Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Structural change and human rights norms : identity-based socialization processes in the international… Dragojlovic, Nicolas Isak 2004

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_2004-0434.pdf [ 4.44MB ]
JSON: 831-1.0091537.json
JSON-LD: 831-1.0091537-ld.json
RDF/XML (Pretty): 831-1.0091537-rdf.xml
RDF/JSON: 831-1.0091537-rdf.json
Turtle: 831-1.0091537-turtle.txt
N-Triples: 831-1.0091537-rdf-ntriples.txt
Original Record: 831-1.0091537-source.json
Full Text

Full Text

STRUCTURAL C H A N G E A N D H U M A N RIGHTS NORMS: IDENTITY-BASED SOCIALIZATION PROCESSES IN THE INTERNATIONAL S Y S T E M by NICOLAS ISAK DRAGOJLOVIC B.A. , Yale University, 2003 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF M A S T E R OF ARTS in THE F A C U L T Y OF G R A D U A T E STUDIES Department of Political Science We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH C O L U M B I A August 2004 © Nicolas Isak Dragojlovic, 2004 i U B C L THE UNIVERSITY OF BRITISH COLUMBIA FACULTY OF G R A D U A T E STUDIES Library Authorization In presenting this thesis in partial fulfillment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Nicolas Isak Dragojlovic 16/08/2004 Name of Author (please print) Date (dd/mm/yyyy) Title of Thesis: Structural Change and Human Rights Norms: Identity-Based Socialization Processes in the International System. Degree: Master of Arts Year: 2004 Department of Political Science The University of British Columbia Vancouver, BC Canada page 1 of 1 last updated: 15-Aug-04 Abstract This thesis examines the international norms banning torture and the death penalty, codified in the Convention against Torture and Cruel and Unusual Punishment and the Second Optional Protocol to the International Covenant on Civil and Political Rights aiming at the abolition of the death penalty. It seeks to explain the differing systemic outcomes experienced by the two norms: why the Convention against Torture is adhered to by most of the state system, while the Second Optional Protocol remains largely a European institution. The operating hypothesis emerges from the recent literature on norms in international relations, and argues that while the norm against torture has undergone a "norm cascade" and initiated a process of socialization in the international system, the norm abolishing the death penalty remains in the "norm emergence" phase, where countries adopt it only as a result of domestic processes. A quantitative analysis is used to test the empirical plausibility of the hypothesis. Binary time-series cross-section (BTSCS) data with a country-year unit of analysis are analyzed using logit regression, with temporal dependence accounted for by a series of temporal dummy variables. A number of identical models are estimated for both norms, comparing the relative significance of the variables in each case. The results appear to support the hypothesis that accession to the Convention against Torture is much more associated with systemic socialization variables than accession to the Second Optional Protocol, which is primarily associated with domestic and regional variables. This points to the relevance of the norm "life-cycle" theoretical framework to the evolution of the norms examined in the thesis, and suggests that it is a generalizable social dynamic. ii Table of Contents Abstract i i Table of Contents i i i List of Tables v List of Figures vi List of Abbreviations vii Acknowledgements viii INTRODUCTION 1 CHAPTER I Norm Diffusion and the Universal Human Rights Regime 4 1.1 Torture: Historical Background 7 1.2 Death Penalty Abolitionism 11 1.3 But are the Norms Comparable? 17 CHAPTER II Theorizing Structural Change and Human Rights Norms 20 2.1 Neo-liberal Theory 21 2.2 Hegemonic Stability Theory 23 2.3 Republican Liberalism 25 2.4 Constructivist Theory, Norms, and Structural Change 26 2.5 Conclusion 37 CHAPTER III Method and Data 40 3.1 Method: Event History Analysis v. Binary Time-Series Cross-Section 40 3.2 Data and Coding 43 3.3 Conclusion 55 CHAPTER IV Empirical Analysis: Contrasting the Rates of Adhesion to the C A T and the SOP 57 4.1 Basic Model 57 4.2 Temporal Dependence 63 4.3 Conclusions 69 CONCLUSION 71 Appendix A 73 ii i Appendix B Data and Coding 77 Appendix C Analysis and Results 84 References 87 iv List of Tables Table A.l State Parties to the Convention against Torture 73 Table A.2 State Parties to the Second Optional Protocol 75 Table B.l Systemic Variables 80 Table B.2 Regional Proportion of States Party to the C A T 83 Table B.3 Regional Proportion of States Party to the SOP 83 Table C l Correlations of Systemic Variables 84 Table C.2 Logit estimates for Models 2 and 2.1 85 Table C.3 Logit estimates for Models 3, 3.1, and 3.2 86 v List of Figures Figure A . l Rate of Adhesion to C A T 74 Figure A.2 Proportion of System Party to C A T 74 Figure A .3 Rate of Adhesion to the SOP 76 Figure A.4 Proportion of System Party to the SOP 76 Figure B.l Growth in System Membership 77 Figure B.2 Systemic Material Power Associated with C A T 78 Figure B.3 Systemic Material Power Associated with SOP 79 Figure B.4 Scatter-plots of Systemic Independent Variables 81 vi List of Abbreviations BTSCS - Binary Time-Series Cross-Section C A T - Convention against Torture and Other Cruel, Inhuman, or Degrading Punishment CINC - Composite Index of National Capability COW - Correlates of War IGO - International Governmental Organization ILO - International Labor Union INGO - International Non-Governmental Organization IR - International Relations (the discipline) NGO - Non-Governmental Organization SOP - Second Optional Protocol to the International Covenant on Civi l and Political rights aiming at the abolition of the death penalty T A N - Transnational Advocacy Network UDHR - Universal Declaration of Human Rights WWII - World War II vii Acknowledgements First and foremost, I would like to thank my supervisor, Richard Price, for his help and direction throughout the research and writing process. I would also like to thank Fred Cutler for his help in designing the statistical part of the project, and for looking over the methods and analysis sections in draft form. Last and not least, I would like to express my gratitude to my father, who read an early draft and provided valuable feedback, and to my mother, who's always provided good advice and much appreciated encouragement and support. viii Human rights issues have come to influence an increasing proportion of international relations in the post-WWII period. As Donnelly (1986) notes, "[h]uman rights are regularly addressed today in bilateral foreign policy and in a variety of multilateral schemes" (605), and ultimately they comprise an issue-area that lies at the core of the United Nations system. The human rights regime is highly formalized, and includes a variety of legal instruments, including the Universal Declaration on Human Rights and the International Covenants on Economic Social and Cultural Rights and on Civil and Political Rights, as well as a large number of'"lower level' regional and single-issue human rights" arrangements (620). While the behavioral impact of such treaties is a valuable area of inquiry in its own right (see Goodman and Jinks 2003, for example), this thesis focuses on the dynamics of norms. Situated in the second wave of norm scholarship in international relations, it examines the dynamics of what Wendt (1999, 185-6) refers to as "structural change" - change in the ideational macro-structure of the international system - rather than on whether norms "matter". In as much as the human rights issue-area tends to be dominated by normative as opposed to functional discourse and focuses on what is fundamentally the domestic behavior of states, traditional IR theories such as neo-realism and neo-liberalism, which focus on material interests, are largely irrelevant. The empirical focus on human rights norms, then, allows this thesis to focus on their evolution and diffusion. Specifically, the norms against torture and against the use of the death penalty are chosen as case studies. While they draw on general human rights principles, they are codified primarily by the Convention against Torture and Other Cruel, Inhuman or 1 Degrading Treatment or Punishment (CAT), which was adopted by the UN's General Assembly in December of 1984, and the Second Optional Protocol to the International Covenant on Civi l and Political rights aiming at the abolition of the death penalty (SOP), adopted in December of 1989. Accession to or ratification of the treaties can be considered an indication of a state's acceptance of norm's status as a legitimate international social norm, even if it is only part of a broader process, and may not indicate full internalization of - and certainly not compliance with - a norm (Goodman and Jinks 2003,172-3). As such, the C A T has gained widespread acceptance, with 133 state parties and 12 non-party signatories as of November 2003, while the SOP has only garnered 50 parties and seven signatories (United Nations High Commissioner for Human Rights 2003). The norm against torture, moreover, is becoming increasingly . institutionalized, with the adoption in 2003 of a hew optional protocol to the C A T that provides for regular inspections of state parties' "places of detention" (Dennis 2003, 371). The SOP, on the other hand, remains a highly contested norm even within the "western" community, with strong opposition by the United States to a moratorium on the death penalty. This thesis investigates why the norm against torture has become an essentially uncontested component of the systemic collective identity of states, while there is little agreement on the death penalty. Ultimately, it argues that the different outcomes can be explained in reference to the socialization processes underlying the spread of the two norms, and investigates the hypothesis empirically using quantitative historical country-unit data. 2 Chapter 1 provides a historical background of the origins of the two norms, tracing the emergence of the two treaties, and lays out the empirical puzzle that lies at the heart of this thesis. Chapter 2 considers alternative explanations for the different outcomes of the C A T and the SOP, ultimately laying out a social constructivist theoretical framework based on Finnemore and Sikkink's (1998) sketch of the life-cycle and evolution of norms in the international system. Chapter 3 discusses the choice of model - binary time-series cross-section analysis - and documents the data structure and coding employed in the compilation of the datasets. Chapter 4 conducts the empirical analysis of the data, finding significant support for the social constructivist account. Finally, the conclusion considers the broader implications of the results for the study and practice of norms in the international system. 3 Chapter 1 Norm Diffusion and the Universal Human Rights Regime The advent of international campaigns to secure and protect the rights of whole classes of individuals preceded the twentieth century. Human rights as a coherent issue-area, however, did not emerge as an integral and legitimate component of foreign policy until after the Second World War. A number of studies document the emergence in the nineteenth century of transnational movements aimed at the abolition of abusive practices such as slavery (Keck and Sikkink 1998, 41-51; Nadelmann 1990, 491-498), footbinding and female genital mutilation (Keck and Sikkink 1998, 59-72), and prostitution or "white slavery" (Nadelmann 1990, 513-516), as well as campaigns in support of positive rights such as women's suffrage (Keck and Sikkink 1998, 51-58; Ramirez, Soysal and Shanahan 1997). Nadelmann (1990) also focuses on the emergence of "prohibition norms" aimed at criminalizing behavior such as privateering and drug trafficking that, while not concerned with human rights per se, are similar to subsequent human rights norms that aim to abolish given practices in domestic contexts. Finally, organizations such as the International Labor Union (ILO) - an international non-governmental organization tasked with promoting equal rights and solidarity for workers - emerged in the late nineteenth and early twentieth centuries (Donnelly 1986, 614). A l l these early efforts, however, took place in a transnational and non-governmental context, involving private activists and individual governments at the domestic level. In other words, human rights issues - defined by Keck and Sikkink (1998, 27) as issues "involving bodily harm to vulnerable individuals" or " legal equality of opportunity" - did not become a formal component of foreign policy for states until after World War II. 4 Donnelly (1986) notes that the war precipitated a "decisive break" with the existing understanding of human rights. The abuses perpetrated by the Axis powers "shocked the conscience of the international community," resulting in "a brief period of enthusiastic action, culminating in the passage in 1948 of the Universal Declaration [of Human Rights (UDHR)]" (614-15). That is, the emergence of a human rights regime as a supplement and elaboration of humanitarian norms is generally seen as a direct reaction to the barbarity of the war, an account bolstered by the signing of the Convention on Genocide - a response to the horror of the Holocaust - only a year later. The UDHR - which Risse and Sikkink (1999,1) argue was seen as "not a binding treaty, but rather a statement of principles" at the time of its signature - comprises the core of a universal human rights regime formed under the umbrella of the United Nations, which remains the principal forum for the elaboration of new human rights norms. Along with the two international Covenants on Civi l and Political and Social and Cultural rights and their respective protocols, they "comprise what is termed the 'International Bi l l of Rights'" (Schabas 2002, 23). They codify a whole raft of rights, ranging from the "right to life, liberty and security of person" (Universal Declaration, Article 3) to the "right to take part in the government of [one's] country, directly or through freely chosen representatives" (Universal Declaration, Article 21). To an extent, they codify customary international law (Schabas 2002, 23), and enunciate norms that "are widely viewed as more or less binding international standards" (Donnelly 1986, 608). In short, these foundational treaties define the human rights issue area as an international concern. They also underpin various regional human rights regimes, notably those in Europe and the Americas (Schabas 2002, 23), and are supplemented by a 5 large number of single-issue regimes - such as the 1964 International Convention on the Elimination of A l l Forms of Racial Discrimination (Donnelly 1986, 630) - that aim "to induce universal participation in a single area of mutual interest" (Donnelly 1986, 628). The universal human rights regime, then, consists of a wide and generally coherent body of norms - most of which are declaratory, but some of which eventually evolve into more substantive norms - that, given their concern "with how a government treats inhabitants of its own country," aim to regulate what is essentially a national issue (Donnelly 1986, 616). That is, unlike 'global prohibition regimes' (Nadelmann 1990) and most other international norms, human rights norms "are not designed primarily to regulate... societal interactions across borders, but to hold governments accountable for purely internal activities" (Moravcsik 2000, 217).' As such, they emerge, by definition, from normative and ideational concerns that are mostly cosmopolitan - concerned with the interaction between individuals or between the state and the individual, and not with interstate relations - the traditional purview of foreign policy (Nadelmann 1990, 483). In short, they aim "to provide a set of rules for the relationship between the individual and government, bearing in mind their fundamental inequality of power" (Burgers and Danelius 1988, 5). Despite mostly deriving their origins in Western practice, human rights norms present an interesting case for studying the way in which norms are created and diffused in the international system. The relative unimportance of functional imperatives for the adoption of a human rights norm allows the analyst to concentrate on the process of normatively motivated adoption and internalization that ultimately governs the 1 This concern is further considered in the discussion in Chapter 2 of how neo-liberal institutional theory might be applied to the human rights issue-area. 6 emergence of all norms. Single-issue regimes, moreover, which tend to be defined by specific legal instruments, avoid the uncertainty associated with treaties of a broader scope in that they tend to enunciate specific, substantive human rights norms that can be observed as they are adopted by increasing numbers of states in the system. This thesis chooses to focus on two recent treaties that comprise part of what Risse and Sikkink (1999, 2) call the "central core of rights - the right to life... and freedom from torture." They are: 1) the Convention Against Torture and Other Forms of Cruel, Inhuman or Degrading Treatment or Punishment of 1984 and 2) the Second Optional Protocol to the International Covenant on Civil and Political Rights Aiming at the Abolition of the Death Penalty of 1989. The two instruments define norms banning two state practices in the arena of criminal justice - torture and the death penalty - that have been framed as human rights issues due to the fact that they can cause "bodily harm to vulnerable individuals" (Keck and Sikkink 1998, 27). Torture: Historical Background The practice of torture as a means of interrogation and punishment, like so many other human rights abuses, has taken place in all historical periods. For centuries torture was regarded as a fundamental part of the criminal justice system, and became alarmingly prevalent in Europe during the late Middle Ages. Ultimately, the first movement to abolish the practice, condemning it "on both legal and moral grounds", emerged in era of the Enlightenment, in which it was largely discontinued as a "routine element of criminal procedure" (Burgers and Danelius 1988, 10). Indeed, by the end of the nineteenth century, most of the continent, in an example of an early customary human rights norm, 7 had eliminated provisions in criminal statutes for the use of torture. That said, the emergent norm against torture suffered a severe setback with the rise of the totalitarian regimes in the first half of the 20 t h century. They "deliberately employed torture as one of the mainstays of their power," primarily as a means to "spread terror among the population" (Burgers and Danelius 1988,10). Unsurprisingly, considering that the modern human rights regime largely emerged as a reaction to the WWII-era abuses, the 1948 UDHR specifically targets the practice of torture, declaring that "[n]o one shall be subjected to torture or to cruel, inhuman or degrading punishment" (Universal Declaration, Article 5; see also Barrett 2001a, 1-2). This prohibition is echoed