RHETORIC AND REVENGE: BRINGING PUBLIC OPINION BACK IN

To develop a theory of cross-national variation in mass attitudes towards war, I begin with what we know about how those attitudes are formed. A significant body of research on public opinion formation has identified core values as a fundamental source of political attitudes (Feldman Reference Feldman1988; Hurwitz and Peffley Reference Hurwitz and Peffley1987; Jacoby Reference Jacoby2006). Core values refer to an individual's deeply held beliefs about what is good, right, and important in life (Rokeach Reference Rokeach1973), and in the realm of politics, these beliefs serve as shortcuts for forming opinions about issues that are often forbiddingly complex and highly uncertain (Goren Reference Goren2000; Sniderman et al. Reference Sniderman, Brody and Tetlock1993). This is particularly true in the realm of foreign affairs, which is often remote from everyday life and shrouded in official secrecy. Lack of experience and information make it difficult for ordinary citizens to form their opinions about a policy by weighing its costs and benefits, leading them to rely instead on assessments of whether that policy accords with their core values (Hurwitz & Peffley Reference Hurwitz and Peffley1987; Meernik and Ault Reference Meernik and Ault2001).

A growing body of evidence points to the importance of revenge as a determinant of individuals’ support for the use of military force (Liberman Reference Liberman2006, Reference Liberman2013; Liberman and Skitka Reference Liberman and Skitka2008; Pagano and Huo Reference Pagano and Huo2007; Stein Reference Stein2011).Footnote ¹ By revenge, I am referring to the belief that wrongdoers deserve to be repaid for their crimes. In this view, taking “an eye for an eye” is a moral good in and of itself, regardless of its impact on future outcomes. The relationship between endorsement of revenge and support for war at the individual level relates to differences in support for war across countries because the prevalence of revenge beliefs exhibits strong cross-cultural variation.

Numerous studies have documented cultural differences in beliefs about revenge and linked this variation to the frequency of lethal violence (Boehm Reference Boehm1984; Chagnon Reference Chagnon1988; Elster Reference Elster1990; Hinton Reference Hinton1998; Nisbett and Cohen Reference Nisbett and Cohen1996). The only multicountry survey to ask directly about revenge also shows substantial cross-national differences. In Gallup International's Voice of the People Millennium (VPM) Survey, which was conducted in the year 2000 in 59 countries,Footnote ² respondents were asked what they thought the main aim of imprisonment should be and I classified as vengeful those who said “to make those who have done wrong pay for it.”Footnote ³ Figure 1 reports the percentage of people in each country who chose that response option, ranging from a low of 13% in Denmark to a high of 54% in South Korea, with the United States falling at the average of 30%.Footnote ⁴ These data demonstrate that even in a highly globalized world, the prevalence of revenge as a core belief varies substantially across national populations.

FIGURE 1.

Variation in Endorsement of Revenge in the VPM Survey

Note: These data are from Gallup International's Voice of the People Millennium Survey (2000). Endorsement of revenge is measured by the percent of respondents who said the main aim of imprisonment was “to make those who have done wrong pay for it.”

This variation is consequential for the behavior of democratic leaders because in countries where vengefulness is widespread, leaders can build popular support for war more easily, making conflict initiation a less risky prospect. Although early research cast doubt on the structure and stability of mass attitudes towards foreign policy and concluded that leaders had little incentive to attend to public opinion on international issues, there is now substantial evidence demonstrating that this pessimism was unwarranted (Holsti Reference Holsti1996). Numerous studies have demonstrated that individuals have stable and coherent opinions about foreign policy (Hurwitz and Peffley Reference Hurwitz and Peffley1987; Wittkopf Reference Wittkopf1990), that they consider their leaders’ foreign policy records when making their vote choice (Aldrich, Sullivan and Borgida Reference Aldrich, Sullivan and Borgida1989; Krosnick and Kinder Reference Krosnick and Kinder1990), and that in turn, leaders take public opinion into account when making decisions about the use of force abroad (Baum Reference Baum2004; Sobel Reference Sobel2001).

