Abstract
The cultural content of international human rights standards has inspired much discussion and debate, but cross-cultural levels of compliance with these standards remains understudied. Using Samuel Huntington’s controversial ‘clash-of-civilizations’ thesis as a point of departure, this article analyzes whether human rights practices differ systematically across dominant cultural and religious traditions, with a focus on two sets of rights: bodily integrity and civil/political. Net of standard political, social, economic, and demographic factors, countries identified as Western countries and Japan generally have better rated practices than other countries, but differences are especially pronounced with respect to Buddhist, Catholic/Latin, Islamic, and Orthodox nations. These countries are closely scrutinized by watchdogs such as Amnesty International, which might account for their lower human rights scores. Variation in human rights scores also tends to be durable over time; in contrast with Huntington’s thesis, the end of the Cold War mattered little for cross-cultural differences in countries’ practices. Finally, cultural and religious effects are more pronounced for bodily integrity rights than for civil and political rights. All told, the analyses offer little support for the clash-of-civilizations approach, insofar as it predicts increased cultural conflict over putatively ‘Western’ human rights standards.
‘In the battle for democracy and human rights, words matter, but what we do matters much more’.
In June 1993, leaders from 171 nation-states gathered with thousands of academics and activists in Vienna for the World Conference on Human Rights. The conference sought to ‘devise ways and means to remove the current obstacles and meet challenges to the full realization of all human rights’. 1 What it showed, however, was that old obstacles had merely been replaced by new ones. Convened less than 2 years after the dissolution of the Soviet Union, the conference highlighted the rapidly shifting nature of geopolitical dynamics in the nascent post-Cold War order. Gone were the days of ideological squabbles between capitalists and communists over the substance and application of human rights norms. New battle lines, ostensibly cultural in nature, emerged in their stead.
Representatives from a handful of East Asian countries – China, Singapore, and Vietnam in particular – stressed that ‘while human rights are universal in nature’, they must also account for ‘the significance of national and regional particularities and various historical, cultural and religious backgrounds’. 2 Human rights standards, they argued, must be adapted to fit local conditions and traditions. This view found support among leaders in many Islamic nations, but it met with staunch opposition from the United States and other Western countries. Speaking at the Vienna conference, US Secretary of State Warren Christopher declared that ‘we cannot let cultural relativism become the last refuge of repression’.
Proponents of cultural relativism question ‘whether it is justifiable or desirable to apply universal standards’ of human rights to non-Western countries (Donnelly, 1999: 62). Critics typically advance one of two arguments. Some insist that human rights standards are universally valid regardless of their sociocultural provenance (Donnelly, 1999, 2007a; Lupp, 2005). Mary Robinson, former president of Ireland and United Nations (UN) Human Rights Commissioner, advocates this perspective. She contends that international human rights norms distill ‘many of the values inherent in the world’s major legal systems and religious beliefs including the Buddhist, Christian, Hindu, Islamic, and Jewish traditions’ (quoted in Franck, 1999: 259). Others point out that non-Western societies have their own ‘home grown’ human rights traditions that are fundamentally commensurable with international human rights standards (e.g. Midlarsky, 1998; Sen, 1997; Tatsuo, 1999; Voll and Esposito, 1994). Those who deny the universal applicability of human rights standards contend that they are inescapably Eurocentric (e.g. Berger, 1977; Mutua, 2002; Nisbett, 2003; Zakaria and Lee, 1994). Some advance this argument even further, alleging that Western powers impose human rights standards on non-Western countries (Huntington, 1996a: 92).
Such questions have been discussed and debated ad nauseam (see, for example, Bauer and Bell, 1999; Donnelly, 1982, 2003; Kausikan, 1993; Mutua, 2002; Neier, 1993; Panikkar, 1982; Sen, 1997). The literature is rife with examples and counterexamples designed to support or refute the notion that non-Western cultural traditions are at odds with international human rights standards. A similar state of affairs characterizes the literature on democracy. Philosophical arguments emphasizing Islam’s ‘democratic essence’ (Voll and Esposito, 1994) coexist with empirical research showing that predominantly Muslim countries are ‘democratic underachievers’ (Fish, 2002) – even though a majority of Muslims, when surveyed, endorse democratic principles (Norris and Inglehart, 2002). Unfortunately, commentators rarely bring systematic empirical data to bear on the debate over cultural relativism and human rights. This article reports one of the first statistical analyses exploring whether human rights outcomes differ systematically across major geocultural and religious divides.
Cultural differences seem to matter little for countries’ purely rhetorical endorsements of human rights standards (Cole, 2005). 3 Still unclear is whether cultural and religious differences explain what countries do as opposed to the words they say (or the treaties they ratify). To what extent does the ‘us’ of generic nation-state identities fail to align with the ‘we’ of particular national or civilizational identities? The ‘we/us’ distinction I have in mind is analogous to the distinction between the ‘I’ and the ‘me’ in Meadian social psychology, but writ large to nation-states. The ‘us’ of nation-states, like the ‘me’ of individuals, is constituted by the norms, expectations, and standards of the broader community – in this case, the world community. The ‘us’ ratifies human rights treaties ceremonially, uncritically, and perfunctorily, for the expressive benefits of doing so (Cole, 2005; Hafner-Burton et al., 2008; Hathaway, 2002, 2007). In contrast, the ‘we’ of nations, like the ‘I’ of individuals, actively engages with and responds to these external structures and frameworks. Nations do not passively accept or conform to world-cultural models; rather, they select, adapt, modify, and at times even resist those models, based at least in part on their unique national or civilizational cultures.
Ultimately, the degree of consonance or dissonance between the ‘constitutive’ culture of the postwar human rights standards and the ‘expressive’ cultures of particular regional or religious groups is an empirical question and should be evaluated as such (on these distinct conceptions of ‘culture’, see Boli and Lechner, 2001; Jepperson and Swidler, 1994). But instead of focusing on treaty commitments, which are easily decoupled from actual practices (Hafner-Burton and Tsutsui, 2005, 2007), I analyze whether countries’ human rights practices are shaped by their dominant cultural and religious traditions, as predicted by Samuel Huntington’s (1993, 1996a, 1996b) controversial clash-of-civilizations perspective. I conclude that ‘civilizational’ and religious differences are not robustly associated with countries’ human rights practices, once confounding factors are taken into account. Although human rights protections initially appear stronger in Western countries than in non-Western countries, a more nuanced analysis linking countries’ religious composition to their human rights practices undermines Huntington’s framework.
