The Politics of International Judicial Appointments: Evidence from the European Court of Human Rights

Abstract

Theories of government–international court relations assume that judges share an interest in expanding the reach of their court. Yet, casual observation suggests that international judges vary in their activist orientations and that governments selectively appoint judges. This article explores a new data set of dissents in the European Court of Human Rights (ECHR) to estimate the ideal points of judges. The results show that activism-restraint is indeed the main dimension of contestation among judges. Variation in judicial activism cannot be accounted for by different legal cultures of judges or by levels of domestic human rights observance in the judges' countries of origins. Instead, aspiring European Union (EU) members and governments more favorably disposed toward European integration appoint more activist judges. These results imply that politics matters in the appointment of international judges and that EU expansion was an important driving force behind the ECHR's increased activism. More generally, the analysis suggests that agent selection is an important and understudied tool for influencing international organizations.Earlier versions of this article were presented at seminars at the University of Chicago, Northwestern, Vanderbilt, George Washington, the University of Wisconsin, William and Mary, Princeton, and Georgetown as well as the 2006 Annual Meetings of the American Political Science Association. I much appreciate comments and suggestions from the participants in those seminars, two anonymous referees, Karen Alter, Freek Bruinsma, Rachel Cichowski, Allison Danner, Darren Hawkins, Larry Helfer, Christopher Joyner, Charles Lipson, Emily Meierding, Andrew Moravcsik, Kimberly Morgan, Eric Posner, Mike Tierney, and Andreas von Staden. I thank Jamie Druckman, Andrew Roberts, and Paul Warwick for making data on cabinet composition available.

On 6 October 2005, the Grand Chamber of the European Court of Human Rights (ECHR) ruled that Britain's law that prevents convicted prisoners from voting violates fundamental rights protected by the 1950 European Convention for the Protection of Human Rights and Fundamental Freedoms (hereafter: the Convention).¹

Hirst v. the United Kingdom (2), 74025/01, 6 October 2005.

The judgment in the case, brought by a man who was convicted for murdering his landlady with an axe, elicited predictably strong denunciations from the British tabloids and Conservatives. For example, shadow Foreign Secretary Liam Fox reacted that: “This is an outrageous decision and a perfect example of how Europe is intruding in areas of our national life where it has no business.”²

The Sun, 7 October 2005, 1.

Yet, the ECHR's stance was also controversial among its judges themselves: five of the seventeen Grand Chamber judges warned in a dissenting opinion that “it is essential to bear in mind that the Court is not a legislator and should be careful not to assume legislative functions.”³

Hirst v. the United Kingdom, 6 October 2005, joint dissenting opinion of judges Luzius Wildhaber, Jean-Paul Costa, Peer Lorenzen, Anatoly Kovler, and Erik Jebens.

Such splits between proponents of a more “activist” and a more “restrained” role for the court are common among ECHR judges.⁴

See, for example, Jackson 1997; Morrisson 1981; and Bruinsma and Parmentier 2003.

This runs counter to the theoretical literature that has treated international courts as unitary actors, typically assuming that international judges seek autonomy and are motivated to make law. Governments, on the other hand, are supposed to jealously guard their sovereignty and to ensure that international judges are not overly zealous in their interpretation of treaties. Scholars differ in their estimation of who generally prevails in the resulting struggle for authority. Some argue that threats of noncompliance and exit exert a powerful constraining force on international courts.⁵

See, for example, Carrubba 2002; and Garrett, Kelemen, and Schulz 1998.

Others claim that although international judges are not unconstrained, they have considerable freedom in interpreting treaties and have used this to impose new obligations on states. Thus, scholars have argued that the rulings of the European Court of Justice (ECJ) have fundamentally transformed the European Union (EU) legal system,⁶

See, for example, Alter 1998; Burley and Mattli 1993; and Stone Sweet and Brunell 1998.

that decisions by the World Trade Organization's (WTO) Appellate Body have amounted to judicial policymaking,⁷

Steinberg 2004.

and that judgments by the International Criminal Tribunals for the former Yugoslavia (ICTY) and Rwanda (ICTR) have helped to establish a substantial new body of international law.⁸

Danner 2006.

Despite these debates, however, the assumptions that judges share an interest in expanding the reach of their court and that governments seek to prevent such occurrences have remained unchallenged.

This article opens the black box of judicial decision making on international courts. Using a new database of all dissents by ECHR judges between 1955 and June 2006, the analysis not only demonstrates that ECHR judges have diverse preferences but also that differences between judges are indeed about the reach of the court itself. Using the same methods employed to place U.S. Supreme Court judges along a liberal-conservative continuum, I estimate the ideal points of judges along an activism-restraint dimension. I then show that the judicial ideology of judges is linked to the political ideology of the governments that appointed them. This latter point is especially important as it suggests that there is a political logic underlying the increased activism of the ECHR. Most notably, aspiring EU members used activist judicial appointments to signal human rights commitments. Moreover, governments more favorably disposed toward European integration tended to pick activist judges.

These findings have important implications for one's understanding of the ECHR and international judicial behavior more generally. The ECHR has received little attention from political scientists⁹

Some notable exceptions include Moravcsik 2000 and Cichowski 2006. Neither of these articles, however, is about how the court makes decisions.

even though it is quickly becoming a significant actor on the European scene. The court allows private access to more than 800 million potential claimants in forty-six Council of Europe member states,¹⁰

The twenty-seven EU members and Albania, Andorra, Armenia, Azerbaijan, Bosnia and Herzegovina, Croatia, Georgia, Iceland, Liechtenstein, Moldova, Monaco, Norway, Russia, San Marino, Serbia and Montenegro, Switzerland, the former Yugoslav Republic of Macedonia, Turkey, and Ukraine.

each represented by a single judge. It has by far the largest caseload of any international court and has issued more than 7,000 judgments. The ECHR scores high on all dimensions of legalization as identified in a recent issue of International Organization: obligation, precision, and delegation.¹¹

Abbott et al. 2000, 404. For more detail, see Keohane, Moravcsik, and Slaughter 2000.

Moreover, its judgments are widely perceived to have markedly influenced domestic legal systems in most Council of Europe countries.¹²

See, for example, Blackburn and Polakiewicz 2001; and Helfer and Slaughter 2005.

The results imply that EU expansion played a significant role in the increased activism of a court whose reach extends well beyond the boundaries of the EU. The ECHR offered an opportunity for candidate countries to demonstrate their commitment to an international human rights regime before entering the EU. Moreover, the desire to integrate new and not yet stable democracies gave other Council of Europe member states incentives to strengthen the court institutionally and to put judges with an expansionary view of the convention on the bench.

Second, remarkably little is known about how judges with different nationalities, from different legal cultures, and appointed by different principals resolve disputes about alleged treaty violations. To my knowledge, this is the first effort to estimate the judicial ideology of judges on any international court.¹³

Existing studies of judicial behavior on international courts have almost exclusively focused on bias; for example, Posner and de Figueiredo 2005.

This is also, then, the first effort to examine the correlates of judicial ideology. Perhaps surprisingly, variation in domestic legal culture does not explain variation in observed judicial ideologies. There is also no substantiation in the data for the idea that judges from countries with poor domestic human rights records or ill-functioning legal systems are less likely to take an activist stance on the court. There is, however, evidence that the observed behavior of ECHR judges correlates with the preferences for European integration held by the governments that appointed them. This implies that at least to some extent, governments that prefer a more activist international court appoint more activist judges.

Although this may seem self-evident, the literature on the relationship between international courts and governments has paid scant attention to the screening role of governments. More generally, the burgeoning literature that applies principal-agent theory to the relationship between governments and international organizations focuses heavily on governmental abilities to sanction poor performance as opposed to their capacity to shape international organization behavior through the appointment process.¹⁴

For an overview of principal-agent applications to international organizations, see Hawkins et al. 2006. For a critique, see Alter 2006 and forthcoming.

