Sovereignty, Substance, and Public Support for European Courts’ Human Rights Rulings

Is the public backlash against human rights rulings from European courts driven by substantive concerns over case outcomes, procedural concerns over sovereignty, or combinations thereof? We conducted preregistered survey experiments in Denmark, France, Poland, Spain, and the United Kingdom using three vignettes: a foreigner who faces extradition, a person fighting a fine for burning Qurans, and a home owner contesting eviction. Each vignette varies with respect to whether a European court disagrees with a national court (deference treatment) and whether an applicant wins a case (outcome treatment). We find little evidence that deference moves willingness to implement judgments or acceptance of court authority but ample evidence that case outcomes matter. Even nationalists and authoritarians are unmoved by European court decisions as long as they agree with the case outcome. These findings imply that nationalist opposition to European courts is more about content than the location of authority and that backlash to domestic and international courts may be driven by similar forces.


INTRODUCTION
T he European Court of Human Rights (ECtHR) and the Court of Justice of the European Union (CJEU) have come under intense public scrutiny over some of their human rights rulings. First, critics charge that European courts have interfered on issues that should be decided by domestic institutions. This is a procedural argument about the proper location for authority. Second, critics have condemned European courts on substantive grounds-for example, because courts protect the rights of immigrants who have been convicted of a crime. Although these critiques can be complementary, they are conceptually distinct and they have different implications. However, the rapidly growing literature on backlash against international institutions has not yet disentangled the extent to which people object to international authority on procedural or substantive grounds.
We conducted preregistered survey experiments in Denmark, France, Poland, Spain, and the United Kingdom to disentangle these motivations in the context of human rights jurisprudence. 1 We used three case vignettes in which an individual contests a government action: a foreign criminal who faces expulsion, a person fighting a fine for burning the Quran, and a home owner contesting eviction. Each vignette had two treatment conditions: whether a European court rules differently from a national court (deference treatment) and whether an applicant wins or loses (outcome treatment). We evaluated whether the treatment affects agreement with the decision, support for compliance, and support for European Court authority. Moreover, we examined whether individuals who strongly identify as exclusive nationalists respond differently to disagreement between European and national courts.
We find substantial outcome treatment effects. However, with the partial exception of Denmark and the United Kingdom, we find no significant procedural sovereignty effects. Most respondents appear to be just fine with a European court disagreeing with a national court as long as they get the outcomes they like. This is equally, and sometimes more, true for those who are generally skeptical of European institutions, such as individuals with exclusive nationalist and authoritarian leanings. These findings are consistent with the idea that the correlation between exclusive national identity and support for European institutions runs through the effect these institutions are perceived to have on immigration and other policies that ethnocultural nationalists dislike rather than an innate preference for national over European institutions.
These findings shed light on two theoretical debates. First, scholars have long contested whether the subjective legitimacy of (international) institutions, including courts, is driven by satisfaction with outcomes or procedural fairness (Caldeira and Gibson 1995;Dellmuth and Tallberg 2015;Dellmuth, Scholte, and Tallberg 2019;Esaiasson et al. 2019;Follesdal 2020;Tallberg and Zürn 2019;Tyler and Rasinski 1991). Most empirical studies focus on procedural features such as transparency, impartiality, or representativeness. Instead, we examine a procedural principle that features prominently in multilevel governance: subsidiarity. Subsidiarity prescribes that the most proximate competent institution should decide controversial issues. However, we find that support for compliance and court authority are unaffected by a European court overruling a national court even among respondents in consolidated democracies and among those who are satisfied with their national legal systems.
To the best of our knowledge, we are the first to examine this question for international institutions. The closest is a study that experimentally manipulates the institutional design of a global climate conference (Anderson, Bernauer, and Kachi 2019). Other experimental studies of backlash against courts focus on framing effects (Dinas and Gonzalez-Ocantos 2021; Gonzalez-Ocantos and Dinas 2019; Zvobgo 2019; Zvobgo and Chaudoin 2021). In addition, there is an experimental literature that finds that policy preferences motivate support for international agreements more than concerns about consistency and reciprocity (Beiser-McGrath and Bernauer 2019;Chaudoin 2014).
A second contribution is to the literature on the nationalist and authoritarian backlash against globalization and international institutions (Ballard-Rosa et al. 2021;Colantone and Stanig 2018;Copelovitch and Pevehouse 2019;de Vries 2018;Hobolt 2016;Jensen, Quinn, and Weymouth 2017;Mutz 2018). Our findings help disentangle the sources of nationalist opposition to international institutions. International relations scholars often use the term "sovereignty costs" to denote both the procedural and policy losses of delegating authority to international human rights institutions (Hafner-Burton, Mansfield, and Pevehouse 2015; Moravcsik 2000). Public opinion scholars also often conflate the two. For example, Pippa Norris and Ronald Inglehart recognize nationalism, mistrust in global governance, opposition to immigration, and authoritarian values as separate variables that together define the cultural values that are associated with support for populist right-wing parties and opposition to globalization (Norris and Inglehart 2019).
Consistent with the literature, we examine how an exclusive national identity affects support for European institutions (Hooghe and Marks 2004;2005). If an exclusive nationalist identification is rooted in an affection for a territorial political unit, then we would expect that nationalists object to international court involvement. Yet, if exclusive nationalism is more about an identification with ethnicity and/or cultural values, then we expect that nationalists willingly support the authority of a European court that limits the rights of immigrants or otherwise reaffirms cultural values, even if that means overruling a national court. Our evidence is consistent with the view that nationalist opposition to European human rights adjudication is more about the content of decisions than the location of authority. Nationalists (and authoritarians) oppose implementation of both national and international human rights judgments that they dislike.