Motivating the analysis
The point of departure for my analysis is Richards’s (2003) study examining the ‘civilizational geography of respect for human rights’, which explored whether levels of government respect for various categories of human rights varied across Huntington’s (1993, 1996a) civilizational groupings. Based on descriptive evidence, Western countries and Japan had the highest overall levels of respect for human rights, whereas Islamic nations and ‘Sinic’ countries such as China, the Koreas, the Philippines, and Vietnam evinced the lowest levels of respect. Richards also showed that respect for different kinds of rights varied across civilizations. Latin American countries, for example, had mediocre physical integrity rights scores, but their civil and political rights scores were among the highest in the world.
My study offers a rigorous, statistically controlled test of the same fundamental questions addressed in Richards’ (2003) descriptive study. Using regression analysis, I isolate the effect of civilizational membership and religious composition on bodily integrity and civil/political rights outcomes by controlling for the potentially confounding effects of political, social, economic, and demographic characteristics. Multivariate designs are important for adjudicating among potentially competing explanations of human rights outcomes. It is often alleged, for example, that autocrats such as Singapore’s Lee Kuan Yew and Malaysia’s Mahathir Mohamad advance cultural relativist arguments merely to justify their authoritarian rule (e.g. Sen, 1997; Tatsuo, 1999). But do civilizational and religious differences continue to shape human rights outcomes once relevant political factors such as authoritarianism are taken into account?
In another example, leaders in developing countries sometimes argue that national economic development takes precedence over, and may even require the abrogation of, individuals’ civil and political rights. Lee Kuan Yew, whose heavy hand presided over Singapore’s rapid economic ascent, suggested that ‘political and civil rights hamper economic growth’ (Sen, 1999: 91). An alternative view, advanced by Inglehart and his colleagues (Inglehart, 1997; Inglehart and Baker, 2000; Inglehart and Norris, 2003; Norris and Inglehart, 2002), predicts that as countries modernize economically, they also liberalize politically and culturally. Non-Western countries should therefore begin to show greater levels of respect for human rights as they develop. In a similar vein, Donnelly (2007b) argues that human rights ideas and practices arose not from any deep Western cultural roots but from the social, economic, and political transformations of modernity. They thus have relevance wherever those transformations have occurred, irrespective of the pre-existing culture of the place. (p. 287)
Purely descriptive analyses are not equipped to disentangle cultural explanations from political or economic ones. My analysis is designed to show whether the baseline differences reported by Richards (2003) persist in the presence of other factors.
Dimensions of comparison: Culture, time, and rights
My analysis addresses three questions. First, does respect for human rights vary across civilizational or religious divides, net of other consequential factors? Second, do these civilizational and religious differences, where they obtain, change over time? Third, are some rights subject to greater cultural and religious variability than others? I address each of these questions in turn.
Cultural dimension
Samuel Huntington’s (1993, 1996a) clash-of-civilizations thesis predicted that the end of the Cold War would produce a dramatic geopolitical realignment, with culturally and religiously motivated conflicts among countries belonging to different ‘civilizations’ replacing ideological conflicts between the two superpowers and their allies. Civilizations, in Huntington’s usage, constitute ‘the highest cultural grouping of people and the broadest level of cultural identity people have short of that which distinguishes humans from other species’ (p. 43). The most salient civilizational distinction, he contends, is often between Western and non-Western societies. Accordingly, ‘the central axis of world politics in the future is likely to be … the conflict between ‘the West and Rest’ and the responses of non-Western civilizations to Western power and values’ (Huntington, 1993: 41).
Although the clash-of-civilizations perspective is primarily a theory of physical conflict, it also speaks to normative conflicts over ideas and values. According to Huntington, Western civilization is firmly anchored in a matrix of intellectual traditions – the separation of church and state, the rule of law, social pluralism, and individualism – that distinguish it from non-Western countries (Huntington, 1996b). These traditions, in turn, allegedly provided the cultural foundations for the development of human rights. Donnelly (1982) went so far as to argue that ‘most non-Western cultural and political traditions lack not only the practice of human rights but the very concept’ (p. 303). This perspective, which replaces the ideological polarity of the Cold War with a new form of cultural bifurcation, motivates the following hypothesis:
Hypothesis 1a: Human rights practices will be better in Western than in non-Western countries.
Another line of thought, derived from Huntington’s more nuanced view that civilizational conflict is multipolar rather than bipolar, suggests that some non-Western countries will show greater levels of respect for human rights than others. Huntington carved the globe into eight non-Western civilizations, together with a residual ‘Other’ category: African, Buddhist, Hindu, Islamic, Japanese, Latin American, Orthodox, and Sinic. These civilizations, Huntington (1996a) argues, ‘share little in terms of religion, social structures, institutions, and prevailing values’, although some have greater cultural or religious affinities with the West than others (p. 33). He singles out the predominantly Catholic Latin American civilization, in particular, as a Western cognate (Huntington, 1996a: 46). Elsewhere he suggests that the monotheistic civilizations descending from Mesopotamia – Islamic, Orthodox, and Western – differ fundamentally from civilizations with Indic and Sinic progenitors (Huntington, 1996a: 48–50). I therefore offer an alternative hypothesis based on the premise that non-Western civilizations with closer religious or cultural affinities to the West will have human rights records resembling Western countries:
Hypothesis 1b: Human rights practices will be better in non-Western civilizations with closer historical, cultural, or religious ties to the West (e.g. Latin American and Orthodox) than in other civilizations (e.g. Sinic, Buddhist, and Hindu).
Huntington’s civilizational scheme has been contested by scholars, and Huntington (1996a: 47–48) himself equivocated as to whether sub-Saharan Africa and Buddhist countries, in particular, constitute distinct civilizations. Critics argue that Huntington’s procrustean framework ignores heterogeneity – and even conflict – within civilizations (Chiozza, 2002; Fox, 2005; Henderson, 2005; Tusicisny, 2004). Even though Huntington (1996a) notes the existence of intra-civilizational divisions – Islam, for example, comprises Arab, Turkic, Persian, and Malay ‘subcivilizations’ – he seems to downplay their import (p. 45). Likewise, although Western civilization by Huntington’s (1996a: 46) own admission ‘combine[s] Catholic and Protestant cultures’, he treats the West as monolithic.