in a number of other international legal instruments relating to human rights, including the Covenant on Civi l and Political Rights, the European Convention for the Protection of Human Rights and Fundamental Freedoms of 1950, the American Convention on Human Rights of 1969, the African Charter on Human and Peoples' Rights of 1981, and the Universal Islamic Declaration of Human Rights (Burgers and Danelius 1988,11). In short, while it is clear that torture is still practiced by a large number of states (Cook 1991, 172), the practice is widely condemned and almost universally accepted as a reproachable human rights abuse. One might argue, then, that there exists a customary norm against torture in the international system, such that it is "international law applying equally to States which are not parties to any of these conventions" (Burgers and Danelius 1988,12). That is, while the prohibition of torture "naturally appears in the various international instruments, the mention is to some extent superfluous, because there are also peremptory norms, rules of jus cogens, enshrined by international custom" (Schabas 2002, 20). While the condemnation of torture is near 8 universal, however, none of the foundational instruments provides a useful definition of the practice (Barrett 2001b, 2), and as such, it is unclear that the customary norm exerts a satisfactory constraint on states with respect to torture. This deficiency is largely rectified by more recent international treaties, most significantly the Convention against Torture of 1984. The Convention Following intense long-term lobbying by European and Latin American states in general, and Sweden and the Netherlands in particular (Burgers and Danelius 1988, 12, 103), the C A T was approved by consensus by the General Assembly of the United Nations on December 10, 1984. It not only provides a rather detailed definition of torture that builds on that expressed in the Declaration on the Protection of A l l Persons from Being Subjected to Torture and Other Cruel, Inhuman or Degrading Treatment or Punishment of 1975 (Barrett 2001b, 19), but it also outlines procedures for monitoring and enforcement in its second part, which establishes a Committee against Torture to "receive and review periodic reports from states parties every four years" (Donnelly 1986, 631). Moreover, in as much as the C A T is a convention - a distinct class of instrument - , it creates a legally binding obligation for state parties to respect the absolute prohibition established by Article 2, which commits the State to abolish torture even during war or in a state of emergency. Unlike declarations, which "address themselves to all members of the organization under whose auspices they have been adopted or even to the entire world," conventions "impose obligations of positive law" on state parties (Burgers and Danelius 1988, 8). In short, the Convention against Torture represents a 9 specification and strengthening of the customary norm against torture (Burgers and Danelius 1988,1), recasting it as a formal norm in the international system. In as much as it establishes a "single-issue regime" (Donnelly 1986, 631), moreover, the Convention has come to define the international norm prohibiting torture, and as such, one might measure support for the norm with respect to the number of state parties to the convention itself. Adhesion to the norm - indicated by ratification of a state's signature or accession by non-signatory states - has progressed to include the majority of the system since the treaty was agreed to in 1984. Table A . l in Appendix A lists the number of states acceding by year, the rate of adhesion expressed in terms of the proportion of the non-signatory population joining in a given year, and the total number of states and the proportion of the system adhering to the CAT. Over the 17 year span between 1986 and 2003, we can observe an initial rush of ratifications - with 24 state parties acceding to the treaty in 1987 - followed by a slightly decreasing number of accessions per year, with a maximum of eleven per year. As Figure A . 1 indicates, however, the rate of accession is more or less constant after 1987, with between 6 and 8% of the non-party population acceding per year. In other words, following an initial rash of ratifications - likely consisting of those states that lobbied for the treaty and had probably already internalized the customary norm against torture - , adhesion to the C A T grew at a more or less constant rate, such that by 2003, two thirds (69.3%) of the system, consisting of 133 states, were party to the norm. In short, the rather strong norm against torture embodied 2 See Meyer, Ramirez, and Soysal (1992, 137) for a precedent in using this type of indicator. 3 Note that this data was compiled using the standard Correlates of War project definitions for system membership, which excludes some parties like the Vatican City. See Chapter 3 for a more detailed account of the coding guidelines used. 10 by the Convention against Torture has garnered widespread support in the international system, is currently accepted by a large majority of the society of states, and is likely headed towards eventual universal acceptance. The CAT, then, is an example of the successful emergence and gradual univeralization of a single-issue human rights norm beyond the group of states that lobbied for its formalization in the treaty. While not quite there yet, it can be argued that participation in the explicit ban on torture has become a defining characteristic of membership in the international system. The movement to abolish the death penalty presents an alternative example of norm emergence and diffusion. While exhibiting important differences with torture as an issue, it has become highly established in Europe and some Latin American countries - the regions in which it originated - but has found significant resistance to its expansion to states and other regions. Ultimately, comparing the evolution of these two norms may yield valuable insights into the conditions under which emergent international legal norms succeed or fail. Death Penalty Abolitionism Much like torture, capital punishment has historically comprised a prominent and routine component of states' criminal justice systems. Unlike torture, however, the abolitionist movement that began in late eighteenth century Europe has led neither to the development of a customary international norm prohibiting it, nor to universal condemnation. But while it has not garnered consensus, it has established strong regional footholds, and continues to make headway in the system as a whole. 11 Humanitarian and philosophical arguments against the death penalty, like that put forward forcefully by the Italian criminologist Cesare Beccaria (Schabas 2002, 5), began to gain ground in the nineteenth century. Unlike torture, which as a customarily proscribed practice tends to be practiced extra-legally by definition, the effectiveness of the abolitionist movement can be gauged by the elimination of provisions for capital punishment from legal statutes, since the death penalty is by definition practiced within a state's legal framework.4 As such, its first success was attained in 1846 when the State of Michigan abolished the death penalty on a permanent basis (Schabas 2002, 5). This first was followed by its abolition in a number of legal jurisdictions - mostly Latin American countries - in the following years, although only eight had abolished the death penalty for all crimes prior to WWII (Prokosch 1999, 17). Once again, the progress of this abolitionist wave was temporarily dampened - though possibly eventually spurred on as a reaction - by the rise of the fascist and communist regimes in Europe in the interwar period, as well as by the emergence of criminological doctrine that emphasized the deterrent effects of capital punishment (Schabas 2002, 6). The emergence of international human rights discourse in the post-WWII period, however, spurred efforts to define the death penalty as a human rights abuse, and the abolitionist wave gained momentum. An increasing number of countries decided to abolish the death penalty for all crimes - the number stood at 63 in 1998 (Prokosch 1999, 17) - while an even larger number have abolished it de facto, refusing to carry out executions. As such, a majority of states - 117 in 2004, according to Amnesty International - might be considered to have dropped the practice of execution as an 4 Extra-judicial executions are a different, if related, human rights issue. 12 element of the criminal justice system.5 In short, the advancement of the abolitionist movement might be conceptualized as the emergence, diffusion, and internalization of a norm banning the death penalty, which, while certainly not as universal or embedded as the norm against torture, has come to encompass much of the state system. What this thesis is concerned with, however, is not simply imitation or norm 'diffusion', but rather with the explicitly international element of the norm, and how it has been incorporated into international society and its rules. The issue of capital punishment emerged as an element of human rights discourse from the very beginnings of the post-war universal human rights regime. From early on, abolitionist campaigners aimed to frame the practice in the context of general, more established norms. Some choose to argue that the "death penalty is an extreme example of torture" (Hood 1999, 9), attempting to borrow the authority of the customary norm against torture. More common, however, is the argument that it violates the right to life enshrined in the Universal Declaration of 1948, which affirms it "in absolute fashion, any limitations being only implicit" (Schabas 2002, 6-7). In fact, "[t]he provisions [in the human rights conventions and treaties] that deal explicitly with capital punishment are generally included within the articles on the right to life" (Schabas 2002, 18). In either case, discussion of the death penalty as a human rights abuse clearly falls within Keck and Sikkink's (1998, 27) criteria of the practice causing "bodily harm to vulnerable individuals." Within international human rights law, provisions relating to the death penalty have evolved over time, progressively restricting the bounds of legitimate practice, 5 See "Facts and Figures on the Death Penalty", available from Amnesty International at 13 including the limitation enshrined in the Covenant on Civil and Political Rights, which calls for its application only in response to "the most serious offences", and the restriction of applicable offenders, which aims to ban executions of minors, pregnant women, and the insane, among others (Prokosch 1999,19-22). Efforts have also been made to ensure procedural fairness in cases involving possible sentences of death. Most states - whether they apply them in practice or not, and with the notable exception of the United States, which continues to allow the execution of minors and the insane - tend to accept the validity of these types of standards. Outright abolition, however, a concept that first emerged in Latin America and Western Europe, remains more controversial, and less accepted as an international human rights norm. That said, a number of treaties include provisions for the abolition of the death penalty, including Protocol No. 6 to the European Convention on Human Rights of 1982, the Protocol to the American Convention on Human Rights to Abolish the Death Penalty of 1990, and the Second Optional Protocol to the International Covenant on Civi l and Political Rights of 1989 (Prokosch 1999,24). This thesis chooses to focus on the latter instrument because it is the only one of universal scope. The Second Optional Protocol While Uruguay and Colombia were most active in lobbying for the death penalty provisions in the original Covenant (Schabas 2002, 64), the European states have since taken over the activist mantle, pushing to "[transplant] the seed of abolitionism (that in a European context has grown into a sturdy and fruit-laden tree) into a global, United Nations framework" (Toscano 1999, 102). Following from strong domestically inspired 14 abolitionist sentiment in a number of countries as well as an emerging abolitionist norm in the well developed European human rights regime (Kruger 1999), Austria, Costa Rica, Dominican Republic, Federal Republic of Germany, Italy, Portugal and Sweden submitted the draft Second Optional Protocol (SOP) to the General Assembly at the 1980 session (Schabas 2002, 174). Its introduction was followed by nearly a decade of tough negotiation within the United Nations, as the European bloc faced stiff opposition from a number of countries, but most notably from Islamic states, who argued that the death penalty was enshrined in their religious law - shar 'ia (Schabas 2002, 16). Ultimately, the instrument was adopted on December 29, 1989 by the General Assembly, but not by consensus. It proceeded to a vote, receiving 59 votes for, 26 against and 48 abstentions (Schabas 2002, 180). Unlike other provisions relating to the death penalty in the universal regime, the SOP is an explicitly abolitionist instrument, as intended by its sponsors, and clearly states its purpose in Article 1: 1. No one within the jurisdiction of a State Party to the present Optional Protocol shall be executed. 2. Each State Party shall take all necessary measures to abolish the death penalty within its jurisdiction. In short, the SOP establishes a powerful new single-issue norm in the international system, calling for the total abolition of the death penalty, and by virtue of its status as a formal treaty, it imposes the obligations of positive law on state parties. States that join the treaty, then, indicate their acceptance of the absolute abolitionist norm, and, within the international legal framework, bind themselves not to reestablish the practice in the future. For the purposes of this thesis, then, the dynamics of accession to the SOP are 15 understood to mirror acceptance of the positive international norm banning the use of the death penalty by states. Table A.2 presents the annual statistics for ratification and accession to the Second Optional Protocol. Except for an initial raft of ratifications in 1991 (Australia and New Zealand in addition to eight European states) and relative spikes in 1993 and 1999, the SOP has experienced between 1 and 5 annual accessions since it was adopted by the G A . The rate of accessions, as shown in Figure A.3, was considerably more varied than that for the CAT, exhibiting no immediately obvious trend. The growth of the proportion of the state system adhering to the norm, however, is again, either constant or slightly decreasing, as shown in Figure A.4. The total extension of the norm against the death penalty, moreover, does not approach anywhere near the level of universality achieved by the norm against torture encapsulated in the Convention against Torture. In 2003, there were only 49 state parties to the SOP, accounting for 25.5% of the system. This is not only a far smaller proportion than the CAT's 69.3% in 2003, but lower than the CAT's proportion in it's thirteenth year of ratifications, 1998, when 59.4% of the system ascribed to the norm. In fact, the C A T crossed the 25% threshold in 1989 - its fourth year of ratification - when it had 28.6% adherence, as opposed to the SOP's numbers in 2003. In sum, the SOP has been met with much less acceptance in the international system, with a relatively small proportion of states - mostly European and Latin American states - adhering to it. This is particularly striking in comparison to the CAT, which has been adhered to by a large majority of states, and is clearly subject to much greater agreement. 16 The aim of this thesis is to explore and compare the different evolutions of the two norms, and the dynamics of their expansion to incorporate new states. It seeks to identify the factors and conditions that might explain the varying success of the two treaties. A comparative study of the two norms, moreover, should yield insights into the general evolution and diffusion dynamics of norms in the international system. But Are the Norms Comparable? It is important at this point, however, to address the question of whether these two norms are, in effect, comparable. After all, the simplest explanation for the different degree of acceptance of the two norms is that the norm against torture was already a widely accepted customary norm, while the norm abolishing the death penalty is relatively new. This is undoubtedly true to a degree, but does not rule out the comparability of the explicit norms defined by the two treaties. On the one hand, the two treaties are similar on several counts. They both define prohibitionist human rights norms relating to the "core" rights of the universal regime as tenets of positive international law, basing themselves on the fundamental rights to life and to freedom from torture. The treaties define single-issue human rights regimes in the international sphere, and were adopted at approximately the same time - that is, immediately prior to the end of the Cold War, and just as the human rights regime began to cascade (Risse and Sikkink 1999, 21). Finally, in as much as ratification or accession 17 by states to the treaties is well documented, they provide a highly reliable means of measuring acceptance of the norms they define.6 On the other hand, the two treaties are fundamentally different in that the Convention against Torture builds on a well-established norm of customary international law and on a general consensus among states on the condemnation of torture as an illegitimate practice. The Second Optional Protocol, in contrast, reflects the much more contested norm aiming for the absolute abolition of the death penalty, and is largely the result of vocal activism by European states at the United Nations rather than general consensus (as indicated by the G A vote that adopted the treaty). That said, these differences do not mean that the comparison of the ratification and accession process for the two treaties is not useful or even possible. While the rate of adhesion to the SOP (as shown in Figure A.3) does exhibit greater variability than that for the C A T (Figure A . l ) , as we would expect for a younger and less established norm, the progress of adhesions to the two treaties over time, except for the rather larger initial spate of ratifications of the CAT, appears to be strikingly similar. As can be gleaned from the approximately linear scatter-plots in Figures A.2 and A.4, the expansion of the norms as a proportion of the system appears to grow at a constant rate. In other words, while the baseline degree of support enjoyed by the norms immediately following the conclusion of the treaties was probably affected by the greater establishment of the norm against torture, the rate of expansion of the norms - over a period of approximately fifteen years for both norms - appears to be essentially similar. While the difference in initial support is significant (25 v. 12), i f we consider the first two 6 Alternative indicators could be used for the death penalty (such as the elimination of provisions for capital punishment from criminal codes), but these are unlikely to be available for torture, as this was already generally considered to be an illegitimate practice. 