The implications of our study are limited by the specific substantive context of our vignettes. We chose our vignettes to mimic the kinds of human rights cases that tap into conflicts over social and moral values. This is justifiable in a study of public backlash. If we are interested in examining whether nationalist opposition to European institutions is driven by an innate preference for national over European institutions or outcomes that ethnocultural nationalists dislike, then we must examine issues that trigger ethnocultural identity. We did find similar patterns on the eviction vignette, which taps into long-standing allegations of the CJEU's neoliberal bias. Yet, our findings may well differ if we had focused on (non-human-rights) regulatory issues or other issues where people have weaker predispositions.
A second limitation is that our vignettes jurisdictionally resemble the ECtHR rather than the CJEU. Our vignettes set up a potential conflict between an international and a national court. In the ECtHR, applicants must first exhaust domestic remedies before bringing their cases. This means that a violation finding by the ECtHR always implies a disagreement with a national court. CJEU rulings do not directly agree or disagree with national courts. Some CJEU judgments on fundamental rights do in practice go against legal interpretations suggested by the referring courts or preexisting national legal practice. For example, the CJEU's Junqueras judgment (on the parliamentary immunity of a pro-Catalan independence politician convicted for sedition) was widely criticized in Spain for ruling in direct contradiction with the Spanish Supreme Court's interpretation of EU law and resulted in a drop in public support for the European Union (Turnbull-Dugarte and Devine 2021). Although we draw inspiration from controversies surrounding CJEU fundamental rights cases, the vignettes themselves are modeled on the ECtHR.
We first offer background for debates about public support for European courts. We then derive testable hypotheses from theoretical frameworks before introducing the research design and data. After presenting the results, the conclusion returns to the institutional implications of our findings and possible extensions to other institutional settings. politicization includes European courts. Especially, human rights litigation has become controversial as domestic and international courts increasingly resolve high-salience disputes, including cases about identity, immigration, and economic redistribution (e.g. , Helfer 2018;Hirschl 2008;Sandholtz, Bei, and Caldwell 2017).
The ECtHR offers direct access for individuals in all 47 Council of Europe (CoE) member states. The Court issued only 200 judgments between 1960 and 1990 but well over 20,000 judgments since. All CoE member states have incorporated the European Convention on Human Rights into domestic law. As explained above, ECtHR violation findings imply a disagreement between a national and an international court over whether a government action violated human rights.
The CJEU examines a more diverse set of cases, mostly concerning the single market and economic freedoms. Its human rights jurisdiction expanded with the adoption of the Charter of Fundamental rights in the 2009 Lisbon Treaty (Craig and De Búrca 2020). The CJEU hears most of its human rights cases through its preliminary references procedure. 2 A national court must refer a question of EU law to the CJEU if there is doubt about its interpretation (Broberg and Fenger 2014). While the CJEU never directly overrules national courts, the preliminary reference procedure creates situations in which conflicts between the CJEU interpretation and prevailing national judicial practice or preferred legal interpretations is apparent and controversial, as in the aforementioned Junqueras judgment. Such controversies should ultimately be resolved in favor of EU law due to the long-standing principle of the supremacy of EU law. Yet, some national apex courts have carved out exceptions to the autonomy and supremacy of EU law for fundamental rights (Gill-Pedro 2019; Mohay and Tóth 2017). Thus, both the location of authority and the content of fundamental rights judgments are subject to contestation in the CJEU context.
Politicians and other critics have increasingly accused the ECtHR of overstepping its authority (Flogaitis, Zwart, and Fraser 2013;Hofmann 2018;Madsen, Cebulak, and Wiebusch 2018;Popelier, Lambrecht, and Lemmens 2016;Stiansen and Voeten 2020). The CJEU has long been criticized for its alleged neoliberal bias-that it prioritizes the free movement of goods, services, labor, and capital over competing objectives, such as social policies and protections (Höpner and Schäfer 2012;Pollack 2017;Scharpf 1998). More recently, right-wing parties have adopted nationalist and authoritarian rhetoric to criticize human rights rulings on both European courts (Madsen 2020). This includes mainstream parties and government officials. For example, Theresa May stated repeatedly that the UK should leave the ECtHR (and not the EU), famously illustrating her pleas with the example of a case of an "illegal immigrant who cannot be deported because, and I am not making this up, 3 he had a pet cat." The UK Conservative Party (the governing party) promised in its 2015 manifesto to "curtail the role of the [ECtHR] so that foreign criminals can be more easily deported from Britain." 4 Similarly, the Spanish populist party Vox stated in its 2019 manifesto that "The institutions of the Union and especially the Court of Justice of the Communities have abused their competences" and proposes a new treaty to curb the court. 5 Vox's parliamentary spokesman called the Junqueras decision: "a stick to Spain, an attack on national sovereignty," 6 and the party threatened a Spaxit after the ruling. 7 In Madrid, street protests erupted over an ECtHR ruling that reduced the effective prison sentences of convicted ETA terrorists. 8 Governments have adopted a series of multilateral declarations from the Brighton Declaration (2012) to the Copenhagen Declaration (2018) that have urged the ECtHR to apply greater restraint in its jurisprudence (Helfer 2018). There is evidence that the Court has responded to especially the concerns from governments in consolidated democracies (Stiansen and Voeten 2020). Similarly, some scholars suggest that the CJEU has started to exercise more restraint in policy areas where both the public and member state governments are opposed to further legal integration, such as citizenship law (Blauberger and Martinsen 2020). Moreover, CJEU judgments are increasingly (and negatively) reported in the media (Blauberger et al. 2018;Dederke 2021).
That European courts receive increased scrutiny does not necessarily mean that people dislike the principle of European Court authority. It may be that some vocal groups are dissatisfied with the the outcomes of particular rulings, whereas they would appreciate European courts setting aside national court rulings on other matters. Our article examines this question.