To address these criticisms, I also analyze the effect of countries’ religious composition on human rights outcomes. With few exceptions, religion constitutes the basis of civilizational identity (Huntington, 1996a: 47); indeed, Huntington’s approach has been criticized for conflating civilizational identity with religion (e.g. Fox, 2001). The Buddhist, Hindu, Islamic, and Orthodox civilizations are defined by their eponymous religious traditions, while ‘Western Christianity, first Catholicism and then Protestantism, is the single most important historical characteristic of Western civilization’ (Huntington, 1996b: 30). Religion also distinguishes Latin America, which is predominantly Catholic, from Western civilization, which is both Catholic and Protestant. Huntington (1993) originally called the Sinic civilization ‘Confucian’, and argues that Confucianism continues to be a ‘major component’ (Huntington, 1996a: 45) of the civilization comprising China, Vietnam, Singapore, and the Koreas. Only the (sub-Saharan) African civilization has no apparent religious basis, leading Huntington (1996a) to question whether Africa constitutes a ‘true’ civilization (p. 47).
Analyzing the relationship between religious composition and human rights outcomes addresses many of the criticisms of Huntington’s framework. First and foremost, Huntington’s civilizations are mutually exclusive and exhaustive: with the exception of a few ‘cleft’ and ‘torn’ countries, 4 entire nation-states are classified as belonging to one and only one civilization. His blunt classification of Lebanon as Islamic, for example, ignores that country’s confessional diversity, where roughly equal percentages of Lebanese profess faith in Sunni Islam, Shi’a Islam, and Christianity. Religious composition, in contrast, is operationally more precise. Countries can be more or less Islamic, Buddhist, Hindu, and so on; they can also, for example, be both Islamic and Christian – not to mention Sunni, Shi’a, Protestant, Catholic, and Orthodox – in different proportions.
Another related shortcoming in Huntington’s approach is his assumption that intra-civilizational differences are less pronounced than inter-civilizational differences. A focus on religion allows for greater diversity within so-called civilizations. Predominantly Sunni and Shi’a countries, for example, need no longer be lumped uncritically into a single ‘Islamic’ category, just as ‘Western’ countries can be delineated into predominantly Protestant, Catholic, and even secular societies. 5 Using both Huntington’s civilizational categories and finer grained measures of religious composition within countries, I implicitly adopt an ‘ANOVA’-like logic to determine whether there is more variation in human rights practices between or within civilizations.
If Huntington’s approach overlooks religious variation within civilizations, it also ignores important commonalities among peoples and nations across civilizations. According to Inglehart and Baker (2000), predominantly Catholic countries in Europe and Latin America form a ‘super-zone’ with respect to their citizens’ fundamental beliefs, convictions, and values. Countries as diverse as Panama, Paraguay, Poland, and Portugal are unified culturally in important ways. But Huntington drives a wedge between them, claiming that the legacy of the Reformation in Europe and the mélange of Iberian and indigenous cultures in the ‘New World’ distinguishes European and Latin American Catholics. My approach considers whether such a distinction is warranted, at least with respect to countries’ human rights practices.
Temporal dimension
Time is a critical component of Huntington’s approach, which predicts that inter-civilizational conflicts will intensify only after the Cold War, when ideological tensions yield to cultural ones. Most empirical tests of this perspective, however, use data with very few post-Cold War observations (Huntington, 2000). Henderson and Singer (2000) found that political factors were more important than cultural ones in predicting the onset of civil wars, but their dataset ended in 1992. As such, too little time had elapsed since the fall of communism as a geopolitical force to evaluate the main clash-of-civilizations hypothesis.
Similar limitations beset other studies. Russett et al. (2000) and Henderson and Tucker (2001) concluded that interstate wars were more likely to break out within rather than between civilizations, but their observation periods also ended in 1992. Using only slightly more updated data, Chiozza (2002), Fox (2001, 2002), Henderson (2005), and Roeder (2003) reached similar conclusions. The most current study to date, by Fox (2005), includes data through 2001. He found that conflicts are indeed deadlier when fought across civilizational boundaries, but that this was true during as well as after the Cold War. All told, empirical evaluations of the clash-of-civilizations perspective are, at best, inconclusive due to data limitations.
Although my analysis focuses on human rights practices within countries rather than armed conflicts between them, the clash-of-civilizations approach implies that civilizational dynamics will affect a country’s commitment to human rights principles more strongly after the Cold War than during it. The Vienna conference lends prima facie credence to this supposition. In fact, Huntington (1996a) cites this conference as an example of a unified Asian–Islamic alliance in opposition to what he calls the West’s ‘human rights imperialism’ (pp. 192–198). My analysis subjects Huntington’s offhand musings to rigorous empirical scrutiny, based on the following hypothesis:
Hypothesis 2a: Cultural variation in human rights practices will be greater after the Cold War (post-1991) than during the Cold War.
To evaluate this hypothesis, I use data that are more balanced temporally than previous tests of the clash-of-civilizations perspective have been. The dataset I assemble includes a roughly equal number of observations during and after the Cold War. The analyses begin in 1980, when data on the human rights outcomes of interest became available, and end in 2005. I consider the fall of the Soviet Union in 1991 to signal the definitive end of the Cold War.
Divergence in countries’ human rights practices is just one possible outcome. A contradictory prediction, implied by the world society perspective (Meyer et al., 1997), is that human rights practices will converge over time. World society theorists often predict superficial or ‘decoupled’ convergence, such that formal policies become increasingly isomorphic while actual practices continue to differ. Were this the case, cross-cultural rates of human rights treaty ratification would be similar – as indeed they are (Cole, 2005) – despite persistent differences in empirical levels of respect for human rights. Yet, recent work finds long-term and widespread ‘drift’ toward convergence in outcomes as varied as educational enrollments (Ramirez and Boli, 1987; Schofer and Meyer, 2005), environmental conditions (Schofer and Hironaka, 2005), police reporting of rape (Frank et al., 2009), and human rights practices (Cole, 2012). In these cases, a process of diffuse ‘acculturation’ (Goodman and Jinks, 2004) gradually prods countries to accept, adopt, and implement world-cultural norms. This view informs an alternative hypothesis:
Hypothesis 2b: Cultural variation in human rights practices will diminish after the Cold War.
Yet another possibility is that cross-cultural differences in human rights practices remain durable over time. Inglehart and Baker (2000), who found that the cultural and religious heritage of a country leaves a lasting imprint on its citizens’ values, concluded that ‘as the twenty-first century opens, cleavages along religious lines remain strong’ (p. 46). According to this view, deep cultural and especially religious differences in values do not converge or diverge over time, but rather move in path-dependent fashion on ‘parallel trajectories shaped by their cultural heritages’ (p. 49). I therefore propose an alternative to both the divergence and convergence hypotheses:
Hypothesis 2c: Cultural variation in human rights practices will remain stable over time.
