Group-Based Injustice,but Not Group-Based Economic Inequality,Predicts Political Violence Across 18 African Countries

Abstract

Political violence causes immense human suffering. Scholars pinpoint economic inequalities between ethnic groups as a major cause of such violence. However, the relationships between group-based inequality, group-based injustice, and political violence are not fully understood. Combining insights from social psychological research on collective action and political science research on civil conflict, we underscore that it is group-based injustice that motivates violence. A perception that one’s group has been treated unfairly tends to produce conflict-related emotions (e.g., anger). By contrast, a mere perception that one’s group is of lower economic status rarely produces such emotions. Furthermore, perceived economic disadvantage negatively relates to perceived political efficacy, which may dissuade engagement in political violence. To assess these arguments, we analyzed attitudes toward, intentions to engage in, and self-reported engagement in political violence, utilizing probability samples from 18 African countries (N > 37,000). We found that measures of group-based perceived injustice, whether controlling or not for group-based economic inequality, predicted all violent outcomes; whereas measures of perceived group-based inequality predicted (negatively) self-reported participation in violence but not the other outcomes. We advance both social psychological and political science literatures, suggesting that group-based injustice and inequality are distinct constructs, relating to political violence via different pathways.

Keywords

civil conflict intergroup relations political psychology political violence injustice inequality

Introduction

Coalitional violence for a political cause, or political violence, is a major source of human suffering. Large-scale organized political violence, such as civil conflict (Gleditsch et al., 2002), constitutes a particularly destructive form of violence (Ghobarah et al., 2003). The Tigray War in Ethiopia is a recent example of a civil conflict, causing more than 100,000 battle-related deaths in just 2 months of fighting in late 2022 (Davies et al., 2023). Other forms of political violence include international wars (e.g., the all-out Russian invasion of Ukraine), terrorist attacks (The September 11th attacks), coups or coup attempts (the 2016 Turkish coup attempt), and violent protests (the 2022 Kazakh unrest).

The study of organized political violence, in particular, research on civil conflict has been heavily influenced by political science (e.g., Fearon & Laitin, 2003) and economics (e.g., Collier & Hoeffler, 2004). Over the last decades, this research has increasingly emphasized inequalities between ethnic or culturally defined groups, the so-called “horizontal inequalities” (Cederman et al., 2011; Stewart, 2008; Østby, 2008, 2013). Research on horizontal inequalities has generated extensive theoretical debate and voluminous empirical evidence, becoming one of the most influential lines of research in the study of political violence (Cederman & Vogt, 2017).

Research on civil conflict has traditionally focused on objective conditions (e.g., group-based wealth distribution) and on states or groups as units of analysis. This macro-level research has typically assumed that inequalities cause grievances that motivate violence. However, psychological research shows that inequalities alone are insufficient to generate conflict-related emotions, such as anger: “people often recognize and accept their relative disadvantage as appropriate” (Smith et al., 2012; Starmans et al., 2017). If anger, hatred, or contempt are required to motivate violence (Becker & Tausch, 2015), and if perceived inequalities alone do not cause them, then analyses focused on inequalities per se are problematic. This concern has recently been voiced in micro-level studies of political violence (e.g., Bartusevičius & van Leeuwen, 2022; Dyrstad & Hillesund, 2020). However, we still poorly understand the relationship between group-based inequality and group-based injustice on the one hand and motivations for political violence on the other. Furthermore, and related, political science research on organized violence has seen limited cross-disciplinary integration, particularly with psychological research, where the injustice-collective action link has been studied for as long as the inequality-conflict link in civil conflict research. The study of why people engage in political violence inevitably involves the consideration of psychological variables and processes (e.g., perceptions and motivations); yet psychological research does not feature prominently in civil conflict research.

In what follows, we present an integration of psychological research on relative deprivation and collective action with the study of horizontal inequality and civil conflict. The causal role of inequality-related injustices, as contrasted to perceived inequalities as such, is well known in psychological research but less well understood in the study of civil conflict. Furthermore, both literatures have yet to clarify the role of perceived inequalities independently from injustices. Psychological research often considers perceived inequalities as (weak) proxies of injustices. Departing from this perspective, we suggest considering group-based perceived inequality as a distinct construct from group-based perceived injustice. Group-based inequalities can, but far from always, produce injustice perceptions (a point we address in the following sections), yet the former, unlike the latter, also likely undermines groups’ political efficacy. This implies that perceived group-based inequality as such might dissuade disadvantaged groups from engaging in political violence. In sum, we posit—our core theoretical contribution—that perceived group-based inequality and perceived group-based injustice have divergent effects on political violence, produced via distinct pathways.

We test our arguments in a multinational analysis of large probability samples from 18 African countries. Psychological research has traditionally focused on non-violent collective action (van Zomeren et al., 2008). Only recently, violence has attracted a more considerable interest among psychologists (Agostini & van Zomeren, 2021; Becker & Tausch, 2015). In addition, the psychological study of collective action has typically centered on Western countries. According to Ayanian et al. (2021)—who have surveyed protesters in highly repressive countries, such as Russia—the dynamics of collective action are profoundly different in non-Western autocracies and Western democracies because protesting in the latter is much safer (see also the work of Ayanian & Tausch, 2016). Concurring with this, we suggest that group-based inequality and group-based injustice likely influence political violence, and collective action more generally, in different ways depending on the political and cultural contexts. Initiating an antigovernment action for a low-resource group may entail dramatically different risks and success probabilities in repressive and non-repressive regimes. Similarly, whether a disadvantaged group acts to address some injustice may be greatly influenced by political norms (e.g., submission to authorities) and system-justifying beliefs (Osborne et al., 2019). We thus cannot assess the generality of the links between inequalities, injustices, and political violence if we solely focus on Western contexts.