However, the public's foreign policy views are not set in stone and elected leaders do not behave as though they are (Canes-Wrone Reference Canes-Wrone2010; Rottinghaus Reference Rottinghaus2009). Instead, recognizing the importance of securing public support for war, leaders devote serious effort to using the advantages of the bully pulpit to sway public opinion in their favor (Jacobs and Shapiro Reference Jacobs and Shapiro2000; Rosenblatt Reference Rosenblatt1998). The more effective these efforts are, the less leaders will fear the consequences of an unpopular war and the more likely they will be to decide that the benefits of initiating a conflict outweigh the costs. A vengeful populace enhances a leader's ability to generate public support for war because it allows him to take advantage of a versatile and powerful rhetorical strategy: framing the use of military force as a punishment that evildoers deserve to suffer.

Framing is one of the most powerful tools that leaders use to influence public opinion, and its effects have been consistently demonstrated for both domestic and foreign policy issues (Chong and Druckman Reference Chong and Druckman2007; Western Reference Western2005). Broadly speaking, framing refers to the way in which an issue or event is described in communication. Framing influences individuals’ attitudes by changing the concerns that they have in mind when forming their opinion, and frames that resonate with deeply held values are particularly effective (Chong Reference Chong2000; Druckman Reference Druckman2001). Framing war as a punishment involves portraying the use of military force as repayment for some act of wrongdoing on the part of the adversary. Because vengeance is conditional—the crime must come first in order for the punishment to be deserved—leaders must provide their citizens with a narrative that has all the necessary raw materials: a victim, a perpetrator, and a punisher. These pieces must be in place in order to trigger the logic of revenge.Footnote ⁵ This argument is supported by studies that find individuals who endorse revenge are more likely to support the use of military force only when it is framed as a punishment for a clear act of wrongdoing. In the absence of that framing, their attitudes are not distinguishable from those of nonvengeful individuals (Liberman Reference Liberman2006; Stein Reference Stein2011).

Thus, it is not the case that a highly vengeful public will seize upon the slightest provocation to demand war from a reluctant leader. Rather, broad public endorsement of revenge is a latent source of support for war that leaders can activate by deploying strategically crafted rhetoric (Jacobs and Shapiro Reference Jacobs and Shapiro2000). I also assume that leaders do not initiate conflicts solely for reasons of personal or national revenge.Footnote ⁶ The rationales that leaders offer to their citizens are not necessarily the same as the reasons they decide to go to war, and the theory presented here concerns only the former. Other literatures have dealt with the issues that give rise to conflicts of interest between states (e.g., Levy and Thompson Reference Levy and Thompson2009) and the reasons why they may fail to find a peaceful bargain that both sides prefer to war (e.g., Fearon Reference Fearon1995), and I do not challenge their claims about these proximate causes of war. However, when conflicts of interest do arise between states, democratic leaders who can increase public support for war by tapping into their citizens’ desire for revenge will be less constrained from using force to resolve those conflicts.Footnote ⁷

The claim that democratic leaders’ decisions about conflict initiation are influenced by the vengefulness of their populations does not require that leaders have access to perfect information about their citizens’ beliefs about revenge, that any and all potential conflicts can be shaped into a victim-perpetrator-punisher narrative, or that leaders always use this frame when it would be advantageous. The use of force can be framed in a variety of ways, and the framing that leaders choose may also be influenced by other factors such as their own beliefs about the nature of international threats (Saunders Reference Saunders2009) or the prevailing norms surrounding the use of force (Finnemore Reference Finnemore2003). Certainly, given the current normative climate, leaders have an incentive not to appear aggressive even when making the first move. A punishment frame is one way to achieve this aim, but this does not mean that most or all conflicts are framed in this way (Western Reference Western2005).Footnote ⁸ A humanitarian intervention frame, for example, can serve this purpose as well (Wheeler Reference Wheeler2000). Furthermore, for leaders in very low vengefulness countries, there may also be costs to employing a punishment frame, since their populations may react negatively to the specter of revenge in order to affirm their own alternative values.