Rights dimension
Despite the claim, articulated during the Vienna conference in 1993, that ‘all human rights are universal, indivisible and interdependent and interrelated’, respect for some categories of rights may show greater cross-civilizational variation than others. Tensions surrounding the conference point to the nature and scope of cultural opposition to different human rights norms. In preparatory meetings, ‘many Asian and African governments [sought] to redefine human rights away from political and civil rights’ (Riding, 1993). Asian leaders, in particular, argued that individualistic rights and freedoms fail to comport with ‘Asian’ values privileging order, community, and deference to authority. They further maintained that civil and political rights can impede collective economic development, which takes precedence over individual freedoms (cf. Sen, 1999).
Islamic conceptions of human rights also elide if not outright reject basic civil and political rights provisions (Mayer, 1994). The Cairo Declaration on Human Rights in Islam (1990), drafted in advance of the Vienna conference, is noticeably silent with respect to the freedoms of speech, assembly, press, and religion or the right to vote. 6 The freedom of religion, in particular, is especially problematic for the many Islamic countries that forbid apostasy. But cultural differences over civil and political rights are not limited to religious freedoms. According to Peter Berger (1977), ‘freedom of speech and of the press, the inviolability of the electoral process, equal protection of the law, due process – all these are rights that find little warrant outside the orbit of Western history’ (p. 62). Respect for basic civil and political rights, which are often thought to derive from the Western liberal tradition (Mutua, 2002), is therefore likely to show a good deal of cross-cultural variation.
Compared with civil and political rights, I expect physical integrity rights to be less affected by civilizational or religious differences. Unlike most human rights protections, freedom from torture and other forms of bodily harm is framed in distinctly fundamental, unambiguous, and absolute terms (Lutz and Sikkink, 2000). Consequently, whereas states are legally authorized to suspend civil and political rights during periods of national emergency or crisis, the torture prohibition is legally non-derogable (Hafner-Burton et al., 2011; Lutz and Sikkink, 2000).
Freedom from government-inflicted physical pain and suffering is one of the most basic and universally recognized rights claims. Ishay (2008) notes that ‘with the injunction “thou shall not kill”, Judaism had already instructed people to honor individuals’ right to life, an entitlement not only respected by other monotheistic religions, such as Christianity and Islam, but also by Buddhism, Hinduism, and Confucianism’ (p. 84). Even in Singapore, whose leaders have been at the forefront of the debate over cultural relativism and human rights, high-ranking officials concede that Asians and Westerners can agree on minimal standards of civilized behavior that both would like to live under. For example, there should be no torture, no slavery, no arbitrary killings, no disappearances in the middle of the night, no shooting down of innocent demonstrators, no imprisonment without careful review. (quoted in Cerna, 1994: 745)
Likewise, although the Cairo Declaration ignores basic civil and political rights, it explicitly recognizes the ‘God-given’ rights to life and safety from bodily harm. In short, very few countries argue that torture and other physical integrity violations are justifiable on cultural or religious grounds, but they readily criticize civil and political rights as suffering from a Western bias. Thus, I offer the following prediction:
Hypothesis 3: Cross-cultural differences in human rights practices will be most pronounced with respect to civil and political rights and least pronounced with respect to bodily integrity rights.
Data and method
Dependent variables, estimators, and the issue of cultural bias
Using data for up to 137 countries between 1980 and 2005, I analyze civilizational and religious variation in respect for bodily integrity and civil/political rights. I include two measures of bodily integrity rights: the physical integrity rights index from the Cingranelli-Richards (CIRI) Human Rights Dataset (Cingranelli and Richards, 2008) and the political terror scale or PTS (Gibney et al., 2007). Both variables focus on governmental or government-sanctioned abuses and are coded from the same source materials: the US Department of State’s Country Reports on Human Rights Practices and Amnesty International’s Annual Report. The physical integrity rights index, ranging from 0 to 8, comprises four subcomponents – torture, extrajudicial killing, political imprisonment, and disappearances – with higher scores representing better practices (i.e. less abuse). The PTS is a 5-point measure gauging the overall scope, intensity, and range of a similar set of abuses. In the analyses that follow, I inverted the PTS so that higher scores correspond to improved practices. Despite important differences in these measures (cf. Cingranelli and Richards, 2010; Wood and Gibney, 2010), they are nevertheless highly correlated in my dataset (r = .83).
I also analyze two measures of fundamental civil and political rights. The first, the CIRI’s empowerment rights index (Cingranelli and Richards, 2008), assesses basic civil and political rights protections in a country: the freedoms of movement, speech, assembly, and religion; workers’ rights; and electoral self-determination (i.e. the right to vote and to run for public office). This index ranges from 0 to 10 and is coded from the same sources as the two bodily integrity measures. The second measure, Freedom House’s (2008) civil liberties score, ranges from 1 to 7 and rates the extent to which countries protect the freedoms of expression, assembly, association, and religion; respect the rule of law; permit free economic activity; and strive for equality of opportunity. Freedom House derives this score from a variety of sources, including annual editions of the Political Handbook of the World and the Worldmark Encyclopedia of Nations, media outlets and regional journals, and the human rights reports issued by the US State Department and Amnesty International. Unlike the CIRI indices and the PTS, Freedom House’s civil liberties score ‘does not rate governments or government performance per se, but rather the real-world rights and freedoms enjoyed by individuals’. 7 Still, the empowerment rights and civil liberties measures are highly correlated (r = .82). As with the PTS, the civil liberties scale is recoded such that higher values indicate lower levels of abuse.
Due to the ordinal nature of these human rights measures, I estimate ordered logistic regression models with standard error estimates adjusted for clustering within countries over time and with time-varying independent and control variables lagged 1 year relative to dependent variables.
Given the nature of my analysis, one must be wary of cultural biases that may affect or infect these measures, especially as they are coded from Western source materials that may focus disproportionately on non-Western countries’ abuses (Mutua, 2002). Researchers have probed these sources and the measures derived from them for political, ideological, and informational biases. State Department human rights reports and Freedom House assessments, in particular, historically favored US allies, but they were critical of communist regimes and even welfare states; however, these biases seem to have diminished over time as data collection and coding procedures became increasingly standardized (Bollen, 1986, 1993; Bollen and Paxton, 1998; Clark and Sikkink, 2013; Innes de Neufville, 1986; Poe et al., 2001; Steiner, 2012). Likewise, whereas early Amnesty International reports focused on the world’s worst abusers, later reports are far more comprehensive in both scale and scope, and a detailed analysis by Hill et al. (2013) demonstrates that the organization ‘rarely succumb[s] to incentives to exaggerate abuse’ (p. 219). For their part, Clark and Sikkink (2013) found that more detailed country reports correlate with lower human rights scores, suggesting that all else being equal, countries receiving greater scrutiny from human rights ‘watchdogs’ tend to have worse human rights ratings – if not necessarily worse human rights practices – than other countries.