Although recent psychological research on collective action has expanded focus to violence and non-Western samples, extant studies remain limited to only a handful of countries and small (often non-representative) samples. By contrast, research on civil conflict has traditionally analyzed violence using data from many diverse countries. However, this research has only recently started testing individual-level theories using individual-level data (e.g., Bartusevičius & van Leeuwen, 2022; Dyrstad & Hillesund, 2020). Focusing specifically on political violence, and using a large multinational sample of non-Western countries, we thus empirically advance both psychological research on collective action and political science research on organized political violence.

Theory

Horizontal Inequalities and Civil Conflict

Organized political violence, such as civil conflict, is typically understood as an outcome of motivations and opportunities (Collier & Hoeffler, 2004; Fearon & Laitin, 2003): Civil conflicts occur whenever non-state actors, such as rebel groups, have reasons and opportunities to fight governments or vice versa. Inequalities between ethnic or culturally-defined groups,¹ the so-called horizontal inequalities, are understood as contributing to both motivations and opportunities (e.g., Cederman et al., 2011). For example, income differences between ethnic groups can motivate members of the disadvantaged groups to challenge the status quo distribution of resources. Such differences can also aid the mobilization of disadvantaged individuals for collective action. Ethnic groups share a common social identity, and horizontal inequalities can strengthen the salience of these identities, which in turn can facilitate the mobilization of solitary individuals for collective action (Gurr, 2000; Østby, 2008; Stewart, 2008). Thus, macro research postulates a positive relationship between objectively measured horizontal inequalities and civil conflict. However, whether this relationship holds depends on multiple micro processes that need to be analyzed in individual-level research. In what follows, we delineate such processes utilizing insights from psychological research on relative deprivation and collective action (Agostini & van Zomeren, 2021; Smith et al., 2012).

What Actually Motivates Political Violence?

Psychologists have rarely analyzed large-scale events of political violence; however, they have analyzed a range of other outcomes that likely constitute precursors of political violence, such as participation in collective actions (e.g., protests) and intergroup attitudes (e.g., outgroup hostility; Smith et al., 2012). Psychological evidence suggests that anger—which may or may not be caused by group-based inequality—is the primary factor that motivates collective action (Agostini & van Zomeren, 2021). This affective aspect is embedded in standard definitions of relative deprivation in social psychology: “(a) People first make cognitive comparisons, (b) they next make cognitive appraisals that they or their ingroup are disadvantaged, and finally (c) these disadvantages are seen as unfair and arouse angry resentment. If any of these three requirements is missing, RD [relative deprivation] is not operating” (Pettigrew, 2016, p. 9).

Theoretical analyses of civil conflict stress the importance of felt grievances (Cederman et al., 2011). Yet, empirical analyses typically rely on measures that are agnostic to emotions, implicitly assuming that the presence of inequality implies the presence of felt grievance. Psychological research demonstrates that this assumption is problematic: people often recognize and accept their relative disadvantage. For example, facing societal threats, people often justify unequal systems (Osborne et al., 2019). Hence, whether objective or perceived, inequality alone is insufficient to cause anger or other conflict-related emotions such as hate or contempt (Becker & Tausch, 2015). This implies that measures of objective or perceived inequalities may not relate to motivations for violence.

What, then, causes conflict-related emotions? Psychological research points to inequalities that are perceived unjust (Smith et al., 2012; Starmans et al., 2017). In a recent individual-level study of political violence, Dyrstad and Hillesund (2020) underscored, and empirically demonstrated, the importance of injustice perceptions for people’s attitudinal support for political violence. However, Dyrstad and Hillesund (2020) did not assess how inequalities influence political violence independently from injustices. Furthermore, Dyrstad and Hillesund (2020) analyzed only three countries, and the authors suggest that country-level characteristics influence individual-level associations between injustices and political violence. Finally, these authors examined attitudinal support for but not motivations to engage or engagement in political violence.

Synthesizing the Core Argument and Hypotheses

Drawing on the above discussion, we thus suggest that measures of group-based inequalities and group-based injustices relate to two distinct motivations for collective action, as delineated, for example, by Agostini and van Zomeren (2021): perceived injustice and perceived group efficacy. Group-based inequality may or may not produce injustice perceptions; however, it very likely relates to perceptions of a group’s political efficacy. While deliberating collective action, disadvantaged groups consider their potential to affect the desired change (e.g., to reduce group-based inequality). Within psychology, such perceived potential is referred to as “political efficacy” and is considered one of four key motivations for collective action (others being identity, injustice, and morality; Agostini and van Zomeren, 2021).

Similarly, civil conflict research considers economic resources of conflict actors as key determinants of their military capacity (Collier & Hoeffler, 2004; Fearon & Laitin, 2003). This implies, then, that horizontal inequality per se may undermine perceived efficacy of disadvantaged groups: People may deem their resources too low to confront governments. Hence, perceived disadvantage and injustice likely have divergent effects. If this holds true, then analysis of one variable should consider holding constant the other. That is, since group-based inequality may simultaneously influence injustice and efficacy perceptions, controlling for group-based injustice may help isolate the “pure” effects of group-based inequality.

