Public preferences for international law compliance: Respecting legal obligations or conforming to common practices?

Abstract

Despite significant debate about the ability of international law to constrain state behavior, recent research points to domestic mechanisms that deter non-compliance, most notably public disapproval of governments that violate treaty agreements. However, existing studies have not explicitly differentiated two distinct, theoretically important motivations that underlie this disapproval: respect for legal obligations versus the desire to follow common global practices. We design an innovative survey experiment in Japan that manipulates information about these two potential channels directly. We examine attitudes towards four controversial practices that fall afoul of international law—same-surname marriage, whaling, hate speech regulation, and capital punishment—and find that the legal obligation cue has a stronger effect on respondent attitudes than the common practices cue. We also show subgroup differences based on partisanship and identification with global civil society. These results demonstrate that the legal nature of international law is crucial to domestic compliance pull.

1 Introduction

Scholars of international relations have long debated whether and why international law alters state behavior in the absence of external sanctions or material penalties.¹ Although early empirical studies suggested that international human rights treaties do not improve domestic practices (Keith , 1999; Neumayer , 2005), recent work reports a positive correlation between treaty ratification and human rights compliance (Hill Jr , 2010; Fariss , 2014). Many point to a variety of interconnected, domestic mechanisms through which governments are pressured to abide by international agreements (Moravcsik , 2000; Dai , 2005; Hafner-Burton & Tsutsui , 2005; Hafner-Burton , 2008; Simmons , 2009; Lupu , 2013; Tsutsui , 2018). While there is significant variation across issue areas, from human rights to jus in bello, there is a growing consensus on the existence of a domestic “compliance pull” for international law.

One factor that should strengthen government adherence to international law is public opinion in support for compliance. A growing body of work has tried to assess this public sentiment empirically, but experimental studies have produced mixed evidence. On the one hand, studies in the United States suggest that voters are more likely to support a policy change if the status quo violates international law. Across issue areas, information that current U.S. policies do not conform to international law increases respondents’ approval for their alteration (Wallace , 2013; Chaudoin , 2014; Chilton , 2014; Wallace , 2014; Chilton & Versteeg , 2016; Chilton , 2015; Kreps & Wallace , 2016; Wallace , 2019; Strezhnev et al. , 2019; Linos , 2011; Linos , 2013).² On the other hand, experiments conducted outside of the U.S. show that there may be backlash effects to international law (Lupu & Wallace , 2019; Chapman & Chaudoin , 2020; Cope & Crabtree , 2020), albeit with exceptions (Anjum et al. , 2020). These results imply that accusations of international law violations can actually cause citizens to dig in and strengthen their support for the policy or the government in question, at least under some circumstances.

One reason why existing experimental manipulations produce contradictory results may be that they do not fully identify which aspect of international law makes voters think that compliance matters. International law, much like its domestic counterparts, is a set of rules that legally binds its signatories. However, its specific manifestations, particularly multilateral treaties on transnational issues, can also convey information about common practices that states follow. Existing studies have not explicitly differentiated whether voters prefer compliance because they value adherence to legal commitments, or because they desire conformity with common practices in other countries. Put differently, when (only) informed that a status quo policy or government action violates international law, respondents may view the policy as illegal, anomalous, or both.

In order to understand why the public prefers compliance in some cases but not in others, individual-level motivations for compliance must be disentangled. To examine the relative salience of these two facets of international law, we conduct a survey experiment that randomly assigns information regarding legal obligations and common practices separately. We ask Japanese respondents for their views on four domestic practices and policies that have been criticized for violating international law or are rare among developed countries. Our treatments are designed to distinguish the effect of being told that these practices violate binding legal commitments, and the effect of being told that these deviate from common practices in peer countries. To the best of our knowledge, this is the first experiment where the experimental manipulation is directly relevant to the motivation underlying preferences for international law compliance.

More specifically, our experiment randomly presents a nationally representative sample of Japanese voters with one of the following information treatments about domestic practices: (1) they violate international law; (2) they violate international law, but the Constitution of Japan requires compliance with ratified international treaties; (3) they are rare in other industrialized countries. Our outcomes of interest are attitudes toward the following four status quo practices: (1) the family law clause under which married couples must choose a common surname, (2) commercial whaling, (3) the lack of penalties on hate speech, and (4) the death penalty.

Our research design has several important advantages over existing survey experiments. First, our treatments seek to distinguish the importance of international law in the minds of citizens. The first treatment is similar to those in previous studies and does not distinguish the “legal” and “common practices” connotations of international law compliance explicitly. By contrast, the second treatment underscores the constitutional importance of international law to emphasize the legal dimension, while the third treatment notes deviations from common global practices without providing any normative judgements.

Second, Japan is a useful laboratory to address caveats relating to the generalizability of previous experimental studies. Japan has ratified numerous international treaties and conventions, but it has several controversial laws and social customs that do not conform to the norms stipulated in the treaties. This is partially related to the fact that Japan is a non-Western society, and therefore certain traditional customs are more likely to conflict with international standards (Anghie , 2007). In terms of political context, Japan is a democracy where public opinion is expected to matter in policy-making; our results thus have real world implications for the likelihood of political reforms. It is also the third largest economy in the world, and so mechanisms that may matter in developing countries, such as the importance of international law compliance to obtain foreign aid, are less consequential.

The results of our experiment suggest that citizens’ views on international law compliance are driven by their respect for legal obligations, rather than by the desire to conform to other states’ practices. While the second (constitution) treatment has statistically significant effects on some of the policy items, the estimated effects of the third (common practices) treatment are not significant for any outcome. In addition, we find that the effect of the second treatment is driven by a critical view of these controversial practices’ underlying values and by partisanship. Those who identify with global civil society, and who do not support the governing conservative party, are more likely to favor treaty compliance under the constitution treatment. Overall, we find evidence that voters favor compliance with international law because they value honoring legal commitments rather than conforming to common practices.

Public opinion, by itself, may not determine whether governments ratify and abide by international laws. The interest of political actors are influenced by other factors, including pressure from domestic lobbyists and foreign governments. However, this paper points to when governments can generate buy-in from the public and lessen the costs of international law compliance. This may be particularly important when compliance requires changes to domestically-contentious practices that may penalize incumbent governments electorally, such as requisite surname changes upon marriage in Japan. Furthermore, our focus on Japan also expands the literature’s empirical scope to non-Western cases, contributing to our understanding of cross-national variation in the impact of international legal commitments.

The remainder of the paper proceeds as follows. The next section articulates our argument and hypotheses on why the public prefers compliance with international law. In the third section, we describe the case context and explain the specifics of our survey design. The fourth section presents our experimental results, and the fifth section discusses how they fit into the literature’s prevailing understanding of international law compliance. We conclude the paper with remarks on the limitations and broader implications of our study.