Procedural Justice, Subsidiarity, and Deference
Procedural justice theories posit that people are willing to accept unpopular court decisions as long as these are made through a process that they perceive as fair (e.g., Tyler 2006;Tyler and Rasinski 1991). Procedural fairness reflects values that are unrelated to the outcome of cases, such as impartiality, transparency, and honesty (Tyler 2006, 7).
We examine procedural concerns about the location of authority. We first scrutinize the idea that people, on average, inherently value proximate national-level decision making, for example because they have an attachment to the nation state or because they feel more involved with or trusting of national-level decisions regardless of the outcome of these decisions. The assumption is often implicit in the statements of politicians. For example, Kenneth Clarke, Lord Chancellor and Secretary of State for Justice, said in the 2011 Izmir ECtHR reform conference that If the Strasbourg Court is too ready to substitute its own judgment for that of national parliaments and courts that have through their own processes complied with the Convention, it risks turning the tide of public opinion against the concept of international standards of human rights, and risks turning public opinion against the Convention itself.
Individual-level preferences for national over equivalent international institutions need not be unconditional. This idea is reflected in the subsidiarity principle, which prescribes that an international institution should only have authority when it can perform a task better and when more decentralized (national) units cannot reach decisions on their own (Follesdal 1998). This principle presumes a preference for national institutions, all things equal. The subsidiarity principle is written into the Treaty of the European Union-Article 5(3) of the Treaty on European Union (TEU) and Protocol (No 2)-and was added to the preamble of the European Convention on Human Rights through Protocol No. 15 (2013). 9 The ECtHR's newly elected President, judge Robert Spano, has argued that the court has entered an "age of subsidiarity" (Spano 2014), claiming that "in this process-based mechanism, the Court may grant deference if national decision-makers are structurally capable of fulfilling that task" (Spano 2018, 493). The 2018 Copenhagen Declaration mentions subsidiarity six times and states that the ECtHR should "[‥]not substitute its own assessment for that of the domestic courts, unless there are strong reasons for doing so." 10 The assumption that people value subsidiarity also underlies many international relations theories. For example, Andrew Moravcsik argues that international human rights courts help democratizing states' make their commitment to implement rights reforms more credible. By contrast, democracies with independent and effective domestic courts have no reason to accept intrusions by international courts (Moravcsik 2000). The implicit underlying behavioral theory is that backlash results from the unwillingness of people in developed democracies to accept the sovereignty cost of international courts overruling competent national courts.
Despite the centrality of the subsidiarity principle in European integration and in international relations theory, there is little research that explicitly examines its behavioral implications. We test two hypotheses. The first, baseline, deference hypothesis is that people are less willing to implement and support decisions reached through an international court disagreeing with a national court either because they inherently value proximate (national) authority or because they live in an established democracy where national courts are working well.
H1a (Deference): Support is lower if the European court disagrees with the national court.
The subsidiarity argument implies that H1a may be less pronounced in Poland than in the more consolidated democracies. But the subsidiary rationale could also operate at the individual level, which implies a heterogeneous treatment effect. Are respondents who look upon their domestic institution more favorably more affected by a European institution disagreeing with a national institution? This question has been studied for political institutions, although not experimentally. For example, Rohrschneider (2002) finds that citizens who perceive that their national democratic institutions are performing well have lower levels of EU regime support and are less likely to support reforms toward an EU-level government. We examine this in an experimental context for legal institutions.
H1b (Subsidiarity): The more satisfied an individual is with the domestic legal system the more support decreases if the European court disagrees with the national court.
Thus, whereas H1a tests whether individuals inherently prefer decisions that come from national courts, H1b examines whether support depends on individual assessments of whether national courts are competent.

Substantive Legitimacy
Substantive legitimacy theories posit that publics afford international legal institutions at best a minimal store of procedural legitimacy that would help controversial decisions gain acceptance (Caldeira and Gibson 1995;. Because people know very little about how these courts make decisions, trust tends to be derivative of trust in international institutions more broadly and of general legal values. However, this trust is not sufficiently specific to induce people to accept unfavorable judgments (Caldeira and Gibson 1995;. This theory implies that people may be perfectly fine with international courts overruling competent national courts as long as they agree with the outcome. Indeed, outcome favorability may color perceptions of procedural fairness (Esaiasson et al. 2019), thus increasing support not just for the implementation of the decision but also the authority of the court.
Politicians frequently resort to outcome-based criticisms. For example, British Prime Minister David Cameron argued in a parliamentary debate over the implementation of the ECtHR Hirst ruling on prisoners' voting rights that: "it makes me physically ill to even contemplate having to give the vote to anyone in prison" (Hough 2011). Yet, European court judgments are not necessarily unpopular. For example, some of the CJEU's opinions on privacy rights, such as Google v. Spain that established a "right to be forgotten," appear quite popular: 11 75% of Europeans support the right to delete personal information from any website. 12 The substantive legitimacy thesis implies that European courts could increase or decrease support by disregarding national courts, depending on whether the public on average sympathizes with an applicant.
H2a (Outcome effect): Support is lower if an unsympathetic applicant wins or a sympathetic applicant loses a case.
H2a raises the obvious issue of heterogeneity in preexisting beliefs. Clearly individuals who are more sympathetic to immigrants should be more sympathetic to a Court that protects immigrant rights than individuals who are less favorably disposed toward immigrants' human rights.
H2b (Outcome heterogeneity: sympathy): Individuals who value the applicant's rights unfavorably are less likely to support a court (ruling) supporting the applicant.

