After major ruptures of international norms, such as foreign invasions, breaches of the laws of war, or atrocities, states may emerge as pariahs, reviled by their victims and by third parties. These negative international perceptions may be not only embarrassing but also costly, by influencing other states’ willingness to engage or cooperate. As Robert Jervis famously noted, “A desired image ... can often be of greater use than a significant increment of military or economic power.”Footnote 1 How, then, can states rehabilitate their tarnished images after trampling on international norms?
We explore one increasingly common tool: apologies. International apologies have surged in recent years, prompting some to call today the “age of apology.”Footnote 2 For example, states apologized for violating foreign citizens’ human rights 137 times between 2000 and 2022, compared to only 69 times between 1947 and 1999.Footnote 3 Prominent examples include Germany apologizing for the Holocaust and World War II, Japan for World War II and its military occupations, Serbia for atrocities during the Bosnian War, and the US for abusing Iraqi prisoners at Abu Ghraib.
Research has typically studied whether apologies promote reconciliation between offender and victim.Footnote 4 However, apologizers may have another goal: repairing their broader international image.Footnote 5 This possibility raises intriguing questions. When countries apologize, do third parties adjust their opinions positively, negatively, or not at all? Moreover, how do reactions in both the apologizing and recipient countries affect third-party views?
We focus on the effects of apologies on the international images of states. Major human rights transgressions implicate not only the government in charge at the time but also the state apparatus and society.Footnote 6 Official state apologies constitute efforts to accept state political responsibility for past injustice. Indeed, they are nearly always given by successor governments rather than the leaders who were directly responsible.Footnote 7 We thus study whether official state apologies can allay concerns about the intentions and values of the country, rather than of individual leaders or governments.
We argue that apologies, even when they consist of mere words, can provide reassurance against future transgressions and signal appropriate moral values. We thus hypothesize that they improve foreign perceptions and encourage future cooperation. We contrast these ideas with a point sometimes raised by domestic critics: that apologizing could backfire by making the sender look weak. We further theorize that negative reactions by two important actors—the victim and the apologizer’s domestic audience—could undermine apologies’ image-rehabilitation effects. However, we expect that even apologies that fall short on these dimensions are better than silence.
Our empirical tests leverage the advantages of different types of evidence, pairing survey experiments with an important historical case: Germany’s 1951 Holocaust apology.Footnote 8 The case suggests that apologies do indeed produce image benefits, while also highlighting empirical challenges, including confounding, selection effects, and measurement. Apologies often coincide with events like democratic transitions, making it difficult to isolate them from other image-enhancing developments. Moreover, countries may apologize primarily when they expect substantial benefits, leading to exaggerated estimates of image repair while obscuring whether apologies convey weakness. Finally, consistent longitudinal measures of international reputations are often elusive.
Our experiments, fielded with nearly 8,000 US-based subjects, help us sidestep these challenges. Randomly varying the existence and context of apologies helps us isolate apology effects from strategic and contextual factors, examine relevant counterfactuals, and gather consistent measures of key outcomes. Our experiments take advantage of a plausible and tangible hypothetical, varying whether Russia apologizes for invading Ukraine and its subsequent atrocities. Some respondents further learned whether Ukraine accepts or rejects the apology and whether the apology produces backlash inside Russia. We also varied Russia’s regime type to gauge whether apology effects differ for democracies versus nondemocracies. We then measured four outcomes, including not only the kind of broad favorability elicited in polling and studied in the public diplomacy literature,Footnote 9 but also material benefits such as support for interstate cooperation and willingness to buy Russian products. We further measured perceptions of weakness to capture potential reputational downsides.
We find that apologies can improve a country’s international image and even produce tangible benefits, regardless of the apologizer’s regime type. In both studies, an apology boosted US public favorability toward Russia, support for US–Russian cooperation, and willingness to purchase Russian products. We found no evidence that apologizing made Russia seem weak in US eyes. The reassurance and good-moral-values mechanisms received support, but the idea that apologies diminish status did not. Finally, rejection by the victim (Ukraine) and backlash within Russia mildly diminished the reputational benefits—but even then, apologies produced image repair.
We focus on US-based respondents because the US is a political, economic, and military juggernaut that can shape other countries’ fortunes, both directly and indirectly: it can offer valuable cooperation opportunities, and its support can enhance countries’ broader international status. We examine mass public opinion because the views of ordinary US citizens have important consequences for US foreign policy. A growing body of research shows that public opinion affects leaders’ foreign policy choices.Footnote 10 Furthermore, citizens can affect the economic outcomes of foreign countries directly through their purchasing behavior.Footnote 11
Studying ordinary citizens can also provide clues about whether apologies shape elite sentiments. Joshua Kertzer shows that elites and masses respond to experiments in remarkably consistent ways, and that the inconsistencies reflect mostly demographic differences.Footnote 12 We examine subsamples of “elite-like” respondents and find little evidence that key conclusions would differ had we fielded these experiments with US decision makers.