Taken together, we hypothesize that group-based perceived injustice, controlling for group-based perceived economic inequality, positively relates to motivations to engage in political violence (H1) and, conversely, that group-based perceived economic inequality, controlling for group-based perceived injustice, negatively relates to such motivations (H2).

Some Qualifications

First, concurring with conventional models of civil conflict (Collier & Hoeffler, 2004; Fearon & Laitin, 2003) and existing micro research (Bartusevičius & van Leeuwen, 2022; Dyrstad & Hillesund, 2020; Miodownik & Nir, 2016; Rustad, 2016), we focus on ordinary citizens, not political or military leaders. Violent collective actors critically rely on recruitment and popular support (Dyrstad & Hillesund, 2020). Hence, attitudes and motivations of would-be participants play a central role in civil conflict.

Second, we focus on the antigovernment side. While members of non-state actors often exercise individual agency in their choice to partake in violence, soldiers of state armies typically do not. Therefore, felt injustices or perceived inequalities should play a less important role among the latter.

Third, inequalities between ethnic groups can have several dimensions: representation in political institutions, distribution of wealth, or access to public services. We acknowledge the importance of studying such different dimensions; yet, here we will exclusively focus on economic inequality (Cederman et al., 2011; Stewart, 2008). Arguably, economic inequality constitutes the most commonly analyzed dimension of horizontal inequalities in the study of civil conflict. Furthermore, a unidimensional focus better aligns our concepts, theories, and measures (Coppedge et al., 2008). Analyzing several dimensions would involve specifying the theoretical mechanisms pertaining to each and how they interact in motivating violence.

Fourth, recent psychological research distinguishes between perceptions of injustice (cognitive injustice) and feelings of injustice (emotional injustice) and suggests that the latter matters more for collective action motivations (Agostini & van Zomeren, 2021). Furthermore, research shows that people can react with stronger, weaker, or no anger to injustices (Osborne et al., 2019), which suggest the importance of identifying those injustice types that are more likely to cause anger. However, we see such typology as the subsequent step, after we establish the role of cognitive injustice in political violence. Meta-analytic evidence shows that measures of injustices that account for emotional injustice, compared to those that only account for cognitive injustice, more strongly relate to collective action outcomes (Agostini & van Zomeren, 2021); yet the former remain reliable predictors of collective action motivations. Hence, although we acknowledge the importance of distinguishing between cognitive and emotional injustices, the present article exclusively focuses on the former.

Fifth, we do not examine mobilization processes through which solitary individuals form collective violent actors. This does not imply that psychological factors are irrelevant for macro events of violence. Individuals motivated to use violence constitute “rebel recruitment pools” (Collier & Hoeffler, 2004). The larger the share of such individuals within a country, the higher the chance of violence in the country. We posit that such pools grow with the population share of people perceiving injustice.

Finally, we do not investigate the underlying processes through which perceived injustices are formed. A large literature exists on this, pointing, for example, to entitlement and deservingness (see literature review in the study by Feather, 2015).

Materials and Methods

In the following sections, we follow a similar research design to that of Bartusevičius and van Leeuwen (2022) and Bartusevičius et al. (2020). These studies analyzed similar outcomes of political violence using Afrobarometer data (specifically, Afrobarometer Rounds 2 and 5), but they focused on a different set of explanatory variables, namely decremental deprivation and autocratic political orientations.

Data

We identified two large multinational surveys that asked questions about group-based inequalities, group-based injustices, and political violence: Afrobarometer Round 2 collected in 2002–2004 (Afrobarometer, 2004) and Afrobarometer Round 3 collected in 2005–2006 (Afrobarometer, 2005). We are not aware of any other multinational surveys containing questions about group-based inequalities, group-based injustices, and political violence. This includes all other rounds of the Afrobarometer, Latinobarómetro, Arab Barometer, Eurasia Barometer, and Asian Barometer Survey.

The Afrobarometers conduct face-to-face interviews of citizens of voting age on various topics related to the economy, governance, and living standards. The surveys use clustered, stratified, multi-state probability sampling, applying random selection with a probability proportionate to the population size at every stage. The coverage of Afrobarometer datasets is exceptional: Round 2 spans 16 countries with ns from 1,198 to 2,400 per country (total N = 24,301; relevant for our analysis N = 16,937 to N = 17,703), and Round 3 spans 18 countries with ns from 1,048 to 2,400 per country (total N = 25,397; relevant for our analysis N = 19,968). The geographical distribution of the survey countries is illustrated in Figures 1 and 2. Specifically, the following 16 countries are covered in Round 2: Botswana, Cape Verde, Ghana, Kenya, Lesotho, Malawi, Mali, Mozambique, Namibia, Nigeria, Senegal, South Africa, Tanzania, Uganda, Zambia, and Zimbabwe. Round 3 covers the same countries, except for Zimbabwe, where the survey did not include questions relevant for our analysis. In addition, Round 3 also includes Benin and Madagascar. Altogether, we utilized data from 18 different countries with over 37,000 respondents.