Theoretically, the selection of agents becomes especially important when the principals' ability to monitor and sanction agent behavior is relatively weak,¹⁵

See Fearon 1999.

a situation that characterizes many government–international organization relationships. This study shows that it is possible to empirically examine the politics behind the appointments of agents and that such a study can challenge long-held assumptions about what drives international organization behavior. For example, the findings of this study imply that governments may deliberately appoint judges that hold an expansive view of the extent to which a treaty binds governments. Thus, not all governments jealously guard their sovereignty and seek to limit the expansionist drifts of international organizations. Moreover, the increased activism of international courts is not necessarily irreversible: governments could, and sometimes do, choose to stack international courts with diplomats predisposed to the raison d'état.

The article proceeds by briefly detailing the ECHR and the usage of activism and restraint in the ECHR's context. I then introduce plausible theoretical explanations for variations in levels of judicial activism. The empirical strategy of this article is to first estimate levels of judicial activism from observed vote choices and then regress these on measures for the various theoretical concepts. The conclusion offers some thoughts on the implications for the ECHR as well as on the generalizibility of the results to other international courts and organizations.

Activism and Restraint in the ECHR

Institutional Detail

The ECHR evaluates complaints by individuals¹⁶

The ECHR can also evaluate interstate disputes but those are rare and are beyond the scope of this article.

that their national government has violated one or more of the provisions of the European Convention for the Protection of Human Rights and Fundamental Freedoms (the Convention) and its amendments (protocols). The Convention was inspired by the 1948 Universal Declaration of Human Rights but has greater enforcement mechanisms attached to it.¹⁷

For a political analysis of the origins of the ECHR, see Moravcsik 2000.

The ECHR has issued judgments on such controversial issues as abortion rights in Ireland, administrative review procedures in Sweden and the Netherlands, gays in the British military, property rights of East Germans, torture in Turkey, slow court proceedings in Italy and Poland, the legality of extraditing prisoners to countries where they may be tortured, the expulsion of Russians from Latvia, and Russian human rights abuses in Chechnya. Understandably, then, the ECHR's rulings frequently invite strong reactions from governments and the occasional intrigue, such as when the ECHR's former president Luzius Wildhaber charged that he had been poisoned by agents of the Russian government in retribution for the court's findings on Chechnya.¹⁸

See Luke Harding, “I Was Poisoned by Russians, Human Rights Judge Says,” The Guardian, 31 January 2007. Available at: 〈http://www.guardian.co.uk/russia/article/0,,2002997,00.html〉. Accessed 12 June 2007.

The ECHR's relevance increased substantially with the adoption of Protocol 11, which went into force on 1 November 1998. Previously, states were allowed to exempt themselves from compulsory jurisdiction and direct access for private litigants. France, for instance, did not accept the ECHR's compulsory jurisdiction until 1974 (when it finally ratified the Convention) and waited until 1981 to declare that French citizens could directly apply to the ECHR. Greece (1979 and 1985) and Turkey (1990 and 1987) waited even longer to accept these provisions. Protocol 11 made both private access and compulsory jurisdiction mandatory. Moreover, the Protocol implemented further institutional reforms that amongst others made the ECHR a full-time court. As a result of easier individual access and expansion of the court's membership, its caseload increased considerably: whereas the ECHR issued only seventy-two judgments in 1996, it issued 889 judgments in 2001 and 1,560 in 2006.¹⁹

All summary statistics come from the Court's annual Survey of Activities. Available at: 〈http://www.echr.coe.int〉

The number of applications rose from 12,700 in 1996 to 50,500 in 2006.

In the post–Protocol 11 ECHR, each application is first evaluated by the registry. About one-fourth of all applications is dismissed at this stage for administrative reasons.²⁰

Precise proportions by year available from the Survey of Activities.

The application then goes to a rapporteur, generally the national judge of the respondent country. The rapporteur can refer the case to a committee of herself and two other judges who may unanimously decide to dismiss a case. Most applications are dismissed at this stage. In case no unanimity is reached on dismissal, the case goes to one of the ECHR's five sections,²¹

The fifth section was added in 2006. Section composition is balanced by geography, gender, and the different legal systems of member states.

which assigns the case to a Chamber of seven judges, including the respondent state's national judge or an ex officio judge assigned by the respondent state. After the Chamber's judgment, parties can request a rehearing by a Grand Chamber of seventeen judges that transcends the sections of the court.

ECHR judgments can demand that violating states pay “just satisfaction” to a victim and can request remedies for the violating offense. There are few credible sanctions that the Council of Europe can apply to ensure compliance, except “naming and shaming.”²²

The Council of Ministers oversees the execution of judgments and publishes regular reports on compliance. Expulsion from the Council of Europe can be used as a last resort.

Enforcement also takes place through the national courts, either because international law carries direct effect or because countries have adopted the Convention and its protocols into national law.²³

Especially noteworthy in this regard is the UK Human Rights Act of 1998, 2 October 2000.

Scholars generally perceive actual levels of compliance as high,²⁴

See, for example, Helfer and Slaughter 1997 and 2005; and Zorn and Van Winkle 2001.

although there is little systematic evidence.²⁵

Posner and Yoo 2005.

Even so, there is little doubt that its rulings have had important effects, leading some scholars to suggest that the ECHR has de facto become a constitutional court.²⁶

Shapiro and Stone Sweet 2002.

A 2001 evaluation of thirty-two member states found that although the influence of the ECHR varied, each state had adopted some form of major legislative reform in response to an ECHR ruling.²⁷

Blackburn and Polakiewicz 2001.

Judgments may also have an impact beyond the immediate case. For example, the previously mentioned judgment that Britain's outright ban on prisoner voting constitutes a violation of the Convention indirectly affected twelve other countries with similar bans.²⁸

Department of Constitutional Affairs 2006.

In anticipation of cases being brought by prisoners, some of these governments have adopted new voting rights legislation to come into compliance with the ECHR judgment.²⁹

See, for example, on the Irish case, Senan Molony, “Inmates to Vote by Post in Next Election,” Irish Independent, 23 October 2006. Available at: 〈http://www.iprt.ie/ireland/1845〉. Accessed 12 June 2007.

ECHR rulings have also been cited in domestic courts as benchmark interpretations for developing international norms. For example, a recent survey by Zaring found that ECHR rulings have been cited twenty-nine times by U.S. federal courts between 1945 and May of 2005, including four citations by the U.S. Supreme Court.³⁰

Zaring 2006.

As such, Zaring concluded in an empirical survey that those worried about the influence of foreign courts on U.S. law “shouldn't be worried about foreign citation as much as citation to the European Court of Human Rights.”³¹

Ibid., 24.

Divisions on the Court

Observers commonly interpret divisions within the ECHR as concerning the size of the margin of appreciation that should be left to respondent states.³²

See, for example, Bruinsma 2006; Bruinsma and Parmentier 2003; Jackson 1997; and Morrisson 1981.

The ECHR's margin of appreciation doctrine holds that each country has some latitude in resolving conflicts that arise between individual rights and the perceived national interests or values of that country.³³

Yourow 1996.

Those who believe that this margin should be broad stress that the subsidiarity principle suggests that it is appropriate to grant a great deal of deference to national practices, policies, and perceived interests. Judges on the other side of the spectrum tend to believe that states have less room to hide behind local customs and stated national interests when it concerns the implementation of the Convention.

For example, the ECHR ruled in 2005 that Turkey could legitimately ban women from wearing headscarves in public education institutions.³⁴

Leyla Sahin v. Turkey, 44774/98, 11 October 2005.