2 Literature review and theory

While most scholars agree that there is no central authority to enforce international law, they remain divided on whether states nevertheless behave differently in the presence of international law than they would in its absence. Some argue that states adjust their behavior to conform to international law once they acquire the capacity to do so (Chayes & Chayes , 1993). Others contend that states join only those treaties that serve their national interest, and therefore their behavior would be identical regardless of international law (Downs et al. , 1996). This debate is difficult to resolve through empirical studies at the state level due to unobserved counterfactuals, namely how states would have behaved had they not become signatories (Simmons , 2000; von Stein , 2005). Instead of trying to measure compliance at the state level, recent research has paid greater attention to the domestic mechanisms through which governments are pressured to abide by international law commitments. Theoretically, constituencies that stand to benefit from compliance should exert electoral leverage on their government to do so (Dai , 2005). Even when elected branches are apprehensive, domestic courts can facilitate the enforcement of ratified treaties (Lupu , 2013). Treaties also create normative pressure for compliance by empowering rights activists (Hafner-Burton & Tsutsui , 2005; Tsutsui , 2018) and by exposing states to international naming and shaming (Hafner-Burton , 2008).

Many of these domestic mechanisms are plausible if public opinion supports compliance. For international law to empower rights activists, the general public must consider violations as something that should be rectified. Similarly, the significance of international naming and shaming presupposes that voters are sensitive to their nation’s international reputation. To establish whether governments risk electoral penalties for violating international agreements, it is critical to first determine whether the public actually cares about international law compliance.

One popular approach has been survey experiments that test whether voters are more likely to support a policy change when they are told that the status quo violates international law. However, the results of these tests have been mixed. On the one hand, the majority of studies conducted in the United States confirm this hypothesis. In one of the earliest such works, Wallace (2013) reports that respondents are more likely to oppose the use of torture in the U.S. War on Terror if they are informed that torture is prohibited by international law. Similar experiments have been conducted across issue areas, including trade policy (Chaudoin , 2014), the rights of prisoners (Chilton , 2014), the use of torture (Wallace , 2014; Chilton & Versteeg , 2016), the conduct of military operations (Chilton , 2015; Kreps & Wallace , 2016; Wallace , 2019), refugee policy (Strezhnev et al. , 2019), and welfare policy (Linos , 2011; Linos , 2013).³ On the other hand, experiments conducted outside of the U.S. show that there may be backlash effects from international law. Informing citizens that a policy violates international law sometimes increases support for that policy, perhaps due to nationalistic sensitivity to criticism from outsiders. For example, Lupu and Wallace (2019) find that government approval increases among Israeli respondents when they are told that their government’s repression of opposition groups violates international law. Similar backlash has been reported in experiments on International Criminal Court rulings in Kyrgyzstan (Chapman & Chaudoin , 2020) and refugee policy in Turkey (Cope & Crabtree , 2020). However, Anjum et al. (2020)’s study on women’s rights in Pakistan demonstrates public preference for compliance. In sum, the current state of the literature suggests that public support for international law compliance may vary significantly across both countries and issue areas.

The mixed nature of existing evidence suggests the need for caution in generalizing experimental results from the U.S. and calls for further studies on the conditions under which international laws influence public opinion positively or negatively (Chilton & Linos , 2020). To understand when citizens support international law and when they do not, it is important to decompose the reasons why they value compliance with international law in the first place. Although previous studies propose several mechanisms through which international law affects citizens’ preferences, they have yet to propose a unified explanation as to why information about international law changes public opinion. In particular, they tend to ignore the distinction between respect for legal obligations and the desire to conform with common practices in other countries.

2.1 Reasons for public preference for compliance

The mechanisms through which international law shapes citizens’ preference can be broadly categorized into two groups, each of which stems from different aspects of international law. The first set of mechanisms derives from the legal aspect of international law, while the second arises from its informational aspect about common practices in other countries.

First, the public may prefer compliance with international law because reneging on treaty obligations can harm the nation’s international reputation. This argument underlies previous studies that identify international law as an instrument for credible commitment. If citizens believe that violations of codified international law can harm their state’s reputation as a law-abiding country and impact its diplomatic capabilities, then they should penalize leaders who renege on treaty obligations, which in turn should deter leaders from doing so. A key underlying assumption is that public preference for compliance is due to the expectation of reputation loss. Empirical studies of audience costs verify a similar claim that the public penalizes inconsistencies of leaders for fear of losing reputation (Tomz , 2008), and this mechanism may also be applicable to commitments to international laws (Chaudoin , 2014). In particular, while arguments linking commitment incentives to public disapproval is common in studies of crisis bargaining and international cooperation, these mechanisms should be even stronger for international treaties, which lay out the obligations of signatories explicitly (Simmons , 2010).

The legal aspect of international law also connects to procedual explanations for compliance. Some citizens may be skeptical about the legal legitimacy of international law, as implied by the backlash effects found in empirical studies outside of the United States. Nevertheless, its legalistic appeal is still relevant to skeptics because international law is, to varying degrees, integrated into domestic law and legal practices. According to data from the Comparative Constitutions Project, 92% of national constitutions include provisions pertaining to international law (Elkins et al. , 2009). In addition, states often change domestic laws to comply with international agreements, and international laws themselves can be used in domestic courts, though their specific treatment varies across states and legal traditions (Simmons , 2009; Lupu , 2013; Tsutsui , 2018). Even when international law itself may not generate strong pull for compliance, international law as an integrated part of domestic legal systems is likely to be perceived as obligations that should be followed. In sum, these mechanisms—expected reputation loss as a law-abiding country and integration into domestic legal systems—arise from the legal nature of international law.

The second set of arguments holds that international laws can influence public opinion because they serve as signals of common practices in other countries. According to information processing theory, people use heuristics to learn about things on which they have little prior knowledge (Kinder , 1998). International law can function as an informational cue about practices in other countries, even if there is a sizable gap between treaty obligations and compliance. This is because the public tends be less knowledgeable of specific practices in foreign countries, let alone about general levels of compliance. For example, Aalberg et al. (2013) argues that public interest in international news is consistently lower than domestic news across 11 countries, including Japan. In our survey, which will be discussed in greater detail below, we asked respondents to match the names of heads of government to their countries for Brazil, Germany, France and Canada. 73.8% chose “Do not know” to at least one of the names, and 32.9% chose “Do not know” for all four names. This suggests that the general public may have limited knowledge about foreign countries, and thus may be reliant on other informational cues. If people are informed that a treaty bans a particular policy, it is reasonable for them to infer that such a policy is uncommon among other treaty signatories. This may, in turn, lead people to prefer policies that conform to the practices of other countries.

A related insight from social psychology is that people conform to the behavior of others whom they identify as “in-group” (Schultz et al. , 2008). This is relevant at two distinct levels. At the country level, this implies that Japan should have incentives to comply with international laws since it is a status-quo power in the current international system. Schoppa (1999) argues that the Japanese government has historically been sensitive to international criticism of its trade policies, particularly when it comes from trusted allies. This tendency is also evident in Japan’s behavior regarding commercial whaling, which we discuss in greater detail in later sections.