Figure 1

Map of Surveyed Countries (Afrobarometer Round 2)

Figure 2

Map of Surveyed Countries (Afrobarometer Round 3)

Such coverage provides several advantages. First, political violence represents an extreme form of political behavior, and so the number of those who participate in political violence constitutes only a fraction of the population. Small samples from single countries are unlikely to contain the number of participants in violence sufficient for a statistical analysis. Second, some country-level characteristics (e.g., political repression) may confound the individual-level associations between group-based inequality, injustice, and political violence. Multinational analyses allow us to account for the influences of such country-level characteristics.

Measuring Political Violence

We used three types of outcome measures: behavioral self-reports of participation in political violence (PVparticipation), behavioral intentions to participate in political violence (PVintentions), and attitudinal support (or endorsement) of political violence (PVsupport). Items suitable for measuring all three outcomes are available in the Round 2 data. The Round 3 data contain identical items on PVsupport but no items to measure PVparticipation and PVintentions. Thus, Round 3 allows replicating the tests of H1 and H2 with different data, although only for one outcome (i.e., PVsupport). To measure PVparticipation and PVintentions, we utilized the following question:

Here is a list of actions that people sometimes take as citizens. For each of these, please tell me whether you, personally, have done any of these things during the past year. If not, would you do this if you had the chance: Used force or violence for a political cause? (Beard, 2006, p. 18)

The original response options were “No, would never do this,” “No, but would do if had the chance,” “Yes, once or twice,” “Yes, several times,” and “Yes, often” (Beard, 2006, p. 18). With PVparticipation, we aimed to measure engagement in violence (as contrasted to attitudes or intentions); therefore, we aggregated the two no-responses into one. As such, our PVparticipation measure takes the following scores: 0 = “No,” 1 = “Yes, once or twice,” 2 = “Yes, several times,” and 3 = “Yes, often.” Supplemental Material provides additional analyses using the full original scale, as well as analyses using a dichotomized version of PVparticipation. These additional analyses generate the same substantive conclusions (Tables S9 and S10 in Supplemental Material).

To measure behavioral intentions, we derived a binary indicator that focuses on the vast majority of respondents who did not report participation in political violence, that is, all those who answered “No, would never do this” (0) or “No, but would do if had the chance” (1) to the above quoted question. Measured in this way, PVintentions allows analyzing exclusively behavioral intentions, as contrasted to behavioral self-reports, which are often considered as distinct types of measures in psychological research on collective action (Agostini & van Zomeren, 2021). In addition, actual engagement in political violence may influence how people perceive inequalities and injustices, and so an analysis of a measure that includes past participation in violence may reflect the effects of such participation on inequality and injustice perceptions, rather than vice versa. By contrast, since PVintentions excludes all (self-reported) participants in violence and asks about intentions to engage in violence at the time of interview, this measure alleviates such reverse-causality concerns.

The attitudinal outcome, support for political violence (PVsupport), was measured using the following item: “. . . which of the following statements is closest to your view . . .,” “A. The use of violence is never justified in [country name] politics.,” “B. In this country, it is sometimes necessary to use violence in support of a just cause.” The original response options were “Agree Very Strongly With A,” “Agree With A,” “Agree With B,” “Agree Very Strongly With B,” and “Agree with Neither” (Beard, 2006, p. 49; Carter, 2008, p. 23). We generated a five-point scale setting “Agree with Neither” as the middle score.

The three outcome measures are based on respondents’ self-reports, not actual or observed behavior. This must be considered while evaluating our following analyses. We return to limitations of self-reported data in Discussion.

Measuring Group-Based Injustice and Inequality

Research on horizontal inequalities and civil conflict has typically focused on inequalities between ethnic groups, where “ethnic” commonly implies linguistic, religious, or geographic groups (see the definition in footnote 1). We see our theoretical argument as general, applying to social groups that define themselves based on other characteristics, for example, socioeconomic status or profession. However, the data we have at hand (especially Round 3) largely, but not exclusively, focus on what civil conflict scholars would normally consider “ethnic groups” (see Tables S5 and S6 in Supplemental Material).

The questions we utilized for measuring group-based injustice and inequality were similar but not identical across Rounds 2 and 3. In Round 2, the respondents were first asked:

We have spoken to many [e.g., Kenyans] and they have all described themselves in different ways. Some people describe themselves in terms of their language, ethnic group, race, religion, or gender and others describe themselves in economic terms, such as working class, middle class, or a farmer. Besides being [Kenyan], which specific group do you feel you belong to first and foremost? (Beard, 2006, p. 40f)

The response options were open-ended. Afrobarometer then classified each response into specific categories (e.g., “Language/tribe/ethnic group”). In turn, the following question was asked about economic inequality (hereafter, inequality): “Are [R’s IDENTITY GROUP] economic conditions worse, the same as, or better than other groups in this country?”(“Much better,” “Better,” “About the same,” “Worse,” “Much worse”; Beard, 2006, p. 44). Then followed a question about perceived injustice (injustice): “How often are [R’s IDENTITY GROUP] treated unfairly by the government?” (“Never,” “Sometimes,” “Often,” “Always”; Beard, 2006, p. 45).

In Round 3, the respondents were similarly first asked about their identity, but the question placed a greater emphasis on ethnicity: “What is your tribe? You know, your ethnic or cultural group” (Carter, 2008, p. 43). As in Round 2, respondents answered the question freely. Subsequently, the questions about inequality and injustice followed: “Think about the condition of [R’s IDENTITY GROUP]. Are their economic conditions worse, the same as, or better than other groups in this country?” (“Much better,” “Better,” “About the same,” “Worse,” “Much worse”) (inequality); “How often are [R’s IDENTITY GROUP] treated unfairly by the government?” (“Never,” “Sometimes,” “Often,” “Always”) (injustice) (Carter, 2008, p. 44-45).