The majority reasoned, in the words of its President Luzius Wildhaber, that “We did take up what the Turkish constitution said. Secularism is at the root of the Turkish constitution. The Turkish constitution is also for gender equality, and sees the ban as very important there. We cannot fault these aims.”³⁵

Irish Times, 22 November 2005, 10.

The dissenting opinion, on the other hand, argued that the majority held too broad a margin of appreciation and that Turkey “should not be granted the right to deviate so sharply from the practices of the other countries subject to the Convention.”³⁶

Dissenting opinion, Judge Tulkens, Leyla Sahin v. Turkey.

This division is generally labeled as being between those who favor “judicial activism” and those who prefer “judicial restraint.” I stick to these terms while noting their specific operationalization in the context of the ECHR: given the legal facts of a case, an activist judge is more likely to rule in favor of the applicant than a judge on the self-restraint side of the spectrum. Thus, the division is ultimately about the degree of deference a judge prefers to grant governments.

Theory: The Politics of the Appointment Process

Why, then, are some judges allegedly more activist than others? A first set of explanations focuses on the politics of the appointment process. In the post–Protocol 11 system, governments no longer have absolute control over judicial appointments. Each government submits three candidates, who they may rank order. The Council of Europe's Parliamentary Assembly then votes on the list. The assembly has occasionally selected a candidate other than the government's favorite and has refused to accept a few candidate lists for want of gender-balance or proper qualifications. Generally, however, the government's preferred candidate is elected. Judges are appointed for six-year renewable terms.

That governments screen candidates for international judicial appointments is at least plausible. Aside from human rights activists, academics, and former national judges, the list of ECHR judges includes former ambassadors, representatives to international organizations, parliamentarians, ministers of justice, and an undersecretary of state. These backgrounds may be quite informative about attitudes and vote choices of ECHR judges.³⁷

Bruinsma 2006.

Ministers of foreign affairs and justice, who are generally responsible for selecting candidates, are likely aware of the tendencies of candidates for high-profile positions. In some cases political motivations are obvious. For instance, the Austrian judge Willi Führman, a former Social-Democratic parliamentarian, was replaced after his party lost domestic elections. Likewise, the Moldovan judge Tudor Pantiru was ousted by the newly elected communist government, which vowed to “send real patriots” to Moldova's diplomatic missions.³⁸

“Communists Announce Possible Recall of ECRH Judge Tudor Pantiru,” Moldova Azi, 6 April, 2001. Available at 〈http://www.azi.md/news?ID=1415〉. Accessed 12 June 2007.

More commonly, politics may matter in less visible ways. International judicial appointments rarely incite much public scrutiny. This leaves government officials relatively free to browse their preferred networks for suitable candidates.³⁹

For example, Bruinsma 2004 reports that each of the three final candidates for the 2004 Dutch vacancy received a personal invitation to apply.

As such, an independent evaluation of the ECHR appointment process concluded that: “Even in the most established democracies, nomination often rewards political loyalty more than merit.”⁴⁰

Limbach et al. 2003, 4.

Any theory that posits a relationship between the preferences of national governments and the observed behavior of international judges needs to establish an explicit link between the desires of national politicians and heterogeneity on the court under investigation.⁴¹

See also Alter forthcoming, who argues that many existing principal-agent models of state–International Court relations are too broad to be testable due to an absence of specific assumptions about preferences.

For example, U.S. Supreme Court justices can be located along the same liberal-conservative continuum that dominates U.S. politics. This establishes a straightforward link between the judicial arena and the political bodies responsible for judicial appointments. While judicial appointees do not always faithfully apply the wishes of their principals, there is a robust correlation between the observed behavior of judges and the political ideology of those who appointed them.⁴²

See, for example, Songer and Ginn 2002; and Rowland and Todd 1991.

It is less clear what this theoretical link is in the ECHR context. Below I derive hypotheses based on three plausible sources of variation in preferences for national governments: variation in levels of domestic human rights protection, partisan left-right divisions, and variation in levels of political support for European integration.

Democracy and Domestic Human Rights Protection

Perhaps the most straightforward hypothesis is that governments that are most vulnerable to negative judgments by the ECHR seek to limit their vulnerability by exercising caution in appointing judges that are likely to be activist. Instead, governments with few concerns that the ECHR will overturn the domestic status quo may favor activist judges, who could aid them in submitting other states to a set of liberal objectives. This perspective could be labeled “realist” in that it posits that liberal states induce or coerce other states to sign human rights treaties in an effort to extend national ideals derived from national pride or geopolitical self-interest.⁴³

See also Moravcsik 2000, tab. 1, 222.

As such, one may hypothesize that states with poor levels of domestic human rights protection are less likely to appoint activist judges than are states with good domestic rights records.

Republican liberalism expects a more subtle relationship between domestic human rights practices and preference for an activist court. In an analysis of the ECHR's origins, Moravcsik found that governments in new democracies sought to lock-in their commitments to human rights against future political change by subscribing to the control mechanisms of the Convention.⁴⁴

Moravcsik 2000.

On the other hand, established democracies and nondemocracies saw little use for an independent international court as it would merely result in unwanted interference with the domestic status quo. Governments in young democracies, on the other hand, worried more about potential future domestic regimes with authoritarian inclinations than about foreign interference. By extension, I hypothesize that democracies in transition are more likely to appoint activist judges than are established democracies and nondemocracies.

Left-Right Politics

An obvious extension from the case of the U.S. Supreme Court, which dominates the literature on judicial behavior, is that governments appoint international judges based on their ideological orientation. Left-right conflict is pervasive in European politics, including in European supranational institutions.⁴⁵

For example, Hix, Noury, and Roland 2007.

Thus, an obvious hypothesis is that left-wing governments tend to appoint judges that are more activist in their orientation whereas right-wing governments favor judges more on the self-restraint side.

This classification may have some merit on social issues, such as cases that concern sodomy laws, gays in the military, abortion, and rights for transsexuals. There are also issues on which an activist ECHR fits comfortably within an ideological framework that should appeal to the European socioeconomic right. Several ECHR decisions have served as a check on state power vis-à-vis individual exercises of economic rights. An example is the set of Article 6 (“right to a fair trial”) cases, led by Benthem v. The Netherlands, that helped improve the ability for private persons and businesses to challenge administrative decisions, such as the rejection of permits.⁴⁶

Benthem v. the Netherlands, 8848/80, 23 October 1985.

Similarly, one may expect those on the right of the socioeconomic spectrum to be supportive of an activist court on Article 1 of the First Protocol, which protects property from improper government expropriation. These cases are the second-most frequent subject of ECHR judgments, after only Article 6 cases.⁴⁷

Cichowski 2006.

The left-right hypothesis thus also suggests that activism-restraint may not be the proper label for the main dimension of contestation in the ECHR, as leftist judges should be activist on some cases but not on others. This issue will be addressed directly in the measurement section.

European Integration

An alternative thesis is that governments use judicial appointments to signal their human rights commitments to interested parties, especially the EU and its member states. The EU is a community of liberal states who view expansion as an attempt to broaden that community.⁴⁸

Schimmelfennig 2001.

Hence, the EU seeks assurances from prospective members regarding their commitments to a set of liberal values.⁴⁹

Kelley 2004.

The so-called “Copenhagen criteria” defined these requirements in an abstract manner: “Membership requires that the candidate country has achieved stability of institutions guaranteeing democracy, the rule of law, human rights and respect for and protection of minorities… .”⁵⁰

European Council in Copenhagen, 21–22 June 1993, “Conclusions of the Presidency,” SN 180/1/93 REV 1, 13.