Nationalism
Scholars have long found that people who exclusively conceive of their identity in national rather than European terms are more likely to oppose European integration (Clark and Rohrschneider 2019;Fligstein, Polyakova, and Sandholtz 2012;Hooghe and Marks 2004;2005;McLaren 2005). Although national and European identities may reinforce each other, EU skeptics have increasingly framed the debate as being about the defense of the nation against control from Brussels, Luxembourg, and Strasbourg. Liesbeth Hooghe and Gary Marks find evidence for their hypothesis that "citizens who see themselves as exclusively national are particularly receptive to elite warnings that European integration harbors unacceptable foreign influence" (Hooghe and Marks 2005, 426). Others have shown that people with exclusively nationalist identities are less likely to favor international cooperation (Herrmann, Isernia, and Segatti 2009) and much more likely to support radical right-wing parties (Dunn 2015), and parties that use exclusive nationalist rhetoric in their manifestos also strongly contest international authority (Ecker-Ehrhardt 2014).
While the correlation between exclusive national identity and support for European institutions is well established, the precise nature of this relationship remains somewhat unclear (Hobolt and de Vries 2016). Among the open questions is whether defining national identity in exclusive terms implies an objection to the principle of locating authority away from the nation or whether it is about the substance of European integration. The theoretical expectations depend on whether nationalists identify their attachment to the nation in civic terms or ethnic and cultural terms.
A civic conception is that nationalism is a group identity defined by an affective connection with the political territorial unit in which people reside (e.g., Wright, Citrin, and Wand 2012). The in-group consists of fellow nationals regardless of ethnicity or cultural demarcations. In this understanding, "unacceptable foreign influence" would include a European court that sets aside a national court judgment. Thus, following this logic, we would expect that those who have an exclusive national identity will react more negatively to a European court overruling a national court.
H3a (Procedural heterogeneity: nationalism): Perceived sovereignty costs increase the more nationalist an individual is.
Exclusive nationalism can, however, also be understood as a demarcation of cultural in-groups and outgroups, or nativism, which has been a defining feature of many European right-wing populist parties (Bar-On 2018; Mudde 2004). An ethnic or cultural conceptualization of nationalism follows from beliefs about the nature of the nation rather than just its territorial borders (Dunn 2015). In this conception, the link between nationalism and European integration runs through substantive issues that are perceived as challenges to traditional conceptions of the nation rather than foreign interference per se. For example, selfdefined national identity correlates strongly with fear of especially Islamic immigration (Sides and Citrin 2007). Moreover, attitudes about immigration correlate strongly with preferences over European integration, as the EU is perceived to encourage immigration (de Vreese and Boomgaarden 2005;Hobolt 2016). A nativist might support European institutions that protect them from non-European immigration, as suggested by the "fortress Europe" language that some right-wing parties have adopted (Lamour and Varga 2020). This more cultural conception of nationalism implies that exclusive nationalists should be moved primarily by the outcome treatment, especially on the vignettes referencing immigration and Islam.
H3b (Outcome heterogeneity: nationalism): Nationalists are more likely to support court rulings against someone outside of the cultural or ethnic in-group regardless of whether the ruling comes from an international court.

Authoritarianism
Like nationalism, authoritarian values have been strongly associated with opposition to globalization and international institutions (Ballard-Rosa et al. 2021;Norris and Inglehart 2019). Interest in authoritarian values dates back to the 1950s when scholars sought to understand the personality traits of individuals who supported the Nazis and other fascist movements (Adorno et al. 1993). Authoritarians place great value on conformity with in-group norms and desire uniformity. Authoritarians have punitive and intolerant attitudes toward threats that may fragment society and often favor aggressive, even violent, suppressions of those threats.
There are different ways to conceptualize and measure authoritarianism. One approach is to assess personality traits by asking respondents about appropriate ways to teach and discipline children at home (Dunn 2015;Feldman and Stenner 1997;Stenner 2005). An alternative, adopted in this article, is to assess right-wing authoritarianism more directly as a combination of aggression, submission, and conventionalism (Altemeyer 1988;Dunwoody and Funke 2016). The literature has already found a robust correlation between these measures and opposition to immigration, Islam, ethnocentrism, political intolerance, and antidemocratic attitudes (Altemeyer 1988;Ballard-Rosa et al. 2021;Dunwoody and Funke 2016;Dunwoody and McFarland 2018).
In accordance with this literature, we hypothesize that people with strong authoritarian values are more likely to be moved by the outcome treatment, especially on the immigration and Quran-burning vignettes of our study. Indeed, authoritarians may be skeptical of the very idea of human rights litigation. Because authoritarianism is also associated with respect for traditional authority, which is located at the national level, we also examine whether people with strong authoritarian values are less accepting of a European court contradicting a national court.
H4a (Procedural heterogeneity: right-wing authoritarianism): Perceived sovereignty costs increase the more authoritarian an individual's values are.
H4b (Outcome heterogeneity: right-wing authoritarianism): Support for a court (ruling) will vary depending on how authoritarian an individual's values are.

RESEARCH DESIGN
We designed survey experiments that disentangle responses to information about a European court disagreeing with a national court and case outcomes. Our research design is similar to an emerging experimental literature that examines whether and to what extent the public punishes democratically elected leaders for process violations, such as sidestepping legislatures through unilateral actions (Becher and Brouard 2019;Christenson and Kriner 2017;Graham and Svolik 2020;Reeves and Rogowski 2018).
Qualtrics conducted the survey in Denmark, France, Poland, Spain, and the United Kingdom to a sample of around 3,000 respondents per country. The five countries were chosen to assess the generalizability of our findings rather than to examine cross-country hypotheses. These five countries represent variation across both the North-South and East-West axes of Europe. Poland is the only newer democracy and a country that is widely perceived to be backsliding. In all countries, right-wing authoritarian and nationalist movements are relevant and have challenged international court judgments. We used one vignette in all countries for half of the sample and administered three different vignettes to the other half (see Table 1). This approach balances concerns about our findings' dependence on country and vignette context while ensuring that our main vignette has sufficient power.
Unfortunately, a translation error prevents us from interpreting the data on the LGBT rights vignette (on pride parades) in a coherent way. Appendix C contains more details.