Respondents who answered “Don’t know,” did not provide any reply, or responded “Won’t differentiate/National Identity only” to the above quoted questions about group identity did not receive the questions about injustice and inequality (19% of the Round 2 sample and 10% of the Round 3 sample).

We used continuous versions of inequality and injustice, with higher values indicating higher levels of perceived inequality and injustice. The correlation between injustice and inequality was r = 0.19 in the Round 2 data and r = 0.28 in the Round 3 data; Tables S3 and S4 in Supplemental Material provide a bivariate correlation matrix for all variables. The correlations imply that, although perceived group-based inequality and perceived group-based injustice correlate, the two operational variables share only a small fraction of covariation and thus likely represent distinct constructs.

Modeling

We opted for models to estimate “pure” within-country effects, as our primary interest was in individual-level associations, not higher-level variables that moderate or confound such associations. Specifically, we used country fixed-effects estimators that are common in political science and economics (Bell & Jones, 2015) but less familiar in psychological research (McNeish & Kelley, 2019). These models are mathematically equivalent to conventional multilevel models with random intercepts and group-demeaned predictors (Bell et al., 2019; McNeish & Kelley, 2019). Unlike multilevel models, fixed-effect models do not require large number of higher-level units, and our data contain only 16 (Round 2) and 17 (Round 3) such units (McNeish & Stapleton, 2016). Mathematically, country fixed-effect estimators subtract country-level means of each variable from each observation and then regresses the demeaned outcomes on demeaned predictors.² As such, these models fully remove country-level influences, resulting in coefficients that represent the individual-level associations of interest. We used a logit variant of the fixed-effects estimator to analyze the binary outcome (PVintentions) and a linear variant with country fixed effects to analyze continuous outcomes (PVparticipation and PVsupport). We normalized all variables in the analyses to range from 0 to 1 to aid interpretation.

Regarding individual-level controls, we opted for parsimonious specifications with only age and gender as control variables in the main analysis. Parsimonious models are less vulnerable to the so-called post-treatment bias, whereby the effects of predictors are underestimated or overestimated because the controls include mediators through which the predictors exert their effects. In Supplemental Material, we present analyses with additional demographic controls (education and income) and models with controls for other predictors of political violence (trust in government, ethnic identification, and freedom of expression).

Transparency and Openness

In previous sections, we have reported all data exclusions and relevant measures. We did not report how we determined the sample size because we used secondary data, and we used all available observations in these data. The data and analysis code are available at: https://osf.io/naw29/. Data were analyzed in Stata (v17) using various packages, all of which are reported in the replication code. This study’s design and its analysis were not preregistered.

Results

First, we tested H1 using the Afrobarometer Round 2 data. If the hypothesis holds true, then group-based perceived injustice should positively relate to motivations to engage in political violence. As shown in the right-hand panes of Figures 3 –5, we found support for H1. Group-based injustice positively relates to all three outcome measures, PVparticipation (b = 0.01, 95% CI = [0.01, 0.02]), PVintentions (log-odds = 0.47, 95% CI = [0.31, 0.63]), and PVsupport (b = 0.07, 95% CI = [0.05, 0.08]), while controlling for group-based inequality; see detailed results in Table S7 in the Supplemental Material. These results remain robust if we add additional individual-level controls (income, education, trust in governments, ethnic identification, and freedom of expression) and use alternative model specifications—including those without inequality as control—estimators, or measures of outcomes; for details, see Tables S9–S15 in the Supplemental Material. Substantively, a min-to-max increase in injustice is associated with an estimated 1-percentage-point increase (from 0.02 to 0.03) in the predicted value of PVparticipation, a 4-percentage-point increase (from 8% to 12%) in the predicted probability of PVintentions, and a 7-percentage-point increase (from 0.25 to 0.32) in the predicted value of PVsupport. To put these estimates in perspective, men, compared to women, are 1-percentage point more supportive of political violence (0.28 vs. 0.27).

Figure 3

Participation in Political Violence, Round 2

Figure 4

Intentions to Participate in Political Violence, Round 2

Figure 5

Support for Political Violence, Round 2

Next, we tested H2 using the same Round 2 data. If the hypothesis holds true, then group-based perceived inequality should negatively relate to motivations to engage in political violence. The left-hand pane of Figure 3 shows that inequality is negatively associated with PVparticipation while controlling for injustice (b = −0.01, 95% CI = [−0.02, 0.00]; see Table S7 in the Supplemental Material). In turn, the association of inequality with the other two outcome measures, PVintentions and PVsupport, is also negative, but the estimates are imprecise: log-odds = −0.14, 95% CI = [−0.34, 0.05]; b = −0.02, 95% CI = [−0.03, 0.00] (see Figures 4 and 5). We thus find mixed support for H2 in the analysis of the Round 2 data, and the results remain substantively the same if we do not control for injustice (see Table S7 in Supplemental Material).