Our studies may also offer insights about how citizens in countries other than the US react. We explore whether potential moderators—such as commitment to democratic norms, authoritarian or right-wing values, or preexisting views of Russia—influence the effects of apologies.Footnote 13 While only suggestive, our findings increase confidence that a Russian apology might receive a warm public reception far beyond the US, and in turn affect how foreign governments react.
Overall, our findings suggest that apologies may be a powerful tool of image repair while not undermining a country’s reputation for strength, even when victims reject the apology or citizens in the sending state criticize the gesture. Thus our findings could help explain why states often make apologies that fall flat at home or in the target country. While such apologies might fail at reconciliation, they may still contribute to foreign-image repair, potentially encouraging cooperation.Footnote 14 Our study thus joins the burgeoning scholarship on “public diplomacy”—attempts to communicate with foreign publicsFootnote 15—as well as the literature on norms-based behavior in international politics,Footnote 16 by showing that international aspirations can sometimes encourage moral acts and by illustrating the potential international benefits of doing “what is right.”
The Image Effects of International Apologies
History provides many examples of governments apologizing with the hope of rehabilitating their country’s international reputation. After World War II, for example, West German leaders believed public atonement was crucial for restoring Germany’s global image.Footnote 17 Similarly, in the 1990s, Japanese leaders felt it was “pragmatic” to apologize for “‘the past’ because doing so would affirm the nation’s current and future ties with its Asian neighbors.”Footnote 18 In the US, the Bush administration followed a similar playbook, apologizing for the Abu Ghraib prison abuse in part to soften the scandal’s damage to the US’s reputation.Footnote 19 Serbia apologized for crimes during the Bosnian War, and only two and a half months later enjoyed progress on its path to EU membership as well as valuable trade and funding opportunities.Footnote 20
These examples suggest that policymakers believe that international apologies can improve their international image, and that apologies do, at least sometimes, have such effects. However, we lack a good understanding of when and why apologies influence third parties. Here we theorize the potential positive and negative effects of apologies on a state’s image. We then examine how contextual factors—the reaction of the victim and of audiences in the sending state—could diminish such benefits.
Positive Image Effects
We define international apologies as official statements by a country’s representatives (the sender) that acknowledge that the victim (the target) experienced harm, take responsibility, acknowledge moral wrongdoing, and express remorse.Footnote 21 Such apologies are typically delivered publicly at high-profile events and garner significant media attention. For example, West German Chancellor Brandt’s kneeling at the Warsaw Ghetto memorial in 1971 was reported on the front page of major newspapers in the US, Britain, Italy, and France.
Apologies, even when mere words, can send costly signals of future behavior and moral valuesFootnote 22 due to both “sunk costs” and “tying hands” logics.Footnote 23 There are sunk costs because apologies produce a loss of face and undermine individuals’ positive feelings about their ingroup.Footnote 24 This is particularly acute for the leaders responsible for the original violation, but sunk costs also apply to successor governments and ordinary citizens. Likewise, apologizing risks backlash from citizens who resent impugning their country’s honor.Footnote 25 Once an apology is issued, the loss of face and backlash cannot be taken back. These sunk costs help make the apology’s sentiments credible.Footnote 26
Apologies can also tie the apologizer’s hands. Jennifer Lind argues that by delegitimizing past aggression, apologies can limit leaders’ ability to rally citizens behind future aggression.Footnote 27 Moreover, the apologizer will lose international credibility if they later repeat similar behavior. Apologies that are not adhered to, like peace treaties that are violated, damage reputations and cast doubt on future commitments.Footnote 28
These costs help send two signals. The first is reassurance against similar future transgressions. Lind’s pathbreaking work argues that apologies help reassure the victim.Footnote 29 We add that this reassurance goes beyond a promise not to re-offend a particular victim. Interstate apologies typically condemn the behavior, not its targeting toward a specific country. The apology thus signals forbearance with respect to any future target. Third parties should value such reassurance, whether they fear attack themselves or simply value peace more generally.
Second, apologies signal appropriate moral values and readiness to be a respectable member of the international community. This signal appeals especially to states aligned with the values underlying the apology. Given our focus on apologies for mass atrocities and other universally condemned violations, the apologies we study should thus have broad appeal, and could increase not only favorable views of the offender but also support for cooperation with it and willingness to engage in private economic transactions with it.
These considerations yield the following hypotheses:
Effect of apology: Relative to no apology, apologies increase
-
a. favorable views of the offender (favorability);
-
b. support for state-to-state cooperation with the offender (international co-operation); and
-
c. willingness to transact privately with the offender (private transactions).