18 years of ratifications), it is not sufficient to account for the tremendous difference in 2003 (130 v. 49). In short, the explanation for the relative success of the two norms lies in the relatively constant growth rate for both norms. As such, this thesis proposes to investigate the dynamics governing the growth of the two norms, and argues that a difference in these dynamics can account for the different outcomes. 19 Chapter 2 Theorizing Structural Change and Human Rights Norms The study of norms in international relations emerged largely from the burgeoning regime literature of the 1980s. The concept of a "regime" was developed to describe and explain "instances of rule-based cooperation in the international system" and how "international institutions... affect the behavior of state and non-state actors" (Hasenclever, Mayer, and Rittberger 1997, 1), both phenomena that had been largely ignored by neo-realism - the dominant school in IR. Norms, often defined as "standards of behavior defined in terms of rights and obligations" (Stephen Krasner in Kratochwil and Ruggie 1986: 767), and are considered to be "one of the four analytical components of the concept of regimes" (Kratochwil and Ruggie 1986, 767) - in short, norms are one of the fundamental phenomena underpinning international regimes. Theoretical approaches to regimes can be classified as "power-based," "interest-based," and "knowledge-based" - which generally mirror the neo-realist, neo-liberal, and constructivist schools of international relations - and vary primarily with respect to the explanatory variables they highlight (Hasenclever, Mayer, and Rittberger 1997, 1-2). The following section attempts to apply variants of the first two approaches to this thesis' empirical focus, arguing ultimately that traditional theory is unable to adequately describe the diffusion dynamics of the Convention against Torture and the Second Optional Protocol, laying the ground for the elaboration of a social constructivist alternative. 20 Neo-liberal Theory Neo-liberalism's focus on international regimes originally emerged as a self-conscious extension of neo-realist principles to explain the empirical phenomenon of sustained international cooperation, and is therefore often categorized with neo-realism as part of the "rationalist" school of international relations. Keohane's (1986b, Ch. 6) exposition of "functional regime theory" in After Hegemony - one of the foundational texts of the neoliberal school - is based on neo-realist assumptions of state rationality and egoism, and aims to describe how international cooperation emerges in the context of a Prisoner's Dilemma collaboration game. He argues that international regimes - broadly understood - can facilitate cooperation by performing a number of functions to reduce uncertainty in the environment and discouraging defection. These tasks include reducing transaction costs (89), enforcing issue-linkage (91), and reducing asymmetries of information (94). While he tends to rely on hegemonic stability theory as an explanation of the origins of regimes, Keohane argues that a wide range of institutional components, ranging from rules and formal organizations to simple norms, are created by actors to facilitate the persistence of cooperation in the anarchic international environment. Donnelly (1986) adapts neo-liberal regime theory to the human rights issue-area, noting that human rights "are regularly addressed today in bilateral foreign policy and in a variety of multilateral schemes" (605), and that they represent a highly institutionalized issue area with a number of agreements at the global level as well as a multitude of'"lower level' regional and single-issue human rights" arrangements (620).7 While the latter, which include the highly institutionalized European and Western 7 See also the United Nations High Commissioner for Human Rights website for a comprehensive listing of legal instruments relating to human rights, <> 21 Hemispheric systems, might be considered to be "largely autonomous," they are nevertheless generally understood to be "relatively coherently 'nested'" components of a global human rights 'regime' (Donnelly 1986, 620). Ultimately, norms and institutions may well form a coherent regime in the broad sense of the term, but applying neo-liberal conceptions of regimes to the issue area is of doubtful utility. Donnelly's portrayal of "moral interdependence" is unconvincing, however, as human rights issues are not areas where 'joint benefits' might be attained by cooperation and is therefore simply not subject to rational-choice formulations per se - although Risse and Sikkink (1999) do argue that other instrumental calculations arising from issue-linkage are a dominant mechanism leading to norm adoption. In other words, unlike Nadelmann's (1990,481) description of "prohibition regimes," which address "the inadequacy of unilateral and bilateral law enforcement measures in the face of criminal activities that transcend national borders," human rights issues do not generally include a functional "transnational dimension." Moreover, as Reus-Smit (2004, 19) notes the "progressive 'cosmopolitanisation' of international law," including, by definition, human rights law, makes explanatory frameworks that focus on interstate behavior as opposed to the relationship between the state and its citizens increasingly irrelevant. Finally, Donnelly's (1986, 624) discussion of 'moral interdependence' relies on the "strong national commitment of the European states to human rights" as the fundamental factor allowing for the greater institutionalization of the European regime. This commitment, however, is a normative attribute that does not sit well with neo-liberalism's largely materialist and explicitly interactionist formulation. In short, in as much as the success of the human rights regime arises out of purely normative preferences and ultimately only 22 depends on national enforcement, it simply does not fit with neo-liberal functional regime theory. As such, this thesis will no longer consider neo-liberal theoretical explanations in its investigation of the growth in C A T and SOP adherence. Hegemonic Stability Theory While traditional realists and neo-realists tend to dismiss the relevance of regimes in general and norms in particular (see Mearsheimer 1994, for example), others have advanced power-based alternatives to neo-liberal regime theory. They tend to "emphasize relative power capabilities as a central explanatory variable and stress states' sensitivity to distributional aspects of cooperation and regimes" (Hasenclever, Mayer, and Rittberger 1997, 84). While the former does not rule out norm diffusion, the latter aspect is particularly problematic with respect to the human rights issue-area. Once again, in as much as power-based theories are couched in the terms of rationalist theory -the collective action problem, "public goods," as well as coordination versus collaboration games (Hasenclever, Mayer, and Rittberger 1997, 84-89,100-02) - their applicability to the human rights regime is questionable. That said, it is possible to apply the power-based approach to this type of issue. Moravcsik (2000, 221) outlines a realist explanation of the human rights regime, nevertheless, noting that it would imply that "[governments accept international obligations because they are compelled to do so by great powers, which externalize their ideology" and stressing the "distribution of interstate bargaining power." Establishing a human rights regime, then, depends on "a hegemonic group of great powers wiling to coerce or induce recalcitrant states to accept, adjust to, and comply with international human rights norms" (221). Notably, this conceptualization moves away from the 23 explicitly unipolar power configuration implied by traditional hegemonic stability theory (Hasenclever, Mayer, and Rittberger 1997, 84), relying instead on a dominant coalition. Moreover, it distances itself from the "public good" reasoning behind the formation of regimes, arguing instead that dominant states seek to impose their own ideology and values on the rest of the international system. In a general sense, this explanation resonates with the argument that most of the norms defining international society originated with the Western states, which have been dominant players in the international system for centuries (the Cold War notwithstanding). Specifically, both the campaign against torture and that against the death penalty have undoubtedly benefited from the status of Europe - where both norms largely originated - as a major world player. Similarly, the United States' strong and active opposition to the SOP has likely significantly hampered the expansion of the norm. That said, it is possible to question the utility of the power4}ased explanation in a similar fashion. Given that the United States, which enjoys relative hegemony, opposes the ban on the death penalty, why has it continued to expand? Moreover, why did the highest rate of accession to the C A T take place in the late 1980s and early 1990s, even though the U.S. did not join the treaty until 1995, and why did the significant shift in the power configuration of the system that followed the Cold War not visibly affect the rate of accession to the treaty? In short, in as much as realist theories tend to focus on systemic phenomena like polarity and the distribution of power in the system, they are not well positioned to explain variation in the adoption of purely normative rules like human rights norms. Moreover, in as much as the specific cases considered in this thesis occurred at about the same time historically, it is not clear how the same systemic 24 transition could account for the successful spread of one set of norms championed by a set of actors (European states), and the stagnation of another set backed by the same actors. Republican Liberalism A final alternative explanation - grounded in liberal institutionalism (see Moravcsik 1997) - emerges from Moravcsik's (2000) application of "republican liberal" theory to explain relative support by European states for a European human rights regime. He argues that national governments of young democracies use international regimes as a means of "locking in" the new institutions and reducing political uncertainty. To that effect, new democracies are more likely to support human rights regimes than more established democracies as these are more secure in their political stability, and will not be willing to accept the "sovereignty costs" associated with the regimes (Moravcsik 2002, 225-230). This fits in with Finnemore and Sikkink's (1998, 906) discussion of "legitimation" as a motivation for joining a norm. While this is ultimately also a constructivist (ideational) concept, it is aimed at the domestic level, and, more importantly, operates at the national level, and is therefore not necessarily dependent on other states' behavior. Importantly, however, given that the norms against torture and the death penalty are closely tied in normative human rights discourse, it is not clear why a given state would adopt one and not the other. If it depends on the degree of acceptance of the norms in the system, however, it is essentially an argument about socialization, which is discussed below. In other words, while republican liberalism might well explain why an individual state joins human rights treaties, it does not explain 25 why that state would ascribe to one norm and not to another, and why one norm would be more successful at a systemic level than another. In short, it does not provide a satisfactory explanation for the empirical puzzle outlined in Chapter 1. Having examined the potential applicability of traditional IR theories and found them wanting, then, the next section examines the constructivist approach to norms, and proceeds to elaborate a theoretical framework with which to analyze the growth dynamics of the Convention against Torture and the Second Optional Protocol. Constructivist Theory, Norms, and Structural Change Broadly speaking, the constructivist approach to norms focuses on their ideational and cognitive components. Unlike neorealist and neoliberal theories, in which norms derive their influence over behavior from the power of the actors involved or from the logic of functional imperative, constructivists conceive of norms - which Jepperson, Wendt, and Katzenstein (1996, 54) define as "collective expectations about proper behavior for a given identity" - as fundamentally ideational phenomena. They shape and homogenize states' beliefs, identities, interests, and ultimately affect their behavior (see Reus-Smit 2004 for a brief summary). Early constructivist work on international norms in international relations focused primarily on whether "norms matter," and was aimed at countering the realist contention that norms and international institutions are epiphenomenal to the material forces governing state behavior (see Mearsheimer 1995, for example). A number of studies examine the role of norms in shaping the choices of state actors, focusing on bureaucratic structure (Finnemore 1993), the deployment of weapons systems (Price 1998; Legro 1997), the formation of national identity (Checkel 26 1999), and security and economic policy (Cortell and Davis 1996). In sociology, the insitutionalist or "world polity" school examine the role of world cultural norms in generating global "isomorphism," arguing that states adopt increasingly homogenous organizational forms over time as a result of cultural diffusion (Meyer et al. 1997, 152). This literature includes studies examining the diffusion of women's suffrage (Ramirez, Soysal, and Shanahan 1997) and mass education (Meyer, Ramirez, and Soysal 1992), as well as the growth and role of international non-governmental organizations (Boli and Thomas 1997). This broad range of literature, spanning academic fields, provides convincing empirical evidence for the causal and constitutive effects of norms in world politics. While these studies also address the dynamics and evolution of the norms in question, a second wave of scholarship does so in a more general sense. Finnemore and Sikkink (1998) and Risse and Sikkink (1999) draw heavily from social psychology and sociology, providing accounts of the general socialization processes through which states become part of international normative regimes and how norms expand and entrench themselves over time. Axelrod (1986), Florini (1996), and Bicchieri, Jeffrey, and Skyrms (1997) approach the issue from a different angle, proposing non-intentional social mechanisms that might explain the emergence and evolution of norms. Broadly speaking, this later body of work tries to explain what Wendt (1999) calls "structural change" in the international system's ideational structure. That is, they address the emergence of norms "from the standpoint of the system and do not seek to explain the behavior of individual actors" (Wendt 1999, 152) - how and why they become systemic attributes as a result of "patterns of coordinated, system-wide 27 redefinition of interests" (Finnemore 1996, 2). Crucially, while norms can be conceived of as "macro-structure", they ultimately "are an ongoing effect of process, at the same time that process is an effect of structure" (Wendt 1999, 186) such that "[ajctors create structures which take on a life of their own and in turn shape subsequent action" (Finnemore 1996, 30). Keohane (1986, 152) echoes this formulation in his critique of Waltz (1979), arguing that "ultimately, in any social system, structural change has no source other than unit-level processes." In short, in as much as norms exist as collective understandings of individual actors, it is adherence to norms by agents that ultimately creates and sustains normative structure, while at the same time this structure emerges as a systemic attribute that shapes the behavior of agents embedded in it. Structural change, moreover, takes place as a consequence of agent-level behavior, and as such explanations of structural change require explanations of agent-level choice. This study, then, aims to examine the structural change represented by the emergence and evolution of the norms against torture and the death penalty. While they originated in specific states and geo-political regions as a result of largely endogenous processes, increasing adherence to the norms, as indicated by the increased proportion of state parties to the CAT and the SOP, led to a redefinition of the "[cjultural or institutional elements of states' environments" (Jepperson, Wendt, and Katzenstein 1996, 54) - or in other words, it led to a change in the normative macro-structure of the international system. The question, then, is why the change represented by the norm against torture is more widespread and has been generally accepted, whereas the effort to de-legitimize the use of the death penalty by states has met with significant opposition. The following section begins to address this question by presenting a more detailed 28 theoretical account of normative structural change, relying on existing understandings of normative evolution and identity change among states. First, however, we must consider an alternative explanation for the success of different norms in international society. While environmental factors of all types are generally used to explain the growth of norms within the international system, an alternative explanation might rest on the substantive content of the norms themselves. Analogous to message-based persuasion in social psychology (Chaiken, Wood, and Eagly 1996, 707-723), the issue appears in the norms literature as a question of norm "resonance" and "adjacency" (Finnemore and Sikkink 1998, 908). Checkel (1999, 87), for example, in his analysis of national identity in Europe, argues that "[diffusion is more rapid when a cultural match exists between a systemic norm and a target country.. .where it resonates with historically constructed domestic norms." One might hypothesize, then, that the different successes of the anti-torture and anti-death penalty norms are the result of differing resonance with national norms. This explanation is problematic because the two norms are relatively similar. They both proscribe practices within the "core" of the human rights issue-area, relating to the right to life and freedom from torture (Risse and Sikkink 1999, 2), and more generally, to "issues involving bodily harm to vulnerable individuals, especially when there is a short and clear causal chain... assigning responsibility" (Keck and Sikkink 1998, 27). In short, they are normatively very similar practices, and as such, we would not expect to see a systematic difference in normative resonance within individual states. That is, while contingent culture within the United States, for example, might value the death penalty more highly than torture, it appears unlikely that states with cultures as 29 different as China, Russia, and India would systematically exhibit similar variation in resonance. Arguments based on functional explanations develop in much the same way. Even i f one accepts the deterrent effect of the death penalty and/or the investigative utility of torture, it is unclear that individual countries ranking high on an "insecurity" scale would choose to adhere to one and not the other, and especially not systematically, which is exactly what has happened with the C A T and the SOP. 8 In short, in an argument analogous to the world polity theorists' claim that isomorphism in the international system cannot be explained by functionalist theorists due to variation in alleged domestic functional imperatives (Meyer et al. 1997, 152-153), this thesis argues that unit-level variation in domestic norms governed by independent mechanisms is highly unlikely to have macro-structural effects. At most, we would expect that a contingent effect on a key player like the United States might significantly affect the socialization dynamic, preventing the more general diffusion of the norm, but the primary explanation for structural change would continue to depend on factors (whether power- or ideas-based) operating at the systemic level. Normative Structural Change Finnemore and Sikkink (1998, 891) synthesize the literature on social norms -which they define as "standard[s] of appropriate behavior for actors with a given identity" - , and present their findings in the context of international relations. They 8 Note also that the relevance of the "deterrence" argument in support of the death penalty is also ideational. The functional effectiveness of the deterrent effect is highly contested, and as such, it would be a combination of the crime situation in a country with the domestic cultural context - i.e., the level of belief in the deterrent effect of capital punishment - that would ultimately determine a state's behavior. As such, exclusively functional indicators would not be useful. The domestic indicator presented in Chapter 3, however, should account for both aspects of the endogenous domestic process, as it focuses on practice. 