At the individual level, this insight further suggests that information about international law violations may prompt different responses based on citizens’ prior values. For one, citizens who identify as members of a wider international community may have a stronger desire to conform with global standards than those with more nationalistic orientations. Kertzer and Brutger (2016) find that cosmopolitanism moderates how much voters penalize governments for inconsistent policy claims, and Bayram (2017) shows that cosmopolitan identity influences the desire to be in compliance with international law. Chilton (2014) similarly argues that such pro-compliance attitudes influence how information about international law violations shape preferences for policy reforms, although—as we expand on below—its experimental design does not capture the purely informational role of international law. In addition to cosmopolitan identity, partisanship is likely to be an important moderator. Those with a strong affinity towards governing parties—the Liberal Democratic Party, in the case of Japan—may be less willing to accept criticisms of status quo practices, either because of their ideological agreement with the parties’ policies, or because of partisan motivated reasoning (Bolsen et al. , 2014). These effects of partisanship are greater when voters are less informed about the issue (Lupia , 1994), which is likely to be the case for international law violations.

The distinction between the mechanisms derived from the legal versus information aspects of international law is more than a matter of nuance. Theoretically, it helps us distinguish whether legally-binding commitments are necessary to influence public opinion, or whether it is sufficient to inform citizens about common practices in other countries. Although the legalization literature has argued that legal norms have distinct features from non-legal norms that facilitate inter-state cooperation (Abbott & Snidal , 2000), this contrast has yet to be addressed in terms of its effects on public opinion. One exception is Wallace (2013), which tests how the degree of legalization causes different levels of public response. However, this study uses various design information about treaties to compare variations within legal norms, and does not explicitly address the salience of informing citizens about the behavior of others.

Furthermore, this distinction may help explain why public responses to international law violations are positive in some cases but negative in others. For instance, some people may perceive information about common practices in other countries as undue foreign interference in domestic matters, particularly when they do not identify themselves or their country with other, typically Western, countries. However, they may be more amenable to information that emphasizes violations of legal obligations, as these were entered into by their government’s own volition.

As clarification, we should note that the informational role of common global practices is distinct from arguments relating to normative pressures for compliance. For example, studies of elite level behavior argue that states copy policies and institutions of peer countries to enhance their status in the international community (Meyer & Rowan , 1977; Finnemore & Sikkink , 1998). In the case of Japan, Gurowitz (1999) explains how the government strategically adopted global norms on immigration in order to claim legitimacy in the international community. At the same time, membership in reputable international organizations (Gray , 2013), forging trade agreements with reputable countries (Gray & Hicks , 2014), or even adopting electoral gender quotas (Bush & Zetterberg , 2021) can improve a country’s standing among foreign publics. If domestic voters care about their government’s foreign reputation, then international norms may shape their views. Indeed, studies of norm diffusion emphasize the role of norm advocates in promoting the value of international law among the public (Sikkink , 1993). If norm advocates can argue convincingly that the adoption of global standards can reduce sovereign credit risk (Gray , 2013) or increase foreign aid (Bush & Zetterberg , 2021), then elites and citizens alike may agree on its material benefits. Distinguishing the effects of international law as information heuristics versus norms is an important endeavor. In this paper, however, we focus on developing an empirical strategy that distinguishes the legal versus informational mechanisms, and leave the normative dimension to future studies.

2.2 Empirical strategy

Our study can be distinguished from previous research in three important respects, each of which addresses shortcomings in the literature mentioned above.

First and foremost, our experiment manipulates information about legal obligations and common practices separately. This allows us to test which of these two dimensions of international law is more salient to public opinion formation. Before measuring respondents’ support or opposition to four domestic legal and social practices, we randomly assign respondents into three treatment groups, as well as one control group. The first treatment group receives information that the practice is in violation of international law, without further elaboration. This treatment is equivalent to those in canonical studies (Wallace , 2013; Chilton , 2014). The second group receives similar information about these violations, but with a further emphasis on the legally-binding nature of international law. To do this, we provide additional (true) information that the constitution mandates compliance with international law. An important note is that this treatment only implies procedural violations of the constitution, rather than a substantive violation of a right that is explicitly enumerated in the constitution itself.⁴ In contrast to the first two treatments, the third treatment does not mention violations of international law per se. Instead, respondents receive factual information about the rarity of these practices in peer countries. Its purpose is to assess whether voters desire conformity with common global practices, even in the absence of legal obligations. This approach is distinguished from Chilton (2014), whose causal mediation analysis attempts to address whether citizens prefer compliance because they value conformity with international standards. In the mediation analysis, however, this preference for international conformity is measured after the respondents are informed of international law violations. Therefore, its results do not necessarily mean that inconsistencies between domestic practices and international standards, independent of legal obligations, cause preferences to change. Our design excludes the information of international law violations from the information treatment on globally common practices.

Prior to the main outcome questions with the manipulated information, we also ask for respondents’ political beliefs, which may reasonably produce heterogeneus treatment effects. First, we ask whether they see themselves as members of the international community, i.e. whether they identify with the values that international law represents. Second, we inquire whether they are supporters of the long-ruling Liberal Democratic Party, under which the controversial domestic practices have been fostered or tolerated. These two questions, which we will describe in greater detail below, speak to the inherent value that citizens (may) place on international law and that underlies existing theories about the domestic pull for compliance.

This study’s second innovation is its topical scope. We inquire about opinions on a wide variety of issues beyond national security matters, which have been the primary subject of analysis in previous experiments. These include practices that are domestic, social matters that do not impinge on the material welfare of other countries, and thus are unlikely to engender calls for penalties or economic sanctions from the international community. This should allow for a more straightforward test of the inherent “legal commitment” versus “common practices” perceptions of international law across issue domains.

Third, our study is conducted in Japan, which we consider to be an ideal laboratory for this subject matter. Japan has ratified numerous international treaties, but several of its legal and social practices have been criticized for being in violation of these commitments. This is due in part to the fact that Japan is a non-Western society, whose domestic social practices evolved outside of the debates and movements that shaped Western nations, which have had an outsized influence on the establishment of international norms and law (Anghie , 2007). In addition, Japan is a consolidated democracy, where public attitudes matter for policy choice. If information on treaty violations does change public opinion towards the government’s competence or morality, then it is also more likely to translate into actual policy changes in democratic than authoritarian countries. Finally, Japan is the third largest economy in the world, and so its citizens are unlikely to support treaty compliance for economic purposes, such as receiving foreign aid. While we do not discount the importance of this factor in developing economies, such countries vary in the degree and nature of aid desired, raising challenges for external validity. Given mixed evidence on the effect of international law in non-Western countries, we believe that this new case can contribute to our understanding of the generalizable implications of international law.

