas

Knoch

Daria

. 2014. “Intergroup Bias in Third-Party Punishment Stems from Both Ingroup Favoritism and Outgroup Discrimination.” Evolution and Human Behavior 35 (3): 169-175.

61.

Schwarz

Susanne

Coppock

Alexander

. 2018. “What Have We Learned about Gender From Candidate Choice Experiments? A Meta-Analysis of 30 Factorial Survey Experiments.” Unpublished Manuscript.

62.

Shayo

Moses

Zussman

Asaf

. 2011. “Judicial Ingroup Bias in the Shadow of Terrorism.” The Quarterly Journal of Economics 126 (3): 1447-1484.

63.

Sierra

Jerónimo Ríos

García

Jaime Zapata

. 2019. “Democratic Security Policy in Colombia: Approaches to an Enemy-Centric Counterinsurgency Model.” Revista de humanidades 36: 129-154.

64.

Soendergaard

Maren

(6 March 2014). How Political Exclusion Affects Colombia’s Black Minority. https://colombiareports.com/amp/afro-colombian-political-exclusion-effects-social-indicators/. Accessed 26 May 2022.

65.

Stantcheva

Stefanie

. 2022. How to Run Surveys: A Guide to Creating Your Own Identifying Variation and Revealing the Invisible. Technical report, National Bureau of Economic Research.

66.

Steele

Abbey

. 2011. “Electing Displacement: Political Cleansing in Apartadó, Colombia.” Journal of Conflict Resolution 55 (3): 423-445.

67.

Strömbom

Lisa

. 2014. “Thick Recognition. Advancing Theory on Identity Change in Intractable Conflictsr.” European Journal of International Relations 20 (1): 168-191.

68.

Tajfel

Henri

. 1974. “Social Identity and Intergroup Behaviour.” Social Science Information 13 (2): 65-93.

69.

Taylor

Luke

(25 July 2023). ‘Insult to His Victims’: Outrage as Warlord Appointed ‘Peace Manager’ in Colombia. The Guardian. https://www.theguardian.com/world/2023/jul/25/colombia-salvatore-mancuso-peace-manager-outrage. Accessed 15 August 2023.

70.

Teele

Dawn Langan

Kalla

Joshua

Rosenbluth

Frances

. 2018. “The Ties that Double Bind: Social Roles and Women’s Underrepresentation in Politics.” American Political Science Review 112 (3): 525-541.

71.

Tellez

Juan Fernando

. 2019a. “Peace Agreement Design and Public Support for Peace: Evidence from Colombia.” Journal of Peace Research 56 (6): 827-844.

72.

Tellez

Juan Fernando

. 2019b. “Worlds Apart: Conflict Exposure and Preferences for Peace.” Journal of Conflict Resolution 63 (4): 1053-1076.

73.

Thrasher

Michael

Borisyuk

Galina

Rallings

Colin

Webber

Richard

. 2017. “Candidate Ethnic Origins and Voter Preferences: Examining Name Discrimination in Local Elections in Britain.” British Journal of Political Science 47 (2): 413-435.

74.

Uberti

Ornella

. 2022. “Un Análisis Del Reclutamiento Forzado De Niños, Niñas Adolescentes y Jóvenes (nnaj) en el Departamento de Chocó, Colombia/an Analysis of the Forced Recruitment of Children, Adolescents and Young People (nnaj) in the Departamento of Chocó, Colombia.” Encuentros 20 (02-Julio-Dic.): 141-153.

75.

Universidad Nacional de Colombia . 2017. “Caracterización Comunidad Farc-Ep: Resultados Generales.”

76.

Wucherpfennig

Julian

Metternich

Nils W.

Cederman

Lars-Erik

Gleditsch

Kristian Skrede

. 2012. “Ethnicity, the state, and the duration of civil war.” World Politics 64 (1): 79-115.

77.

Yordi

Lucas

. 2022. Colombia Elections 2022: An Uncertain Political Future for the Former Farc Guerrilla. LSE Latin America and Caribbean Blog. https://blogs.lse.ac.uk/latamcaribbean/2022/03/15/colombia-elections-uncertain-future-farc/ (Accessed 15 August 2023).

78.

Yudkin

Daniel A.

Rothmund

Tobias

Twardawski

Mathias

Thalla

Natasha

Van Bavel

Jay J.

. 2016. “Reflexive Intergroup Bias in Third-Party Punishment.” Journal of Experimental Psychology: General 145 (11): 1448.

79.

Zschirnt

Eva

Ruedin

Didier

. 2016. “Ethnic Discrimination in Hiring Decisions: A Meta-Analysis of Correspondence Tests 1990–2015.” Journal of Ethnic and Migration Studies 42 (7): 1115-1134.

80.

Agneman

Gustav

. 2023. How Ethnic Discrimination Shapes Political Reintegration After War.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.03 MB

0.00 MB

0.54 MB