Enforcement was problematic as the ECJ had no jurisdiction over most human rights issues and would only obtain jurisdiction over Eastern European states after their accession to the EU. Thus, the Council of Europe's ECHR became a convenient alternative. For example, Austrian Chancellor Franz Vranitzky noted that “[t]he economic and political realities of the present time, particularly in the former communist countries, are such that the Council of Europe is in fact the only organization capable of admitting these states to full membership without undue delay and so making them part of the European dynamic.”⁵¹

Irish Times, 11 October 1993, 8.

Shortly after the agreement on the Copenhagen criteria during the June 1993 European Council meeting, the Council of Europe's heads of state agreed in principle to the Protocol 11 reforms, in the Vienna Declaration of 9 October 1993. These reforms created the full-time independent ECHR with compulsory jurisdiction, something that various states, most notably Britain, had opposed prior to the issue of enlargement.⁵²

For a concise view of British opposition, see “Where Europe Rules,” Guardian, 11 October 1993, 19.

Thus, it appears that enlargement and the integration of new democracies were a catalyst behind efforts to create a stronger ECHR.

For aspiring EU members, appointing judges that appear willing to actively apply international standards, perhaps even against their governments, is a way to signal their commitments to a set of rules and conflict resolution procedures that are integral to the EU. As such, this argument is similar to Moravcsik's Republican Liberal thesis, except that the incentive here is signaling rather than lock-in. Judicial appointments are inadequate as lock-in mechanisms given that they are inherently short term. Yet, appointing a judge who is known to be independent or even a human rights advocate may well be an effective signaling mechanism given that it could impose short-term costs on a domestic government. Some qualitative evidence is that the CVs submitted to the parliamentary assembly for candidates from Europe's new democracies frequently stressed the independent nature of the candidate.⁵³

See also Flauss 1998.

For example, both Czech judge Karl Jungwiert and Slovak judge Bohumil Repik added paragraphs of text to their CVs to stress that they lost their formal positions in 1970 due to their activities in protesting the 1968 Warsaw Pact occupation of Czechoslovakia.

Among EU members there is considerable variation in how eager governments and national publics are to expand the reach of European institutions. EU membership is accompanied by ever increasing adjustments of domestic rules and regulations in response to supranational decision making and ECJ rulings.⁵⁴

See, for example, Stone Sweet and Brunell 1998; and Stone Sweet and Sandholtz 1998.

Some governments and national publics are more favorably disposed toward supranational commitments whereas others are more skeptical of such commitments.⁵⁵

For alternative explanations of the variation in public opinion, see Gabel 1998; and McLaren 2002.

This is especially relevant as the EU currently does not include an extensive set of formal human rights commitments. This leads to the hypothesis that governments that are more favorably disposed toward the EU are also more likely to appoint activist international judges. Similarly, there are a number of non-EU members that would qualify for EU membership but have not applied. These countries, such as Iceland, Norway, and Switzerland, have traditionally been particularly cautious about making commitments to supranational institutions. As such, if preference toward supranational integration is the driving force behind the court's activism, then one would expect governments from those countries to appoint less activist judges.

Theory: Domestic Legal Systems

There are reasons to be skeptical that governments appoint ECHR judges for political reasons. First, while the link between the ideologies of governments and judges seems obvious from an U.S. perspective, judicial appointments may be much less motivated by political considerations in many European countries. To the extent that research exists, there is little evidence that divisions on Europe's domestic constitutional courts are determined by partisanship.⁵⁶

See Von Brünneck 1988; and Schwartz 1993, although Magalhães 1998 provides evidence for the importance of partisanship on the Portuguese Constitutional Court.

Second, judges have strong professional norms that tend toward independence from the political process. Such norms may be strengthened in the collegial and relatively isolated setting of Strasbourg. Third, influencing individual judges is of limited consequence in controlling court decisions that are taken by simple majority votes. This point is modified somewhat by the important gatekeeping role that national judges play in the judgment process. The national judge usually serves as the rapporteur on cases that concern his or her country. The rapporteur makes a recommendation on the admissibility of cases, which can lead to a unanimous decision by a committee of three judges (including the rapporteur) to dismiss a case before deciding on its merits. About 93 percent of all cases are dismissed. As such, a national judge predisposed toward the raison d'état may have greater utility to governments than appears at first glance. Still, governments may simply not care enough to scrutinize the political preferences of appointees to international judgeships.

A plausible alternative is that variation in the activism of international judges stems from variation in the domestic legal systems of ECHR member states. The literature has focused mostly on the distinctions between common law and civil law countries, suggesting that many features of contemporary legal systems are inherited and thus exogenous to other aspects of the political systems of countries.⁵⁷

See, for example, Merryman 1985; and Glaeser and Shleifer 2002.

Due to colonial exploits, military adventures, and other reasons, the legal systems of France and England have had a vast influence on systems adopted in many other countries. Divergence in these legal systems was shaped by developments in the twelfth and thirteenth century, when France moved toward adjudication by royally controlled professional judges, while England developed a system in which courts enjoyed greater independence from royal interference.⁵⁸

Glaeser and Shleifer 2002.

As a consequence, judges have played a subordinate role to legislatures in legal systems designed after the French civil law system.⁵⁹

Merryman 1985 dates this back even further to the Roman judex.

In common law countries, on the other hand, judges are regularly asked to engage in broader interpretations of legal principles. This, the argument goes, would make judges from common law countries more likely to be activist in their role orientation than judges from French civil law countries.⁶⁰

Alivizatos 1995. It should be noted that much of the “new constitutional politics of Europe” has been the result of activism by judges from civil law countries, so it is thus not clear that the above characterization is even informative about judicial behavior at the domestic level (see, for example, Shapiro and Stone 1994; and Stone Sweet 2000.)

The divergence between legal origins is recognized in the internal organization of the ECHR in that its sections are distributed to maintain a balance between legal traditions.

More generally, domestic legal systems vary in at least four potentially relevant ways. First, in some systems judges have greater independence from the political process than in others. Potential institutional sources of independence are lifelong tenure, which shields judges from ex post political evaluations, and the use of case law as a source of law, which increases the implications of judicial decisions.⁶¹

La Porta et al. 2004.

Judges who are accustomed to independence from political interference may be more activist in their orientations. Second, not all legal systems formally assign a role to courts as evaluators of the constitutionality of legislative initiatives. Judges from countries where constitutional review is absent are perhaps less inclined to engage in more open interpretations of textual provisions than judges who commonly review the legality of initiatives from democratically elected officials.

Third, countries vary in the extent to which national judges are used to interpret international law. In monist legal systems, domestic judges are bound to apply rules of international law whereas in dualist countries, this is so only when international law is incorporated into national legislation.⁶²

Cassese 2005.

Although the Convention has been adopted into national law in virtually all member states, it may still be true that judges from monist legal systems interpret international law more broadly than judges who are accustomed to deferring to national level legislation.⁶³

I thank an anonymous referee for suggesting this and the following hypotheses.

Fourth, countries vary in their experience with activist international courts, especially the ECJ. Alter has argued that national judges came to see an interventionist ECJ as an ally in strengthening the domestic rule of law.⁶⁴

Alter 1998.

It may be, then, that judges from countries with a long tradition of ECJ activism are less restrained in their understanding of what an international court should do.

There is some ground for skepticism about these hypotheses. The view that judges who are “activist” on the national level would translate that activism to an international court relies on the notion that judicial behavior is driven by role conceptions into which judges are socialized. If judges more instrumentally determined their interests based on an assessment of the effects of their actions, one may well reach the exact opposite predictions: National judges who enjoy a high degree of independence may find that activism by an international court constitutes undesirable interference whereas judges from countries where courts are less secure from political interference may view an activist international court as a potential ally.

Estimating Levels of Activism and Restraint

The empirical strategy is first to estimate levels of judicial activism and restraint from observed dissents and then regress these on measures for the theoretical concepts introduced in the previous section. This section introduces the methodological issues, the data, and finally presents the estimated ideal points and checks these for face validity.