3 Research design

3.1 Case context

Our outcomes of interest are responses to four Japanese customs and practices that have received international criticism. Their domestic and global salience vary: marriage surnames, hate speech, and the death penalty are mostly domestic matters, while commercial whaling involves global commons with cross-border implications. Let us begin by describing the background of these items briefly. The distribution of responses to related questions among our control group respondents is shown in the Appendix.⁵

3.1.1 Mixed-surname marriage and CEDAW

Since 1896, Japan’s civil code has required married couples to adopt a common surname. This practice is rare globally: as of 2020, Japan is the only OECD country with this mandate. In 96% of cases, the wife changes her surname to that of the husband (Nippon.com , 2019). This legal mandate has long been controversial in Japan, even as the informal use of separate, pre-marriage surnames has become de facto accepted in professional life. Proponents of the status quo warn that separate surnames will weaken familial unity. Critics, including the Japan Federation of Bar Associations (2017), contend that this requirement runs afoul of the Convention on the Elimination of All Forms of Discrimination against Women (CEDAW), which Japan signed in 1985.

The provision has been contested in domestic courts since the late 1980s. However, the Supreme Court ruled in 2015 that that the law does not discriminate against women because the adopted surname can be that of either spouse, the practice is well established in society, and there is a rational basis for its requirement.⁶ However, in a supplementary opinion, Justice Kiyoko Okabe argued that this practice is in tension with CEDAW; this conflict was also referred to in Justice Yoshiki Yamaura’s dissenting opinion. While the Ministry of Justice has drafted bills to allow separate surnames, this has been blocked repeatedly by the conservative Liberal Democratic Party, which has argued that the family is the fundamental unit of society, and common surnames contribute to a stronger sense of familial unity.

On this issue, we ask respondents whether Japan should amend the civil code so that couples can choose their preferred surname. In our control group, the mean outcome value, measured on a 1 to 6 scale where higher values denote stronger agreement, is 4.08. This practice has the strongest baseline support for change among our four dependent variables.

3.1.2 Whaling and the IWC

Japan, along with Iceland and Norway, is one of the few countries that engage in commercial whaling as of 2020.⁷ While some regions of Japan have a long tradition of whaling, the government has, in the past, bowed to international pressure and curbed these practices. As a member of the International Whaling Commission (IWC) since 1961, Japan halted commercial whaling when the IWC placed a moratorium in 1982, but it continued to hunt whales by utilizing the scientific research exception.⁸ Catalinac and Chan (2005) argue that this balance—agreeing to the moratorium on paper but utilizing the scientific exception as a work around—reflects Japan’s sensitivity to international norms. In principle, the government was opposed to the banning of commercial whaling, due to the whaling industry’s political clout and its belief that whaling was necessary for Japan’s resource security. The weakness and scattered nature of environmental civil society movements also limited domestic pushback. However, Catalinac and Chan (2005, p. 152) also contend that Japan could not object to the IWC for two reasons: the desire to be seen as an equal by Western powers, and to remedy its poor reputation with regard to environmental issues.

Commercial whaling has been criticized by the European Union and Australia, as well as by environmental NGOs such as Sea Shepherd. Strausz (2014, p. 457) argues that commercial whaling, and the subsequent usage of the scientific whaling exception, was seen domestically as hindering Japan’s attempts to flex its soft power globally. However, after the International Court of Justice ruled in 2014 that its continued whaling violated IWC regulations, Japan withdrew from the IWC in 2019. The ICJ dispute was widely reported in both domestic and international media and caused a nation-wide debate about the Japanese government’s response. Given high public salience, Japan’s withdrawal from the IWC was seen as a strategic move by the government to avoid further public accusations for violating international rules.

We include one item on this practice: whether the respondent thinks that Japan should stop whaling. In our control group, the mean outcome value, measured on a 1 to 6 scale where higher values denote stronger agreement, is 2.73.

3.1.3 Hate speech and CERD

While hate speech laws have become commonplace among OECD nations, there are significant differences in enforcement and compliance. Japan passed an anti-hate speech law in 2016, motivated by racist protests and demonstrations by extreme-right organizations—dubbed the “Action Conservative Movement” (ACM)—against ethnic Koreans, Taiwanese, and Chinese who are long-term residents of Japan (Smith , 2018). However, this law does not impose legal penalties on hate speech itself, and its salience remains somewhat low. In their text analysis of media coverage of hate speech incidents, Merklejn and Wiślicki (2020) find that newspapers point to official statistics that show an increase in such crimes, but they rarely delve into underlying structural causes. Conservative outlets primarily treat these as isolated criminal cases, rather than as manifestations of deeper, societal issues. Even ostensibly center-left newspapers relegate hate speech incidents to the back page.

The lack of legal penalties has been criticized as being in violation of the International Convention on the Elimination of All Forms Racial Discrimination (CERD), of which Japan has been a member since 1995. In a 2013 report, its Committee on the Elimination of Racial Discrimination noted that Japan’s lack of a comprehensive law banning hate speech prevented victims of racial discrimination from seeking legal relief. This was reiterated in a 2017 report, which demanded that Japan pass specific legislation banning direct or indirect racial discrimination, including hate speech (Committee on the Elimination of Racial Discrimination , 2013; Committee on the Elimination of Racial Discrimination , 2017).

We include one survey item for this practice: whether the respondent thinks that hate speech should be criminalized in Japan. As a baseline, the mean outcome value, measured on a 1 to 6 scale where higher values denote stronger agreement, in our control group is 4.04.

3.1.4 Death penalty and ICCPR

Japan’s criminal justice system currently allows for capital punishment, making it one of two OECD countries—along with the United States—that both permits and carries out death penalties against civilians. Notably, Japanese executions are carried out by hanging. Both the practice and its method have been protested by the European Union and human rights NGOs such as Amnesty International (European Union External Action Service , 2018; Amnesty International , 2019). Although Japan has not signed the second optional protocol of the International Covenant on Civil and Political Rights (ICCPR) that abolishes the death penalty, the manner in which executions are implemented has been criticized as a violation of the ICCPR.

In a comparative analysis of Japanese, Chinese, and American public opinion, Jiang et al. (2010) find that support for the practice is lowest in Japan, but also that belief in its deterrent effects is highest. However, Muramatsu et al. (2018)’s time series analysis demonstrates that neither death sentences nor executions reduces homicides. There is significant secrecy and silence surrounding the actual timing of executions (Johnson , 2006) not to say sentencing decisions (Johnson , 2013), producing a public that is both uninformed and misinformed about its effects.

We include one survey item for this practice: whether the respondent thinks that the death penalty should be abolished in Japan. The mean outcome value, measured on a 1 to 6 scale where higher values denote stronger agreement, in our control group is 2.48. It evinces the lowest level of support among our four dependent variables.

3.2 Treatment

Before asking questions about these items, we assign respondents to one of the four following groups randomly. The first group receives information about international law violations in general, which is in line with the design of most previous studies. We create two additional groups who receive information that emphasizes the legal nature of international law and information about common practices in other countries separately. To emphasize the domestic linkage of international law, the second group receives information that highlights constitutional obligations to comply with international law, in addition to the information about international law violations. By contrast, the third group receives information about common practices in other countries, without being informed about international law violation. The fourth group is our control, with no additional information.