30 characterize the existing IR norm literature as "static", arguing that "[c]laims that actors conform to the 'logic of appropriateness' say little about how standards of appropriateness may have changed" (888). They are more concerned, instead, with how norms emerge and change as opposed to how existing norms affect behavior. As such, they present the concept of a norm "life cycle," in which "different behavioral logics dominate different segments of the life cycle" (888). It consists of three parts: 1) norm emergence, in which norm entrepreneurs pressure states domestically to espouse a given norm (896-898), 2) a norm cascade, where a different dynamic takes over, in which "[m]ore countries begin to adopt new norms more rapidly even without domestic pressure" (902), and 3) norm internalization, in which norms "become so widely accepted that they are internalized by actors and achieve a 'taken-for-granted' quality" (904).9 Crucially, the transition between the first and the second stages takes place when a "tipping point" is reached, where enough states - Finnemore and Sikkink mention 33% of the system as a possible threshold - have adopted the norm to induce an increased effect for international factors in prompting states to adopt the norm (901). They note that empirical studies in sociology and other fields - like Ramirez, Soysal, and Shanahan (1997) investigation on the spread of women's suffrage - provide strong evidence for a change in the dominant mechanisms motivating the spread of international norms. Both of these studies refer to "socialization" as the new process operating in the 'cascade' phase, distinct from the unit-level dynamic in the 'emergence' phase. Conceptions of 'socialization' in the literature are quite varied, however, ranging from a desire for conformity and a "dynamic of imitation" (Finnemore and Sikkink 1998, 895) to 9 Most of these concepts emerge out of sociology, and especially empirical and theoretical work in the legal academy on social norms. See, for example, Sunstein (1996), Picker (1997), Posner (2000), and Ellickson (2001). 31 concepts like "world culture and historical isomorphism" drawn from the world polity school of sociology (Ramirez, Soysal, and Shanahan 1997, 740), and tend to be theoretically under-specified. Notably, while some studies aim to identify the mechanisms underlying world cultural "diffusion" (see Strang and Meyer 1993, for example), the concept suffers from a lack of specificity with regards to the source and agency of change in the system (Keck and Sikkink 1998, 33). As such, it is important to carefully define the concept of "socialization" for the purposes of this thesis. Risse and Sikkink (1999, 11) define socialization as "the 'induction of new members ... into the ways of behavior that are preferred in a society,'" and note that this understanding assumes the existence of a society of states like that proposed by Hedley Bull (1977). This is reflected in Wendt's (1999) theory of structural idealism. While he acknowledges the operation of endogenous mechanisms in shaping state identities and interests (Wendt 1999, 181), he argues that structural change occurs as a result of broader processes based on interaction (185). Specifically, he cites natural selection and cultural selection as the mechanisms of macro-level change (319-325). The former is not relevant to our problem, as it depends on the survival of 'fitter' individuals to propagate traits, and state death was essentially not an issue in the late 20 t h century.10 The latter process is further divided into two categories. On the one hand, imitation processes involve the adoption of "the self-understandings of those whom [actors] perceive as 'successful'" (Wendt 1999, 325). On the other hand, social learning processes determine how "identities and their corresponding interests are learned and then reinforced in response to how actors are 1 0 Note that this natural selection mechanism is distinct from Florini's (1996). 32 treated by significant Others" (327). In short, imitation involves an actor's direct perception of macrostructure, while social learning happens in the context of interaction between actors, which may include active pressure on states, ranging from the use of status by activist IGOs like UNESCO (Finnemore 1993), to transnational civil society (INGOs) exerting pressure on national governments by means of coalitions with domestic opposition groups (Risse and Sikkink 1999, 18; Keck and Sikkink 1998), and the shaming and isolation of norm violators by both civil society and prominent states (Wexler 2003). As such, the imitation dynamic illustrates a less purposive mechanism at the systemic level - in that actors are not actively seeking to spread norms -, and while he argues that social learning leads to 'deeper' change in actor identities, Wendt (1999, 354) notes that collective identities - ideational macro-structure - are founded upon an "'objective' homogeneity" at the unit level that facilitates subjective categorizations. In other words, imitation may not indicate internalization of a given practice by actors, but in as much as it leads to homogeneity of practice, it lays the foundation for the subsequent formation of collective identities and the evolution of customary international law through "argument, learning, and repeated practice [italics added]" (Reus-Smit 2004). This approach is rooted in social identity theory - a school in social psychology that, amongst other issues, aims to explain social influences in attitudinal change -, and assumes that "influence is normatively regulated by social group membership" (Chaiken, Wood, and Eagly 1996, 725). That is, individuals are understood to be motivated by a desire for "positive identification" that is pursued by a process of self-categorization. Crucially, "[a]ttitudes and opinions are common category characteristics that are inferred 33 from the available exemplars or prototypes of a social category and are automatically assigned to all members including oneself (725). When others categorized as "similar" disagree on a given attitude, then, the resulting cognitive uncertainty can be reduced by 1) re-categorizing oneself or the other, 2) redefining the stimulus (i.e., redefining the norm such that it is no longer associated with the category, or 3) exerting mutual influence in an effort to bring oneself into line with the social consensus among those who are similar. In short, individuals "motivated by social identity concerns... align with positively valued social groups and... differentiate from derogated groups" (726). While the third response might involve social learning, then, social identity theory implies that actors (states in our case) will tend to pursue imitation, conforming to the normative consensus exhibited by "like" states without explicit pressure being exerted by international actors. This understanding is echoed by Strang and Meyer (1993, 491), who argue that "reference groups" and social comparison among states are a presupposition for "rational mimicking practices." In other words, from a diffusionist perspective, "[pjractice [not only] diffusefs] along the lines of social relations, but also to other actors broadly considered similar" (492) - collective identities act as a permissive variable in explaining the imitation of behavior in the international system. The principal result of the "identity mechanism", then, is that international society is opened up "to rather rapid penetration" (Meyer 2000, 243), allowing for the possibility of the rapid expansion of a norm - its cascade - without the need for a significant intentional component. Both Finnemore and Sikkink's (1998, 901) mention of "critical states" and Chaiken, Wood, and Eagly's (1996, 726) discussion of the importance of "status hierarchies," moreover, suggest that 34 some states might be more influential than others in defining relevant identities, supporting Wendt's conception of "significant Others." It is important at this point, however, to consider the distinction between the imitative socialization processes this thesis deals with and the myriad of intentional sources of norm change and socialization in the international system. Keck and Sikkink (1998) provide a detailed look at "transnational advocacy networks", which they argue play a crucial role in pressuring states both from inside and from outside their structure to change their practices in a whole range of issue-areas. They characterize networks as a third type of organizing principle in addition to anarchy and hierarchy (Keck and Sikkink 1998, 8), and note, moreover, that their analytical approach depends on conceptualizing a genuine transnational civil society, as opposed to simply Bull's (1977) society of states (Keck and Sikkink 1998, 34). This approach appears to be quite relevant in the human rights issue area as a whole (Keck and Sikkink 1998, Ch. 3), and certainly with respect to the movements against torture and the death penalty, in which major international non-governmental organizations like Amnesty International have played a central role (Cook 1991)." As such, intentional transnational actors have clearly been involved in the process of norm diffusion for both the C A T and the SOP. The question, then, is whether this presents a fundamental threat to this thesis' examination of non-intentional, imitative processes. On the one hand, the effect of transnational advocacy networks (TANs) can be measured at the systemic level. That is, the effect of the pressure exerted by the networks 1' In fact, the "Death Penalty" and "Stop Torture" campaigns comprise one sixth (two of twelve) of Amnesty's current international campaigns. See <> for a listing. 35 on states and other actors can be measured in part as an increase in the global profile and salience of the issues they champion, which leads to the facilitated diffusion of related norms in international society. As Risse and Sikkink (1999) note, however, the transnational cultural and institutional structure underpinning the modern human rights regime was built in the 1970s and early 1980s, and was largely completed by 1985, when "the world began a process of a genuine international 'norms cascade', as the influence of international human rights norms spread quickly" (21). In other words, the emergence of a significant transnational cultural structure - the systemic effect of TANs - preceded the advent of the C A T and the SOP, and as such cannot account for variation both within the norm expansion dynamics and between different norms. If the scope of the international campaigns against specific practices is examined, however, it is more difficult to bracket the effect of TANs. While Amnesty International's major campaigns against torture and the death penalty began at roughly the same time - 1972 and 1977 respectively - , it may be reasonable to argue that the amount of resources devoted by TANs in general is greater for torture than for the death penalty. This remains an empirical question that is beyond the scope of this thesis. Nevertheless, it is also reasonable to suggest that in as much as T A N efforts rely on practices like shaming, the relative effectiveness of the campaigns is likely to depend on the existing "objective homogeneity" among different groups of states, and as such the 1 2 The existence of the Coalition of International NGOs Against Torture, for example, suggest that this might be the case. 36 intentional socialization processes they comprise should reflect the underlying imitative processes.13 On the other hand, TANs can have a strong domestic effect, framing and reinforcing the struggle of domestic intentional norm entrepreneurs. That said, given the existence of the networks, it is difficult to strictly differentiate between domestic and transnational sources of change, as domestic actors such as NGOs tend to define their issues and frame their mobilization in terms of existing or incipient international norms and networks. Domestic actors, nevertheless, have to initiate the process by providing a point of entry and attachment for TANs, and as such retain substantial initiative in change at the domestic level. Variation in the extent of transnational mobilization towards a specific country, moreover, is by definition likely to cause unit-level variation in adherence to the norm, and while this might present a significant problem if the level of mobilization is systematically biased towards a specific subgroup of countries, a moderate level of unit-level variation will simply increase the variance and error in evaluating the impact of imitative mechanisms, but should not fundamentally bias observation of the trend at the systemic level. Conclusion We find, then, that an identity-based approach complements Finnemore and Sikkink's (1998) description of the norm life cycle. During norm emergence, states' identities change to incorporate a given norm as a result of endogenous processes driven 1 3 As noted in Chapter 3, it would be ideal to elaborate an indicator for the degree of pressure exerted by both domestic and international NGOs on states, but as is also noted, it would have to be compiled from primary sources, and is therefore beyond the scope of this study. 37 by domestic norm entrepreneurs, which provides the unit-level variation necessary to initiate structural change (Wendt 1999, 319). This sets up an imitation-based socialization process, operating with respect to both the general population of 'like' states and with respect to 'significant' states. Crucially, even when the socialization dynamic has emerged, endogenously-driven unit-level variation is likely to persist, and may reinforce or extend the socialization process. Ultimately, "macro-level phenomena... become unstable only when enough important actors change their behavior that a tipping point is crossed" (Wendt 1999, 340), setting off a norm cascade driven by the imitation process that engulfs the system. Importantly, while this conceptualization "presupposes the prior existence of a collective identity to which states are trying to gain membership" (341), it does not require that there exist only one collective identity or that it be global in scope (though this may be the case). Unlike world polity models, where cultural change tends to take place in a diffuse, global, and generally underspecified manner, a norm may originate within a particular collective identity, but even i f it initiates a socialization dynamic within the identity in question, it may be unable to 'break out' to the wider population i f it does not reach a state with an 'overlapping' identity that would then 'transmit' the norm. We might hypothesize, then, that while the torture norm has progressed through most of its "life-cycle", the death penalty norm has yet to begin its cascade in earnest (indeed, the proportion of the system adhering to the SOP remains just below the 33% threshold mentioned by Finnemore and Sikkink 1998), and that this explains the difference in scope and acceptance of the two norms. The norm against the death penalty, furthermore, may be finding it difficult to 'break out' from its general acceptance 38 by states with a 'European' identity to other subsets of states, such as 'democracies' in general, whereas the norm against torture has been relatively successful in being adopted at the level of the world polity school's general "state" identity. This might be the result of a lack of "bridging states" (like Turkey, for example, which combines Muslim, European, and democratic identities), or because, due to idiosyncratic reasons, there has been an insufficient number of extra-regional endogenously driven adoptions of the norm in order to instigate a socialization dynamic in broader identities. The following chapters examine this set of hypotheses empirically, basing their argument on the imitative theoretical framework outlined above, as well as testing alternative hypotheses. They aim to test whether parties to the C A T joined increasingly for external reasons relating to changes in collective identities as measured by "objective homogeneity," and whether this socialization dynamic is ultimately lacking or limited to the "European" identity in the dynamics of accession to the SOP. 39 Chapter 3 Method and Data Given the relatively sweeping character of the hypotheses presented in this thesis, and the fact that they focus on the dynamics of norm evolution from a systemic perspective, it proposes to use macro-comparative quantitative analysis to test the propositions elaborated in Chapter 2 and to explain the different dynamics underlying growth in state adhesion to the Convention against Torture and the Second Optional Protocol. This chapter begins with an examination of the method used, including a discussion of its pros and cons, and then proceeds to outline the sources and coding decisions involved in building the datasets relating