Method

I employ a probabilistic spatial model of judicial decision making to estimate ideal points from observed vote choices. This model has also been used to estimate ideal points of U.S. Supreme Court justices,⁶⁵

See, for example, Bafumi et al. 2005; Bailey, Kamoie, and Maltzman 2005; and Martin and Quinn 2002.

legislators in the U.S. Congress,⁶⁶

Poole and Rosenthal 1997.

and actors in international (semi-)legislative bodies such as the European Parliament⁶⁷

Hix, Noury, and Roland 2007.

and the United Nations General Assembly.⁶⁸

Voeten 2000 and 2004.

Ideal point models simultaneously estimate characteristics of judges and cases from observed vote choices. Moreover, similar to factor-analytic models, the estimates allow for an examination of the underlying continuum that divides judges rather than make strong assumptions about this. Unlike factor-analysis, the ideal point model is based on an explicit model of how judges translate their judicial ideologies into observed vote choices: the spatial model of voting.

Figure 1 illustrates the one-dimensional spatial model. Figure 1 looks at five judges who can indeed be depicted by their ideal points along a continuum from activism to restraint. Three hypothetical issues are represented by their cut-points. The probabilistic spatial model assumes that the further a judge's ideal point is to the left of an issue cut-point; the more likely that judge is to rule against the government. Conversely, the more a judge is to the right of the cut-point; the less likely it is that the judge will find a violation. There will be issues, such as issue 1, on which only the highly activist judge A is likely to find a violation (an example is the Turkish headscarf issue discussed earlier). Similarly, there will be clear violations of the Convention, such as issue 3, on which even most judges on the self-restraint side of the spectrum are likely to find a violation. The most contentious issues are like issue 2, where judges are more or less evenly split on whether an alleged violation warrants a negative judgment. Ignoring variability in the nature of cases could seriously affect estimates of judicial ideology. By the luck of the draw, some judges may vote on a disproportionate number of cases in which there were clear violations. Those judges would be labeled “activists” even if a moderate judge would have voted the same way, given the nature of the cases. This problem is exacerbated by the fact that ECHR judges vary considerably in the number of cases they voted on.

Illustration of the spatial model of judicial decision making

Unfortunately, one observes neither the judges' ideal points nor the issue cut-points. Item-response theory (IRT) can be used, however, to estimate both issue parameters and the ideal points of judges from what one does observe: whether judges ruled in favor of or against the government. IRT models were developed in the literature on educational testing, where they are used to simultaneously estimate the difficulty and discriminatory nature of test items and the ability of students from observed right and wrong answers on standardized tests. This allows evaluators to estimate standardized scores for students on tests such as the Graduate Record Examination (GRE) without requiring that all students answer the same questions. It is well understood that the probabilistic spatial model of voting is mathematically equivalent to the IRT model.⁶⁹

See Jackman 2000; and Poole 2005.

The basic intuition is that IRT models estimate the configuration of judicial ideal points and case cut-points that are most consistent with the coalition patterns observed in the data. One has n observed responses y (vote choices), where i = 1,..,n, by J judges on K issues. y_i = 1, if the vote choice is in favor of the government (that is, a finding of “no violation” by a judge), 0 otherwise. One can model these responses using a two-parameter item response model:

In this equation, θ_j reflects judge j's ideal point and α_k represents the cut-point on issue k as outlined in Figure 1. β_k is the discrimination parameter (similar to a factor loading). If β_k equals 0, variation in judicial ideal points is not informative about how judges vote on issue k. If β_k is large and positive, judges with positive values of θ have a high probability of voting in favor of the government. If β_k is large and negative, judges with positive estimates of θ are likely to vote against the government on this particular issue. Thus, the method offers an opportunity to explicitly test whether activism-restraint is indeed the main dimension of contestation: If the main division between judges is activism-restraint, then the discrimination parameters across issues should be positive. If, left-right were the dominant cleavage, then the sign of the discrimination parameters should vary across issues, as left-wing judges are activist on some issues (for example, gay rights), but not on others (for example, property rights). Thus, one does not need to assume what the main dimension of contestation is. Instead, one can verify this by examining the discrimination parameters after estimating the model.

Data

Although there are analyses of episodes of dissenting behavior,⁷⁰

See Arnold 2001; Bruinsma 2006; Bruinsma and de Blois 1997; Jackson 1997; and Schermers 1998. For legal analysis of dissenting opinions until 2000, see Rivière 2004.

this is the first study that examines the entire record between 1960 and 30 June 2006. While most judgments were unanimous or included only separate concurring opinions, dissents were not rare events: 900 of the 6,749 ECHR judgments included at least one dissenting opinion.⁷¹

Based on a search in the online catalog HUDOC, available at 〈http://www.echr.coe.int/ECHR/EN/Header/Case-Law/HUDOC/HUDOC+database〉. Accessed 12 June 2007. The search identified judgments that include the word “dissenting” in the separate opinions portion of the case. Since the full text of some cases is only included in one of the two official languages, a complementary search was performed in the French language portion of the database.

Many of the unanimous judgments were routine. For example, between 1 January 2000 and 20 June 2006 there were 1,377 judgments considering Italian violations of Article 6, paragraph 1. Almost all of these cases were charges of excessive delays in court proceedings, an issue for which the ECHR has developed standardized procedures. Dissents were common among cases that did not simply apply case law: 469 of the 1093 judgments that “make a significant contribution to the development, clarification or modification of its case-law” elicited at least one dissenting opinion.⁷²

These are judgments with importance level 1 in HUDOC catalog: “Judgments which the Court considers make a significant contribution to the development, clarification or modification of its case-law, either generally or in relation to a particular State.”

I excluded votes by judges on their home countries. The spatial model assumes that judges sincerely express their ideological position when voting. One may expect judges to use a different and perhaps strategic set of motivations when they vote on cases that concern their country of origin. Whether and why this is so is the topic of a separate paper.⁷³

Voeten 2006.

Some judgments drop out of the database, then, because they elicited only a dissenting opinion from the judge sitting on behalf of the respondent government. These were primarily judgments on standard applications of ECHR case law. For example, on 28 February 2002, the judge sitting on behalf of Italy dissented on a set of 133 judgments on alleged Italian Article 6 violations using the same dissenting opinion. Such judgments cannot reasonably be expected to contain information about ideological divisions between judges.

In many judgments, separate decisions were made on admissibility, the merit of violations of individual articles of the Convention, and just satisfaction (monetary compensation). Splits on just satisfaction decisions often followed naturally from decisions on the merit of the case. To the extent that they did not, judges did not always motivate why they voted differently on the just satisfaction decision than on the merit question. As such, I decided to exclude decisions on just satisfaction. Sometimes, decisions on violations of individual articles within a judgment exhibited differential patterns of disagreement between judges. For example, a government may have clearly violated Article 6 of the Convention, creating an issue such as issue 1 in Figure 1, but its alleged violation of Article 8 is much less clear and looks more like issue 2 in Figure 1. On twenty-four cases, at least half the judges expressed a dissenting opinion. Often, these were cases where some dissenters argued that no violation had occurred whereas others believed that violations of multiple articles had occurred, even though a majority could only be found on the occurrence of a subset of these violations. Treating such cases as single entries in the data set would ignore information about splits between judges. Therefore, different splits between judges were treated as separate entries in the data set.

The primary analysis contains only data on judgments that that were not unanimous. This is common in studies of judicial and legislative voting, given that unanimous votes do not provide information about variation in ideal points. Running the spatial model on unanimous votes is equivalent to running a logit model on a data set without variation in the dependent variable. One aside is that differently composed panels could have different propensities to be unanimous. Thus, unanimous votes could contain some information about ideal points.