The first group is informed that domestic rules and practices have been criticized as violations of specific international laws. For instance, in the case of the mixed-surname marriage, the respondents first see basic information about the current policy: “In Japan, married couples must use a common surname.” Then, they see the following text: “However, this system is said to be in violation of the ‘Convention of the Elimination of All Forms of Discrimination against Women’, of which Japan is a member.” This treatment measures whether information of international law violations changes public opinion, but it does not allow us to differentiate the underlying mechanisms. In other words, we expect this information to imply both a violation of legal obligations and a deviation from common global practices. We label this the “International Law” treatment.

The second group receives information that emphasizes the legally-binding nature of international law at the domestic level. In addition to the information about international law provided in the first group, the second group receives additional information that Article 98.2 of the Constitution of Japan mandates compliance with international laws. This clause stipulates, “The treaties concluded by Japan and established laws of nations shall be faithfully observed”; as noted earlier, this type of procedural requirement is relatively common in national constitutions.⁹ This information is included to highlight the domestic link of international law. In the case of mixed-surname marriage, the respondents see the following text: “Despite the fact that the Constitution of Japan stipulates the obligation to observe international treaties, this practice is said to be in violation of the ‘Convention on the Elimination of All Forms of Discrimination against Women’, of which Japan is a member.” We label this the “Constitution” treatment.

Note that this constitutional information is about the procedural obligation to comply with ratified international laws, not that the domestic policy is in direct violation of constitutionally-enumerated rights. In this sense, our treatment is different from that in Chilton and Versteeg (2016), where one of the treatments states that the use of torture substantively violates the U.S. constitution. We focus on the procedural element for two reasons. First, it is factually appropriate with respect to our four cases, insofar as the Supreme Court has never ruled that the constitution prohibits these status quo provisions and practices. Second, we believe it emphasizes the legal aspect of international law in a relatively matter-of-fact way, without provoking a partisan or ideological response.

This latter point is crucial, because many democracies venerate the constitution as the supreme law of the land and backbone of the polity. We avoid an explicit suggestion that the constitution includes a direct proscription of the current practices, which may cause respondents to view them more negatively regardless of their status with respect to international law. Of course, public opinion towards national constitutions is rarely uniform, including in Japan. While the constitution has been held in high esteem as the cornerstone of post-WWII democracy (Sakaiya , 2017; McElwain , 2021), conservative actors, including the long-ruling Liberal Democratic Party, have accused it of being an outdated relic of the postwar Allied Occupation (Winkler , 2012). However, the most heated criticism of the constitution has centered on Article 9’s “Peace Clause”, which disavows war and proscribes the establishment of a military. By contrast, Article 98, which discusses international law, has received scant attention in political discourse.¹⁰ In sum, we designed our treatment to avoid priming respondents’ prior beliefs about the merits of the constitution itself, but we later discuss whether party identification, which is correlated with constitutional views, moderates the treatment effect.

The third group receives information about common practices that notes the rarity of Japanese practices in peer countries. Importantly, respondents in this group do not receive information about treaty violations. In the case of mixed-surname marriage, the respondents see the following text: “However, more than 80% of OECD countries do not have such a system.”¹¹ This treatment is purely informational: it conveys facts about the rarity of Japanese practices globally, but it does not mention treaty violations or include any normative judgments. At the same time, it leaves out any connotation of legal obligations. Table 1 describes the treatment and control texts in greater detail.

Table 1 An example survey question, translated from Japanese (bolded emphasis in original)

What does this treatment design tell about how international laws influence public opinion? Most importantly, comparisons between the first two treatments that mention legal obligations and the third treatment that only includes common practices point to the relative importance of our two key dimensions of international law. If the third treatment causes a stronger change in respondents’ attitudes than the first and second treatments, we can conclude that the most important driver for opinion change is the desire of citizens to conform to global practices. By contrast, if the third treatment elicits no change while the first or second treatment does, then this suggests that the legal element is an essential facet of international law. Furthermore, a comparison of the first and second treatments points to the specific pathways through which legalization affects public opinion. If both are statistically and substantively significant, we are agnostic about the exact reasons why citizens care about legal obligations. By contrast, if only the second treatment that references constitutional obligations is effective, we believe that the explicit linkage between the international and domestic legal spheres is critical for international legal obligations to have consequences on domestic public opinion.¹²

We should note that other ways to operationalize the informational role of international law were considered. One was criticism from a specific country, such as a major ally or superpower, and another was recommendations from international organizations or NGOs, which earlier research has flagged as being relevant (Linos , 2011). However, we chose to use factual information about practices in peer countries for the following reasons. First, providing information about the identity of the critic can confound respondents’ opinions. For instance, people who have favorable attitudes towards the U.S. may be more likely to change their opinion when the U.S. criticizes the pertinent domestic policy, but the same criticism may generate backlash among those who view the U.S. negatively. These would be the effect of affinity towards the U.S., not of international law per se. Second, while we ask for respondents’ opinions about multiple policies, we could not identify real examples where the same actor criticized Japanese policies across domains consistently. For example, the U.S. has criticized Japan’s whaling activities but not its death penalty. Third, explicit foreign disapproval or suggestions about “ideal” behavior would incorporate normative connotations into our treatment. While the normative dimension is important, it would deviate from our design goal of distinguishing the legal versus informational functions of international law.

Another potential treatment we considered but did not implement was priming respondents about the “soft law” aspect of international law. While hard law, which binds parties and can be enforced in courts, tends to define which specific practices are prohibited and/or permitted, soft law, which denotes agreements and principles that are not legally binding, only contains general statements in a precatory way (Abbott & Snidal , 2000). Although exploring how hard and soft laws differs in the eyes of the public is an interesting research avenue, we chose to focus on treatments with clearer distinctions between law and practices, given the breadth of possible wordings and contents that soft law can take.¹³ Moreover, similarly to our reason for not using criticisms or recommendations from specific actors, we were not able to find existing international laws that were equally “soft” across the practices we include in our survey. Accordingly, our common practices treatment was designed to minimize legal or normative connotations, which is at least an implicit element of both hard and soft laws.

3.3 Treatment effect heterogeneity based on individual beliefs

The effect of our three treatments should vary based on respondents’ prior beliefs about the value of international law and the consequences of non-compliance. These may relate to the meaning they attach to international law, such as their identification with global civil society, and their support for the domestic government that has preserved the controversial domestic practices. To explore these differences, we include two related questions in the pre-treatment section of our survey instrument. The distributions of the outcome variables conditional on the responses of the pre-treatment questions, as well as the raw distributions of the conditioning variables themselves, are reported in the Appendix.