This argument is also not meant to suggest that framing the use of force as a punishment is the only strategy that leaders use to sway public opinion. Punishment framing is best thought of as one item in the toolbox of strategies that can be employed alone or in combination (Edwards Reference Finnemore2009). My argument is that leaders with more vengeful publics will utilize a punishment framing strategy more often and with greater effect than leaders with less vengeful publics, and that on average, this advantage will result in a greater propensity to initiate conflict. Thus, I hypothesize that democracies with highly vengeful publics will be more likely to initiate the use of military force than democracies with less vengeful publics.

This is a story about conflict initiation rather than reciprocation or escalation because the advantage enjoyed by leaders with highly vengeful populations is largely eliminated once the adversary makes the first move. Consider a situation in which state A has initiated a dispute against state B. The leader of state B can now make the case for going to war on the basis of self-defense. Given the widespread legitimacy of self-defense as a justification for both interpersonal and interstate violence (Alexandrov Reference Alexandrov1996; Ramirez Reference Ramirez2007), this is a potent argument that will often generate enough public support to overcome leader B's reticence about waging an unpopular war, regardless of the vengefulness of his constituents. However, if state A has not yet made the first move, then the self-defense argument becomes much more difficult to sustain and leader B's ability to appeal to the public's desire for revenge becomes much more important in generating a sufficient level of public support. If leader B faces a highly vengeful population, he can generate support for initiation against state A by framing the use of force as a punishment for acts of wrongdoing that state A has committed against state B in the past, against a third state, or against its own people. Having this tool in his tool kit means that leader B has greater liberty to take military action against state A compared to a leader who must wait for a plausible case for self-defense to emerge in order to surmount public opposition.Footnote ⁹

I do not expect the same logic to apply to autocratic leaders. While recent research has argued that autocratic leaders do in fact face a need to maintain support among their “selectorate,” or the group of people who have a say in leadership selection, that selectorate tends to be both smaller and less diverse in terms of interests than the selectorate in democracies (Bueno de Mesquita et al. Reference Bueno de Mesquita, Smith, Siverson and Morrow2003; Weeks Reference Weeks2008). Weeks (Reference Weeks2008) argues that for most autocratic leaders, the relevant selectorate is composed of regime insiders, individuals who hold power within the institutions of government, rather than the mass public, as is the case for democracies.

There are several reasons to believe that in autocracies, leaders’ willingness to initiate the use of military force is not conditional on the prevalence of vengeful attitudes in their selectorate. First, with a small selectorate, autocratic leaders can afford to maintain support by doling out rewards and punishments (Bueno de Mesquita et al. Reference Bueno de Mesquita, Smith, Siverson and Morrow2003). An autocratic leader facing a less vengeful selectorate has the ability to build sufficient support for war by paying off those who back him and eliminating those who do not, and he is therefore not at a disadvantage compared to an autocratic leader facing a more vengeful selectorate. In contrast, the reward and punishment strategy is too expensive for democratic leaders with large selectorates, and as a result, they must rely on cheaper but less certain methods of ensuring support, such as rhetorical persuasion.Footnote ¹⁰

Second, given their status as political elites, regime insiders will rely less on their core values when evaluating the wisdom of going to war than the average citizen. In general, these individuals should have access to higher quality information about foreign policy than is available to the general public, combined with the experience and knowledge required to process that information and determine which policy is in their best interest. Thus, even without the payoffs and punishments provided by autocratic leaders, vengefulness should play no more than a minor role in shaping the opinions of well-informed regime insiders.