Although it is impossible to eliminate these sorts of biases, I attempt to mitigate them with the inclusion of several control variables. To account for the possibility that human rights reports and measures are biased in favor of US allies, I include separate measures of American military and economic aid (logged) received by a country each year, obtained from the US Overseas Loans and Grants Database (U.S. Agency for International Development, 2012). To adjust for possible anticommunist and even anti-leftist biases – Bollen (1986: 585), for example, detected a ‘conservative bias against welfare states’ in Freedom House ratings – I include dummy variables for communist regimes (Easterly, 2001) and countries in which a (noncommunist) leftist party (e.g. Socialist, Social Democrat, and Labor) controls the executive branch (Beck et al., 2001). Finally, to guard against ‘information paradoxes’ (Keck and Sikkink, 1998) that occur when increased scrutiny of countries’ practices gives the impression that levels of abuse are worse than they actually are, I compute a name-and-shame index (Hafner-Burton, 2008) summing the standardized scores of four variables: media reporting of human rights abuses in (1) The Economist and (2) Newsweek (Ramos et al., 2007), (3) Amnesty International press releases targeting a country’s human rights abuses (Ron et al., 2005), and (4) UN Commission on Human Rights resolutions condemning a country’s human rights performance (Lebovic and Voeten, 2006). This index also helps to adjust for any cultural biases in human rights coverage on the part of media outlets or human rights organizations, whose assessments inform the human rights indices analyzed in this study. 8
It is of course possible – and even likely – that causality between these control variables and countries’ human rights practices operate in both directions. While the US State Department may paint an unduly optimistic picture of its allies’ human rights conditions, the government may also use financial aid to reward countries with favorable practices. Similarly, while anticommunist biases may have tainted human rights assessments, it may well be that conditions are objectively worse in communist regimes than elsewhere. And while information paradoxes can inflate perceptions of abuse, media outlets may simply pay more attention to countries with the worst abuses. As control variables, however, reverse or reciprocal causality of this sort is not especially problematic.
Independent variables
The analyses include two sets of independent variables: indicators of civilizational membership and measures of religious composition. Nine civilizational indicators – African, Buddhist, Hindu, Islamic, Japanese, Latin American, Orthodox, Sinic, and Western – are based on the map reported in Huntington (1996a: 26–27) or as described in his text, with a residual ‘Other’ category for countries that are difficult to classify. Examples include Ethiopia, which, according to Huntington (1996a: 47), ‘constituted a civilization of its own’ historically; Haiti, whose language, ethnic composition, and religious traditions distinguish it from other Latin American countries (Huntington, 1996a: 136–137); Israel, which Huntington (1996a: 90) classified as a non-Western country; and several microstates that lack a majority religious tradition. (Appendix A groups countries into their respective civilizations.) In the analyses that follow, Western civilization is the omitted reference category, such that the effect of each non-Western civilization is interpretable relative to Western countries. Although these civilizational groupings are multinational (with the exception of Japan, which Huntington argues is a civilization unto itself), the unit of analysis for this study is the nation-state. Huntington (1996a) argues that ‘nation states remain the principal actors in world affairs’, but claims that states’ behaviors are increasingly shaped by ‘cultural preferences, commonalities, and differences’ (p. 21).
A second set of variables from the World Religion Dataset (Maoz and Henderson, 2013) identifies the percentage of people in each country belonging to each of the following religions or religious families: Christianity (Protestant, Catholic, and Orthodox), Islam (Sunni and Shi’a), Judaism, Buddhism, Hinduism, Confucianism, and Shintoism. Another variable gives the percentage of people with no religion. 9 These measures are available at 5-year intervals from 1945 to 2010; due to missing data on the dependent and control variables, the time period covered by my analysis is 1980 to 2005.
Table 1 interrogates the relationship between civilizational identity and religious composition. Although there is often a close correspondence between the two, they are not synonymous. The strongest correlation (r = .98) is between Japanese civilization (i.e. Japan) and Shintoism. Correlations between the Orthodox, Buddhist, and Hindu civilizations and their eponymous religious traditions are also very strong, registering at .80 and above. The correlation between Islamic civilization and percent Muslim (not reported) is .94, but disaggregating this measure into percent Sunni and Shi’a produces lower correlations (.65 and .35, respectively). Although 87 percent of Latin Americans, on average, are Catholic, the Latin indicator correlates with percent Catholic at only .64, in large part because Western countries also include a substantial share of Catholics (47%, on average). Western countries are largely Protestant and Catholic, with a nontrivial percentage of people reporting no religious faith (roughly 11%); Sinic countries are more atheist than Confucian, due to the influence of communist ideologies there; and African countries do not correlate strongly with any religious tradition, which accounts for Huntington’s reluctance to deem it a civilization of its own. Finally, percent Jewish is associated most strongly with the residual ‘Other’ category, which includes Israel.
Correlation matrix for civilizational indicators and religious composition measures.
Control variables
In addition to the set of control variables designed to partial out potential biases in the human rights measures, I also include several measures shown in previous research to affect human rights outcomes (Atkinson, 2010; Englehart, 2009; Greenhill, 2010; Hafner-Burton and Tsutsui, 2005, 2007; Henderson, 1991, 1993; Keith, 1999; Melander, 2005; Mitchell and McCormick, 1988; Poe et al., 1999; Poe and Tate, 1994; Whitten-Woodring, 2009). A standard measure of democracy from the Polity IV database (Marshall and Jaggers, 2007) ranges between −10 (most autocratic) and 10 (most democratic). From this variable, I computed age of democracy, defined as the consecutive number of years since 1800 (or independence) a country scored 6 or higher on the democracy scale. Another variable from the same database, regime durability, gives the number of years since the last substantive change in a country’s authority pattern, defined as a 3-point or greater deviation in its democracy score.
Dummy variables for former British colonies (Norris, 2009) and military dictatorships (Hadenius and Teorell, 2007) tap additional historical legacies and regime characteristics. Former British colonies generally have better human rights practices than other countries. Military regimes, on the other hand, often deploy physical repression as an instrument of rule; at the same time, they are less likely than other kinds of authoritarian regimes to restrict civil liberties (Davenport, 2007). Thus, whereas military regimes are less likely to curtail the rights and freedoms that permit individuals to express dissent, they are more likely to repress individuals who exercise these rights.