The final data set contains votes on 709 issues with thirty-eight different respondent states. It includes ninety-seven judges who voted at least ten times, fifty-nine of whom voted at least fifty times on a contentious issue. The vast majority of issues are from recent years: 84 percent of all issues are from a judgment in 1990 or later, 54 percent of all issues come from the “new” post–Protocol 11 court. The data match the overall distribution of cases fairly well, with the notable exception that Article 6 cases (right to a fair trial) invited fewer dissents than their preponderance in the universe of judgments would suggest.⁷⁴

For a more detailed analysis of the universe of cases, see Cichowski 2006.

Validity of Activism-Restraint Estimates

I estimated the models using MCMC methods within a Bayesian framework, using the robust logistic model specified by Bafumi and colleagues.⁷⁵

See Bafumi et al. 2005; Pr(y_i = 1) = ε₀ + (1 − ε₀ − ε₁)logit⁻¹β_k(θ_j − α_k), with ε₀ ∼ dunif(0,.1),ε₁ ∼ dunif(0,.1) θ_j ∼ N(μ_θ,σ_θ²), for j = 1,…, J,α_k ∼ N(μ_k,σ_k²) and β_k ∼ N(μ_β,σ_β²), for k = 1,…, K. The parameters were normalized after estimation was complete: θ_j^adj = (θ_j − θ)/s_θ,α_k^adj = (α_k − θ)/s_θ,β_k^adj = (β_k) * s_θ.

The polarity of the scale is identified by restricting one judge, Thor Vilhjálmsson, to have an ideal point to the right of another judge, Jan De Meyer. These two judges have served together for a long time and infrequently vote together, so their ideal points can be expected to be far apart, regardless of the content of the dimension. The one-dimensional model classifies 84.9 percent of all vote choices correctly.⁷⁶

This is the mean of the posterior distribution of classification percentages.

Figure 2 plots the estimated ideal points and the 95 percent posterior credible intervals of the ninety-seven judges that voted at least ten times. The larger the posterior credible interval (the line in the figure) the more uncertainty there is about the precise location of a judge's ideal point. Large posterior intervals mostly reflect the small number of vote choices of some judges, meaning that the ideal points of some judges can be estimated more precisely than those of others.

Estimates of levels of activism ECHR judges (95 percent credible intervals)

There are two ways to verify that the plotted divisions indeed reflect variation in levels of activism and restraint. First, to the extent that the positions of judges are known, the estimates have considerable face validity. The British judge Sir Gerald Fitzmaurice is estimated to be the most “extreme” judge on the restraint side. Fitzmaurice was a legal adviser in the British foreign office who became a well-known academic and a judge on the International Court of Justice (1960–1973) before coming to the ECHR in 1973. He is often cited as the prototype of the “tough conservative” judge,⁷⁷

Merills 1988.

who sometimes angered his colleagues with long and opinionated dissents.⁷⁸

Bruinsma and Parmentier 2003.

Another judge with a precisely estimated ideal point on the restraint side, the Austrian judge Franz Matscher, spent seventeen years in the Austrian diplomatic service before joining the ECHR in 1977 and has openly expressed his concern about the court's activist tendencies by writing that the ECHR has “entered territory which is no longer that of treaty interpretation but is actually legal policy-making.”⁷⁹

Matscher 1993, 70.

The Swiss president of the court, Judge Luzius Wildhaber, proclaimed himself to be “slightly more to the self-restraint side” in comparison to his colleagues,⁸⁰

Ibid.

which is confirmed by his ideal point estimate.

Similar observations can be made about many of the judges with estimated ideal points on the activist side of the spectrum. The Belgian judge Françoise Tulkens (the lone dissenter on the Turkish headscarf case) asserted in a recent interview that: “One can speak of judges who are concerned about problems of the raison d'état and others who sympathize with the applicants. The raison d'état is more present here than I would have thought possible.”⁸¹

Bruinsma 2006, 211.

The Italian judge Josep Casadevall (serving for Andorra) declared: “Personally I am a judicial activist as to Article 6 with a bent to enlarge its scope.”⁸²

Ibid., 225.

The French judge Louis-Edmond Pettiti, who has a background as attorney (avocat) and human rights activist, had been singled out as the prototypical “activist” in an earlier study.⁸³

Jackson 1997, 25.

Similarly, the Maltese judge Giovanni Bonello has defended 170 human rights lawsuits in the Maltese courts as a private practitioner and is a self-identified activist.⁸⁴

Bruinsma 2006.

The “most” activist judge, Cypriot judge Loukis Loucaidis, has published widely on human rights, motivated primarily by concerns over individual rights in Turkish Cyprus.⁸⁵

Loucaides 1995, 2003, and 2004.

A second verification mechanism is the examination of the discrimination parameters. All votes were coded as 1 if they were decisions in favor of the government and 0 if they went against the government. If activism-restraint is the main dimension of contestation, then one would expect a positive relationship between the ideal points of judges and their propensity to vote in favor of the government. As such, one would expect the discrimination parameters to be positive. If the main dimension of contestation reflected some other meaning, there is no a priori expectation of such a positive relationship. For example, if it reflected left-right conflict, one may well expect that judges on the right have a tendency to vote against governments on property rights issues but not on abortion or gay rights. There are no issues on which the 90 percent credible interval of the discrimination parameter falls below 0. Thus, one can comfortably assert that the main dimension indeed reflects activism-restraint.

Why Are Some Judges More Activist Than Others?

Why are some judges more activist than others? This section seeks to answer this question by regressing indicators derived from the various theoretical perspectives on the estimates of judicial activism introduced in the preceding section. Given that the theories are not mutually exclusive, they are evaluated in multiple regression analyses. The dependent variable is the mean of the posterior distributions for each judge, as depicted in Figure 2. Higher scores indicate more restrained judges and lower scores more activist judges. I use weighted least squares to adjust for the variation in precision of the ideal point estimates.⁸⁶

The weight is 1 divided by the standard deviation of the posterior.

There are five judges from Andorra, Liechtenstein, and San Marino for whom there is little data on the independent variables, so they have been excluded from all analyses. There is also limited coverage on the independent variables for the four judges in the data who were elected before 1970.⁸⁷

Note that given the small number of votes before the 1980s, most of the early judges were dropped from the sample because they had not participated in at least ten controversial votes.

Table 1 presents the main results. All independent variables are measured at the time of a judge's election or are time invariant (such as legal origins). Model 1 includes indicators for all hypotheses. The other models use alternative indicators for the various theoretical concepts. Generally, these models are estimated on a subsample of judges due to limited data availability. For reasons of economy and clarity, I present the results in separate sections for each of the individual hypotheses, starting with an exploration of overall trends.

Weighted least squares regressions on levels of activism

Trends in Levels of Activism

A key observation that motivates this article and much of the literature on international courts is the perception that international courts tend to become more activist as time progresses. Early on in the life of a court, judges may be hesitant in their activism as they are uncertain as to the implementation of the court's decisions as well as the institutional future of the court itself. Judges that enter into a more mature court may well perceive an increased sense of institutional security of the court and hence feel more liberated to make decisions that may be perceived as unpopular by governments. To capture this effect, each model includes a linear time counter. This time counter is negative and significant in virtually all specifications.

Figure 3 plots the mean ideal point estimate for ECHR judges from 1978 to 2006.⁸⁸

1978 was chosen as it was the first year with at least twenty judges in the data set.

This presents an alternative way of looking at the underlying trends in that it focuses on the impact of replacement on the court as a whole. The overall trend toward increased activism concurs with observations from legal scholars⁸⁹

For example, Mowbray 2005.

and is consistent with finding that the ECHR increasingly rules for the applicant (and thus against the government).⁹⁰

Cichowski 2006.