to both C A T and SOP accessions. Method: Event History Analysis v. Binary Time-Series Cross-Section With notable exceptions like Finnemore's (1996, Ch. 2) use of quantitative data on the adoption of science bureaucracies, most work on norms in international relations has tended to eschew this methodological approach, and as such, little specific guidance can be drawn form the IR literature. The world polity school of sociology, however, has endeavored to provide a sound quantitative empirical basis to their claims of cultural norm diffusion and isomorphism effects. Ramirez, Soysal, and Shanahan (1997) - in one of the studies cited by Finnemore and Sikkink (1998) in their discussion of the norm life cycle - provide compelling analysis of the diffusion of women's suffrage, examining a whole host of factors associated with the adoption of the practice. Since it is one of the few quantitative articles that deals directly with this thesis' theoretical focus, the general structure of their model and their choice of variables will strongly influence the 40 elaboration of the model in this chapter and the next. This does not extend, however, to the choice of technique. Ramirez, Soysal, and Shanahan (1997), along with much of the diffusionist literature in sociology, use a type of event-history methodology in their analysis. Alternatively termed survival or duration analysis, this family of techniques examines the length of time that elapses before like units experience an 'event' or 'failure', focusing on the variation in the duration and on the 'hazard rate' - the probability of a unit failing given that it has 'survived' a certain amount of time. Ramirez, Soysal, and Shanahan (1997, 739) note that "[e]vent history methods allow [them] to use cross-national data over a long historical period and to analyze the historical timing of women's suffrage across countries." Indeed, while this type of method is virtually unused in the IR literature, Box-Steffensmeier and Zorn (2001) argue that much of the existing work on the liberal peace can be improved by adopting hazard-rate duration models. That said, there are a number of downsides to this type of model, not least of which is the relative rarity of its use in IR, and the lack of familiarity among researchers in the discipline in comparison to more widely used methods. Moreover, the use of time-varying covariates - a fundamental element of this project, with respect to socialization dynamics - is much more problematic than in more traditional methods (Beck 1998, 206-7). Finally, continuous time event-history and duration models tend to encounter further problems when more than one event takes place in the same time-period (Beck, Katz, and Tucker 1998, 1266, fn. 21). The structure of Ramirez, Soysal, and Shanahan's (1997) data - in which a relatively small number of events are spread out over a hundred-year period - allows them to largely avoid this issue. Given the abbreviated time frame for 41 C A T and SOP ratifications and accessions, however, the data used in this analysis consistently has more than one accession in any given year. A suitable alternative to specialized event-history analysis is binary time-series cross-section (BTSCS) analysis, which provides a means of analyzing discrete duration data (Beck 1998, 205). Indeed, BTSCS data records the same type of "counting process" as duration, event-history, and event count data (Beck 1998, 191). As Beck, Katz, and Tucker (1998, 1264-5) put it, BTSCS data are, more generally, grouped duration data, in which "we only know whether a unit has failed in some discrete time interval (with independent variables only measured to the fineness of that interval)." That is, we know that a given state has acceded to one of the treaties in a given year, and can observe independent variables for that state for each year.14 Beck, Katz, and Tucker (1998, 1267-1271), furthermore, outline how to adapt logit estimation to analyze grouped duration data by using temporal dummy variables or cubic splines as an analogue to the continuous time baseline hazard function that aims to account for temporal dependence in the process. While the systemic socialization variables may pose a threat to this type of analysis - as outlined towards the end of this chapter -, they are not an insurmountable problem. In short, BTSCS analysis incorporating the corrections suggested by Beck, Katz, and Tucker (1998) should provide a suitable framework with which to analyze the expansion in state party adherence to the C A T and the SOP. 1 4 The data could, theoretically, have been coded to the day instead of to the year. A number of factors argue against this, however: 1) multiple instances exist in which states joined one of the treaties on exactly the same date, 2) most of the independent data could not be collected, or would be meaningless when measured in time units smaller than a year, and 3) it is not clear that the specific order of accession when separated by days or months is theoretically meaningful, since a whole host of confounding factors, like the institutional structure charged with ratification, can cause this type of variation. 42 The rest of this chapter documents the structure of the datasets compiled for this project, detailing the coding and operationalization criteria used to construct the variables, the empirical analysis of the data is conducted in Chapter 4. Data and Coding The two datasets were compiled using widely available sources, including staples of the quantitative international relations literature like the Correlates of War (COW) data and the Polity IV Project regime type data. Following from the above discussion regarding the structure of binary time-series cross-section data, both datasets use the "country-year" unit of analysis, allowing one observation per year for each state in the system. For the purposes of this study, membership in the international system is defined according to the criteria established by the Correlates of War project, in which all independent polities with populations greater than 500,000 are included, as well as all United Nations members-states (Correlates of War 2 Project 2003).15 Ultimately, this typology can be conceived of as defining membership in international society, which is a prerequisite for involvement in the creation and perpetuation of its norms.16 Figure B . l in Appendix B illustrates the growth in total membership in the international system. Following Beck's (1998, 194) discussion of discrete event-history analysis, observations are included for all states starting in the year of the first ratification of each treaty (1986 for the C A T and 1991 for the SOP), up until the year of ratification for the 1 5 All COW data used in this study are available online at 1 6 Along these lines, it is worth noting that Taiwan and Monaco were eliminated from the datasets. On the one hand, Taiwan's membership in the society of states is highly contested and as such, it is unlikely that it has had the opportunity to join formal instruments such as the CAT and the SOP. On the other hand, Monaco joined the CAT in 1992, a year before its entry into the dataset (by virtue of U.N. membership), and was purged for consistency's sake. 43 state in question. Data for dates of ratification or accession were obtained from the website of the United Nations High Commissioner for Human Rights.1 7 The dependent variable, Joined, then, is a dummy variable coded "1" if a state joined in a given year 18 and "0" otherwise. As such, the data is right censored with respect to time, but this is a fundamental property of grouped duration data describing one-time events, and does not present a problem for the analysis (Beck 1998, 203). Independent Variables The independent variables can be classified into three groups: 1) those reflecting the alternative hypotheses, 2) variables indicating state-level (i.e. domestic) predisposition to the norms, and 3) variables reflecting socialization processes in the international system. Alternative Hypotheses The first variable in this group is derived from the hegemonic stability theoretical standpoint. As indicated in Chapter 2, most power-based norm theory would be concerned less with a hegemon asserting its position than with a coalition of powerful states (an oligarchy) that aim to impose their values on the rest of the system. This 1 7 See "Status of Ratifications of the Principal International Human Rights Treaties" at This study uses the previous edition, valid up to November 2, 2003. See also United Nations High Commissioner for Human Rights (2003) in the references section. 1 8 Note that Ukraine and Belarus enjoyed U.N. membership during the Soviet era, and joined the CAT along with the Russian Federation in 1987. However, given their political dependency on Moscow at the time, only Russia is coded as having joined in 1987 and the Ukraine and Belarus are not included in the dataset. They are, however, included in the calculations of total state parties in the post-1991 period. Other former Soviet republics are included in the dataset as newly independent states. Similarly, the former Yugoslav successor states are included, with the rump Yugoslavia inheriting the SFRY personality in the dataset. The Philippines were not included in the 2003 version of the "Status of Ratifications", but did in fact join the CAT in 1986, an oversight that was corrected in the dataset. Czechoslovakia joined the CAT in 1988 and was coded as such, while the Czech Republic and Slovakia are included in the calculations of total state parties. Finally, the German Democratic Republic joined in 1987, while the Federal Republic of Germany joined in 1990. The GDR, however, was eliminated as a state party in the years following 1990. 44 argument also resonates with the world polity literature, which identifies the origin of most international norms (and indeed, the constitutive rules of modern international society itself) with the powerful Western states (Meyer et al. 1997,158,168). The indicator should therefore reflect the relative power represented by the states adhering to a norm. As such, PowerChange is defined to indicate the proportion of total systemic power comprised by all newly-acceding states in a given year, and is lagged, taking into account the accessions or total state parties in year x-1 when calculating the value for year x . 1 9 Data on state power are drawn from the COW National Military Capabilities v3.01 dataset (Singer, Bremer, and Stuckeyl972; Singer 1987).20 The Composite Index of National Capability (CINC) variable, which itself measures the proportion of total systemic power represented by a given state, and data on state system membership were used to calculate the proportion of systemic power represented by the states acceding to the treaties in the previous year. Ultimately, this indicator should account for the effect of a particularly powerful state joining a norm in a given year, and i f the hegemonic stability hypothesis is accurate, we would expect this type of increase to be associated 21 with increased adherence to the norm in following years. Figures B.2 and B.3 illustrates the proportion of power associated with the two norms. The second alternative hypothesis is Moravcsik's (2000) republican liberal framework in which relatively new democracies or states adhere to international human rights instruments as a means of consolidating domestic progress in human rights issues. 1 9 For reasons discussed later on in this chapter, the cumulative proportion of power is not used. 2 0 This data is available online at 2 1 Note that this type of explanation might become closely entwined with a "status"-based imitation dynamic, as the more powerful states would probably be more likely to have a higher "status" in the system (although this would by no means be a perfect correlation). It is not clear that this presents a major problem per se, but if the empirical effect of power were important, it should prompt further work to examine whether an imitation or a coercive dynamic is dominant. 45 The Democratization variable aims to account for this process, identifying states in any given year that have experienced a recent democratic transition. It is constructed using the Polity IV Project dataset, "Political Regime Characteristics and Transitions, 1800-99 • 2002" (Marshall and Jaggers 2003). A state is considered to have undergone a relevant transition when coded a '+2' ("minor democratic transition") or a '+3' ("major democratic transition") in the Polity IV Regtrans variable (Marshall and Jaggers 2003, 34-5). If a state has experienced this sort of democratic transition, Democratization is coded a ' 1' for the year in which it occurred and the four subsequent years, and a '0 ' otherwise.23 The five-year time lag allows for a delay between polity change and a regime's engagement in international affairs, but nevertheless reflects the non-consolidated condition of the democratizing state. If republican liberal theory is accurate, then, we would expect this variable to be highly correlated with accession to the CAT, at the very least, in that it codifies a customary norm of the international system, and in fact, this association should be higher than that for generic or stable "democracies." As noted in Chapter 2, moreover, this may relate to the socialization components discussed below in that we would expect the incidence of adhesion to the norms associated with democratization to vary with the degree of objective homogeneity among democracies. This suggests the possible relevance of an interaction term including both This dataset and documentation are available online at- 2 3 Note that the Polity IV data are limited to states with population greater than 500,000, and as such, a number of small states are not accounted for (Andorra, Antigua and Barbuda, Barbados, Belize, Brunei, Cape Verde, Dominica, Grenada, Iceland, Kiribati, Liechstenstein, Luxembourg, Maldives, Malta, Marshall Islands, Micronesia, Nauru, Palau, St. Kitts and Nevis, St. Lucia, St. Vincent and the Grenadines, Samoa, San Marino, Sao Tome and Principe, Seychelles, Solomon Islands, Suriname, Tonga, Tuvalu, Vanuatu, Bahamas). Data for these states are drawn from the Freedom House "Freedom in the World Country Ratings" dataset available at See also Freedom House (2004) in the references section. States are coded for transitions if they experience a change from "NF" QNot Free) or "PF" (Partly Free) status to "F" status. This supplement to the Polity IV dataset is also used for the democracy indicator variable below, with "F" states coded as democracies. 46 Democratization and the indicator for the proportion of democracies adhering to the norms. Socialization Hypothesis The next set of variables aims to test the effects of international socialization variables. Following Ramirez, Soysal, and Shanahan's (1997, 741) example, both domestic-level and socialization indicators are examined in an effort to separate out the dominant processes in the early part of the norm life-cycle (primarily domestic mobilization of norm entrepreneurs and idiosyncratic country-level commitment to the norms) and the latter phase of the norm cascade and consolidation (involving international and transnational socialization mechanisms). As such, this thesis elaborates country-level variables and a set of international and primarily systemic variables. Domestic Variables The first state-level variable is regime-type - perhaps the most obvious domestic characteristic. Human rights are generally considered to be strongly associated with political liberalism and democracy (Poe and Tate 1994, 855-7), and as such, it would appear likely that democratic polities would be more likely than autocratic ones to join international human rights instruments like the C A T and the SOP. 2 4 Regime-type is defined by DEMOC, a dummy variable indicating whether a state can be described as a consolidated democracy in any given year. Data for this variable is drawn primarily from the Polity IV dataset, with observations coded a T ' i f the Polity2 indicator is greater than 5 and '0 ' otherwise. While Polity! is defined on a 21-point scale ranging from -10 (extreme autocracy) to +10 (consolidated and stable democracy), states with scores 2 4 Note that this variable is analogous to Ramirez, Soysal, and Shanahan (1997) "Western status" variable. 