Human rights practices are also conditioned by economic factors. Affluent countries generally show greater levels of respect for bodily integrity and civil/political rights, so I control for gross domestic product (GDP) per capita, measured in constant 2000 US dollars and rescaled by 100. Two additional economic measures, trade as a proportion of gross domestic product and foreign direct investment inflows, measure trade openness and global economic integration, both of which are associated with better rights conditions (Blanton and Blanton, 2009; Harrelson-Stephens and Callaway, 2003; Richards et al., 2001). A measure of central government expenditures (as a percentage of GDP) serves to proxy a state’s capacity. These measures come from the World Development Indicators database (World Bank, 2008).
Three measures of fractionalization describe the probability that two randomly selected individuals from a country’s population belong to different ethnic, religious, and linguistic groups (Alesina et al., 2003). These variables control for within-country levels of diversity, to isolate further the effects of between-country civilizational and religious differences. They also account for ‘cleft’ and ‘torn’ countries in Huntington’s scheme – countries with multiple civilizational identities – and consider whether ‘the job of providing [human] rights is more difficult in an ethnically diverse society than in a more homogeneous one’ (Walker and Poe, 2002: 263). Although these measures are time invariant and do not reflect demographic shifts over time, Alesina et al. (2003) surmise that the ethnic, religious, and linguistic composition of most countries is relatively stable in the short term, particularly ‘at the 30-year horizon of the typical cross-country regression’ (p. 160).
The extent of a country’s embeddedness in global civil society, operationalized as the number of international nongovernmental organizations (INGOs) to which its citizens belong, might also affect human rights outcomes. Hafner-Burton and Tsutsui (2005), for example, found that a respect for basic physical integrity rights increases as the number of INGO ties in a country increases. I include the number of INGOs that count at least one citizen or domestic association as a member during a given year, as reported by the Union of International Associations (UIA) (various years). This measure is rescaled by 100.
Two additional factors, population pressures and warfare, are consistently associated with increased levels of human rights abuse. I control for population density (World Bank, 2008) and two 11-point scales from the Major Episodes of Political Violence dataset (Marshall, 2006) that assess the intensity of international and civil warfare involving a country each year. 10
Findings
I report three sets of results: the effects of (1) civilizational membership and (2) religious composition on human rights outcomes, as well as (3) their historically contingent effects during and after the Cold War. I also present findings that gauge whether uneven levels of scrutiny accorded to countries’ human rights conditions at least partially shape observed cross-cultural variation in human rights scores.
Effects of civilizational membership
Table 2 presents ordered logit coefficients for the effect of civilizational membership on each of the bodily integrity and civil/political rights measures. Two models are reported for each outcome: the first includes only the civilizational indicators; the other adds control variables to determine whether civilizational identity predicts human rights outcomes over and above various political, social, economic, and demographic characteristics. In the stripped-down models, the coefficient on each non-Western civilizational indicator except Japan is significantly negative, indicating lower human rights scores relative to Western countries. 11 For the most part, these effects remain robust in the fully specified models, although the size of each coefficient drops appreciably. Only five estimates become insignificant: physical integrity rights scores in African and Sinic countries, political terror scores in Sinic countries, civil liberties scores in Hindu countries, and empowerment scores in ‘Other’ countries without a clear civilizational designation. In general, though, civilizational differences are strong even in the presence of numerous control variables.
Ordered logit models for the effect of civilizational indicators on human rights measures, 1981 to 2005.
INGOs: international nongovernmental organizations; GDP: gross domestic product; FDI: foreign direct investment; BIC: Bayesian information criterion.
Robust standard errors, clustered on countries, in parentheses.
Time-varying control variables lagged by 1 year.
p < .05; **p < .01; ***p < .001 (two-tailed).
Based on the fully specified models, four civilizational groupings have significantly lower human rights scores than Western countries across all measures: Buddhist, Islamic, Latin, and Orthodox. Two civilizations – African and Sinic – have lower scores on both civil and political rights measures; African countries also score lower on the PTS but not the physical integrity rights index. Hindu and Other countries score lower on both bodily integrity rights indices but only one of the civil and political rights measures (empowerment rights and civil liberties, respectively).
To facilitate the interpretation of these results, Figures 1 to 4 plot cumulative probability distributions for physical integrity, political terror, empowerment, and civil liberties scores, respectively, by civilization (see Long, 1997: 133). For each outcome, the top panel (a) displays statistically unadjusted distributions from the reduced models in Table 2, and the bottom panel (b) uses estimates from the fully specified models to plot regression-adjusted distributions, with control variables held constant at their mean values. Curves shifted farther to the right indicate higher predicted human rights scores.
Cumulative distribution of predicted physical integrity rights scores by civilizational grouping.
Cumulative distribution of predicted political terror scale scores by civilizational grouping.
Cumulative distribution of predicted empowerment rights scores by civilizational grouping.
Cumulative distribution of predicted civil liberties scores by civilizational grouping.
Consider the distribution of physical integrity rights scores. As shown in Figure 1(a), the estimated probability that Western countries had physical integrity scores of 6 or lower (on a scale of 0 to 8) was 20.5 percent, whereas the corresponding probability for Hindu countries was 97.5 percent. Probabilities for each discrete outcome are derived by subtracting the cumulative percentage for a given outcome from the one before it. Thus, the probability that Western countries scored a 6 was 10.8 percent (20.5–9.7%), compared with 3.3 percent for Hindu countries (97.5–94.2%). The same interpretations apply, mutatis mutandis, to the regression-adjusted probability distributions in Figure 1(b).
Surveying the trends, one first notices that cross-civilizational differences in physical integrity scores are much greater when control variables are omitted than when they are taken into account. In other words, much of the rather extreme variation in physical integrity rights scores traces to civilizational differences in other factors. After adjusting the estimates for these factors, outliers on both ends of the distribution shift inward, although predicted physical integrity rights scores remain highest in the Western and Japanese civilizations and lowest in Hindu countries. Trends for the PTS, plotted in Figure 2, are broadly similar, such that the large gap between the West and Japan, on the one hand, and the ‘rest’ of the world, on the other hand, become much less pronounced in the presence of control variables. A closer inspection reveals more detailed differences. In the regression-adjusted estimates, Sinic countries’ political terror scores are sandwiched between those for Western countries and Japan; in the unadjusted estimates, the gap between Sinic countries and the West/Japan was much larger. At the other end of the distribution, Hindu countries have the lowest unadjusted political terror scores (Figure 2(a)), but Latin countries perform the worst in the regression-adjusted analysis (Figure 2(b)).