We now turn to testing H1 and H2 using the Round 3 data. As noted earlier, these data were collected several years later, and thus are independent from the Round 2 data, but only contain one of the three outcome measures, PVsupport. As shown in Figure 6, injustice is, once again, positively associated with PVsupport while controlling for inequality: b = 0.07, 95% CI = [0.06, 0.09] (see Table S8 in the Supplemental Material for details). The coefficients are similar in size to the ones found using the Round 2 data. Substantively, a min-to-max change in injustice is associated with a 7-percentage-point increase in PVsupport (from 0.24 to 0.31). Furthermore, and corresponding to the analyses of the Round 2 data, inequality is negatively related to PVsupport, but the estimate is imprecise: b = −0.01, 95% CI = [−0.03, 0.01]. The results of the two analyses, based on two independent datasets, are thus consistent.

Figure 6

Support for Political Violence, Round 3

Discussion

Synthesizing social psychological and political science research, we have argued that group-based injustice motivates, whereas group-based economic inequality dissuades, political violence. We found considerable support for the first part of the argument: group-based injustice relates to increased endorsement of political violence, heightened intentions to participate, and self-reported participation in political violence. By contrast, we found only mixed support for the second part of the argument: group-based inequalities only significantly predicted decreased participation in political violence, but not (or only weakly) endorsement of, and intentions to engage in, such violence.

This latter finding may be explained by considering our earlier discussion about political efficacy. The decision to take part in political violence involves the consideration of the risks and benefits associated with violence. Such considerations are arguably less important for the endorsement of violence in which one does not engage. That is, members of economically disadvantaged groups may deem participation in violence as risky and unlikely to succeed, but that may not undermine their attitudinal support for violence. The likely heterogeneity in associations with different political violence outcomes appears as a prospective topic for future research.

Limitations

First, to enhance external validity, and control for higher-level confounding, we searched for existing multinational surveys that contained questions about inequalities, injustices, and violence. This implies, however, that we were limited to secondary data, containing only a restricted set of predictor and outcome measures. Arguably, the one-itemed predictors and outcomes we used are crude and suffer from considerable measurement error. Although we found that these measures revealed patterns consistent with prior psychological work (i.e., the link between group-based injustice and collective action motivations; Agostini & van Zomeren, 2021), it is likely that future analyses with more precise, multi-item scales would reveal stronger associations. Second, our measure of group-based injustice did not explicitly account for emotions. As underscored earlier, emotional injustice (compared to cognitive) seems a stronger predictor of collective action motivations in existing research. Since our measures did not explicitly measure emotions, yet still related to political violence, future research may expect an even stronger association between felt injustices and political violence. Third, our measure of group-based injustice specifically referred to injustices caused by governments. By contrast, the measure of group-based inequality did not refer to any actor or cause of inequality. Although people tend to attribute their economic conditions to state policies, they may also consider other causes and in turn act against them rather than governments—and our outcome measures proxied antigovernment violence. Future studies should therefore consider using explicitly equivalent measures of inequalities and injustices, specifying (or leaving out) the sources or causes of the two in both. Fourth, we relied on observational survey data, which implies that we cannot rule out alternative explanations of our results and the possibility of reverse causality. We have attempted to address both concerns via statistical control, partialing out higher-level confounds, and using outcome measures that are less likely to influence our predictors (i.e., behavioral intentions). This notwithstanding, future studies could explore innovative ways to experimentally manipulate perceptions of inequalities and injustices, without undermining ethical standards, to establish causation. Fifth, we relied on self-reported data. People may underreport or overreport their support for or motivations to engage in violence, for example, due to fears of reprisals from state governments. Such withheld or exaggerated reports would only bias our results if they systematically related to the key predictors of interests. We can only speculate if underreporting or overreporting was random with respect to our variables of interest; however, if people who are treated unfair tend to underreport motivations for violence (e.g., due to fears of repression), then our presented results constitute underestimates of the effects of injustices on violence. To explicitly address this concern, future studies may consider using survey techniques geared toward sensitive questions, for example, item-count techniques (Tourangeau & Yan, 2007). Sixth, we conceptualized and operationalized political violence as a unidimensional construct. Recently, scholars have developed conceptualizations and measures of related constructs, such as collective violence beliefs, that are multidimensional. For example, Abou-Ismail et al. (2023) proposed to distinguish violence against outgroup members from violence against leaders, showing that some predictors (e.g., sectarian narcissism) predict one dimension but not the other (see also the work of Winiewski & Bulska, 2020). It is probable, therefore, that future studies will reveal divergent effects of group-based injustice and group-based inequality on different forms (or dimensions) of political violence. Finally, although we used large multinational data, they were still limited to one context: Africa. We have no evidence or theory to suggest that the patterns revealed in African samples are different in other contexts. Yet, some socioeconomic or political factors unique to Africa may condition how inequalities or injustices relate to violence, suggesting a need for conceptual replications beyond the African context.

Conclusion

A recent report by the United Nations and the World Bank—Pathways for Peace—underlines that “violent conflict is increasingly recognized as one of the big obstacles to reaching the Sustainable Development Goals” and stresses the “conviction … that the attention of the international community needs to be urgently refocused on [conflict] prevention” (World Bank Group and United Nations, 2018). Our research produces knowledge that can inform such conflict-prevention policies. Conflict prevention, as contrasted to crisis management, requires assistance to vulnerable countries long time before the shooting starts. A key assistance area that Pathways for Peace highlights is inequality reduction. This emphasis draws (see the report's Chapter 4, pp. 109–139) on research on horizontal inequalities. We advance this research by pointing to a specific aspect, or a likely consequence of horizontal inequality, that most likely motivates civil violence: group-based injustice. This may hold important clues on how to design inequality-reduction policies to prevent violence. Advice to merely “reduce inequality because inequality is correlated with violence” provides limited guidance for policymakers. We need to know what it is exactly about group-based inequalities that motivate violence—and which groups are particularly likely to act on their disadvantage.