The trend in Figure 2 is a function of the replacement of restraint judges with more activist ones, not the effect of individual judges becoming more activist given that the data are not sufficiently informative to allow for the estimation of a model in which the ideal points of individual judges vary over time.⁹¹

For such exercises in the context of the Supreme Court, see Martin and Quinn 2002; and Bailey 2006.

In this regard, especially the first election for the post–Protocol 11 court (1998) was important as it introduced a large number of new judges that were more activist than the judges they replaced.⁹²

Note that we can anchor the space because there generally is considerable continuity in the composition of the bench, even in 1998.

(Introducing a dummy variable for the post–Protocol 11 court has no effect in the models presented in Table 1.)

Temporal change in mean activism of ECHR judges

Democracy and Domestic Human Rights Protection

Do governments that offer better protections for individual rights generally appoint judges that are more activist than governments that provide less broad guarantees to their citizens? The main indicator for domestic rights protection is a country's Freedom House civil liberties score at the time of a judge's election, running from 1 (extensive civil liberties) to 7 (very restrictive civil liberties).⁹³

See 〈www.freedomhouse.org〉. Accessed 12 June 2007.

These scores are widely used and have greater coverage than other available indicators. As Table 1 shows, the coefficient on civil liberties has the predicted sign but does not reach conventional levels of statistical significance. This conclusion also holds in bivariate regressions and in various leaner specifications of the model (not shown in table). The result is robust to using alternative indicators of domestic human rights observance: political terror scores based on Amnesty International reports⁹⁴

Taken from Mark Gibney, at 〈http://www.unca.edu/politicalscience/images/Colloquium/faculty-staff/gibney.html〉. Accessed 12 June 2007.

and Polity democracy scores.⁹⁵

Results available upon request from author.

The ECHR also aims to correct deficiencies in the functioning of domestic legal systems. Hence, it may be that countries with legal systems that are generally perceived to function poorly are less likely to pick activist judges. I therefore also included a six-point “law and order” scale developed by the Inter Country Risk Guide published by Political Risk Services.⁹⁶

See 〈http://www.prsgroup.com/ICRG.aspx〉. Accessed 12 June 2007. Data is available from 1980 onwards.

This indicator is also not significantly correlated with levels of judicial activism (results available from author).

In short, there is no evidence that countries that appear vulnerable to an activist ECHR tend to pick judges that favor self-restraint. One reason for this finding may be that there is too little variation: countries that are willing to commit to the ECHR's jurisdiction generally have high levels of domestic rights protection. Most judges come from countries with either a Freedom House score of 1 (54 percent) or 2 (26 percent). On the other hand, even casual inspection of Figure 2 shows that the most self-restraint judges are generally not from the remaining poor performers but rather from countries that have strong records in protecting human rights, such as Iceland, Austria, and the United Kingdom.

Moravcsik posits a more sophisticated relationship between democracy and preferences for binding international human rights courts.⁹⁷

Moravcsik 2000.

An extrapolation of his republican-liberal perspective is that both established democracies and nondemocracies ought to be hesitant toward appointing activist judges, while transitionary democracies have most to gain from active international human rights enforcement. Consistent with Moravcsik, I define a transitionary democracy as a country that has been a democracy for less than thirty years, where democracy is defined as a country with a Polity score of 7 or higher,⁹⁸

Reasonable extrapolations were made to democracies (for example, Luxembourg) that Polity did not cover.

as is common in the literature. There is no evidence for the hypothesized relationship: the coefficient on transitionary democracy has the wrong sign and is insignificant. This basic finding does not change in leaner or alternative specifications of the model, including models that separate the reference categories in Table 1 (“stable democracy” and “no democracy”). In addition, I have estimated models that examine a curvilinear relationship between the number of years a country has been democratic and the relative activism of its appointee. Again, there is no relationship between the duration of democracy and ECHR appointments. These results do not refute Moravcsik's theory about the ECHR's origins but rather show that it cannot be extended to the logic of the appointment process.

Left-Right Politics

Are judges appointed by left-wing governments more activist than judges appointed by right-wing governments? It is difficult to adequately measure the ideological composition of governments in a comparative way. The most straightforward approach is based on party families. A measure for left-wing government ideology is the percentage of total cabinet seats, weighted by days of a calendar year, held by social-democratic and other left parties.⁹⁹

Armingeon et al. 2005. The data for Eastern European countries is computed from Armingeon and Careja 2004. Missing information is imputed from Beck et al. 2001.

Similar indices can be constructed for right-wing and centrist parties. I computed a combined index that takes the value −1 if all seats are taken by leftist parties, and 1 if all seats are taken by parties of the right. This measure has the important advantage over alternatives, such as indicators based on party manifestoes or expert judgments, that data is available for virtually all judges.

The results show that this indicator has the predicted effect. A government composed of only left-wing parties is estimated to appoint a judge who is around .40 points more activist than a government composed of only parties of the right. This effect is substantively important: it represents a shift of almost half a standard deviation in the distribution of judges. Yet, the standard error on the coefficient is large. It is not implausible that left-right divisions matter because left-right correlates (imperfectly) with preferences for international integration in general, and EU integration in particular. More on that in the next section.

European Integration

Do governments use judicial appointments to signal their human rights commitments to interested parties, especially the EU and its member states? I code aspirant membership with a dummy variable that takes the value 1 after a government has announced its intentions to join the EU and there have been at least some formal discussions with the EU with regard to a country's membership.¹⁰⁰

In the absence of a clear announcement date, I coded a country as aspiring in the ten years leading up to their EU membership.

The regression results show that judges from aspirant EU countries are on average located .53 points toward the activist pole of the scale in comparison to the reference group, judges from non-EU countries that do not have aspirant membership status. This is a very sizeable effect: a judge who would be expected at the 25th percentile score based on other characteristics is estimated to have the median activism score due to the effect of EU membership aspirations. The standard error of the coefficient is relatively small. The coefficient is also robust to the inclusion and exclusion of other variables,¹⁰¹

Aside from the variables in Table 1, I have also assessed its robustness to the inclusion of GDP per capita.

as well as to slight alterations in the definition of aspirant membership. Given that the model controls for socialist heritage and democratic transition, one can also exclude that aspirant EU membership is merely picking up the effects of those variables. The coefficients also suggest that aspiring EU members appoint more activist judges than current EU members, holding other things equal. In all, this is strong evidence for the signaling hypothesis.

The second hypothesis regarding European integration is that governments more favorably disposed toward integration are more likely to appoint activist judges. I created a measure for government preferences toward the EU using data from expert surveys from twenty-five of the twenty-seven current EU members, including Bulgaria and Romania.¹⁰²

Expert survey data come from Ray 1999; Hooghe, Marks, and Wilson 2002; and Marks et al. 2006. The expert survey data is available from 〈http://www.unc.edu/∼gwmarks/data.htm〉. Accessed 12 June 2007.

The expert survey rates the main political parties in each country on a 7-point scale ranging from “strongly opposed to European integration” to “strongly in favor of European integration.” It also rated the parties on a similar 7-point left-right scale. I created government scores by averaging the scores for the parties that held the executive (prime-minister or president), ministry of justice, and ministry of foreign affairs.¹⁰³

Based on data collected in Druckman and Roberts 2005; and Druckman and Warwick 2005.

This latter decision is based on a qualitative assessment that these three posts were generally determinative in the actual nomination processes.

The results in Model 2 show strong support for the hypothesis that government support for European integration is an important driving force behind judicial ideologies. A government fully on the anti-European integration side of the scale is expected to appoint a judge about 1.6 points further to the self-restraint side of the scale than a government on the pro-integration side. The standard deviation of the dependent variable is .90, so this is a sizeable effect. In short, governments less favorably disposed toward European integration also appoint judges who are less favorably disposed toward extending the reach of a European supranational court. Thus, there is a demonstrable political logic behind the appointment process.