47 ranging from -5 to +5 are considered to be polities ' in transition.' Data for countries missing from the Polity IV dataset is drawn from the Freedom House ratings dataset, as noted above. The crucial question here, of course, is whether regime type can be characterized as a primarily endogenous phenomenon. After all, adoption of democracy and adherence to international human rights norms might be two aspects of the same socialization process. This is undoubtedly accurate for at least some cases. The inclusion of a democratization indicator, however, should account for this process - i.e., newly democratic states also adopting human rights norms. As such, the DEMOC regime-type indicator should capture primarily the effect of pre-existing status as a stable democracy (which is by no means perfectly correlated with acceptance of the norms), and whether stable democracies were more likely than other types of states to adopt the norms prior to the initiation of a socialization dynamic. Coding the second state-level variable is more problematic, but in essence aims to account for the degree of existing domestic internalization of or commitment to the principles of the norm in question prior to adhesion to the international norm. Ramirez, Soysal, and Shanahan (1997) deal with this question using two variables. On the one hand, "welfare citizenship, refers to the number of key welfare state initiatives in place in the decade preceding acquisition of female suffrage" (740). On the other hand, "national organizations, measures the strength of the domestic or national women's political movement within each country" by counting the number of domestic political NGOs committed to women's rights (740). The former, then, might be said to account for the degree of a state's endogenous commitment to the broader concept of social equality and 48 justice, of which women's rights is a subset, while the latter refers to the degree of active pressure exerted by civil society on a state's power structure. For the purposes of examining the C A T and the SOP, elaborating a domestic human-rights NGO indicator is not practical. To the author's knowledge, no existing dataset includes this type of variable for the whole state system, and it would therefore have to be compiled from scratch, which is beyond this thesis' scope. It is unclear, moreover, that this type of indicator would prove useful in this study. While Ramirez, Soysal, and Shanahan's (1997) timeframe begins in the late 19 th century, the C A T and SOP both emerged in the human-rights "cascade" phase of the late 20 t h century, during which the distinctions between domestic and international NGOs are much more difficult to draw, and as such it is not obvious that a useful "domestic" indicator could be elaborated along these lines. Indeed, it might be more useful to elaborate a general indicator to capture the degree of pressure exerted by civil society groups on governments at both domestic and international levels. Attempting to capture the activities of TANs and NGOs and measure the processes described by Risse and Sikkink (1999) would represent a significant and worthwhile research project. Constructing an indicator to reflect states' existing commitment to the norm in question, however, is both more practical and more useful. With respect to the SOP, we choose to use Amnesty International's data on the date in which states carried out their last execution.25 We argue that refraining form resorting to capital punishment, whether or not it remains in a state's legal code, reflects a state's normative commitment to the illegitimacy of the death penalty. LastExec, then, is defined as a counting variable Available online at 49 indicating the number of decades since a state's last execution.26 We would expect, then, that countries exhibiting a smaller propensity to use the death penalty would be more likely to join the SOP. Additionally, this effect is probably non-linear, since the difference between a state that carried out its last execution in 1850 and one that did in 1870 is likely to be much less important than that between 1960 and 1980. A quadratic term is included to test for this effect. Similarly, with respect to the CAT, the actual practice of states regarding the respect of personal integrity is used, indicating the de facto commitment to not resorting to torture or other cruel or degrading punishment. Data was drawn from the dataset compiled by Poe and Tate (1994). Specifically, they compiled information regarding "coercive activities on the part of the government designed to induce compliance in others," which include "murder, torture, forced disappearance, and imprisonment of persons for their political views," using Amnesty International and U.S. State Department country reports (854-5). Their five-point scale ranges "from 1 for a country with a healthy record of respect for personal integrity (Canada, the United States) to 5 for a human rights disaster (Iran during several years in the 1980s)" (855). Amlnt, then, is defined as a lagged variable using the Poe and Tate (1994) ratings for 1980 and 1987, such that years before 1992 are coded with the 1980 data and 1992 and later years are coded with the 1987 data. The time lag is meant to establish the actual domestic 2 6 Therefore, if it took place in the 1990s, LastExec = 0, 1980s = 1, 1970s = 2, etc... Countries with no data in the Amnesty International data are coded as '0'. For countries that have not had any executions since independence, the actual dates of independence, as opposed to the year of entry into the system provided in the COW system data, are obtained from the CIA World Factbook (Central Intelligence Agency 2004). 2 7 Note that their data, available for the years from 1980 to 1987 includes only 153 countries (the vast majority of the system at the time). While data for the former Soviet and former Yugoslav republics and other newly independent states was drawn from the "mother" country's data, our dataset is still missing data for a small number of countries. 50 practice, as opposed to the change in practice that might result in the years immediately prior to joining the treaties as a result of international incentives. Note that because a " F indicates a country with a strong record, Amlnt should, at least initially, have a negative effect on the probability of accession to the treaties. That is, a country with a larger numerical rating - a worse human rights record - should be less likely to join the treaty. Moreover, for the sake of consistency, we will test for the effect of a quadratic term. Socialization Variables In order to account for the effect of socialization in the international system, this section elaborates variables that aim to represent the "objective homogeneity" that facilitates subjective categorizations and acts as the basis of the intersubjective social identities of states (Wendt 1999, 354). Once again, Ramirez, Soysal, and Shanahan (1997) provide an example - in their world polity sociological institutionalist framework - of how this might be accomplished. They construct a number of variables to reflect "world culture and isomorphism" processes (741), including the regional cumulative percentage of states adopting women's suffrage, the world suffrage count, and a lagged variable measuring when a state gained independence. The first two essentially attempt to capture changing "objective homogeneity" within social identities that states might belong to, while the independence variable examines another aspect of the process, whereby a high degree of norm adherence by newly independent states indicates that the norm in question has become associated with the general "state" identity. A number of analogous variables can be elaborated for the C A T and SOP datasets. 51 First, the independence variable is rather easily compiled. Independence is defined for both datasets as a lagged dummy variable for which observations are coded "1" if the state entered the system that year or in any of the immediately preceding four years. Data on independence dates is drawn from the COW "System" dataset (Correlates of War 2 Project 2003). Ultimately, i f either of the treaties is strongly associated with the general "state" identity, we would expect this variable to have a strong positive effect. Moreover, the alternative variables outlined below and the use of interactions should separate out the effect of the "state" identity from other reasons that might motivate states to join the treaty. The second group of variables aims to measure the degree of homogeneity in given identities with respect to the norms in question. They include the total number of state parties, the regional cumulative proportion of state parties, and the cumulative proportion of democratic states. The purpose is to differentiate between the general "state", regional, and "democratic" identities. The global adherence to the norms is reflected by the SystemPercent variable, which is illustrated in Table B . l and Figure B.4. Also shown is the DemPercent variable, which represents the proportion of stable democracies (as defined by DEMOC, based on the Polity IV and Freedom House data) party to the two norms. Both indicators are lagged, such that the data for 1994, for example, takes into account the state parties as of 1993. While it is not clear that this type of socialization would operate in such a short time-frame, lagging the variable at 5-year intervals is not feasible, given the relatively compressed time-spans (18 and 13 years) of the datasets. 52 Regional adherence to the norms is measured by the Region variable, depicted in Tables B.2 and B.3. This variable is particularly relevant to the C A T and the SOP in that some regions, like Europe and the Americas, exhibit highly developed regional human rights regimes, while others lag far behind, so a strong regional effect should be expected in this issue-area. Given the contested definitions of geo-political regions (as evidenced by the controversy over the hyphenation of "Asia-Pacific"), however, the regions are defined primarily geographically as the Western Hemisphere, Europe, Africa, Asia-Pacific, and the Middle East. This definition should roughly reflect the possible existence of regional identities. As Ramirez, Soysal, and Shanahan (1997, 740, fn. 11) note, however, the effect of these three indicators will not be constant. On the one hand, we would expect that as the proportion of states adhering to the norm increases, socialization effects would strengthen in a non-linear fashion (i.e., 70% adherence is likely to exert more than double the socializing pressure that 35% would). On the other hand, as adherence to the norm becomes more widespread and the pool of non-parties shrinks, an "exhaustion effect" should emerge, such that the probability of the remaining states adhering to the norms would drop, as these are probably particularly "resistant" states. In short, the effects of these variables should roughly follow a logistic, or s-curve. Another, possibly more serious problem, however, that these indicators raise is the fact that they are likely to be highly correlated with each-other and temporally dependent. In as much as these cumulative variables are defined in terms of the same process - the gradual adhesion of states to the norms, which appears to be a smooth, 2 8 Exceptions: Turkey is included in Europe, Egypt in the Middle East. Note that the Central Asian former Soviet republics are included in the Asia-Pacific region. 53 constant process - , they are likely to follow smooth upwards trends (as illustrated in Appendix B), and will therefore all be highly correlated with time and with each-other. Table C l lists the numerical correlation coefficients between these variables as well as the systemic power and IGO variables. Not only does this raise the problem of introducing a collinearity bias into the analysis, but it also makes the application of Beck, Katz, and Tucker's (1998) suggested correction for the problem of temporal dependence in BTSCS more difficult. Ultimately, part of the problem lies per se in using systemic variables that, by definition, do not vary cross-sectionally (Beck, Katz, and Tucker 1998, 1273). This cannot be helped, since the systemic aspect of the socialization variables is vital component of the theoretical framework. That said, an effort can be made to minimize the problem presented by the cumulative indicators by refraining from using all of them directly in the same model estimation. While this is far from ideal, the analysis in the next chapter will work on this basis.30 A final variable aims to account for the effect of social status in the socialization process. As was noted in Chapter 2, most theories predict that the adoption of a practice by high status members will make its diffusion to others more likely. While the power variable introduced earlier might well reflect status, a specifically social variable is IGOStatus. It calculates the proportion of the total number of IGO memberships in the system (IGO memberships are often used represent "connectivity" or "embeddedness" in international society - see Ingram, Robinson, and Busch, Unpublished, for example) The latter is introduced below. 3 0 Note that Ramirez, Soysal, and Shanahan (1997) are able to use this type of systemic indicators due to their choice of a specialized event-history technique, rather than the more familiar logit model. 54 comprised by the states acceding to a treaty in a given year.31 As with the other variables, the data are lagged one year, to allow for an effect in the next year. In short, the accession of a highly connected state (a Western European state, for example) should increase the likelihood of other states adhering to the norm. This is a particularly useful indicator in that it is distinct from power, since many small, highly connected European states possess minuscule amounts of conventional material power. Conclusion Unlike most existing quantitative studies of norm diffusion, this study does not use specialized event-history methods or survival analysis, choosing instead to use binary time-series cross-section analysis. Not only is logit the grouped duration equivalent of hazard rate models, but it is also much more prevalent in the IR literature and more familiar to the IR community. This chapter specifies the variables necessary to test the theoretical framework and hypotheses presented in Chapter 2, moreover, by aiming to operationalize indicators for the processes underlying the alternative hypotheses as well as both endogenous domestic and international socialization aspects of the constructivist account. While using systemic indicators raises potential problems for the analysis, they do not pose a fundamental threat to the comparative nature of this thesis' research design, which aims to compare and contrast the broad processes underlying the expansion of the norms against torture and the death penalty, as opposed to estimating the exact magnitudes of 3 1 Data on IGOs is drawn from the COW "IGO State-unit v2.1" dataset, available online at See Wallace and Singer (1970) and Pevehouse, Nordstrom, and Warnke (2002) for more information. Note that IGO data is only available up to 2000. For subsequent years, state-membership numbers for 2000 are used. 55 substantive effects. Chapter 4 uses the compiled datasets for the Convention Against Torture and the Second Optional Protocol to analyze the dynamics of adhesion to the two treaties, and empirically evaluate the relevance of the different theoretical explanations for the puzzle examined in Chapter 1. 56 Chapter 4 Empirical Analysis: Contrasting the Rates of Adhesion to the CAT and the SOP This chapter conducts an empirical analysis of the dynamics of adhesion to the Convention Against Torture and the Second Optional Protocol. Unlike most quantitative studies, it does not aim to estimate the magnitude of the effects exerted by the independent variables on the probability of states' joining either of the treaties. The key to addressing the empirical puzzle, rather, is to provide an explanation for the difference in the extent to which the two norms have been adopted. As such, the following analysis is primarily a comparative exercise in which identical models are estimated for both the C A T and the SOP, and the direction and significance of the substantive effects, rather than their magnitude, are compared across the two sets of results in an effort to test the success of the theoretical framework presented in Chapter 2. Basic Model The basic model takes into account both the alternative explanations and the domestic and socialization indicators. Ideally, it would incorporate both the proportion of the system, the proportion of democracies and the regional proportional indicators to test for the effects of different identities, and would be: 57 Joined = p0 + Pi (Domestic) + p2 (DEMOC) + p3 (% Dems) + p4 (% Regional) + p5 (% System) + p6 (Indep.) ^ + p7 (Status Chg.) + p8 (Power Chg.) + p9 (Democratization) + u As pointed out in Chapter 3, however, this basic model would present a number of problems, the most important being that the cumulative variables are highly correlated both with each other and with time, as is shown in Table C l . The proportions of the system and of democracies and the year are all correlated with each other at over 0.95, while Regional has a lower correlation, though it is still relatively high with respect to the C A T . These high correlations reflect the fact that growth in system membership and in the number of democracies was sufficiently constant over the 1986-2003 time frame not to create significant variation in the proportional outcomes, and, of perhaps greater importance, that the cumulative nature of the variables induces autocorrelation. In fact, a similar problem was encountered in elaborating the power and status indicators, prompting the use of annual changes as opposed to cumulative proportion. The regional indicator is not as highly correlated because regional growth in adhesion to the norm is uneven, which causes cross-sectional variation. Indeed, for the SOP, in which membership is overwhelmingly European, the correlation is quite low, despite the use of cumulative indicators. For the CAT, which has experienced more uniform growth across the regions, it is more highly correlated. Ultimately, this high correlation between the system proportion and the proportion of democracies is problematic because it does not allow for an accurate 58 estimation of their separate effects, and indeed, may well reflect an unobserved growth process also correlated with time. The variables, however, are crucial aspects of this thesis' theoretical premise, and cannot simply be discarded. The solution lies in further specifying the theoretical conditions under which they should have an effect. On the one hand, in as much as we argue that the proportion of the system adhering to the norm represents the objective homogeneity on which the system-wide "state" identity is based, its effect should be particularly observable in the behavior of newly independent states. That is, i f independent states try to adopt aspects of the accepted "state" identity, we would expect that the rate of accession to the C A P and the SOP would become more pronounced as they became more associated with that identity. Interacting Independence with SystemPercent should then account for the effect of the latter. Similarly, we would expect that effect of the Democratization indicator (which reflects Moravcsik's (2000) republican liberal theory) on the probability of accession to the treaties would depend on the extent to which it is accepted by democracies, and as such interacting DemPercent with Democratization should account the effect of the proportional variable. Model 2, then, is specified as: Joined = p0 + Pi (Domestic) + p2 (DEMOC) + p3 (% Regional) + P4 (Independence) + Ps (Independence * SystemPercent) [Model 2] + p6 (Status Chg.) + p7 (Power Chg.) + p8 (Democratization) + P9 (Democratization* DemPercent) + u As noted in Chapter 2, moreover, it is reasonable to suspect that the effect of the Domestic indicators might be non-linear (in addition to the non-linearity induced 59 by the logit model). As such, Model 2.1 incorporates a quadratic Domestic term. Estimates for both models are presented in Table C.2 in Appendix C. Broadly, the results suggest that the domestic indicators, being a stable democracy, and regional socialization are all significant for the process underlying the growth of both the C A T and the SOP. Democratization is significant at the 0.05 level for the CAT, and appears to vary with respect to the proportion of democracies adopting the norm. Moreover, the domestic indicator does indeed appear to have a decreasing effect on the probability of joining, as demonstrated by the 2.1 models, which appear to fit better than Model 2 (as indicated by the higher pseudo R for the unrestricted models), and in which the indicators exhibit greater significance. Notably, neither PowerChange and StatusChange nor the independence-based indicators are highly insignificant at traditional levels, and the null hypothesis that their actual effect is zero cannot be rejected. In short, this suggests that in the case of both norms, country-unit attributes (Domestic and DEMOC) and regional socialization processes were key to their expansion. Moreover, the C A T appears to exhibit a socialization process within the "democracy" identity, in which the norm has become associated with it. As such, new democracies are more likely to join the CAT as the objective homogeneity increases within the democratic group, while the effect on SOP adhesion is not significant. Finally, the results do not support power-based nor status-based hypotheses for norm expansion, while socialization at the level of the global "state" identity appears not to be significant factor. 