First, previous studies have found that a “cosmopolitan identity” can be an important moderating factor. Bayram (2017) demonstrates that legislators with a cosmopolitan social identity are more likely to believe in the legitimacy of international law and prefer compliance. We replicate this at the level of voters by asking respondents whether they identify as a “citizen of the world”.¹⁴ This survey instrument was taken from the World Values Survey (Inglehart et al. , 2014).

Second, partisanship is also reported to be an important source of heterogeneous effects (Chilton & Versteeg , 2016; Wallace , 2019; Chilton , 2014). Most domestic practices in Japan have been implemented—or at least remained unopposed—under the conservative Liberal Democratic Party (LDP), which has been in government for all but four years since its founding in 1955 (circa 2020). This can generate two complementary effects. First, respondents who support the LDP may be more conservative, and thus less likely to be convinced by global trends or international laws on ideological grounds. Second, they may be less likely to be swayed by implicit criticism of their favored party, under which these controversial practices were fostered or tolerated. While we do not aim to disentangle the motivations proxied by partisanship, our survey instrument asked respondents for their favored party. In our analysis, we examine the conditional effect of the treatment on people who support the LDP versus those who do not.

3.4 Sampling

Our survey experiment was conducted on August 5th–10th, 2020. Respondents were recruited through Nikkei Research, one of the largest survey vendors in Japan. We employed quota sampling by age (20-69), gender, and region to match the most recent national census distribution in 2015. 3212 respondents participated in the survey, and the final sample size in our analysis, after excluding incomplete or irregular responses, was 2954. We used a block randomization scheme, based on the respondents’ gender (male or not) and party identification (supporters of the LDP, of other parties, or independents). Both conditioning variables used in our analysis—identification with global civil society and party identification—were asked prior to the experimental treatment.

4 Results

The results of our survey experiment can be summarized as follows. First, we find evidence that the constitution treatment, which reinforces the legal aspect of international law, changes attitudes about some policy items. By contrast, we did not find any evidence that information about practices in peer countries does the same. This suggests that the legal obligations of international law have a stronger effect on public opinion than the common practices cue. Second, we find heterogeneous treatment effects across items based on prior beliefs. Respondents with stronger identification with global civil society and non-LDP supporters are more likely to be influenced by the constitution treatment.

4.1 Average treatment effect

Figure 1 shows the estimated average treatment effects of the basic international law information (top row), the international law with additional information about constitutional obligations (middle row), and the common practices cue about the rarity of Japanese practices among OECD countries (bottom row). The columns correspond to the policy items in our experiment: mixed-surname marriage, whaling, hate speech, and death penalty. In each column, we show point estimates with the 95% confidence intervals with and without multiple testing correction. The left bars show estimates without corrections. The point estimates are the difference-in-means estimates, and the confidence intervals are based on standard errors accounting for the block randomization. For the right bars, we use the adaptive shrinkage method (Stephens , 2017; Stephens et al. , 2020) to correct for multiple hypothesis tests. In addition, we conducted the Benjamini-Hochberg (BH) procedure with the false discovery rate at .05 (Benjamini & Hochberg , 1995). “BH” is added to an estimate for which the null hypothesis is rejected after the BH procedure. The raw distribution of control group responses can be found in the Appendix.

Fig. 1

Estimated average effect of the International Law (top row), Constitution (middle row), and Practices treatments (bottom row). The columns correspond to the survey items: mixed-surname marriage, whaling, hate speech, and death penalty. In each column, the left bars present the difference-in-means estimates and the 95% confidence intervals without multiple testing corrections. The right bars show the results using the adaptive shrinkage method (Stephens , 2017). Statistically significant estimates are shown in black; those that are not are in gray. The dependent variables range from 1 to 6 (Strongly Disagree to Strongly Agree). The International Law treatment has a significant effect on the hate speech item, and the Constitution treatment has significant effects on the mixed-surname marriage and whaling items. The Practices treatment does not have significant effects on any items. The Benjamini-Hochberg procedure rejects the null hypothesis of no effect for the Constitution treatment on the whaling item only

The main result is that the Constitution treatment has statistically significant effect estimates on some items, while the International Law or Practices treatments do not, after multiple testing correction. For example, the effect of the International Law treatment is estimated to be 0.13 on the hate speech item. While this is statistically significant at the 5% level, it is not after performing multiple testing corrections. Estimates for the Constitution treatment are significant on two items: the legalization of mixed-surname marriage (0.16) and the prohibition of whaling (0.20), and the significance on the whaling persists after multiple testing correction. Regarding the hate speech item, we cannot reject the null hypothesis that the Constitution treatment has no effect, though the difference between the International Law and Constitution treatments is negligible. In contrast to the first two treatments, the Practices treatment does not have any effects significantly different from zero on any of the items. However, all point estimates are either positive or very close to zero, which is the expected direction.

The magnitude of these positive and statistically significant estimated effects is substantively meaningful. For example, the estimated effect for the whaling item (0.20) means that the average response on a six-point Likert scale in the treatment group is higher than the control group by 0.2. To interpret this number in terms of potential outcomes, consider a hypothetical profile of potential outcomes in which the treatment effect on each respondent is either zero or one. The average effect of 0.2 in this scenario implies that the treatment shifts the preferences of 20% of respondents upwards by one point on a six-point scale. In general, it is not easy to change the opinion of 20% of voters simply by providing information.

What do these results connote for our understanding of international law and public opinion? Their most important implication is that information about treaty violations nudges public opinion to favor domestic reforms primarily when respondents are told that the constitution obliges compliance. By contrast, respondents do not appear to be sensitive to information about common practices in peer countries. This suggests that legal obligations, not information about common practices, are the key feature that drives international law’s compliance pull. Furthermore, our results also suggest an important qualification to the power of legal obligations. Our first treatment that noted international law violations did not produce statistically significant changes in public opinion, while our second treatment that explicitly stressed constitutional mandates did. This implies that people may need to be reminded that their own constitution obligates their country to abide by international law. Put differently—and perhaps pessimistically—many citizens may not be fully aware of the domestic legal significance of international law. We return to this point in Section 6.

This result is consistent with most studies conducted in the U.S., but goes against some experiments conducted outside the U.S. While we also explore whether backlash effects exist among respondent subpopulations later, our main result supports the hypothesis that international law shifts public opinion positively rather than negatively, at least at the level of the general population.

Overall, none of the three treatments has a statistically significant estimate on preferences about the death penalty. The null effect on the death penalty may be because of its low political salience in Japan, or because of voters’ strong prior preferences for maintaining the practice. In our survey, the average support for the abolition of the death penalty in the control group is 2.48, which is the lowest among the four dependent variables. 27.1% of control group respondents expressed strong disagreement (1 on the 6-point scale) with abolishing the death penalty, which is the highest proportion among all items.