Deportation Vignette
The deportation vignette is (UK example): Suppose that United Kingdom (UK) authorities decided to deport a foreigner who has been convicted of a crime. The foreigner appealed at a UK court that the decision to deport him violated his human rights. The UK court found that the authorities [CAN/CANNOT] deport the foreigner. The question was then brought before a European court, which [AGREED WITH/DISAGREED WITH] the UK court. The final decision is that the foreigner should [REMAIN IN THE UK/BE DEPORTED].
We chose this stylized case as our main example because the issue is politically salient in all countries and both national and European courts have produced numerous judgments on similar issues. Our interest is not in estimating the effect of a "typical case" but in whether controversial judgments are more acceptable if they come from a national rather than a European court.
In only this vignette, we include a control condition that eliminates the sentence on the European Court. This assesses the possibility that the mere mention of a European court raises sovereignty concerns. This creates the following six assignment categories (see Table 2). In scenario 1, the applicant wins in the national court and the European court agrees. In scenario 3, the applicant loses in the national court but wins in the European court. In scenario 5, the applicant wins in national court and there is no European court ruling. Scenarios 2, 4, and 6 mirror these, but now the applicant loses. The outcome effect is the contrast between conditions 1, 3, and 5 (blue cells) with 2, 4, and 6 (yellow cells). The sovereignty effect is the contrast between 3 and 4 and 1 and 2. We use 5 and 6 as a placebo test to examine whether the mere mention of European court involvement matters. Previous studies have found that setting appropriate benchmarks may matter for evaluating procedural justice perceptions (Werner and Marien 2020).
The exercise is abstract but not unrealistic. We refer to a "European court" and remain deliberately vague about whether this concerns the ECtHR or the CJEU, as we do not presume that respondents know the difference. We abstract away from jurisdictional issues. The ECtHR can only directly hear cases in boxes 2 and 3, where the applicant initially loses in the national court. 13 The CJEU mostly weighs in on deportation cases through its preliminary references procedure and does not formally agree or disagree with national courts but provides the authoritative interpretation of EU law. Yet, politicians and media often frame the role of European courts in ways that reflect all boxes.
For example, in 2016 the Danish Supreme Court ruled that expelling a non-Danish citizen with an extensive and public criminal record, would constitute a violation of the European Convention. The Danish Court argued that: "[‥] despite the severity of his current crime and despite his criminal career, [the request] for deportation would be a disproportionate infringement on his rights in respect to his private and family life." 14 This decision caused uproar among Danish politicians and helped instigate efforts to reform the ECtHR (Madsen 2020). The ECtHR had not actually ruled counter to the national court (as in scenario 1), yet the press portrayed the situation as though it was a scenario 3 case where the Danish commitment to the Court led to the inability to expel the person. Another case, involving a different member of the same family with a long criminal CV, did reach the ECtHR. In this case, the ECtHR found that the Danish court had given "relevant and sufficient reasons" to expel him (scenario 2). 15 Jurisdictionally, the most difficult scenario is the 4th, in which the applicant wins a domestic case but the European court disagrees. Governments do not have the right to appeal national court decisions for alleged violations of European law. Yet, it is entirely possible that a European court is more sympathetic to the government's concerns than a national court would have been.
An example is a German case involving a Turkish national of Kurdish origin (Mr. T). 16 Mr. T was recognized as a refugee and granted an indefinite resident permit in Germany in 1993 based on his political activities in support of the PKK (Lampert 2017). After he was convicted of supporting a terrorist organization, the Karlsruhe regional government determined in 2009 that Mr. T qualified as a "present danger" and that he should be expelled and that his residency permit should be evoked. The High Administrative Court accepted Mr. T's appeal in 2012. The Court expressed doubt about the revocation of Mr. T's residency permit, temporarily restored that permit, and requested a preliminary ruling from the CJEU. Yet, in 2015, the CJEU found that a residence permit to a refugee may be revoked where there are "compelling reasons of national security or public order," including participation in a terrorist organization. 17 The CJEU judgment gave an interpretation that was less favorable to the applicant than the preliminary national court decision.
There are other, similar, examples especially on migration and citizenship law. For example, the CJEU, contrary to the opinion of the national referring court, declared that the notion of "freedom to provide services" in the Additional Protocol to the Turkey-EU Association Agreement must be interpreted as not encompassing freedom for Turkish nationals who are the recipients of services to visit a Member State, thus denying Turkish citizens EU access without a visa (Hatzopoulos 2014). Danish courts have arguably applied a security margin (sikkerhedsmargin) in some expulsion cases fueled by a desire to not be overruled by the ECtHR, making protection standards higher than in ECtHR case law . The 4th scenario is theoretically and substantively interesting even if jurisdictionally unrealistic.

Quran-Burning Vignette
In the second vignette, we chose an applicant that should be relatively more sympathetic to supporters of right-wing authoritarian parties. The vignette is modeled on a series of cases that balance free speech with other rights. For example, in E.S v. Austria, the ECtHR upheld a domestic court's fine for an Austrian woman who had called the prophet Mohammed a pedophile. The European court was widely criticized for this judgment by right-wing politicians and columnists (Smet 2019), suggesting that the authoritarian right wanted the European court to overrule the national court. Our vignette is the following (using DK example):

Eviction Vignette
The vignettes above appeal to cases that are controversial due to identity issues. Yet, economic issues are also an important driving force behind dissatisfaction with globalization, including in the European court context (e.g., Höpner and Schäfer 2012). Major corporations or banks have prevailed, but individuals have also won major cases against Google, Facebook, or against state laws that were seen as insufficiently protective. In these instances, the applicant may be more appealing to the political left. We based our vignette roughly on Mohamed Aziz v. Catalunyacaixa, a CJEU case that ruled that the Spanish civil code and civil proceedings law violates EU law because it does not give applicants the right to file lawsuits on the basis that the terms of the initial loan include provisions that are unfair. 18 While Aziz was eventually evicted, the case changed the procedural challenges that mortgage holders can raise before eviction. The human rights dimension of this case derives from article 47 of the EU Charter of Fundamental Rights that provides fair trial rights to all rights arising from EU law, in this case Council Directive 93/13/EEC on unfair terms in consumer contracts. The CJEU thus ruled that the Spanish legal system made it impossible for mortgage holders to enforce their consumer rights against major banks. 19 The ECtHR also has an extensive case law on eviction, sometimes concluding that individuals were wrongly evicted. 20 Fundamentally, this is a case about an individual's right to be protected by the state from potentially predatory practices by banks. Both national and European courts play an important role in guaranteeing these rights. We construct the following vignette (using the Spanish example): Suppose that Spanish authorities, at the request of a bank, requested the eviction of an individual who had not paid his mortgage. The individual appealed the decision, charging that the terms and conditions of the bank's loan were unfair and violated his rights. A Spanish court found that the authorities [CAN/CANNOT] evict the individual from his home until a decision is made on the legality of the contract. The question was then brought before a European court, which [AGREED WITH/DISAGREED WITH] the Spanish court. The final decision is that the Spanish authorities [CAN/CANNOT] evict the individual.

DATA AND METHOD Survey
Each survey targeted 3,000 respondents recruited from each country via Qualtrics Panels from January 27 to February 27 of 2020. This service is an online opt-in survey platform. All respondents were informed that they were taking part in a research study and were given the names and contact information of the principal investigators. All respondents were asked for their consent. No identifying information was presented to the investigators, nor was any deception used.
We used quota sampling to ensure that the marginal distributions of important covariates (age, gender, and education) in our sample match known census benchmarks for each country. The census shares, the implied sample quotas, the actual recruitment in each quota group, and the full questionnaire are in the methodological appendix.