It should be noted that governmental preferences toward European integration could only be determined for slightly more than half the sample. I estimated a similar model with an alternative measure of governmental preferences toward the EU, based on party manifesto data.¹⁰⁴

This measure took the percentage of positive minus negative statements on EU integration for parties. Data are from Armingeon et al. 2005.

This analysis reveals similar results but on an even smaller sample of judges.¹⁰⁵

Where coefficient −0.165, significant at 10 percent level, only thirty-three cases.

These subsamples are not representative for all Council of Europe countries, in that there is no data for countries who do not meet the criteria for membership (such as Russia and Moldova) or who meet the criteria but do not consider membership (such as Norway, Switzerland, and Iceland). Visual inspection of ideal point estimates suggests that the latter omission may understate the estimated effect of government preferences in that these governments frequently appointed restrained judges and are known to be relatively skeptical of supranational integration.

Domestic Legal Systems

Are international judges from civil law countries, where judges have traditionally played a subordinate role, less likely to be activist than judges from common law countries? The results from Table 1 show no evidence for this proposition.¹⁰⁶

Data on legal origins are from La Porta et al. 1998.

In the regressions, judges from British common law countries are the reference group. There is no indication that these judges are more activist than judges from French civil law countries: the coefficient on French civil law has the wrong sign, although it is statistically indistinguishable from 0. There is some evidence that judges from former Socialist countries and German and Scandinavian civil law countries are more likely to be more on the self-restraint side, although the standard error on these coefficients are large. In an ANOVA analysis, one cannot reject the null hypothesis that judges from different legal origins are equally activist.¹⁰⁷

p = 0.141.

To further scrutinize the potential impact of domestic legal systems, I estimated models that code institutional protections for judicial independence as well as the presence of constitutional review in countries.¹⁰⁸

La Porta et al. 2004.

Both variables are normalized from 0 to 1 where higher values equal a higher degree of judicial independence and constitutional review respectively. Unfortunately, these measures are only available for a subsample of the data. As Model 4 shows, neither variable has an impact on observed judicial activism.

There is also no evidence that judges from countries that have been in the EU for a longer period, and thus, presumably, are more accustomed to an activist international court, are more activist (although the coefficient in Model 1 has the predicted sign). Finally, Model 3 tests whether judges from monist countries were more activist than judges from dualist countries. Unfortunately, this data is also only available for a subsample.¹⁰⁹

Voigt 2006.

While the coefficient on monism is negative, as predicted, it is small and not significantly different from 0.

In sum, then, these analyses do not provide evidence for the thesis that judges transport their supposed domestic roles into the international arena.

Conclusions

A first conclusion from this study is that “activism–self-restraint” constitutes the most prominent divisions between ECHR justices. That is, judges vary in the extent to which they show deference to governments when assessing whether a violation has occurred. That this dimension rather than a more value-based cleavage emerges as the most prominent dimension is telling, and points to an interesting difference with the U.S. Supreme Court. It means, among others, that conflict over the proper reach of the institution is at the heart of divisions between judges. It also suggests that the common assumption that international judges share an interest in the expansion of the reach of their court is unwarranted. This assumption is fundamental in current theoretical explanations for the agency of international courts.

Second, the analysis implies that politics plays a role in international judicial appointments, as it does nationally (at least in the United States). Governments are heterogeneous in their preferred levels of activism of international courts and this warrants attention in one's conceptualizations of the interactions between governments and courts. As such, the results challenge the assumptions underlying both the trustee model,¹¹⁰

See Alter 2006 and forthcoming.

where judges are selected mostly for professional reasons, as well as conventional principal-agent models that assume that international judges have an inherent expansionary drift.¹¹¹

See, for example, Garrett 1995; Garrett, Kelemen, and Schulz 1998; and Kelemen 2001.

Third, the results suggest that governments, at least to some extent, use judicial appointments to fill the ECHR with agents that match their preferences. There are reasons to suspect that such screening also occurs on other international courts. Steinberg reports, for instance, that both the EU and the U.S. Trade Representative conduct extensive interviews with candidates for the WTO's appellate body, with an explicit focus on the degree to which the candidate can be expected to have an expansive view of judicial decision making.¹¹²

Steinberg 2004.

Observers have charged that the ECJ became more restrained after the replacement of four ECJ judges following the controversial 1994 Codorniu judgment.¹¹³

Costa 2003, 744.

Alter found evidence that Germany and France have appointed judges to limit judicial activism, although she did not find similar instances in her other cases.¹¹⁴

Alter interviewed legal scholars and government officials in France, Germany, and the UK, as well as the Italian, Greek, Dutch, Belgian, French, German, British, and Irish judges at the ECJ; see Alter 1998, 139, fn. 62.

So far, however, scholars have not investigated the other side of the story: that some governments deliberately appoint activist judges. This conclusion would also fit recent scholarship that finds that while states did not create bodies such as the international criminal tribunals or the WTO's appellate body to create international legal norms, many governments have tolerated and sometimes even embraced such creative activities.¹¹⁵

See, for example, Broude 2004; and Danner 2006.

Unfortunately, it will be difficult to verify these assertions using the methods developed here. Dissents on other international courts are either shrouded in secrecy (WTO or ECJ) or are too limited in number to allow the type of statistical inquiry engaged in here (International Court of Justice and the International Criminal Tribunals).¹¹⁶

For a statistical analysis of ICJ votes that answers a different set of questions, see Posner and de Figueiredo 2005.

Theoretically, one would expect that selecting judges becomes more important as monitoring becomes more difficult in the absence of public dissents.¹¹⁷

Fearon 1999.

It may be that easily observable ECHR decisions have a screening function themselves: four of the twenty-five current ECJ judges previously served on the ECHR, a career step that is nonobvious given that EU law is a vastly different field of expertise than human rights law.

Studying judicial selection in the absence of public dissenting opinions may still be possible. ECHR judges whose previous careers were primarily as diplomats or bureaucrats are significantly less activist than are judges with other previous career tracks.¹¹⁸

Mean activism score for diplomats and bureaucrats was 0.35 (N=24), for others −0.11 (N=73). P-value is 0.029.

Thus, one could investigate which governments are most likely to appoint diplomats to international courts. More generally, understanding if and how governments influence the behavior of international organizations by carefully selecting the agents that staff them is an understudied topic. There is a considerable literature on the agency of international organizations but relatively little is known about the agents that presumably make the decisions that lead international organizations to escape the reigns of governments. Similar empirical strategies to the one followed in this article have been applied not just to courts but also to study how appointments shape the ideological composition of domestic bureaucratic agencies. Such research is promising for studying agency in the international arena as well.

Finally, the analysis reveals that the ECHR's composition has grown more activist over time as governments have tended to replace more restrained judges with more activist judges. The evidence suggests that this process is driven in good measure by European integration. Governments more favorably disposed toward European integration tend to appoint more activist judges. Aspiring EU members appoint more activist judges than other nonmembers. Neither of these findings disappears when controlling for the most obvious alternative accounts. If this finding survives further scrutiny, it suggests rather strongly that the EU has been the driving force behind the increased activism of the ECHR, even if there is little formal relationship between the EU and the ECHR. This spillover effect of the EU into the role of supranational institutions elsewhere in Europe warrants further attention. I have suggested that the Protocol 11 reform was driven by the issue of European integration but further qualitative evidence could be collected to test that hypothesis more systematically. Nevertheless, from the results presented in this article, it appears that the increased activism of the ECHR has a political logic and is not merely the result of “wayward judicial activism,” as Margaret Thatcher labeled the ECHR decision that overturned the British prohibition for gays to serve in the military.¹¹⁹

The Scotsman, 24 March 2002, 2.