In order to corroborate our finding that information about constitutional obligations reinforces the effects of international law, Fig. 2 shows the difference-in-means between the effects of the Constitution and International Law treatments (as opposed to the difference between each treatment arm and the Control group) on the four items. We find that all point estimates are either positive or close to zero, and the positive estimate of the effect on the death penalty item is statistically distinguishable from zero without multiple testing correction. Although not all items evince statistically significant differences, these results support the argument that additional constitutional cues about the procedural requirement to follow international laws can strengthen the effects of information about international law violations.

Fig. 2

Estimated difference between the effects of the International Law and the Constitution treatments. The columns correspond to the survey items: mixed-surname marriage, whaling, hate speech, and death penalty. In each column, the left bars present the difference-in-means estimates and the 95% confidence intervals without multiple testing corrections. The right bars show the results using the adaptive shrinkage method (Stephens , 2017). Statistically significant estimates are shown in black; those that are not are in gray. The dependent variables range from 1 to 6 (Strongly Disagree to Strongly Agree). All point estimates are either positive or close zero, although the only statistically significant estimate is for the death penalty item without multiple testing corrections

4.2 Heterogeneous effect

Respondents’ underlying sentiments about law and politics are likely to moderate their sensitivity to our treatments. In this section, we explore the conditional effects of information about international law violations on two dimensions: identification with global civil society and party identification. Overall, we find that both factors are relevant.

First, respondents’ identification with global civil society is an important moderator of the treatment effect. Figure 3 presents the estimates of the conditional treatment effects by respondents’ identification with global civil society. Our survey instrument asked respondents whether they identified as “a citizen of the world” on a 1 to 6 scale, which we then dichotomized.¹⁵ Our results confirm those from earlier research (Bayram , 2017)people with stronger identification with global society are more likely to be influenced by the treatments. In particular, the International Law treatment has a significant and positive estimated effect on the hate speech item, and the Constitution treatment does so on the mixed-surname, whaling, and hate speech items. The significant estimated effects of the Constitution treatment on the mixed-surname and whaling persist after multiple testing correction.

Fig. 3

Estimated treatment effects conditional on identification with global civil society. The columns correspond to the survey items: mixed-surname marriage, whaling, hate speech, and death penalty. For each item, the left (right) two bars show the conditional treatment effect among people with higher (lower) identification with global civil society. The unfilled symbols present difference-in-means estimates with the 95% confidence intervals without multiple testing corrections. The filled symbols show the results using the adaptive shrinkage method (Stephens , 2017). Statistically significant estimates are shown in black; those that are not are in gray. The dependent variables range from 1 to 6 (Strongly Disagree to Strongly Agree). People with stronger identification with global society are more likely to be influenced by the treatments

Second, partisanship also moderates the treatment effect. This dimension is widely explored in the literature, particularly in the U.S. context. To the extent that individuals sort into parties based on ideological leanings and/or take policy cues from party elites, we can expect variation in treatment effects by partisanship. Figure 4 shows the estimates conditional on party affinity. The “LDP” column indicates the treatment effect among supporters of the conservative Liberal Democratic Party, while the “Other” column includes all other respondents. While none of the estimates is significant on LDP partisans, there are some notable patterns among non-LDP supporters. The International Law treatment is estimated to have positive and significant effects for non-LDP supporters on the hate speech item, and estimates for the Constitution treatment are positive and significant on mixed-surname marriage and whaling. The significant estimated effect of the Constitution treatment on the whaling item persists after the adaptive shrinkage correction, but none of the estimated effects are statistically significant after the BH correction.

The stronger effect among non-LDP respondents is consistent with previous survey experiments in the U.S., which found that liberals are more likely to change their opinions when shown information about treaty violations (Wallace , 2013). In the Japanese context, our results can be interpreted in two ways. First, LDP supporters are more likely to be conservative, and thus may be less amenable to liberal policy changes such as the legalization of mixed-surname marriages or the penalization of hate speech. Second, LDP supporters may oppose significant changes to the status quo to avoid cognitive dissonance, given that current policies were established or tacitly accepted by the long-ruling LDP. In either case, the Constitution treatment makes even LDP supporters more likely to support liberal policy changes on items other than the death penalty, though the estimates are not statistically significant. As the governing party, the LDP has blocked the proposals of these policy reforms, but our results suggest that its supporters may be amenable to pro-revision arguments when informed of the state’s legal obligations.

Collectively, these tests of conditional average treatment effects further confirm that legal obligations, not conformity to common practices, are the primary channel through which information about international law violations stimulates voters to change their views of status quo practices. The Constitution treatment prompts greater support for legal reforms among those who self-identify with global civil society (Fig. 3) and do not support the governing Liberal Democratic Party (Fig. 4). These findings are, to a large extent, intuitive. People who share the values underlying international law and who disapprove of the government’s tolerance of controversial Japanese practices are more sensitive to accusations that domestic practices run afoul of international legal commitments.

Our non-findings are arguably more consequential to our understanding of public opinion towards international law. First, the Practices treatment does not alter people’s views consistently (Fig. 1). Second, we do not observe strong backlash effects (significant, negative coefficients) from any treatment or subgroup. This suggests that people do not reflexively prefer the status quo because of implied international criticism, at least among the practices we examine in this paper.

Fig. 4

Estimated treatment effects conditional on partisanship. The columns correspond to the survey items: mixed-surname marriage, whaling, hate speech, and death penalty. For each item, the left (right) two bars show the conditional treatment effect among the LDP (non-LDP) supporters. The unfilled symbols present difference-in-means estimates with the 95% confidence intervals without multiple testing corrections. The filled symbols show the results using the adaptive shrinkage method (Stephens , 2017). Statistically significant estimates are shown in black; those that are not are in gray. The dependent variables range from 1 to 6 (Strongly Disagree to Strongly Agree). Non-LDP supporters are more likely to be influenced by the treatments than LDP supporters

5 Discussion

Several issues regarding our results deserve greater scrutiny. First, why does factual information about common practices have no effect on Japanese public opinion? Our expectation was that awareness of their country’s deviation from peer practices would be enough to elicit conformity, particularly among people with strong cosmopolitan identities. However, this was not the case, at least for the issues examined in our survey. One plausible explanation for this null result is that for common practices to be interpreted as a normatively positive “global standard”, deviation needs to be framed negatively. Our treatment intentionally did not attach any judgment to being an outlier in order to test the purely informational effect of status quo practices among peers. As a result, some respondents may have inferred that Japanese laws, while unusual, were not necessarily inferior. This interpretation would not produce greater favorability towards adapting current policies to match that of their peers, absent any binding legal or normative commitments. Liu et al. (2022) report similar null effects for information connoting that Japan is an outlier, albeit in the case of potential reforms to imperial succession rules.

Another possible reason for the null effect is the reference peer group from which Japanese practices deviate. As discussed in Section 3.2, we avoided specifying country names in any of our treatments to keep our statements consistent across the four policy items; we were unable to find a single actor that consistently criticized Japan for international law violations across domains. That said, preferences may have been more responsive—whether positively or negatively—had the reference group explicitly included Japan’s regional competitors, such as China or South Korea, as opposed to Western countries or the United States. Future research may want to explore the effects of information suggesting deviation from specific countries, such as allies versus rivals.