Negative Image Effects
Despite these potential image benefits, decision makers often fret about a potential reputational downside: that apologies convey weakness. For example, in interviews with Israeli officials, Aldar, Kampf, and Heimann repeatedly heard that humiliating diplomatic apologies must be avoided: “If you humiliate yourself in the Middle East ‘you admit weakness, not just that you are weak, but also inferior. This is unforgiveable [behavior].’”Footnote 30 Likewise, US Republicans vigorously criticized President Obama for his so-called Apology Tour, accusing him of projecting US weakness.
Why might apologies convey weakness? Critics have not detailed their causal logic, but one feature that makes apologies costly—loss of face—might harm another facet of reputation, perceptions of the country’s strength. Apologizers accommodate and defer to victims. Senders bow to victims figuratively and sometimes even literally, lowering their own status.Footnote 31 This may suggest that the sender is weak and was forced to give in and apologize. In sum, perhaps apologies signal weakness because, correctly or not, they indicate low status and/or material capabilities.
Effect of apology (continued): Relative to no apology, an apology increases
-
d. perceptions of weakness (weakness).
How Responses by Key Actors Influence the Effects of Apologies
Foreign audiences often learn additional information that could shape their responses to apologies. We focus on reactions in the two states directly involved: victim (target) and offender (sender).
The Victim’s Response
Some apologies, like Serbia’s 2007 apology to Croatia, elicit the victim’s appreciation.Footnote 32 Other times victims reject apologies. For example, China and South Korea spurned Japan’s 2015 apology,Footnote 33 and China rebuffed American apologies for accidentally bombing the Chinese embassy during the Kosovo War.Footnote 34
We theorize that the victim’s response provides cues about the apology’s quality. Foreign observers put stock in the victim’s ability to evaluate the situation. Acceptance confirms that the apology is appropriate and credible, strengthening signals of reassurance and good moral values. Rejection suggests that the target does not think the apology is appropriate or sincere, attenuating these signals. Thus, a rejected apology should produce less image rehabilitation than an accepted one.
Still, we expect a rejected apology to be more beneficial than no apology. Even without the victim’s acceptance, “apologies perform an important function by putting facts on the public record and stating a position with regard to the relevant norms.”Footnote 35 Observers might also expect victims to have unrealistically high standards or even strategic motives, such as to create pressure for material reparations. International audiences should thus appreciate a state that attempts contrition, even if the victim isn’t satisfied.
The victim’s response could also influence perceptions of the sender’s weakness. Rejected apologies may involve greater face loss, compared not only to no apology, but also to accepted ones. The victim is, in essence, demanding that the sender bow down even further. A snubbed sender may seem less deserving of respect, and possibly lacking in material capabilities, generating foreign perceptions of weakness.
Rejected apology vs. accepted apology: Compared to the target accepting the apology, rejection>
-
a. decreases favorability,
-
b. decreases support for state-to-state cooperation,
-
c. decreases willingness to transact privately, and
-
d. increases perceptions of weakness.
Rejected apology vs. no apology: Compared to no apology, a rejected apology increases
-
a. favorability,
-
b. support for state-to-state cooperation,
-
c. willingness to transact privately, and
-
d. perceptions of weakness.
The Sender’s Domestic Response
Apologies can trigger strong backlash inside the sending state. Lind highlights Japan, where almost every government apology provokes fierce criticism by conservatives.Footnote 36 Similarly, Croatian elites decried their president’s Bosnian War apology,Footnote 37 and Russia’s Communist Party criticized President Putin for apologizing to Poland for the Katyn Massacre.Footnote 38
One might expect backlash to reinforce the signal that the country is committed to peace. After all, backlash is politically costly, and only a government serious about turning over a new leaf would willingly bear these sunk costs.Footnote 39 Mild-to-moderate domestic backlash may indeed have a reinforcing effect. However, we predict that strong domestic backlash like that in Japan, Croatia, or Russia, should weaken the apology. As Lind argues, strong backlash undermines the apology’s tying-hands signal: it indicates that the public has not adopted new beliefs, so mobilization for future aggression is possible.Footnote 40 Extensive backlash could prompt worries that a hostile government could soon take over while raising doubts about the current government: perhaps it shares its citizens’ views and just apologized instrumentally. Thus, we predict that substantial domestic backlash undermines signals of reassurance and moral values.Footnote 41
At the same time, observers should prefer apologies accompanied by domestic criticism, even strong criticism, over no apology. Silence shows that the sender is not ready for change, but a domestically contested apology at least raises the possibility that it favors peace and cooperation.Footnote 42
Strong sender backlash vs. no backlash: Compared to apologies without strong sender backlash, backlash decreases
-
a. favorable views,
-
b. support for state-to-state cooperation, and
-
c. willingness to transact privately.
Apology with strong sender backlash vs. no apology: Compared to no apology, apologies with strong backlash increase
-
a. favorable views,
-
b. support for state-to-state cooperation, and
-
c. willingness to transact privately.