60 The results do raise several interesting questions. On the one hand, growing homogenization amongst democracies appears to reduce the effect of democratization, which on first inspection would seem to be the opposite of what we would expect. It is not clear what is causing this effect, although one might speculate that as the norm against torture grew and became increasingly associated with the "state" identity, which due to the temporal process discussed earlier, is correlated with the DemPercent indicator, the need for new democracies to secure this norm would decrease. As such this result might result from temporal dependence. On the other hand, and more surprisingly, both domestic indicators have positive, i f decreasing effects. While this is perfectly reasonable for the SOP, it is certainly not what was expected for the CAT. That is, since the domestic indicator for the C A T is a categorical variable in which the lower numerical values indicate better respect for personal integrity, we would expect the domestic parameter to be inversely associated with the probability of adherence to the norms. States with greater values (less tradition of respect for personal integrity) should be less likely to join the CAT, not more likely as the results indicate. Thus, while the effect is decreasing, states with worse records appear to be more likely to join than those with better records. Needless to say, this result is rather puzzling. A possible explanation is suggested by the Chilean experience, in which a state with an extremely abusive record was quick to join the convention following a regime change and democratic transition, and as such the result would be an artifact of the time-lag in 61 coding the variable. Conceptually, this is plausible, but it is again not clear whether post-observation transitions took place in the systematic fashion necessary to generate the result. Empirically, moreover, when Domestic is interacted with Democratization, the interaction terms consistently fail to show significance. In short, the observed effect appears to be substantive. Indeed, when the models for C A T are run using a series of dummy variables instead of Domestic (an ordinal variable), the trend of the result is robust. Interestingly, the strongest effect appears to occur at the " 3 " value of the scale, which indicates countries in which: [tjhere is extensive political imprisonment, or a recent history of such imprisonment. Execution or other political murders and brutality may be common. Unlimited detention, with or without trial, for political views is accepted. (Poe and Tate 1994, Appendix, 867) This, then, suggests that it may be countries with ambiguous records of respect for personal integrity that most seek to join international instruments such as the C A T as a means of legitimizing their status in the international community, which would suggest a socialization process. That said, in other exploratory analyses, interacting Domestic with SystemPercent showed little significance, and in fact exhibited a negative parameter. In short, this result presents an interesting puzzle that should be explored further. Future work might examine the conditions under which this effect is relevant, or might alternatively seek to elaborate a more valid indicator for the process we aim to capture, 62 which is beyond the scope of this thesis. What the result does indicate, however, is that existing domestic attributes are highly significant, and regardless of the substantive implications, the fact that they account for the influence of idiosyncratic state-level characteristics allows us to more confidently analyze the separate effects of the international and socialization variables. Temporal Dependence As was noted at the beginning of this chapter, Beck, Katz, and Tucker (1998) characterize the possible violation of the temporal independence assumption as a major threat to binary time-series cross-section analysis, as it can "result in overly optimistic inferences (underestimates of variability leading to inflated /-values)" (1261). Temporal dependence with respect to a given model arises from the operation of a temporally correlated unobserved variable that causes a temporal trend in the data that is not accounted for by the independent variables incorporated into the analysis. While event-history techniques such as Cox regression (Cox 1972) deal with this issue by factoring out the varying baseline hazard from the effective hazard rate equation, logit analysis is inherently vulnerable. As such, Beck, Katz, and Tucker (1998) suggest that researchers always test for temporal dependence and correct for it i f necessary. The initial analyses of our data, moreover, suggest that temporal dependence is indeed a possibility, especially for the CAT. One way of testing for temporal dependence is to include a series of temporal dummy variables - one for each year in the data - in the model, and subsequently conduct a standard likelihood ratio test of the null hypothesis that the actual effect of the 63 temporal dummies as a whole is zero (Beck, Katz, and Tucker 1998,1269). Model 3, specified below, incorporates the temporal dummies. Note that PowerChange and StatusChange, which do not vary cross-sectionally, were dropped due to collinearity. This is not a significant disadvantage, seeing as neither variable even came close to statistical significance in the four models estimated in Table C.2. Joined = po + Pi (Domestic) + p2 (Domestic)2 + p3 (DEMOC) + P4 (% Regional) + Ps (Independence) + P6 (Independence * SystemPercent) + P7 (Democratization) + Pg (Democratization* DemPercent) + 8to-t (time dummyto-t) + u Table C.3 shows the estimates of the parameters for the substantive variables in this model. The likelihood ratio test for the C A T yielded a chi-squared statistic of 51.34 and therefore P>y? = 0.0000, while the SOP yielded a statistic of 35.57 and P > = 0.0004. In short, it appears that the CAT and SOP data are indeed temporally dependent. As such, Model 3 should provide more accurate estimates than Model 2 or 2.1. The results tend to confirm and extend the findings of Model 2.1. For the CAT, all the variables except Independence and the interactive term between Democratization and DemPercent are significant at the 0.05 level. The domestic terms and the regional variable, moreover, are highly significant at the 0.001 level, with less than one chance out of a thousand that the parameter estimates would emerge i f the real effects of the variables were nil. The independence term, moreover, is significant at the 0.10 level, and interaction between democratization and DemPercent is only slightly 64 higher, with a p-value of 0.11. In short, with the possible exception of the latter terms, all the variables in the regression appear to be significant, suggesting that both endogenous domestic attributes and socialization along the regional and systemic, and possibly the democratic, identities played significant roles in the expansion of adhesion to the norm against torture enshrined in the CAT. The SOP estimates of Model 3 indicate that the same three variables identified in the Model 2 results - the domestic terms and the regionalization variable - are all significant at the 0.01 or smaller level. A l l other variables, moreover, remain statistically insignificant at traditional levels. The substantive magnitudes of the variables also remain roughly equivalent. The results of Model 3 - which should be more accurate than previous models due to its correction for temporal dependence - then, suggest that the broad theoretical account of the different expansion dynamics for the C A T and the SOP is generally accurate. The SOP estimates indicate that states acceded to the treaty primarily due to pre-existing affinity with the norm as a result of endogenous processes, as well as increasing socialization pressure at the regional level. This is consistent with the theoretical argument that the norm against the death penalty codified in the SOP originated in Europe amongst a core of early abolitionist states, and that once a sufficient proportion of European states had adopted the norm, it became associated with the European identity, setting off a socialization dynamics at the regional level. Importantly, however, while non-European states may have adopted the norm for domestic reasons, the regionalization process has not expanded to other regions, and the norm has not engaged in other socialization processes. 65 The C A T estimates suggest that while state-level factors are clearly a crucial factor in explaining adherence to the convention, socialization process of all kinds, but especially those associated with regional and systemic dynamics, also drove the expansion of the norm. There is also tentative support for the republican liberal hypothesis. Again, these results are consistent with the hypothesis that while the movement to codify the norm against torture in the C A T originated for endogenous reasons primarily in a European context (though some Latin American states also lobbied for the convention), and eventually unleashed a socialization process at the regional level in Europe (such that over 90% of European states currently espouse the norm), it was able to break out of the European context, unlike the SOP. That is, sufficient numbers of democracies adopted the norm that it became associated with the democratic identity, and eventually enough states adopted the norm so as to associate it with the more general global identity of the state. The different outcomes might be explained in a number of ways. On the one hand, idiosyncratic domestic factors may have simply prompted more states to become early adopters of the CAT and not the SOP. On the other hand, the customary or settled nature of the norm against torture suggests that it may have exhibited a higher pre-existing rate of internalization than that for the norm against the death penalty, and as such, and as such, a higher proportion of states were willing to sign up initially, before the socialization processes took hold. In either case, it appears that the larger number of states that adopted the C A T early on led to both a stronger regional socialization dynamic in Europe (where there essentially is objective homogeneity), and, more importantly, was sufficient to prompt norm cascades and the accompanying socialization processes within 66 the democratic and "state" identities. This led to its widespread adoption throughout the system, such that most other regions have C A T membership hovering between 60% and 70%. Thus, despite their similar origins, the fact that the core adopters of the C A T hit the threshold for cascading along systemic collective identities, and that the SOP did not, explains why the C A T has expanded to include a large majority of the system, while the SOP remains a highly contested norm at the systemic level. To further test these conclusions, it is possible to estimate Model 3 for relevant subsets of the population. That is, Model 3.1 estimates the model for non-European states, and Model 3.2 estimates it for states that were original signatories of the convention. For its part, Model 3.1 allows us to examine whether the dynamics for non-European states were indeed substantially different between the two norms.32 As would be expected, only the domestic indicators are significant for the SOP, which suggests that the regional socialization dynamic was in fact limited to Europe. The regionalization indicator is also non-significant for the CAT, while the interactions with SystemPercent and DemPercent - the non-regional socialization indicators - remain significant. This suggests that while the influence of domestic attributes persists, universal socialization mechanisms (especially within the democratic identity) were more prevalent amongst non-European states, and that regional socialization is largely a European phenomenon. In short, the estimates for Model 3.1 support the account suggested by the theory and by Model 3. Note that estimating results only for European states is not practical due to what would be a significantly reduced sample size. 67 Model 3.2 examines the other component of the story. Simply running the model for European states is not practical due to what would be a significantly reduced sample size. Moreover, it is also desirable to test directly for the transition between the norm emergence and norm cascade phases described by Finnemore and Sikkink (1998). Ramirez, Soysal, and Shanahan (1997) use a piecewise approach to accomplish this, dividing the data into pre- and post-tipping point sets. They identify the tipping point by means of a visual inspection of the scatter-plot of the cumulative proportional adherence to the women's suffrage norm (Ramirez, Soysal, and Shanahan 1997, 738). As the graphs in Appendix A illustrate, however, the progression of adherence to the norm does not provide compelling evidence of a tipping point for the CAT and SOP ratification and accession process. This is at least partly attributable to the fact that the norms are defined as treaties, and that the instruments themselves would have been unlikely to emerge without a sufficient critical mass of support in the first place. Finnemore and Sikkink (1998, 901) suggest that "[w]here treaties exist, the entry into force of the treaty may be a useful proxy for the critical mass necessary to say that a norm exists," but in as much as the number of states required for entry into force varies widely, and that generally many more states sign on initially than are required, it is not clear that this is a useful proxy. Moreover, in both empirical cases, dividing the dataset along these lines would result in a severely reduced sample size for the pre-cascade phase. A n alternative, which is pursued in Model 3.2, is to categorize states that joined the treaties according to whether they were original signatories that ratified the treaty or whether they acceded to it thereafter. Theoretically, we would expect that the original signatories would be more likely to be motivated by the domestic and regional 68 socialization factors associated with the emergence of the norms in Europe, while those that acceded later would do so for more external reasons. Model 3.2, then, estimates Model 3 for the subset of the population excluding those states that acceded to the treaties (i.e., Accession = 0). Once again, the results appear to support the account presented above. For the C A T estimation, the domestic terms and the regional indicators are highly significant, with p-values under 0.01, while the Democratization variable appears to be much more weakly significant, at the 0.10 level. For the SOP, once again, the domestic indicators are significant (one at the 0.001 level, the other at the 0.10 level), and the regional variable is also highly significant, at the 0.001 level. In short, both estimates exhibit the characteristics that we would expect from a norm emergence phase sample. Importantly, these estimates are markedly different than those in Model 3.1, which represents the cascade phase, and in which the C A T exhibits a high degree of systemic-level socialization, while the SOP only appears to be influence by domestic factors, since the socialization dynamics has not broken out of the European context. Conclusions In sum, once the correction for possible temporal dependence in the data was made using the temporal dummy variables, all three models - Model 3, Model 3.1, and Model 3.2 - appear to support the account derived from the social constructivist approach to norms presented in Chapter 2. It is true that a higher degree of existing internalization and acceptance of the norm against torture - derived from its long-standing customary status - could be argued to have been the essentially idiosyncratic factor that propelled the C A T across the tipping point and into the cascade phase. That said, the evidence presented in this chapter strongly supports the contention that it is the subsequent 69 socialization and "break-out" of the C A T from its origins in Europe, and not simply the greater incidence of endogenously-driven adoption, that can best explain how it spread so widely. The similar norm against the death penalty codified by the SOP remains a largely an exclusively European phenomenon because the rate and pattern of domestically-driven adoptions has not been sufficient to initiate socialization mechanisms in more universal identities. Thus, while the idiosyncratic causes underlying the endogenous adoption of the norms form part of the causal story, they are not sufficient to fully account for the vastly different outcomes for the two norms, yielding an explanatory gap that is filled by the socialization and norm-cascade framework. 70 Conclusion This thesis finds substantial empirical evidence for the existence of a norm life cycle and the operation of socialization mechanisms. It suggests that Finnemore and Sikkink's (1998) framework is indeed broadly applicable to norms in general, including formal norms codified in international legal instruments. With respect to the specific dynamics of norm diffusion, furthermore, it points to the possibility that a given norm experiences a life-cycle within each collective identity it comes to encompass. Specifically, cascading and socialization may take place within one identity - say, the region - , but fail to spread to other, more universal identities. Successful norms like the Convention against Torture manage to "bridge" the dividing line between different identities, either by means of overlapping identities or due to endogenous domestic factors, and thus spread widely. In short, the findings presented in Chapter 4 suggest that the norm life cycle from a systemic perspective is not a single dynamic process, but rather a combination of a number of interacting life-cycles that may not unfold at the same pace. These conclusions point to several factors that actors seeking to spread a norm should take into account. That is, while 'seeding' a norm within a region is effective in that it allows it to harness a regional socialization dynamic, entrepreneurs also need to focus on 'bridging' the norm to other identities. Whether by means of selective T A N pressure on extra-regional democratic states or by concentrating resources on susceptible states in the system, it is necessary to generate a sufficient critical mass to set of cascading within different identities. This effort might be strengthened, moreover, by framing the issue in the context of an existing customary or settled norm. 71 Ultimately, the analysis does indicate that the customary or 'settled' status of the norm against torture played a crucial role in the evolution of the Convention against Torture and the Second Optional Protocol. The larger number of states initially supporting the C A T - which probably explains why it was able to cascade within the more universal identities - , appears to have been strongly influenced by Poe and Tate's (1994) state-level ratings regarding respect for personal integrity. As noted, however, the direction and form of this variable's effect was puzzling, with a 'worse' score in the rating associated with a higher probability of joining the CAT. That said, this result may well be explained by the customary status of the norm against torture, in that, given the association of the norm with the state identity, violators would be more inclined pay lip service to a codified norm so as to maintain their status as responsible members of international society. Thus, while this dynamic might be considered a form of socialization peculiar to implicit customary norms, what is unclear is what exactly defines this type of norm empirically. Is it the level of homogeneity in the system? Its longevity? Or is it an intangible above and beyond this type of unit-centered measures? Answering these questions is likely to be relevant to understanding how and why some norms experience cascading across multiple identities while others do not. As such, they point to the need for further empirical work in international relations within the framework of normative evolution and the norm life cycle. 