Dependent Variable: MID Initiation

My dependent variable is the initiation of a militarized dispute. In the MIDs data, the initiator is defined as the state that made the first move by threatening, displaying, or using force. Initiation is a dichotomous variable equal to one if state A initiated a MID with state B in a given year. Between 1945 and 2001, there were 1,246 instances of initiation. In order to exclude relatively minor incidents that have little or no chance of ever coming to the attention of the mass public, I look only at high hostility level MIDs, which are those that involve an actual use of force or full-scale war. This leaves 710 instances of initiation in the data.

To identify democratic initiators, I use the combined Polity score from the Polity IV dataset, which ranges from zero (most strongly autocratic) to 20 (most strongly democratic). I code the initiating state as a democracy in a given year if its combined Polity score is greater than 16. In the analyses that follow, I subset the data on this variable in order to examine democratic and autocratic initiators separately.

Measuring Vengefulness

My main independent variable is the vengefulness of the initiating state's population. A rigorous test of my hypothesis that controls for other factors known to affect the likelihood of initiation requires a measure of vengefulness that is widely available both across countries and over time. Ideally, I would use a direct measure of attitudes towards revenge based on public opinion surveys, but there is not sufficient data to take this approach. Instead, I use the legal status of capital punishment as an indicator for whether revenge is broadly endorsed by a country's population. This measure can be coded consistently for all countries in the international system without any missing data or issues of cross-national comparability. I collected data on the legal status of capital punishment worldwide from Amnesty International. Based on this information, I created the dichotomous variable Initiator death penalty, which is equal to 1 in a given year if the initiating state retains the death penalty for ordinary crimes. In the 1945–2001 period, a total of 66 countries have been democratic and retentionist in at least one year, and 41% had a duration of at least 10 consecutive years. Overall, democracies with the death penalty make up 51.2% of all dyad-years with a democratic initiator in my data.Footnote ¹¹ Figure 2 plots the number of retentionist democracies in each year from 1945 to 2001, showing that they have been a constant presence in the international system since 1945, although their number has fluctuated between 10 (in 1945) and 30 (in 1993 and 1994).Footnote ¹²

FIGURE 2.

Number of Retentionist Democracies in the International System, 1945–2001

Note: A country is coded as a retentionist democracy in a given year if it has a Polity score greater than 16 and retains the death penalty for ordinary crimes.

Using law as a proxy for public opinion in democracies is a viable strategy for a number of reasons. First, numerous studies have linked cross-cultural variation in beliefs and values to national-level policy differences among democracies across a variety of issue areas (e.g., Cohen Reference Cohen1996; Inglehart Reference Inglehart1990; Kasser Reference Kasser2011), suggesting that policy often reveals information about values. Second, the global trend in the 20th century has been towards the abolition of capital punishment, with an especially strong wave following World War II (Simmons Reference Simmons2009). As a result, governments that retain capital punishment in the post-1945 era do so as a deliberate choice and against the prevailing tide of international opinion. In democracies, this decision rests on leaders’ assessments of how their political fortunes will be affected by a move to abolish the death penalty. They are unlikely to take the risk if support for capital punishment is high and they expect a strong public backlash (Newburn and Jones Reference Newburn and Jones2005). Finally, research on individual attitudes towards the death penalty has converged on the conclusion that, at the individual level, attitudes towards capital punishment are symbolic rather than instrumental (Ellsworth and Gross Reference Ellsworth and Gross1994; Tyler and Weber Reference Tyler and Weber1982) and that support is driven primarily by a desire for revenge, rather than by utilitarian concerns (Bohm Reference Bohm1992; Carlsmith et al. Reference Carlsmith, Darley and Robinson2002; Cotton Reference Cotton2000; Ellsworth and Ross Reference Ellsworth and Ross1983; McKee and Feather Reference McKee and Feather2008).