We further preregistered hypotheses about how apologies affect mechanisms (reassurance, appropriate moral values, perceptions of status loss, and diminished material capabilities);Footnote 43 how the victim’s reaction and domestic backlash affect these mechanisms; the effects of regime type and democratization; and the interaction between domestic backlash and victim rejection. We summarize our findings and give details in the online appendix.
Research Design
We describe our two survey experiments before turning to an illustrative case. We administered Study 1 to 4,955 US-based respondents in September 2022 via Lucid.Footnote 44 We opted for a vignette involving a future apology from Russia for its invasion of Ukraine. Here we detail our design and then reflect on its advantages and disadvantages.
Participants first answered questions about party affiliation, views on international politics, and attentiveness. Next, we reminded respondents that Russian forces illegally invaded Ukraine and committed atrocities. The scenario continued: “We are now going to describe a hypothetical situation involving this conflict that could occur in the future.” We began by describing (hypothetically) how the war ended. Subjects read that, in 2025, after a “long military stalemate,” Russia and Ukraine make a mutually agreeable peace deal giving autonomy to areas in eastern Ukraine. We held this outcome constant so our treatments did not shape perceptions of military power, which could affect assumptions about whether the apology was voluntary.
Respondents then learned either that Russia remains nondemocratic or that it “transitions to democracy.” We manipulated regime type to assess whether apology effects depend on regime type, and how the image-rehabilitation effects of apologies compare to those of democratic transitions.Footnote 45 Treatment wording appears in Table 1.
Table 1. Treatment wording (Study 1)
* Only a subset of respondents in the apology condition received the sender domestic backlash and victim reaction treatments.
Respondents next read that later in 2030 “a new Russian leader, Andrei Petrov, comes to power. In his first few months in office, President Petrov does not make any major changes to Russia’s domestic or foreign policy. Russia remains a [democracy/nondemocracy].” We described a leadership change to isolate democratization from leader turnover, and because apologies usually come from successor governments.Footnote 46 To prevent respondents from concluding that the apology involves other desirable policies, we say there were no major policy changes after Petrov assumed office and that the scenario occurs “soon after [Petrov’s] coming to power.”
Petrov is described as participating in an international summit and giving a speech. This matches typical historical apologies—which are usually delivered verbally at high-profile eventsFootnote 47 —while also setting up a plausible no-apology condition. Subjects in the apology condition read that Petrov gave an apology, which contained all four elements described earlier: acknowledging the harm, taking responsibility, acknowledging moral wrongdoing, and expressing remorse. In Study 1, those receiving the no-apology control instead learned that Petrov “does not apologize” and “does not comment on the invasion at all.”Footnote 48 (Study 2, described later, operationalizes the no-apology less explicitly).
For respondents in the “short apology condition,” the vignette stopped here. These respondents let us test hypotheses about the main effects of apologies. The rest learned two additional (randomized) pieces of information: whether the apology elicited strong domestic backlash or widespread support inside Russia; and whether Ukraine accepted or rejected the apology. We modeled these statements on real-world apologies.
All subjects then read a bullet-point summary.Footnote 49 To measure favorability, international cooperation, private transactions, and weakness, we asked four questions, each prefaced with “If events in 2030 happened just as we described.” Table 2 provides the wording and our main operationalizations, which are binary for easy interpretation.Footnote 50 Our results can thus be interpreted as the percentage of respondents who had favorable views of Russia, supported interstate cooperation, would not avoid purchasing Russian goods, and/or perceived Russia as weak. (Our conclusions are the same using the raw ordinal scales; see Appendix C). We concluded with questions about mechanisms, demographic characteristics, and political attitudes, and an opportunity for comments.
Table 2. Dependent variables
Note: The first three questions begin with “If events happened just as we described ...”
We fielded Study 2 in February 2024 with 2,852 US-based respondents. Our primary goal was to address the possibility that the explicit no-apology in Study 1 might be interpreted as a criticism of Russia or as implying that Russia categorically refused to apologize. Study 2 thus included two no-apology treatments (Appendix I). The “explicit no-apology” condition is very similar to Study 1’s no-apology condition. The “no mention” condition states that Petrov “discusses global challenges associated with artificial intelligence. He does not comment on the Ukraine war.”Footnote 51 We include the reference to AI, a relatively neutral topic, in all three conditions to make the absence of an apology less glaring in the “no-mention” condition.
Study 2 contained two additional innovations: we preregistered the hypothesis that apologies increase perceptions of military weakness and used a better measure of weakness (Table 2). Study 2 did not provide information about sender backlash or the target’s response.