Outcome Questions
We have three outcome measures. First, we simply ask respondents whether they "agree or disagree with the final decision?" Second, we ask respondents whether they "agree that the final decision should be 18 Case C-415/11, March 14, 2013. 19  implemented?" Third, we ask whether they agree their country "should continue to accept the authority of European courts?" All questions were rated on a sixpoint scale from "strongly agree" to "strongly disagree." As preregistered, we present results on dichotomized versions of the outcome variables, but appendix D includes analyses on the continuous measures.

Treatment Effect Moderator Questions
We preregistered hypotheses about heterogeneous treatment effects. To examine hypothesis 1b, we asked respondents "In general terms, how satisfied are you with the way the rule of law works in [country]." The questions to examine hypothesis 2b vary by vignette. For deportation, we used a question that forced respondents to choose whether they think it is worse to give legal status to immigrants who misrepresented their situation back home (e.g., face a death threat back home) or deny legal status to immigrants who did not misrepresent their situation. Similarly, for eviction, we asked whether it was worse to evict people who were treated unfairly by a landlord or mortgage company or to not evict people who have not paid their rent or mortgage. For the Quran-burning vignette we asked respondents whether they agree that the "public display of Islamic symbols such as burkas, veils, and minarets, should be reduced." Our primary measure for exclusionary nationalism is a common one in the literature: whether people identify themselves as only "British" (using UK example) or also partially or fully as "European" (e.g., Hooghe and Marks 2004). As a robustness check, we also use an additive nationalism scale that in addition to the item above includes the questions: "How proud are you of being British?" In the United Kingdom, our people are not perfect, but our culture is superior to others" and "I would rather be a citizen of the United Kingdom than of any other country in the world." Authoritarianism is measured by an additive scale based on the Aggression-Submission-Conventionalism (ASC) scale (Ballard-Rosa et al. 2021;Dunwoody and Funke 2016). This scale is based on Altemeyer's research on right-wing authoritarianism, but it avoids items that border on the tautological for outcomes of interest. For example, the Altemeyer scale asks whether "It is important to protect the rights of radicals and deviants in all ways" as well as preferences about homosexuality and civil rights. For the same reason, we exclude the submission component of the ASC scale. Instead, we use the four items that are most strongly associated with the aggression and conventionalism parts of the scale (two of which are reverse coded, marked by *): A. Police should avoid using violence against suspects.* B. It is necessary to use force against people who are a threat to authority. C. Traditions are the foundation of a healthy society and should be respected. D. People should challenge social traditions in order to advance society.* Table 3 offers descriptive statistics by vignette and country. There is little cross-country variation in agreement with the decision or implementation. However, as expected, respondents in the UK were much less inclined to accept the authority of European courts. Respondents in the UK and Denmark were on average more likely to be satisfied with domestic courts and more likely to be nationalist: 63% of UK respondents chose exclusively their national identity compared with just 20% of Spanish respondents. We coded respondents who were above 0.5 on the authoritarianism scale (about half of respondents in each country) as "authoritarians." But we also estimated models for the continuous scales for both nationalism and authoritarianism. Authoritarianism and nationalism are only moderately correlated (0.34 in the full sample). The authoritarianism scale is lacking in terms of internal reliability. 21 Yet, the scales do have validity, for example in predicting support for extremeright-wing parties. In Denmark respondents who intend to vote for the Danish People's Party (Dansk Folkeparti) and the New Right (Nye Borgerlige) are much more likely to be authoritarian (72% and 78%) and nationalist (65% and 64%) than voters for any other party. In France this holds for respondents who intend to vote for the Rassemblement National (previously known as the National Front), in Spain for Vox voters, and in Poland for Law and Justice (Prawo I Sprawiedliwość) and the Confederation Liberty and Independence (Konfederacja Wolność i Niepodległość). In the UK, those who intended to vote for UKIP, the Brexit Party, or the British National Party were all about equally nationalist and authoritarian, although they are not statistically distinguishable from Conservative party voters.

Summary Statistics
Overall, the sample is somewhat more Eurosceptic than the most recent Eurobarometer (October 2019). That survey found that 24% of Brits thought EU membership a "bad thing" compared with 28% in our sample. This difference was higher in France (19% vs. 11%), and Denmark (15% vs. 6%). Poland and Spain were below 10% in both samples.

Estimation
We estimate the average effect of each treatment, marginalized over the other treatment condition, using a standard poststratified difference-in-means estimator (and in the case of interaction effects, a difference-in-difference-in-means estimator). We estimated the effect within strata constructed using a set of preregistered pretreatment covariates. Poststratification provides a reasonable alternative to stratified randomization to improve the precision of our estimators (Miratrix, Sekhon, and Yu 2013). The covariates that we use to construct the strata are age, gender, level of education, and survey country, and the specific bins we used are outlined in the methodological appendix. Appendix D produces nonstratified estimates, which yield virtually identical conclusions. Figure 1 plots the mean proportion (and the 95% confidence interval) of respondents who agree on each of the three outcome measures for each of the treatment conditions (denoted by the scenario numbers in Table 2) on each of the three vignettes. This figure is not designed for formal hypothesis tests on treatment effects, but it provides a useful summary of the pattern of responses and allows visual comparisons between mean responses across the treatment conditions. The treatment effects for agreement with the decision are much larger than for implementation and authority. Respondents clearly differentiate between the outcomes. For example, the figure shows that in a number of the treatment conditions where a large share of the respondents disagree with the decision outcome-such as the applicant winning in the deportation vignette-a much smaller share endorse nonimplementation. This suggests that there are respondents who, despite disagreeing with the decisions, still support implementing them. If the ultimate test for subjective legitimacy is whether people believe that decisions that they disagree with should nonetheless be implemented (Caldeira and Gibson 1995;Gibson, Caldeira, and Baird 1998), then European courts do have legitimacy.