The fact that horizontal inequalities are only weakly related to injustices does not imply that the former do not have other adverse societal effects, some of which may promote political violence. For example, the concentration of economic resources in the hands of dominant groups may incentivize intra-elite fighting over political power, increasing the risk of violent coups and political instability more generally. As such, our research by no means promotes the focus shift from group-based inequalities to injustices. Rather, by clarifying the more specific aspects of inequalities that relate to violence, we hope to nuance conflict-prevention policies that often focus on inequalities as such.

Supplemental Material

sj-docx-1-spp-10.1177_19485506241253359 – Supplemental material for Group-Based Injustice, but Not Group-Based Economic Inequality, Predicts Political Violence Across 18 African Countries

Supplemental material, sj-docx-1-spp-10.1177_19485506241253359 for Group-Based Injustice, but Not Group-Based Economic Inequality, Predicts Political Violence Across 18 African Countries by Casper Sakstrup and Henrikas Bartusevičius in Social Psychological and Personality Science

Footnotes

Handling Editor: Alexa Tullett

Authors’ Contribution

Henrikas Bartusevičius developed the study concept. Both authors developed the theoretical arguments. Casper Sakstrup analyzed the data. Both authors wrote the manuscript and approved the final version for submission. The two authors share equal authorship.

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) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Casper Sakstrup

Supplemental Material

The supplemental material is available in the online version of the article.

Notes

Author Biographies

Casper Sakstrup (PhD, Aarhus University, 2019) is an Associate Professor at the Department of Politics and Society, Aalborg University. His research focuses on armed conflict and autocracies.

Henrikas Bartusevičius (PhD, Aarhus University, 2013) is a Senior Researcher at the Peace Research Institute Oslo. His current research focuses on the psychology of political violence, individual-level analysis of political violence, and human evolution.

References

Abou-Ismail

Gronfeldt

Konur

Cichocka

Phillips

SenGupta

N. K.

(2023). Double trouble: How sectarian and national narcissism relate differently to collective violence beliefs in Lebanon. Aggressive Behavior, 49(6), 669–678. https://doi.org/10.1002/ab.22104

Afrobarometer. (2004). Merged round 2 data, 16 countries, 2003–2004. http://www.afrobarometer.org

Afrobarometer. (2005). Merged round 3 data, 18 countries. http://www.afrobarometer.org

Agostini

van Zomeren

(2021). Toward a comprehensive and potentially cross-cultural model of why people engage in collective action: A quantitative research synthesis of four motivations and structural constraints. Psychological Bulletin, 147(7), 667–700. https://doi.org/10.1037/bul0000256

Ayanian

A. H.

Tausch

(2016). How risk perception shapes collective action intentions in repressive contexts: A study of Egyptian activists during the 2013 post-coup uprising. The British Journal of Social Psychology, 55(4), 700–721. https://doi.org/10.1111/bjso.12164

Ayanian

A. H.

Tausch

Acar

Y. G.

Chayinska

Cheung

W.-Y.

Lukyanova

(2021). Resistance in repressive contexts: A comprehensive test of psychological predictors. Journal of Personality and Social Psychology, 120(4), 912–939. https://doi.org/10.1037/pspi0000285

Bartusevičius

van Leeuwen

(2022). Poor prospects—Not inequality—Motivate political violence. Journal of Conflict Resolution, 66(7–8), 1393–1421. https://doi.org/10.1177/00220027221074647

Bartusevičius

van Leeuwen

Petersen

M. B.

(2020). Dominance-driven autocratic political orientations predict political violence in Western, educated, industrialized, rich, and democratic (WEIRD) and non-WEIRD samples. Psychological Science, 31(12), 1511–1530. https://doi.org/10.1177/0956797620922476

Beard

(2006). Afrobarometer data codebook for round 2 Afrobarometer surveys. The Institute for Democracy in South Africa; Ghana Centre for Democratic Development; Michigan State University.

10.

Becker

J. C.

Tausch

(2015). A dynamic model of engagement in normative and non-normative collective action: Psychological antecedents, consequences, and barriers. European Review of Social Psychology, 26(1), 43–92. https://doi.org/10.1080/10463283.2015.1094265

11.

Bell

Fairbrother

Jones

(2019). Fixed and random effects models: Making an informed choice. Quality & Quantity, 53(2), 1051–1074. https://doi.org/10.1007/s11135-018-0802-x

12.

Bell

Jones

(2015). Explaining fixed effects: Random effects modeling of time- series cross-sectional and panel data. Political Science Research and Methods, 3(1), 133–153. https://doi.org/10.1017/psrm.2014.7

13.

Carter

(2008). Afrobarometer data codebook for round 3 Afrobarometer surveys in 18 African countries. The Institute for Democracy in South Africa; Ghana Centre for Democratic Development; Michigan State University.

14.

Cederman

L.-E.

Vogt

(2017). Dynamics and logics of civil war. Journal of Conflict Resolution, 61(9), 1992–2016. https://doi.org/10.1177/0022002717721385

15.

Cederman

L.-E.