It is important to consider the (dis)advantages of a vignette about a hypothetical future apology for a real, ongoing conflict between Russia and Ukraine. We chose this scenario because it offers an attractive balance of internal validity—specifically, our ability to isolate the effect of an apology from other factors that could affect respondents’ reactions—and external validity—in particular, our ability to draw conclusions about how apologies would shape third-party perceptions in natural (“real-world”) settings.Footnote 52
In terms of internal validity, describing a real country and transgression is an economical and effective way of holding background information and past behavior constant. This makes it more likely that the treatment varies only whether the country apologized, improving “information equivalency.”Footnote 53 Respondents do not need to “fill in the blanks” by making unwarranted assumptions based on the treatments, which could introduce confounding.Footnote 54
The specific choice of the Ukraine War also has advantages for internal validity because it features a schema- and treatment-consistent actor.Footnote 55 Russia is schema-consistent—the choice of actor is reasonable for the scenario being described—because Russia has committed transgressions and it is highly plausible that its leader would attend a summit after the war. It is also treatment-consistent—that is, all levels of the experimental treatments are believable for Russia. Russia could plausibly apologize to Ukraine in the future, given its apologies for the Katyn Massacre and abuses against Japanese World War II prisoners; and it could also (re-)democratize.
Regarding external validity, the Russia–Ukraine conflict provides insights about exactly the kind of case we are interested in given the scope of our argument, which is about whether countries can rehabilitate their tarnished images and reverse strong public disapproval after serious norm violations.Footnote 56 As in prominent historical cases, such as the German one we will consider later, the apology comes from a successor government relatively soon after the war’s end and involves widely known violations.
Of course, the choice of the Ukraine War also has features that might influence our findings. One possibility is that, because the war was ongoing during our surveys, Americans were particularly outraged, perhaps inflating estimates of apology effects by providing more “room” for apologies to work.Footnote 57 On the other hand, strong attitudes about Russia may make respondents especially resistant to change, biasing against finding apology effects. It is difficult to know a priori which way such bias, if any, would go. Future research could study apologies in additional experimental scenarios, including fully hypothetical ones with unnamed countries or different historical cases involving a (thus far) unapologetic country. In the meantime, our study can, at minimum, yield important insights about a key contemporary case and potentially tell us something about apologies’ international image effects more broadly.
Experimental Findings
Our analysis proceeds in three parts. Part 1 analyzes the main effects of apologies in the absence of information about target response or sender backlash. We show that apologies have large beneficial effects compared not only to explicit no-apologies (Study 1) but also to when apologies are simply not mentioned (Study 2). Parts 2 and 3 then unpack the more complex design of Study 1, assessing whether the target’s response (Part 2) or the sender’s domestic reaction (Part 3) moderate the effects of apologizing. We then briefly discuss mechanisms, the roles of regime type and democratization, and generalizability to different groups. All of the details of our analyses follow our preregistrations.
Part 1. The Effect of Apologizing
We first investigate apology effects when observers do not know how the target or the sender’s public reacted. Figure 1 plots results for Study 1, in which the no-apology was explicit. The left-hand side shows the level of each dependent variable (DV) by treatment condition, and the right-hand side displays the effects of the apology, testing our hypotheses. We calculated effects separately for democracies and nondemocracies, without control variables, using binary versions of the DVs and OLS models for ease of interpretation.Footnote 58 Our conclusions do not hinge on any of these choices: they are the same when using the full ordinal scales of the DVs, including demographic and attitudinal controls, or using logit (Appendix C).
Figure 1. The effect of an apology versus no apology (Study 1)
The top of Figure 1 shows the results for favorability toward Russia. The first row depicts favorability when Russia remains nondemocratic. Only 23.6 percent of Americans view Russia favorably when Russia explicitly does not apologize (solid black dot). When Russia apologizes, however, favorability swells to 67.1 percent (hollow gray dot). The hollow black dot at far right shows the difference, or in other words our estimate of the causal effect of an apology when Russia is nondemocratic: 43.6 percentage points. In the second row, we see that apologies also produce large and significant effects when Russia transitions to democracy: although 51.7 percent view a democratic Russia favorably even when it fails to apologize, favorability swells to 82.4 percent when Russia tells Ukraine it is sorry, an effect of 30.7 points.
Study 1 likewise reveals significant effects of apologies on support for cooperation with Russia (increases of 15 to 18 points) and on willingness to purchase Russian products (23 to 24 points). Thus, we find that apologies might not only improve favorability but even provide tangible benefits. In contrast, apologies have null effects on perceptions of weakness. Americans don’t doubt that Russia is “a force to be reckoned with,” even when it acknowledges past wrongs. In sum, despite the worries of some political commentators, Study 1 reveals only positive effects of apologies.
Figure 2 shows results from our follow-up study, fielded with a new set of US-based respondents more than a year after Study 1. Recall that Study 2 replicated Study 1 with only minor changes, including a second control condition in which the no-apology was implicit. The hollow dots in the right-hand panels of Figure 2 plot the effects of apologies relative to explicit no-apologies, while the hollow squares plot the effects of apologies relative to the no-mention condition.