Summary Effects by Treatment Block
Overall, there is no evidence that respondents' answers on the placebo treatment that made no mention of the European court are significantly different from the other treatment conditions. To ease presentation, we omit this condition from further tables and figures (see results including the placebo group in appendix D). Figure 2 plots the average treatment effects of the European court disagreeing versus agreeing with the domestic court (full tables are in Appendix E). While in the deportation vignette we find slight negative effects on respondents' agreement with the decision and FIGURE 1. Mean Proportion "Agree" with Each Outcome Question and 95% Confidence Intervals by Treatment Note: Numbers in parentheses correspond to treatments from Table 2. Sovereignty, Substance, and Public Support for European Courts' Human Rights Rulings beliefs that the decision should be implemented, the magnitude of these effects is relatively small-less than 4 percentage points. The treatment effects are not significantly different from zero for any of the outcomes on the eviction vignette. The estimated effects are somewhat larger for the Quran-burning vignette on both agreement and belief that the decision should be implemented, but in all three vignettes we find no treatment effect on respondent's overall belief that their government should continue to accept the authority of European courts.

Deference (H1)
On closer inspection, the differences across vignettes could be a function of the countries in which the vignettes were administered. On the deportation vignette, the estimated effect of the deference treatment on implementation was largest in Denmark, Poland, and the UK, although it did not reach conventional levels of statistical significance in any individual country. Averaging over the deportation and Quran vignettes, the no-deference condition reduced support for implementation (but not court authority) among Danish respondents by 5 percentage points (p = 0.001) and among British respondents by 3.8 percentage points (p = 0.007). Among French and Spanish respondents, the no-deference condition did not reduce support for decisions or implementation in either the deportation or eviction vignettes.
That the British and Danish publics appear more sensitive to European court deference appears consistent with hypothesis 1b: that people who are satisfied with the domestic rule of law are also less enthusiastic about international court interference. However, Figure 3 provides no evidence for an individual-level interaction. In none of the vignettes for none of the outcome measures do people who are satisfied with the domestic rule of law respond differently to a lack of EC court deference than people who are dissatisfied. The absence of an interaction remains if we subset our analysis to the UK and Denmark in particular.
Outcome Effects (H2) Figure 4 plots the average treatment effects of the case outcome treatment: the effect of a win for the applicant versus a loss for the applicant. For this treatment, the effects on both agreement and support for implementation and acceptance of European court authority are much larger in magnitude than for the deference treatment. A win for the applicant reduces agreement with the judgment by 46 percentage points in the deportation vignette and by 22 percentage points in the Quran-burning vignette. By contrast, respondents on average found the applicant more sympathetic in the eviction vignette with a win for the applicant raising support by 24 percentage points.
The effects on support for implementation were more modest but still statistically significant: 8.5, 5.5, and 4.2 percentage points on deportation, eviction, and Quran burning, respectively. Moreover, unlike the deference treatment, the outcome treatment also significantly affects support for the authority of the European court (10.4, 6.1, 5.3 percentage points, respectively). That is, the act of defying a national court does not reduce support for European court authority but producing an outcome where a disliked applicant wins or, in the case of eviction, a likeable applicant loses, does reduce support for European court authority. Closer inspection finds that these effects are remarkably consistent across countries (see appendix D).
We conducted a series of tests to examine whether the coefficient on the case outcome treatment was larger than the deference treatment. We found that the case outcome treatment effects were significantly larger (at p = 0.01) in all instances other than for the implementation and authority outcomes in the Quran vignette, where we could not reject the null hypothesis of no difference between the coefficients. Figure 5 examines hypothesis 2b, which states that the effect of the outcome treatment should be more negative for individuals who are less sympathetic to an applicant's rights. Recall that our measures are pretreatment and do not capture agreement with the decision but general predispositions toward human rights (e.g., would you rather wrongly deport or wrongly admit an immigrant?).
There are sizeable interaction effects on most measures. For example, in all three vignettes support for implementation is 9-10 percentage points lower among applicants who are predisposed against the applicant if they are assigned to the treatment in which the applicant wins. The effects are largest in the eviction vignette. The treatment effect of an applicant win on accepting the authority of European courts is 13 percentage points larger for respondents who believe it is worse to evict a tenant who has been maltreated than to not evict a tenant who just has not paid the rent (about 70% of respondents).

Nationalism
Are respondents who self-identify as exclusive nationalists more responsive to the deference treatment? Consistent with the literature, respondents who selfidentify as only having a national identity are much less likely to agree that their country should continue to accept the authority of European courts regardless of treatment condition. The smallest difference is in Spain Sovereignty, Substance, and Public Support for European Courts' Human Rights Rulings (65% vs. 83%) and the largest in the UK (38% vs. 72%) and France (41% vs. 70%), although a much larger proportion of the British population identifies as nationalist in the UK than in France (64% vs. 40%, respectively). 22 Figure 6 evaluates hypotheses 3a and 3b. Overall, there is no evidence that nationalist respondents disagree more with decisions, their implementation, or the authority of European courts when the European court refuses to defer to the national court. If we evaluate countries individually, we do not find evidence that nationalists are significantly more likely than nonnationalists to adjust their agreement with the decision or support for implementation and court authority based on the deference treatment in any of the vignettes.
By contrast, nationalists are much more responsive than nonnationalists to the outcome of the deportation vignette. The treatment effect of an immigrant win on acceptance of court authority and support for implementation is 7.5 percentage points larger for nationalists than nonnationalists, although it is still also significant for nonnationalists. Note that it does not matter whether the decision came about because the European court overturned or upheld the national court, the outcome is the primary driver of response especially for nationalists. Moreover, when we include the condition that makes no mention of a European court, we find the same effect: nationalists are less likely than are nonnationalists to support implementation of a national court decision when the foreigner avoids deportation.
Nationalists are also relatively more sympathetic than nonnationalists to the applicant in the Quranburning vignette. Nationalists are 10 percentage points more likely to agree with the decision and 9 percentage points more likely than are nonnationalists to favor implementation if the fine gets repealed. There is no significant interaction between nationalism and the treatment effect of the outcome on support for the authority of the court.
Nationalists and nonnationalists do differ slightly in terms of their response to the outcome treatment in the eviction vignette. Interestingly, the effect of the applicant winning in the eviction vignette on agreement is larger for nonnationalists than nationalists (though the effect is positive and statistically significant in both subgroups). However, the effect on support for implementation is slightly larger among nationalists than nonnationalists. This is consistent with a possible "ceiling" effect where nonnationalists may generally be supportive of implementation irrespective of their level of agreement.
We repeated all analyses with the four-item nationalism scale using more flexible interaction models (see appendix D). The findings were consistent with those presented here: no significant interactions for the deference treatment, but we do find significant and substantively important interactions for the outcome treatment. These findings support the idea that the correlation between exclusive national identity and support for European institutions runs through the effect European institutions are perceived to have on immigration and other policies that ethnocultural nationalists dislike rather than an innate preference for national over European institutions. Figure 7 shows the heterogeneity in treatment effect estimates by individuals' authoritarianism. Authoritarians are not more or less moved than nonauthoritarians by a European court sidestepping a national court. Especially on the deportation case, however, the effect of the outcome treatment is much larger for authoritarians on both implementation (11 percentage points) and acceptance of court authority (9 percentage points). Despite professed respect for authority, authoritarians are more likely to support nonimplementation of court decisions they do not like compared with nonauthoritarians.