A second issue is differences in estimated treatment effects across the four issue areas we considered. While the Practices treatment was consistently insignificant across all items, the International Law and Constitution treatments saw some variation in results across issues and estimation strategies. What explains these differences between mixed-surname marriage, whaling, hate speech, and death penalty? First, the null effect on the death penalty item is likely due to strong prior preferences. As discussed in Section 3.1.4, previous surveys have noted that Japanese citizens believe that the death penalty is an important deterrent on violent crime (Jiang et al. , 2010). This conjecture is consistent with low levels of baseline support for abolishing the death penalty among our control group (2.48 out of 6 points). Our experimental treatments were all statistically insignificant, except for the difference between the Constitution and International Law treatments in Fig. 2, and the point estimates were close to zero or even negative for some treatments.

By contrast, the treatment effects were consistently significant on the whaling item; the positive effect of the Constitution treatment was robust even with different multiple testing corrections, as shown in Fig. 1. One likely reason is the issue’s high domestic salience, particularly in relation to diplomatic concerns. The ICJ ruling against Japan was widely reported in Japanese media, and the Japanese public is likely to be aware that whaling caused diplomatic tensions with certain countries, as discussed in Section 3.1.2. Respondents may have been more sensitive to this topic, because they are aware that violations of relevant international agreements can complicate relations with allies and partners. Another reason relates to the lack of a ceiling effect. The baseline response for the whaling item is skewed in a negative direction, meaning that the Japanese public tends to be against the prohibition of whaling. Given that we expected the treatments to cause a positive shift, there was more room for respondents’ preferences to change.

The other items—mixed-surname marriage and hate speech—are likely to lie in the middle of these two issues. The baseline distributions for these items are both skewed positively, suggesting a possible ceiling effect. In addition, these two items are reported in the news media periodically, but they are often framed as a purely domestic matter. While people may be responsive to the informational treatments due to the moderately high salience of these issues, they may not be particularly sensitive to information from international sources.

In sum, the four items vary in terms of the direction and strength of prior preferences, as well as their salience in relation to diplomatic concerns. These variations deserve further scrutiny in future research, as they suggest the limits of international “naming and shaming” strategies. First, accusations of international law violations may not have symmetrical effects across issue areas in the same country, because some policy preferences are more ingrained than others. Second, we may observe asymmetrical treatment effects on the same topic across countries because of differences in issue salience and sensitivity to diplomatic criticism.

6 Concluding remarks

The determinants of state compliance with international law is a long-standing debate in international relations. Recent studies have focused on domestic mechanisms through which governments can be pressured to change status quo policies to conform with international laws. One critical factor is public support for compliance, but the mechanisms that drive voter preferences remain unclear. Our experiment shows that citizens prefer policies that comply with international laws, not because they value conformity with common practices, but because they think the government should abide by legally binding commitments. Information that emphasized the government’s constitutional obligation to comply with international law made respondents more supportive of revisions to current practices that violate international treaties. Given that more than 90% of national constitutions have such provisions, our results are relevant to the vast majority of polities around the world. The effect of this Constitution treatment is stronger among those who identify with global civil society, and thus may be more respectful of international law, or who do not support the long-ruling conservative party, under which status quo practices were maintained. Put differently, voters—at least in Japan—do not necessarily care about conforming with globally-common practices per se; instead, they want their government to honor legal commitments.

Our results also speak to the growing literature on the soft law aspect of international law. We show that emphasizing the constitutional status of international laws can produce significant changes in citizens’ preferences. One derivable insight is that international agreements that have the status of “law” and are approved domestically have more bite on domestic public opinion. That is, “soft” international laws may be more powerful than often assumed. As long as international laws are formally ratified and are given protected legal status by domestic constitutions, citizens prefer to abide by them regardless of the law’s specific contents. However, informal statements and declarations that do not require formal ratification may not have as strong of an impact, since their legitimacy is not backed by the constitution. Since our experiments are agnostic about the specific content of the laws, future research needs to explore how hard and soft laws affect public opinion differently.

As is the case with other survey-based studies, ours is not without limitations, although we believe these identify what steps should be taken next to improve our understanding of a critical research field. First, further work is required to examine public opinion on the specific relationship between constitutions and international law. On the one hand, our results imply an important qualification to the power of international law to shift public opinion. While our first treatment with international law per se did not lead to statistically significant effects across issue areas, our second treatment that added constitutional mandates did so on some. This suggests that for international law to generate public support for reforms, its legal legitimacy needs a supplementary boost from a domestic source whose authority is broadly accepted, i.e. the constitution. On the other hand, we must be careful in how we interpret the constitution effect. Our treatment took pains to emphasize that the constitution procedurally required compliance with international law. However, respondents may nevertheless have interpreted the treatment as a claim about substantive, rather than procedural, constitutional violations. In the survey context, we believe that most respondents know that the controversial policies are not substantively unconstitutional. For instance, Supreme Court rulings that upheld the current civil code’s requirement of same-surname marriage have been widely reported in Japanese media. Still, we cannot completely discount the possibility that the respondents interpreted the treatment in a different manner. Future research should be able to address these points by using a new treatment design that facilitates a more specific interpretation.

Second, our experiment, much like most others on public attitudes toward international law, is conducted within a single country. Although our results confirm U.S.-based findings that citizens prefer compliance with international law, we cannot attest to whether the legally-binding nature of commitments are the primary motivation underlying public support in other countries. Another issue is that the effect of the “constitution” treatment may vary across countries, depending on people’s veneration of the supreme law. However, there is, to our knowledge, limited cross-national work on attitudes towards constitutional performance. To address this question, future research involving similar experiments in other countries is warranted.

Similarly, further work on how “common practices” shape public preferences is necessary. Our experiment found no effect of information that emphasized the atypicality of Japanese practices among other industrialized countries. As discussed in Section 5, we purposefully chose not to signal normative claims in the treatment, such as accusations of international law violations from foreign nations, NGOs, or expert commissions. While our intention was to identify the pure effect of Japan being described as an outlier, the Japanese public may react differently depending on who they think is setting “global standards”. It may be useful to consider including different versions of this treatment in future experiments, particularly when conducted in non-Japanese contexts. For example, comparisons to regional neighbors or to those who share similar colonial histories may be more relevant in other countries.

On a final note, unlike some previous experiments in non-US countries, we did not observe a significant backlash against international law. On all items, none of the treatments significantly increase respondents’ opposition to compliance. This is true even for whaling: the literature has argued that international criticism has been seen as an attack on Japan’s traditional culture, but in our experiment, information that Japanese practices violated international law did not increase support for the status quo. Unfortunately, neither this study nor previous scholarship has further evidence on why domestic publics sometimes react negatively to international law. This is a crucial question that future studies on public attitudes toward international law are encouraged to address.