Figure 2. The effect of an apology versus no apology (Study 2)
We highlight several findings. First, the results from Study 2 (Figure 2) confirm Study 1’s conclusion that apologies have large effects on favorability, cooperation, and willingness to buy products compared to explicit no-apologies. Second, apologies continue to have significant effects on these three variables when the apology is only implicit. Third, we again see no evidence that apologies signal weakness, even using our improved measure. In conclusion, across two large studies fielded more than a year apart, we find that an apology helps repair a country’s image abroad, without undermining its reputation for strength.
Part 2. Does the Victim’s Response Moderate the Effect of the Apology?
We now deepen our analysis of Study 1 to investigate factors that could influence the size of apology effects.Footnote 59 We begin with the target’s reaction. Figure 3 distinguishes three situations: the sender does not apologize; the sender apologizes and the target accepts; and the sender apologizes but the target rejects. As before, the left-hand side plots levels of DVs and the right-hand side shows treatment effects, for democracies and nondemocracies separately. We focus on situations without backlash, to isolate the effect of the target’s negative reaction.
Figure 3. The target’s reaction (Study 1)
Although target acceptance enhances the benefits of apologies, apologies remain potent even when the target rejects them. For example, favorability of a nondemocratic and unapologetic Russia is 23.6 percent, but when Russia apologizes and Ukraine rejects, favorability increases to 61.2 percent (a 37.6-point effect). When Ukraine accepts, favorability rises further, to 74.6 percent (a 51-point effect). The extra effect of Ukraine’s accepting the apology is thus 51.0 – 37.6 = 13.4 percentage points, easily distinguishable from zero (Appendix C). We observe similar results for support for interstate cooperation and buying products: Ukraine accepting is helpful, but rejected apologies are significantly better than no apology. We draw similar conclusions when Russia is described as democratic.
In contrast, regardless of Russia’s regime type, the target’s reaction has no effect on perceptions of weakness. Whether or not Russia apologizes, and whether or not Ukraine accepts that apology, voters rarely see Russia as weak. In sum, victims can influence how international audiences view apologies, but senders enjoy benefits even when victims spurn the overture.
Part 3. Does Sender Backlash Moderate the Effect of the Apology?
Finally, we test whether strong backlash inside the sender undermines the effects of apologies. We do this separately for each combination of regime type and target reaction. The left-hand side of Figure 4 plots the levels of the DVs for vignettes where the target accepts. Relative to no apology, even apologies with significant backlash improve perceptions of favorability, willingness to cooperate, and willingness to purchase products, regardless of regime type. We thus conclude that apologies carry benefits even if they produce significant criticism in the sender.
Figure 4. Backlash in the sender (Study 1)
The evidence on whether severe backlash attenuates the effects of apologies is less consistent. When Russia remains nondemocratic, a domestically contentious apology produces about nine percentage points less favorability than one without backlash. When Russia transitions to democracy, the pattern is in the same direction but not statistically significant with our main (binary) DV. Backlash significantly reduces support for cooperation regardless of regime type, but significantly reduces support for buying products only when Russia is nondemocratic (see Appendix C). Backlash has similar effects when Ukraine rejects the apology (right side of the figure). In sum, we conclude that widespread backlash often, but not consistently, degrades the beneficial effect of an apology relative to no backlash, and that apologies with backlash are consistently better than no apology at all.
Mechanisms, Regime Type, and Democratization
We also preregistered predictions about mechanisms, regime type, and democratization. We refer readers to the appendix for details but provide summary observations here.
Regarding mechanisms, across both studies, we find that apologies consistently reduce perceptions of threat and increase perceptions of moral values relative to explicit no-apologies (Appendix D).Footnote 60 These effects are somewhat attenuated when comparing apologies to no-mentions in Study 2. As expected, moreover, perceptions of threat and values correlate with favorable view, support cooperation, and buy products. Apologies have no effect on perceptions of status or military power, perhaps explaining why apologies did not convey weakness.
As for mechanisms behind the effects of target response, acceptance decreases perceptions of threat relative to rejection but does not significantly improve perceptions of values. At the same time, rejected apologies improve perceptions of both threat and values relative to no apology. Rejection has no effect on perceptions of status.
Regarding sender backlash, evidence that it affects perceptions of threat and values is mixed. However, apologies with backlash are always better than no apology. Altogether, we interpret these findings as suggesting that apologies (and target rejection and backlash) work primarily by affecting perceptions of threat and values.