Weidmann

N. B.

Gleditsch

K. S.

(2011). Horizontal inequalities and ethnonationalist civil war: A global comparison. American Political Science Review, 105(3), 478–495. https://doi.org/10.1017/S0003055411000207

16.

Collier

Hoeffler

(2004). Greed and grievance in civil war. Oxford Economic Papers, 56(4), 563–595. https://doi.org/10.1093/oep/gpf064

17.

Coppedge

Alvarez

Maldonado

(2008). Two persistent dimensions of democracy: Contestation and inclusiveness. The Journal of Politics, 70(3), 632–647. https://doi.org/10.1017/S0022381608080663

18.

Davies

Pettersson

Öberg

(2023). Organized violence 1989–2022, and the return of conflict between states. Journal of Peace Research, 60(4), 691–708. https://doi.org/10.1177/00223433231185169

19.

Dyrstad

Hillesund

(2020). Explaining support for political violence: Grievance and perceived opportunity. Journal of Conflict Resolution, 64(9), 1724–1753. https://doi.org/10.1177/0022002720909886

20.

Fearon

J. D.

Laitin

D. D.

(2003). Ethnicity, insurgency, and civil war. American Political Science Review, 97(1), 75–90. https://doi.org/10.1017/S0003055403000534

21.

Feather

N. T.

(2015). Analyzing relative deprivation in relation to deservingness, entitlement and resentment. Social Justice Research, 28(1), 7–26. https://doi.org/10.1007/s11211-015-0235-9

22.

Ghobarah

H. A.

Huth

Russett

(2003). Civil wars kill and maim people—Long after the shooting stops. American Political Science Review, 97(2), 189–202. https://doi.org/10.1017/S0003055403000613

23.

Gleditsch

N. P.

Wallensteen

Eriksson

Sollenberg

Strand

(2002). Armed conflict 1946–2001: A new dataset. Journal of Peace Research, 39(5), 615–637. https://doi.org/10.1177/0022343302039005007

24.

Gurr

T. R.

(2000). Peoples versus states: Minorities at risk in the new century. United States Institute of Peace Press.

25.

Harff

Gurr

T. R.

(2018). Ethnic conflict in world politics. Routledge.

26.

McNeish

D. M.

Kelley

(2019). Fixed effects models versus mixed effects models for clustered data: Reviewing the approaches, disentangling the differences, and making recommendations. Psychological Methods, 24(1), 20–35. https://doi.org/10.1037/met0000182

27.

McNeish

D. M.

Stapleton

L. M.

(2016). The effect of small sample size on two-level model estimates: A review and illustration. Educational Psychology Review, 28(2), 295–314. https://doi.org/10.1007/s10648-014-9287-x

28.

Miodownik

Nir

(2016). Receptivity to violence in ethnically divided societies: A micro-level mechanism of perceived horizontal inequalities. Studies in Conflict & Terrorism, 39(1), 22–45. https://doi.org/10.1080/1057610X.2015.1084162

29.

Osborne

SenGupta

N. K.

Sibley

C. G.

(2019). System justification theory at 25: Evaluating a paradigm shift in psychology and looking towards the future. The British Journal of Social Psychology, 58(2), 340–361. https://doi.org/10.1111/bjso.12302

30.

Østby

(2008). Polarization, horizontal inequalities and violent civil conflict. Journal of Peace Research, 45(2), 143–162. https://doi.org/10.1177/0022343307087169

31.

Østby

(2013). Inequality and political violence: A review of the literature. International Area Studies Review, 16(2), 206–231. https://doi.org/10.1177/2233865913490937

32.

Pettigrew

T. F.

(2016). In pursuit of three theories: Authoritarianism, relative deprivation, and intergroup contact. Annual Review of Psychology, 67(1), 1–21. https://doi.org/10.1146/annurev-psych-122414-033327

33.

Rustad

S. A.

(2016). Socioeconomic inequalities and attitudes toward violence: A test with new survey data in the Niger Delta. International Interactions, 42(1), 106–139. https://doi.org/10.1080/03050629.2015.1048856

34.

Smith

H. J.

Pettigrew

T. F.

Pippin

G. M.

Bialosiewicz

(2012). Relative deprivation: A theoretical and meta-analytic review. Personality and Social Psychology Review, 16(3), 203–232. https://doi.org/10.1177/1088868311430825

35.

Starmans

Sheskin

Bloom

(2017). Why people prefer unequal societies. Nature Human Behaviour, 1(4), 1–7. https://doi.org/10.1038/s41562-017-0082

36.

Tourangeau

Yan

(2007). Sensitive questions in surveys. Psychological Bulletin, 133(5), 859–883. https://doi.org/10.1037/0033-2909.133.5.859

37.

Stewart

(2008). Horizontal inequalities and conflict: Understanding group violence in multiethnic societies. Springer.

38.

van Zomeren

Postmes

Spears

(2008). Toward an integrative social identity model of collective action: A quantitative research synthesis of three sociopsychological perspectives. Psychological Bulletin, 134(4), 504–535. https://doi.org/10.1037/0033-2909.134.4.504

39.

Winiewski

Bulska

(2020). Beat, ignore, force to conform: Development and initial validation of a multidimensional scale of acceptance of collective violence. International Journal of Conflict and Violence, 14, 1–17. https://doi.org/10.4119/ijcv-3775

40.

World Bank Group and United Nations. (2018). Pathways for peace: Inclusive approaches to preventing violent conflict. World Bank.

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.

0.00 MB

0.38 MB