Authoritarianism
Perhaps surprisingly, we do not observe as strong an interaction in the Quran-burning case: authoritarians are 5 percentage points more likely than nonauthoritarians to favor implementation when the applicant wins. In the flexible interaction models (appendix), we find that this is primarily driven by people in the top third of the authoritarianism distribution. However, we find no effect on court authority. One possibility is that the applicant in this case is not just anti-Islam but also displays behavior that disrespects public authority. On the eviction vignette, we find that authoritarians respond less positively to the applicant winning a case but the heterogeneous effect is only evident when the outcome is agreement with the decision.

Right-Wing Populism and Euroskepticism
We also examined (not preregistered, in appendix D) whether respondents who indicated their intent to vote for right-wing populist parties 23 respond differently to the treatments. The results are very similar to those on authoritarianism and nationalism: large heterogeneous effects on the outcome treatment for the deportation vignette and to a lesser extent the Quran-burning vignette. However, we found no differences on the deference treatment.
Similarly, those respondents who said that EU membership was either "bad" or "very bad" (asked pretreatment) are also unresponsive to the European court disregarding the national court, but they are highly responsive to the outcome treatments. Eurosceptics are 10 percentage points less likely than non-Eurosceptics to favor compliance following the immigrant winning and 10 percentage points more likely to support implementation if the person who burns Qurans and the home owner fighting eviction win. The effects on support for European court authority are not significant, but this could be because baseline support among Eurosceptics is so low to begin with: just 26% agree that European courts should have authority.
In short, our evidence implies that neither respondents with nationalist or authoritarian values nor those who indicate a vote preference for right-wing populist parties or those who think that EU membership is bad for their countries respond to a lack of deference by a European court, but they are responsive to outcomes.

FIGURE 6. Nationalism and Treatment Effects
Note: Lines denote 95% confidence intervals.

CONCLUSIONS
Our central finding is that European respondents in five countries were not less acceptant of human rights judgments that came about through an international court overruling a national court, other than a small effect in Denmark and the United Kingdom. By contrast, support for implementation and even European court authority were significantly lower if an unsympathetic applicant won a case or a sympathetic applicant lost. Moreover, this is equally, and sometimes more, true for the most strident opponents of European courts: nationalists, authoritarians, Eurosceptics, and those who vote for right-wing populist parties. These findings suggest that public opposition to European human rights adjudication is more about the content of decisions than the location of authority and that it follows political preferences.
The question of whether people are inherently more acceptant of decisions that are made by national rather than international institutions is more broadly relevant to understanding the public backlash against international institutions. For example, international investment arbitration is facing strong public backlash in both developing and developed countries (Walter 2021). But we do not know the extent to which this discontent follows from the location of authority (international arbitral panels versus domestic courts) or findings favoring foreign multinational corporations (see Johns, Pelc, and Wellhausen 2019). Similarly, we could apply our experimental design to examine whether people are more acceptant of decisions by domestic war crimes tribunals rather than international ones. An alternative experimental design would be to directly examine public acceptance of a national or an international court striking down or upholding a government action rather than to examine disagreement between two judicial institutions at an international and a national level. 24 Finally, we could take this experimental design outside the court context altogether, such as instances where the European Commission or another nonjudicial international body conflicts with a domestic agency. There is no guarantee that such experiments would yield the same findings.
There may well be substantive settings in which sovereignty concerns loom larger than the vignettes evaluated here. For example, the CJEU evaluates questions of national regulatory competence, which ascertain whether national legislation unjustly restricts EU market freedoms. Such judgments could well raise stronger sovereignty concerns than human rights cases as could matters involving national security. However, there are some reasons to believe that nationalist and authoritarian objections to international institutions may more generally follow the pattern found in this article. Other scholars have found that exclusive nationalism and its link to right-wing authoritarianism is more about cultural and ethnic in-group-out-group conflict than adherence to the territorial political unit of the state (Bar-On 2018; Dunn 2015). We might expect, then, that nationalist opposition to globalization is more about its content, especially rising immigration, rather than the relocation of authority to the international level.
These findings also have implications for efforts at mitigating public backlash. Liberal international institutions, and courts adjudicating human rights in particular, are ill suited to producing outcomes that are favored by nationalists and authoritarians. Human rights courts are designed to protect vulnerable, and sometimes unpopular, minorities from majority repression. Moreover, courts are supposed to build on legal precedent, produce case law that generates legal certainty, and display independence from popular and political pressures. Although we know that courts are sometimes responsive to public opinion, courts ultimately cannot fulfill their core functions if they are too sensitive to what is popular and what is not. Processbased responses to backlash, such as emphasizing a margin of appreciation or subsidiarity doctrine in the ECtHR context, are easier to implement that substance-based responses. Yet, increased deference also results in a denial of legitimate human rights claims and, as our results imply, will not by itself sustain public legitimacy. The dilemma faced by international human rights courts is thus more complex than a choice between self-restraint and backlash.

SUPPLEMENTARY MATERIALS
To view supplementary material for this article, please visit http://doi.org/10.1017/S0003055421001143.

DATA AVAILABILITY STATEMENT
Research documentation and data that support the findings of this study are openly available at the Amercian Political Science Review Dataverse: https://doi. org/10.7910/DVN/WJQITU.