We next consider regime type and democratization (details in Appendix E). Regime type does not consistently moderate the effect of apologizing: Russia garners apology benefits whether or not it transitions to democracy. Moreover, independent of apologies, democratization improves the offender’s international image. This echoes findings that foreign democracies approve when countries embrace liberal values and recoil against backsliding.Footnote 61 The effect of apologizing without democratizing is not consistently larger or smaller than the effect of democratizing without an apology: both effects are large.
Effects Across Subgroups and Generalizability to US Elites or Respondents in Other Countries
We preregistered exploratory analyses to see whether the effects of apologies occur across the population or are unique to specific subgroups. With a few exceptions, apologies generally proved potent, regardless of political party, gender, hawkishness, age, prior views of Russia, interest in politics, education, or authoritarian attitudes (Appendix H).
Our data also let us consider whether our conclusions might generalize to US decision makers. Following Kertzer and Mattes and Weeks, we analyze subgroups with “elite-like” characteristics, such as high income, high education, age forty to sixty-five, and/or strong political interest (Appendix G).Footnote 62 Our findings on the effect of apologies largely hold among subjects with these traits (or combinations thereof), though the findings on victim response and sender backlash are mixed, possibly because of small N. Thus, if elite-like voters shed any light on the expected reactions of political elites, there is some indication that had we surveyed US elites we would have found similar image-repair effects for apologies, though we are less sure how elites would respond to target rejection and sender backlash.
Finally, Study 2 proposed additional explorations to probe the scope of our conclusions. Americans are typically committed to liberal values and perceive Russia as threatening, potentially predisposing them to value a Russian apology. Would individuals less committed to democratic principles and less worried about Russia—such as Hungarian or Chinese respondents—react similarly? We find that an apology improves Russia’s image even among subjects who tend toward authoritarianism, embrace right-wing thinking, have weak commitment to democratic norms, or view Russia as friendly, though effects for some combinations of regime type and DV are attenuated (Appendix H). We find no indication that apologies convey weakness within any groups. Thus, while scholars would need to replicate our experiments in other countries to know for sure, our evidence suggests that apologies may be viewed favorably by audiences with different attitudes and preferences than typical Americans.Footnote 63
Illustrative Case: West Germany’s 1951 Holocaust Apology
We supplement our experimental findings with evidence from Germany’s first apology after World War II. The case reinforces conclusions from our survey experiments: apologies can indeed translate into tangible benefits, and reassurance and perceptions of values are plausible explanations for these effects. At the same time, the case demonstrates the challenges of testing our hypotheses with historical data—including selection bias, lack of consistent measurement of the DVs, and potential confounding—reaffirming our reliance on experimental tests.
West Germany emerged from World War II a “pariah.”Footnote 64 As Chancellor Adenauer put it, Germany “had committed such crimes against [the Jews], that this needed to be atoned [gesühnt] or recompensed [wiedergutgemacht] should we want to win back esteem among the peoples of the earth”Footnote 65 —including from Israel, which had demanded that Germany admit its guilt. On 27 September 1951, Adenauer gave a historic speech acknowledging Germany’s “unspeakable crimes” and stating that Germany owed “moral and material restitution.”Footnote 66
Thus, not long after the transgression, a new leader came to power and offered an apology, just as in our experiment. Adenauer’s apology was simultaneously stronger and weaker than in our experiments: it did not unequivocally fault the German people, but it did offer reparations. The apology did not produce backlash at home, but the target’s response was noncommittal: Israel said only that it would “study the German Chancellor’s declaration and ... in due course make its attitude known.”Footnote 67
To assess international reactions, we focus on the immediate aftermath of Adenauer’s speech, before the September 1952 reparations agreement. Available evidence suggests that the international response was overwhelmingly positive, with the apology reducing perceptions of German threat and increasing confidence that it was undergoing a “moral rebirth” and abandoning the unacceptable values of the Nazi era.Footnote 68 The American High Commissioner for West Germany was reportedly “deeply impressed with the tone and content” of the speech, and in the British Parliament, Adenauer’s speech “was widely admired because of its ‘high moral character’.”Footnote 69 The New York Times called it “the best thing that [has come] from Germany since before 1933,” and the Washington Post referred to “a phase of moral regeneration.”Footnote 70 We found no evidence that third parties thought the apology signaled German weakness.
The scarcity of polling in the 1950s makes it difficult to gather an accurate picture of American public opinion, but polls conducted before and after the apology suggest the apology improved Germany’s image. In April 1950, 74 percent expressed concerns about Nazis regaining power in Germany;Footnote 71 in February 1952, only 30 percent attributed “much chance” to this.Footnote 72 In November 1950, 58 percent believed West Germany should be allowed to build an army,Footnote 73 while, in June 1952, 72 percent approved of “bring[ing] western Germany into the defense of Europe.”Footnote 74 Americans clearly viewed the German threat as diminished after the apology.