Figures 3 and 4 present corollary plots for the civil and political rights outcomes. Once again, large cross-civilizational differences in the unadjusted trends shrink in the presence of control variables, and the relative performance of different civilizational groupings sometimes changes. Take, for example, the distribution of empowerment rights scores in Figure 3. Unadjusted empowerment rights scores are lowest in Islamic countries, but Orthodox countries turn out to have the worst rated practices once confounding factors are controlled. Compared with the unadjusted trends, the relative standing of Islamic and Buddhist countries improves in the regression-adjusted analyses, whereas Orthodox, Sinic, and Hindu countries’ ratings decline. Aside from Japan, which earned a score of 10 throughout the period (and hence has a 100 percent probability of achieving that score), regression-adjusted empowerment rights scores are highest in the West, followed by Latin America.
Probability distributions for the empowerment rights index’s companion measure, the civil liberties scale, are similar but even starker. Inter-civilizational differences in unadjusted scores are quite large, but the regression-adjusted trends cluster together much more tightly. Although Western countries and Japan continue to outperform other non-Western countries, civilizational differences by themselves explain less of the variation in civil liberties scores compared with other human rights measures, over and above the effects of control variables.
For the most part, the effects of control variables accord with previous research. Scores on all four human rights measures increase as the level and age of democracy increases, but they decline as scores on the name-and-shame index increase. Bodily integrity scores increase as a function of economic development and trade openness but decrease as a function of civil and international wars. Only the effect of INGO linkages, which is significantly negative for both bodily integrity rights measures, deviates from established findings (e.g. Hafner-Burton and Tsutsui, 2005). Rather than improve bodily integrity rights outcomes, it seems that INGOs play an important role in monitoring and reporting these violations (Cole, 2010; Keck and Sikkink, 1998; Murdie and Davis, 2012). 12 Only a handful of additional estimates achieved statistical significance: empowerment rights scores were higher in countries with leftist executives but decreased with international war intensity, civil liberties scores increased with GDP per capita but decreased with civil war intensity, and political terror scores decreased with population density.
Effects of religious composition
Table 3 repeats the foregoing analyses after replacing the set of civilizational indicators with religious composition measures. Because these variables are measured at 5-year intervals, the sample sizes drop by about 80 percent, although only a handful of countries are lost.
Ordered logit models for the effect of religious composition on human rights measures, 1980 to 2005 (5-year intervals).
INGOs: international nongovernmental organizations; GDP: gross domestic product; FDI: foreign direct investment; BIC: Bayesian information criterion.
Robust standard errors, clustered on countries, in parentheses.
Time-varying variables lagged by 1 year.
p < .05; **p < .01; ***p < .001 (two-tailed).
Of the two sets of human rights measures, the effect of religious composition is strongest for the bodily integrity rights measures, both of which decrease as the percentage of Catholics, Jews, Muslims (both Sunni and Shi’a), Orthodox Christians, and Buddhists in a country increases. Except for Judaism, however, these estimates are statistically significant only in the fully specified models; the negative effect of Judaism, although present in the reduced models, nevertheless increases with the addition of control variables. This pattern suggests that countries where these faiths are strong exhibit much variation on other factors that affect bodily integrity rights outcomes, thereby confounding the effect of religious composition. Controlling for these factors helps to isolate the independent effects of religion.
In other cases, the effect of religion is significant in the reduced models but not in the presence of control variables, suggesting that the relationship between religion and human rights is spurious. This pattern characterizes the positive effect of Catholicism, Judaism, and Hinduism on civil and political rights scores; the positive effect of Shintoism and Confucianism on bodily integrity scores; and the negative effect of Shi’a Islam on empowerment rights scores. Even more dramatically, Protestantism is associated with higher scores on each human rights measure in the reduced models, but these findings evaporate with the addition of control variables for all but empowerment rights. Yet even here, the estimated ‘controlled’ effect of Protestantism is cut in half. There is thus nothing intrinsic to Protestantism that improves most human rights outcomes, including, especially, bodily integrity rights; instead, its effect is mediated by other factors.
Overall, the main takeaway from Table 3 is that a country’s religious composition is tied much more closely to bodily integrity practices than to respect for civil and political rights, at least in the presence of control variables. With few exceptions, the effects of these control variables remain robust. Level and durability of democracy continue to improve human rights outcomes; the effect of the name-and-shame index remains negative across the board; GDP per capita and trade openness are still associated with higher bodily integrity rights scores; and INGO linkages and civil war intensity continue to correlate with lower bodily integrity scores.
Period-specific effects
Table 4 summarizes the period-specific effects of civilizational identity and religious composition on human rights outcomes, using the end of the Cold War in 1991 as the historical watershed. To economize the presentation, I report only the sign and significance of each estimate, net of the same variables included in Tables 2 and 3; the full models are available upon request.
Summary findings for the period-specific effects of civilizational identities and religious adherents on bodily integrity and civil/political rights measures.
The ‘Cold War’ period encompasses 1981 to 1990 for the analyses using civilizational indicators and 1980, 1985, and 1990 for the analyses using religious composition measures; the ‘Post-Cold War’ period encompasses, respectively, 1991 to 2005 and 1995, 2000, and 2005.
Blanks: not significantly different from (1) Western countries (for civilizational indicators) or (2) zero (for religious adherents measures); pluses: human rights score significantly (p < .05) higher than Western countries or significantly increases as the percentage of adherents increase; minuses: human rights score significantly lower than Western countries or significantly decreases as the percentage of adherents increase. Effects are net of the same control variables reported in Tables 2 and 3.
Of the 36 pairs of estimated civilizational effects (9 civilizational groupings × 4 human rights outcomes), 6 are statistically insignificant both during and after the Cold War, indicating no difference from Western countries in either period. An additional 15 are significantly negative in both periods, while another, the effect of Japan on empowerment rights scores, is significantly positive over time. On balance, then, the majority of cases show no temporal variation in human rights scores, especially with respect to civil and political rights outcomes. Of the effects that differ by time, 10 indicate that practices significantly worsened after the Cold War while 4 suggest that practices improved in the later period.
Performing similar period-specific analyses for the religious composition measures yields 44 pairs of estimates (11 religious families × 4 human rights outcomes). The plurality of cases – 18 – showed no significant effects either during or after the Cold War. In eight cases, religious composition had a significantly negative effect on human rights scores in the post-Cold War period but not earlier; in another five cases, the effect of religious composition was negative during the Cold War but not after. Six pairs of estimates, all pertaining to bodily integrity rights outcomes, were significantly negative throughout the period, whereas Shintoism improved empowerment rights scores in both periods. The remaining six cases were significantly positive during the Cold War but statistically insignificant after it.