72 Appendix A Table A.1 State Parties to the Convention against Torture Year Number Rate(%) Total Parties % System Joined 1986 1 0.6 1 0.6 1987 24 15.0 25 15.5 1988 10 7.4 35 21.7 1989 11 8.7 46 28.6 1990 7 5.9 53 32.1 1991 10 8.1 65 36.7 1992 7 6.0 72 39.8 1993 7 6.1 80 43.0 1994 7 6.5 87 46.5 1995 6 6.0 93 49.7 1996 7 7.4 100 53.5 1997 5 5.7 105 56.1 1998 6 7.3 111 59.4 1999 8 10.1 119 62.6 2000 5 6.9 124 64.9 2001 4 6.0 128 67.0 2002 4 6.3 132 68.8 2003 1 1.7 133 69.3 Source: U N H C H R 2003. See Chapter 3 for coding details. 73 < o • Figure A . 1 Rate of Adhesion to C A T o 0) .c S to 1985 1990 1995 2000 2005 Figure A.2 Proportion of System Party to C A T < o (0 C L E . S^. in . • >. OT o c o r O CM . Q . • O Q. 1985 1990 : 1995 Year 2000 2005 74 Table A.2 State Parties to the Second Optional Protocol Year Number Rate (%) Total Parties % System Joined 1991 10 5.6 10 5.6 1992 2 1.2 12 6.6 1993 7 4.0 19 10.2 1994 4 2.4 . 23 12.3 1995 5 3.0 28 15.0 1996 1 0.6 29 15.5 1997 2 1.3 31 16.6 1998 2 1.3 33 17.6 1999 8 5.1 41 21.6 2000 3 2.0 44 23.0 2001 2 1.4 46 24.1 2002 2 1.4 48 25.0 2003 1 0.7 49 25.5 Source: U N H C H R 2003. See Chapter 3 for coding details. 75 Figure A .3 Rate of Adhesion to the SOP to -TO a: 1990 1995 2000 2005 Year Figure A.4 Proportion of System Party to the SOP CM O CM E 3 , « u, I o 8 •• © e o CO ' a ^ ; 1990 1995 , 2000 > 2005 76 Appendix B - Data and Coding Figure B . l Growth in System Membership o o .® — — e • a © © © © 1 ; 1 , — . , j _ 1985 1990 1995 2000 2005 Year 77 Figure B.2 Systemic Material Power Associated with C A T $00 o ••• L _ co 0) o a. u E T T 2 • (0 >< CO . . . . . o c Q CM • •e o CL O i— Q_ 1985 1990 1995 Year 2000 < CO . O • o CO CL I— <u S.™ o E a> 05 >. co "co c ' 9 T -•c • o CL o Q) o> c ro o O 1985 1990 1995 Year 2000 78 Figure B.3 Systemic Material Power Associated with SOP 1990 1995 2000 Year 2005 OT O Q. CO a> o , CL y CM £ 9 3 UI >, OT 15 j -c <=> o r o . Q. O a> c O 1990 1995 Year .2000 2005 79 Table B.l Systemic Variables CAT SOP Year % System % Democracy % System % Democracy 1986 0.6 0.0 1987 15.5 0.0 — 1988 21.7 21.7 — 1989 28.6 37.5 — 1990 32.1 50.9 — 1991 36.7 53.8 5.6 0 1992 39.8 55.4 6.6 12.2 1993 43.0 56.6 10.2 14.5 1994 46.5 61.0 12.3 23.4 1995 49.7 64.0 15.0 29.3 1996 53.5 68.4 15.5 35.5 1997 56.1 72.7 16.6 35.1 1998 59.4 76.3 17.6 39.5 1999 62.6 77.9 21.6 40.3 2000 64.9 88.8 23.0 47.5 2001 67.0 90.4 24.1 48.2 2002 68.8 89.4 25.0 48.2 2003 69.3 91.8 25.5 50.6 Source See Chapter 3 for details. 80 Figure B.4 Scatter-plots of Systemic Independent Variables oo J O o E a> >. OT O o o o o © e e i i 1 1 1-1985 1990 1995 2000 2005 Year CM O 0 4 OT E CD >. OT 1990 1995 2000 2005 Year 81 « « • CO < o o co • CL-IO g> o co _ o o E CD a 3^ CM O H 1985 1990 1995 Year 2000 2005 O w CO 0. cu >'<J CO J i - pu. O .. O E CD Q 1990 1995 2000 Year 2005 82 Table B.2 Regional Proportion of States Party to the C A T Year Europe % Western Africa % Asia-Pacific % Middle East % Hemisphere % 1986 0.0 0.0 0.0 0.0 0.0 1987 0.0 0.0 0.0 3.3 0.0 1988 36.7 16.7 8.0 6.7 7.1 1989 46.7 30.6 10.0 10.0 7.1 1990 66.7 33.3 15.6 13.3 7.7 1991 57.9 38.9 17.6 13.2 7.7 1992 68.3 41.7 17.6 15.8 30.7 1993 68.9 41.7 23.1 18.4 30.7 1994 73.3 47.2 28.8 17.9 30.7 1995 80.0 50.0 32.7 20.5 30.7 1996 82.2 52.8 34.6 28.2 30.7 1997 86.7 55.6 38.4 30.8 38.5 1998 86.7 58.3 42.3 33.3 46.2 1999 86.7 58.3 46.2 38.1 53.8 2000 88.9 61.1 53.9 41.2 53.8 2001 88.9 61.1 59.6 41.2 69.2 2002 88.9 63.9 65.4 43.2 69.2 2003 91.1 63.9 69.2 45.5 69.2 Table B.3 Regional Proportion o f States Party to the SOP Year Europe % Western Africa % Asia-Pacific % Middle East % Hemisphere % 1991 0.0 0.0 0.0 0.0 0.0 1992 21.9 0.0 0.0 2.6 0.0 1993 24.4 0.0 0.0 2.6 0.0 1994 28.9 11.1 1.9 2.6 0.0 1995 37.8 11.1 1.9 2.6 0.0 1996 44.4 11.1 5.8 2.6 0.0 1997 46.7 13.9 5.8 2.6 0.0 1998 48.9 13.9 5.8 2.6 0.0 1999 48.9 16.7 5.8 4.8 0.0 2000 64.4 16.7 5.8 7.0 0.0 2001 66.7 16.7 7.7 9.3 0.0 2002 71.1 16.7 7.7 9.1 0.0 2003 73.3 16.7 9.6 9.1 0.0 83 Appendix C - Analysis and Results Table C l Correlations of Systemic Variables % System % Dems % Regional Year % System CAT SOP % Dems CAT SOP % Regional CAT SOP 1.0 1.0 0.9900 0.9911 0.8374 0.3061 1.0 1.0 0.8249 0.3115 1.0 1.0 — Year CAT SOP 0.9737 0.9827 0.9409 0.9613 0.8308 0.2926 1.0 1.0 84 Table C.2 Logit estimates for Models 2 and 2.1 Variable Model 2 - CAT Model 2.1 - CAT Model 2 - SOP Model 2.1 - SOP Coefficient Coefficient Coefficient Coefficient (Sid. Error) (Sid. Error) (Std. Error) (Std. Error) Domestic .1625819* 1.320745*** .0275013* .5169175**** .0910489 .4470057 .0/57905 .1047875 (Domestic)2 — - .207569*** — - .0220961**** .079525 .006102 D E M O C .4054966** .589918*** 1.080829*** .2464099 .21123 .2195204 .3794528 .4287098 Regional 1.081702** 1.33670** 2.742559**** 3.841295**** .5226739 .5370932 .7645577 .8182595 Independence -1.776371 -2.30966 -1.94343 -1.052034 2.508148 2.62142 1.632688 1.756911 Independence * 6.342398 7.091685 10.6991 4.122016 SystemPercent 6.19418 11.9994 13.23601 Status Change -3.278264 -3.464766 -17.63198 -16.89522 3.631181 3.653533 12.18686 11.97914 Power Change 2.551724 2.525015 9.191868 6.126623 2.167468 2.177402 16.05249 /6 .J2557 Democratization 1.408729** 1.491862** .4807778 .5946492 .6132407 .6315145 .7c966i5c9 .7509596 Democratization * -1.974732* -2.347496** -1.41155 -1.394164 DemPercent 1.07138 1.119544 2.6624/9 2.604473 N = 1574 1574 1758 1758 Pseudo R 2 0.0311 0.0399 0.0976 0.1786 Prob > x 2 0.0018 0.0002 0.0000 0.0000 * p<0.\0 * * ;?<0.05 *** p<0.0\ **** p< 0.001 Sources: See methods section in Chapter 3. 85 Table C.3 Logit estimates for Models 3, 3.1, and 3.2. Variable Model 3 Model 3 Model 3.1" Model 3.1 Model 3.234 Model 3.2 CAT SOP CAT SOP CAT SOP Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient (Std. Error) (Std. Error) (Std. Error) (Std. Error) (Std. Error) (Std. Error) Domestic 1.519742**** .4361685**** 1.494397*** .6474975**** 1.90322*** .5621232**** .4525866 .1036873 .5316549 .2008861 .6036999 .1733486 (Domestic)2 - .2445831*** -.0179274*** -.2226612** -.0299275** - .3360017*** -.0230952* .0809094 .0059814 .0915434 .0129315 .1135377 .0123151 DEMOC .5129187** .3434813 .4554562* .3973352 .429524 .7721308 .2253692 . 4253921 .2576677 .6615569 .2979084 .6995281 Regional 3.647622**** 5.486922**** 1.220799 -6.129497 6.451556**** 8.387041**** .9376681 . 9791821 1.366971 6.983251 1.409191 1.964572 Independence -3.869683* -2.613135 -2.653957 — -6.246693 -1.92655 2.06888 1.803759 2.196368 5.976621 2.176538 Independence * 11.30588** 16.47134 8.938155* — 14.46637 20.50058 SystemPerc 4.831514 13.26468 5.036785 14.62039 17.35309 Democratization 1.333595** -.3969174 1.780646** -.6681231 1.535762* -.4082491 .6741977 .7890134 .7015264 1.471886 1.535762 1.070504 Democratization -1.85928 2.037487 -2.38758** .6676226 -1.366942 .9245073 * DemPercent 1.187868 2.73001 1.215395 5.366278 -1.366942 4.750439 N = 1671 2042 1517 1120 1157 1450 Pseudo R 2 0.0979 0.1549 0.0771 0.1756 0.1578 0.3173 Prob > x 2 0.0000 0.0000 0.0013 0.0057 0.0000 0.0000 * ;?<0.10 * * /?<0.05 * * * /?<0.01 **** p< 0.001 Note: Estimates for temporal dummy parameters are not included Sources: See methods section in Chapter 3. Model 3 estimated for non-European states. Model 3 estimated for states that did not accede to treaties after initial signature period. 86 References Amnesty International. 2003. Facts and Figures about the Death Penalty. Available online at http.7/ Axelrod, Robert. 1986. An Evolutionary Approach to Norms. American Political Science Review, 80, 4: 1095-1 111. Barrett, Jastine. 2001a. The Prohibition of Torture under International Law: Part 1: The Institutional Organisation. International Journal of Human Rights, 5, 1: 1-36. . 2001b. The Prohibition of Torture under International Law: Part 2: The Normative Content. International Journal of Human Rights, 5, 2: 1-29. Beck, Nathaniel. 1998. Modelling Space and Time: The Event History Approach. In eds. Elinor Scarbrough and Eric Tanenbaum, Research Strategies in the Social Sciences: A Guide to New Approaches. Oxford, U .K. : Oxford University Press. Beck, Nathaniel and Jonathan N . Katz and Richard Tucker. 1998. Taking Time Seriously: Time-Series—Cross-Section Analysis with a Binary Dependent Variable. American Journal of Political Science, 42, 4: 1260-1288. Bicchieri, Cristina, Richard Jeffrey and Brian Skyrms eds. 1997. The Dynamics of Norms. New York, N Y : Cambridge University Press. Boli , John and George M . Thomas. 1997. World Culture in the World Polity: A Century of International Non-Governmental Organization. American Sociological Review, 62, 2: 171-190. Box-Steffensmeier, Janet M . and Christopher J. W. Zorn. 2001. Duration Models and Proportional Hazards in Political Science. American Journal of Political Science, 45,4: 972-988. Bull, Hedley. 1977. The Anarchical Society. New York, N Y : Columbia University Press. Burgers, J. Herman and Hans Danelius. 1988. The United Nations Convention against Torture: A Handbook on the Convention against Torture and Other Cruel, Inhuman or Degrading Treatment or Punishment. Dordrecht, the Netherlands: Martinus Nijhoff Publishers. Central Intelligence Agency. 2004. World Factbook 2004. Available online at 87 Chaiken, Shelley, Wendy Wood and Alice H. Eagly. 1996. Principles of Persuasion. In eds. E.T. Higgins and A . Kruglanski, Social Psychology: Handbook of Basic Principles. New York: Guildford Press. Checkel, Jeffrey T. 1999. Norms, Institutions, and National Identity in Contemporary Europe. International Studies Quarterly, 43: 83-114. Cook, Helena. 1991. The Role of Amnesty International in the Fight Against Torture. In ed. Antonio Cassese, The International Fight Against Torture. Baden-Baden: Nomos Verlagsgesellschaft. Correlates of War 2 Project. 2003. State System Membership List, v2002.1. Available online at http://c-ow2.1 Cortell, Andrew P. and James W. Davis, Jr. 1996. How Do International Institutions Matter? The Domestic Impact of International Rules and Norms. International Studies Quarterly, 40: 451-478. Cox, D. R. 1972. Regression Models and Life-Tables. Journal of the Royal Statistical Society, Series B (Methodological), 34, 2: 187-220. Dennis, Michael J. 2003. Human Rights in 2002: The Annual Sessions of the U N Commission on Human Rights and the Economic and Social Council. The American Journal of International Law, 97, 2: 364-386. Donnelly, Jack. 1986. International human rights: a regime analysis. International Organization, 40, 3: 599-642. Ellickson, Robert C. 2001. The Market for Social Norms. American Law and Economics Review. Finnemore, Martha. 1993. International Organizations as Teachers of Norms: he United Nations Educational, Scientific, and Cultural Organization and Science Policy. International Organization, 47, 4: 565-597. . 1996. National Interests in International Society. Ithaca, N Y : Cornell University Press. Finnemore, Martha and Kathryn Sikkink. 1998. International Norm Dynamics and Political Change. International Organization, 52, 4: 887-917. Florini, Ann. 1996. "The Evolution of International Norms." International Studies Quarterly. 40: 363-389. Florini, Ann. 1996. The Evolution of International Norms. International Studies Quarterly, 40: 363-389. 88 Freedom House. 2004. Freedom in the World Country Ratings. Available online at Goodman, Ryan and Derek Jinks. 2003. Measuring the Effects of Human Rights Treaties. European Journal of International Law, 14, 1: 171-183. Hasenclever, Andreas, Peter Mayer and Volker Rittberger. 1997. Theories of International Regimes. Cambridge, U.K. : Cambridge University Press. Hood, Roger. 1999. Introduction - The Importance of Abolishing the Death Penalty. In Council of Europe, The Death Penalty: Abolition in Europe. Strasbourg Cedex: Council of Europe Publishing. Ingram, Paul, Jeffrey Robinson and Marc Busch. Unpublished. The Intergovernmental Network of World Trade: IGO Connectedness, Governance, and Embeddedness. Working paper at the Center for the Study of Economy and Society, Cornell University. Available online: papers.shtml. Jang, Yong Suk and Xiaowei Luo. 2001. Nation-State Participation in Intergovernmental Technology Organizations. International Journal of Comparative Sociology (Brill), 42, 3: 255-285. Jepperson, Ronald L. , Alexander Wendt, and Peter J. Katzenstein. 1996. Norms, Identity and Culture in National Security. In ed. Peter Katzenstein. The Culture of National Security: Norms and Identity in World Politics. New York, N Y : Columbia University Press. Keck, Margaret E. and Kathryn Sikkink. 1998. Activists beyond Borders: Advocacy Networks in International Politics. Ithaca, N Y : Cornell University Press. Keohane, Robert. 1986a. Theory of World Politics: Structural Realism and Beyond. In ed. Keohane, Robert. 1986. Neorealism and its Critics. New York, N Y : Columbia University Press. . 1986b. After Hegemony: Cooperation and Discord in the World Political Economy. Princeton, NJ: Princeton University Press. King, Gary, Robert O. Keohane, and Sidney Verba. 1994. Designing Social Inquiry: Scientific Inference in Qualitative Research. Princeton, NJ: Princeton University Press. Klotz, Audie. 1995. Norms reconstituting interests: global racial equality and U.S. sanctions against South Africa. International Organization, 49, 3: 451-78. Kratochwil, Friedrich and John Gerard Ruggie. 1986. International organization: a state of the art on the art of the state. International Organization, 40, 4: 753-775. 89 Krttger, Hans Christian. 1999. Protocol No. 6 of the European Convention on Human Rigths. In Council of Europe, The Death Penalty: Abolition in Europe. Strasbourg Cedex: Council of Europe Publishing. Legro, Jeffrey W. 1997. Which Norms Matter? Revisiting the "Failure" of Internationalism. International Organization, 51,1: 31-63. McLaughlin Mitchell, Sara. 2002. A KantianSystem? Democracy and Third Party Conflict Resolution. American Journal of Political Science, 46, 4: 749-759. Mearsheimer, John. 1994. The False Promise of International Institutions. International Security, 19, 3: 5-49. Meyer, John W. 2000. Globalization: Sources and Effects on National States. International Sociology, 15, 2: 233-248. Meyer, John W., John Boli, George M . Thomas, and Francisco O. Ramirez. 1997. World Society and the Nation-State. American Journal of Sociology, 103, 1: 144-181. Meyer, John W., Francisco O. Ramirez, and Yasemin Nuhoglu Soysal. 1992. World Expansion of Mass Education, 1870-1980. Sociology of Education, 65,2: 128-149. Moravcsik, Andrew. 1997. Taking Preferences Seriously: A Liberal Theory of International Politics. International Organization, 51,4: 513-553. . 2000. The Origins of Human Rights Regimes: Democratic Delegation in Postwar Europe. International Organization, 54, 2: 217-252. Nadelmann, Ethan A. 1990. Global Prohibition Regimes: The Evolution of Norms in International Society. International Organization, 44, 4: 479-526. Pevehouse, Jon, Timothy Nordstrom, and Kevin Warnke. 2002. Intergovernmental Organizations, 1815-2000: A New Correlates of War Data Set. Available online at Picker, Randal. 1997. Simple Games in a Complex World: A Generative Approach to the Adoption of Norms. University of Chicago Law Review, 64: 1225. Poe, Steven C. and Neal Tate. 1994. Repression of Human Rights to Personal Integrity in the 1980s: A Global Analysis. American Political Science Review, 88, 4: 853-872. Posner, Eric A . 2000. Law and Social Norms. Cambridge, M A : Harvard University Press. (KA70.S6 P67 2000) 90 Price, Richard. 1998. Reversing the Gun Sights: Transnational Civil Society Targets Land Mines. International Organization, 52, 3: 613-644. Prokosch, Eric. 1999. The Death Penalty Versus Human Rights. In Council of Europe, The Death Penalty: Abolition in Europe. Strasbourg Cedex: Council of Europe Publishing. Ramirez, Francisco O. and Yasemin Soysal and Suzanne Shanahan. 1997. The Changing Logic of Political Citizenship: Cross-National Acquisition of Women's Suffrage Rights, 1890 to 1990. American Sociological Review, 62 (October: 735-745). Reus-Smit, Christian. 2004. The Politics of International Law. Cambridge, U.K. : Cambridge University Press. Risse, Thomas and Kathryn Sikkink. 1999. The socialization of international human rights norms into domestic practices: introduction. In eds. Risse, Thomas et al. (1999) The Power of Human Rights: Interantional Norms and Norms and Domestic Change. Schabas, William A . 2002. The Abolition of the Death Penalty in International Law, Third Edition. Cambridge, U.K.: Cambridge University Press. Singer, J. David. 1987. Reconstructing the Correlates of War Dataset on Material Capabilities of States, 1816-1985. International Interactions, 14: 115-32. Singer, J. David, Stuart Bremer, and John Stuckey. 1972. Capability Distribution, Uncertainty, and Major Power War, 1820-1965. In ed. Bruce Russett, Peace, War, and Numbers, Beverly Hills: Sage, 19-48. Strang, David and John W. Meyer. 1993. Institutional conditions for diffusion. Theory and Society, 22: 487-511. Sunstein, Cass R. 1996. Social Norms and Social Roles. Columbia Review of Law, 96: 903-968. Toscano, Roberto. 1999. The United Nations and the Abolition of the Death Penalty. In Council of Europe, The Death Penalty: Abolition in Europe. Strasbourg Cedex: Council of Europe Publishing. United Nations General Assembly. 1948. Universal Declaration of Human Rights. Available online at United Nations High Commissioner for Human Rights. 2003. Status of Ratifications of the Principal International Human Rights Treaties. Available online at Wallace, Michael, and J. David Singer. 1970. International Governmental Organization in the Global System, 1815-1964. International Organization, 24, 2: 239-87. 91 Waltz, Kenneth. 1979. Theory of International Politics. New York, N Y : Random House. Wendt, Alexander. 1999. Social Theory of International Politics. Cambridge, U.K. : Cambridge University Press. Wexler, Lesley. 2003. The International Deployment of Shame, Second-best Responses, and Norm Entrepreneurship: The Campaign to Ban Landmines and the Landmine Ban Treaty. Arizona Journal of International and Comparative Law, 20, 3: 561-606. 92 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items