Furthermore, two months after Adenauer’s speech, West Germany and the Allies initialed a draft agreement that would ultimately terminate the occupation statute, suggesting tangible benefits. Of course, other factors, such as the growing communist threat, might have helped thaw Western attitudes, but available evidence suggests that the apology eased West Germany’s acceptance into the international fold. Later German acts of contrition further redeemed Germany’s international image. As Berger notes, “Germany is widely seen to have benefited from its penitent stance on history.”Footnote 75
In sum, US and foreign elites, and the US public, reacted positively to Germany’s apology, which provided reassurance and demonstrated desirable moral values. Moreover, the case highlights a possible tangible benefit to Germany by laying the groundwork for the termination of the occupation statute. The appendix further illustrates the value of apologies by discussing a prominent Japanese apology.
Conclusion
Apologies are usually portrayed as gestures affecting the relationship between violators and victims. We highlight apologies’ potential to also shape broader international perceptions. Apologies can improve states’ international image by providing reassurance and signaling information about values, effects that might be moderated by reactions in the recipient and the apologizing state itself. We contrast these ideas with the view, sometimes espoused by politicians, that apologies convey weakness.
Our evidence—two survey experiments supported by exploration of a key historical case—suggests that apologies can have powerful image-repair effects. In our experiments, Russia benefited significantly from apologizing, whether it was described as democratic or nondemocratic. An apology boosted not only favorable impressions but also enthusiasm for cooperation and economic transactions. We find that apologies can be beneficial even when the victim rejects the apology, or the apology sparks significant domestic backlash in the sender. While our experiments provide strong evidence of the international benefits of apologizing—and for the underlying mechanisms of reassurance and good moral values—we found no indication that apologizing leads third parties to view the sender as weak. Our investigation of Germany’s first postwar apology echoed these positive, and not negative, effects.
We also probed the import and generalizability of our findings. First, we found that individuals most closely resembling political elites appreciate apologies similarly to our sample as a whole. Second, even individuals supportive of Russia and aligned with its values tend to welcome a hypothetical Russian apology and do not see it as a sign of weakness. Thus, while future research would be needed to know for sure, our evidence is at least consistent with the conjecture that apologies would be viewed positively not only among the US public but also among US elites, and among publics in countries dissimilar to the US on important dimensions.
Overall, our study suggests that apologies help states improve their international image—and potentially unlock opportunities for cooperation—even if the apology fails at what is typically seen as its primary goal: reconciliation. These broader international benefits may help explain why countries sometimes apologize even when there is real risk that the victim will reject the apology.
Our study raises important questions for future research. Our experiments focus on a specific high-profile case and examine American responses. Future work should theorize about factors that might amplify or attenuate the effects of apologies, and thus potentially study different transgressions involving other countries or hypothetical situations and consider different audiences. One interesting question is whether, when respondents are less well-informed about a country’s offenses, perhaps because more time has passed, apologies have the downside of highlighting a transgression observers might otherwise have forgotten, reducing the net image benefit of the apology. Furthermore, if respondents become aware of the instrumental calculus behind an apology, they may doubt its sincerity and become less inclined to reward the sender. The effects of apologies might also depend on the sender’s level of military power or other attributes. Finally, perhaps “diplomatic” apologies involving less heinous, that is more justifiable, offenses provoke different foreign reactions, eliciting the perceptions of weakness that leaders sometimes worry about.
Our findings also raise a puzzle: if apologies are so beneficial internationally, why don’t states always apologize? We suspect the answer involves domestic politics. Although senders may reap international benefits despite significant backlash at home, this backlash may endanger the government’s domestic political fortunes.Footnote 76 Future studies might evaluate this trade-off.
In the meantime, our results indicate that international apologies are more than an instrument of reconciliation. They are also a tool of public diplomacy. Apologies can soften the attitudes of foreign publics and leaders, paving the way for a comeback for disgraced states that have broken core international norms. Thus, studying international apologies, including when they are offered and what effects they produce, can broaden our understanding of international politics more generally.
Data Availability Statement
Replication files for this research note may be found at <https://doi.org/10.7910/DVN/YQ5KAH>.
Supplementary Material
Supplementary material for this research note is available at <https://doi.org/10.1017/10.1017/S0020818325000086>.
Authors
Acknowledgments
Previous versions of this paper were presented at the annual meetings of the Peace Science Society (Denver, CO, 3–5 November 2022) and the Midwest Political Science Association (Chicago, IL, 13–16 April 2023). For insightful feedback, we thank Lotem Bassan-Nygate, Austin Beacham, Stephen Chaudoin, Steph Haggard, Michael Joseph, Marika Landau-Wells, Jonathan Renshon, Brandon Yoder, and Agnes Yu, as well as panelists at PSS and MPSA and workshop participants at the University of California, San Diego, University of Virginia, Duke University, Harvard University, Stanford University, London School of Economics, and University at Buffalo.
Open access