All this suggests that retentionist democracies are places where revenge is likely to enjoy more widespread endorsement.Footnote ¹³ The available cross-national public opinion data on revenge support this contention. Using the data from the 2000 VPM survey, I coded the variable Percent vengeful, which is the percentage of people in each democracy in the survey (N = 43) who endorsed revenge as the main justification for imprisonment. I then regressed the present day legal status of capital punishment (equal to 1 if the country retains the death penalty in 2014) on Percent vengeful and find that it is positive and significant predictor of retention. Figure 3 illustrates the substantive effect of Percent vengeful by graphing the predicted probability that a democracy will retain the death penalty as Percent vengeful varies over its range. Figure 3 shows that the probability that a democracy will retain the death penalty rises from 4% to 54% as Percent vengeful increases, a difference that is statistically significant (Δ = 50 [4.1, 87.9]). From the same question on the VPM survey, I also coded variables for the percentage of people who chose deterrence as the main aim of imprisonment and the percentage who chose public safety. Both deterrence and public safety are commonly articulated justifications for capital punishment, but neither variable has a significant relationship with the likelihood that a democracy will retain the death penalty.Footnote ¹⁴

FIGURE 3.

The Effect of Endorsement of Revenge on Death Penalty Retention in Democracies

Note: Predicted probabilities calculated using CLARIFY.

I cannot use the data from the VPM to test my hypothesis because it only covers 43 democracies in a single year, but this analysis should increase our confidence that the legal status of the death penalty is a valid indicator of the prevalence of revenge in national populations.Footnote ¹⁵ It shows that retention of the death penalty is positively correlated with popular endorsement of revenge and that it is not correlated with cross-national variation in beliefs about the efficacy of punishment, thereby ameliorating the concern that the legal status of capital punishment might be proxying for the latter rather than the former. An additional concern has to do with other national attributes that may be correlated with both retention and the propensity to initiate conflict. I address this issue below.

Control Variables

I include two sets of control variables in my analysis. The first is a standard set of MIDs predictors based on existing studies of conflict initiation. I include these variables both to account for the influence of factors like proximity, shared interests, and military capabilities on the baseline risk of conflict, and to compare the effect of Initiator death penalty to the impact of these well-established predictors of initiation.

State power affects both the opportunity for conflict and the expected value of going to war, and I include a set of variables to capture both the absolute and relative power of the states in each dyad. These variables include Initiator capabilities score (measures the initiator's material capabilities compared to all other states in the system), Target capabilities score Footnote ¹⁶ (measures the target's material capabilities compared to all other states in the system), Balance of capabilities Footnote ¹⁷ (measures how evenly matched the two states are in terms of their material capabilities), Initiator share of capabilities Footnote ¹⁸ (measures the strength of the initiator relative to the target), Initiator major power (an indicator of the initiator's major power status), and Target major power Footnote ¹⁹ (an indicator of the target's major power status).

To capture the effect of geographical proximity on opportunity for conflict (Gleditsch Reference Gleditsch1995), I include Land contiguity, which is a dichotomous variable equal to 1 if the states in a dyad share a land border. By contrast, states that are allied or that have similar strategic interests should have fewer opportunities for conflict (Gowa Reference Gowa2011). The former is measured by Defensive pact (a dichotomous variable equal to 1 if the states in the dyad have a formal agreement to aid one another militarily in the event of an attack), while the latter is measured by Alliance portfolio similarity (a measure of the extent to which states in a dyad are allied with similar countries).Footnote ²⁰ In addition to these variables, I also include the regime type of the target state to capture any effect of the democratic peace. Target democracy is a dichotomous variable equal to 1 if target state's combined Polity score is greater than 16.

The second set of control variables addresses the problem of omitted variable bias. Having the death penalty is not itself a cause of conflict, rather it is a proxy for popular endorsement of revenge. Other factors will also influence retention, and thus we must carefully consider how retentionist democracies differ from abolitionist democracies and whether those differences are plausibly related to the likelihood of conflict initiation. There is a small literature on the determinants of cross-national death penalty retention, and these studies suggests a number of predictors of retention that are also cited as causes of conflict or that are correlated with conflict generating processes.Footnote ²¹