Information effects
Table 5, which concludes the analysis, explores the possibility that cross-cultural variation in human rights scores is attributable at least partially to uneven levels of human rights scrutiny and reporting. As a rough check on the presence, magnitude, and direction of potential cultural biases, I replicated the analyses in Ron et al. (2005) study of country reporting by Amnesty International with Huntington’s civilizational indicators and measures of religious composition included as additional explanatory variables. Relative to Western countries and controlling for baseline human rights conditions, democracy, armed conflict severity, national wealth, coercive capacities, media attention, foreign aid, and other factors, Amnesty International press releases were significantly more likely to target Buddhist, Islamic, Orthodox, Sinic, and, in some cases, Latin countries, as well as countries with higher percentages of Jews and Muslims. (Japan also received more attention relative to the West.) Perhaps not coincidentally, many of the same countries – Buddhist, Islamic, Jewish, Latin, and Orthodox, in particular – were also identified as consistent human rights abusers in my analyses. These findings gesture strongly toward the role of information politics in shaping perceptions of human rights abuse (Clark and Sikkink, 2013; Hill et al., 2013; Keck and Sikkink, 1998; Ron et al., 2005). Certain groups of countries appear to have worse practices than others simply because they are under greater levels of scrutiny.
Negative binomial population average models of the factors influencing Amnesty International press releases, 1986–2000.
GDP: gross domestic product; ODA: official development assistance; INGO: international nongovernmental organization.
Robust standard errors in parentheses.
p < .05; **p < .01; ***p < .001 (two-tailed).
It is of course possible that Amnesty International decides to train its focus on the world’s worst abusers, such that increased reporting reflects objective cross-cultural variation in human rights conditions. But it seems equally likely, especially when contextualized within a broader pattern of findings uncovered in this study, that countries’ practices merely seem to worsen when they are monitored more closely by external observers. In stark contrast with previous research (Hafner-Burton and Tsutsui, 2005), for example, I found that country embeddedness in world society via linkages to INGOs is inversely associated with human rights practices, especially physical integrity rights practices. Whereas INGOs sometimes serve as ‘teachers of norms’ that engender commitment to and compliance with human rights standards (Finnemore, 1993), they also monitor, highlight, and uncover human rights abuses (Keck and Sikkink, 1998; Murdie and Davis, 2012; Ron et al., 2005). As such, countries with more INGO linkages appear to have worse practices, simply because they are under greater scrutiny. These dynamics are even stronger for Amnesty International and similar organizations, whose raison d’être is to publicize human rights violations.
Interestingly, countries with religious traditions in close cultural proximity to the ‘West’, including predominantly Catholic, Orthodox, and Jewish nations, often have lower scores on standard human rights indicators than do culturally more distal nations, perhaps because Western cultural cognates receive more scrutiny – and hence appear to have worse practices – than countries for which cultural differences with the West are more pronounced. In Simmelian fashion, Western human rights watchdogs and media outlets often seem to reserve their greatest levels of criticism for cultural ‘cousins’, even while they understate abuses occurring within their own ranks. 13 Such pro-Western biases have been documented in other areas of the international human rights system. For example, Cole’s (2010) study of human rights abuse cases decided by the UN Human Rights Committee found that Western countries were less likely than non-Western countries to be found in violation of human rights standards and correspondingly more likely to be exonerated of human rights abuses – even after holding levels of democracy and abuse constant.
Discussion and conclusion
How does each of the hypotheses fare, given these results? Hypothesis 1a, which predicted that the most relevant distinction in terms of human rights practices is between Western and non-Western countries, finds some support, whereas Hypothesis 1b, which predicted that non-Western civilizations with closer historical ties or cultural affinities with the West would have better human rights records, is not supported. Indeed, countries with cultures or religions that are relatively proximal to those of the West, including Latin/Catholic, Jewish, and Orthodox Christian, often have significantly lower human rights scores than culturally more distal civilizations such as Sinic/Confucian and Hindu, at least once confounding factors are controlled. Although difficult to conclude definitively, the weight of the evidence also strongly suggests that informational dynamics bias human right scores in some civilizational groupings downward, such that lower rights scores artificially reflect increased monitoring by external human rights observers. This possibility is deserving of further research.
Even still, support for Hypothesis 1a is strongest using Huntington’s civilizational indicators; support is much weaker using the operationally more nuanced (yet one-dimensional) religious composition measures. Indeed, religious composition is a rather inconsistent predictor of human rights outcomes. To the extent that ‘religion is a central defining characteristic of civilizations’ (Huntington, 1996a: 47), these differences between the civilizational and religious composition analyses call Huntington’s entire framework into question, at least insofar as human rights practices are concerned.
The second set of hypotheses dealt with temporal dynamics. Hypotheses 2a, 2b, and 2c posited that geocultural variation in human rights practices would, respectively, increase, decrease, and remain stable after the Cold War. Here again the findings are mixed. Human rights outcomes generally showed a great deal of stability within civilizations over time. As such, one of the core hypotheses inspired by Huntington’s approach, that cross-civilizational differences in human rights practices would become more pronounced after the Cold War, finds little support even when evaluated using his own framework. Once again, the religious composition analyses were somewhat less conclusive, with a plurality of cases showing no effects either during or after the Cold War, and a roughly equal number of fewer cases showing stability and decline in rated practices over time. All told, none of the hypotheses regarding the temporally contingent human rights effects of culture or religion found unequivocal support in the analysis.
A final prediction, Hypothesis 3, posited that differences between non-Western and Western countries would be greater for civil and political rights than for bodily integrity rights, owing to greater levels of normative conflict over the former and consensus regarding the latter. Here too the conclusions are far from definitive, especially if one scrutinizes each human rights outcome independently. Taking a panoptic view of the findings, however, shows that geocultural differences in bodily integrity scores are especially robust, in direct contrast to what was hypothesized. To qualify as ‘robust’, the estimated net effects of both a civilization indicator and its corresponding religious composition measure(s) must be statistically significant for each cognate human rights measure (i.e. physical integrity rights and the PTS, or empowerment rights and the civil liberties index). Based on these criteria, bodily integrity rights scores are systematically lower in Buddhist, Islamic, Orthodox, Jewish, and Latin/Catholic countries relative to Western nations. If one further requires these estimated effects to be statistically significant throughout the entire observation period, then only one geocultural grouping – Catholic/Latin – passes muster.
Why are human rights scores robustly lower in Latin America than elsewhere? Hafner-Burton and Ron (2013) offer one possible answer. According to their study, Western media pay especially close attention to abusive practices in the geographically (and culturally) proximate Latin American region, yielding a more complete – and less flattering – portrait of human rights conditions there. Once again, informational dynamics seem to play an important if indirect role in this analysis.
A similar if somewhat counterintuitive reason Hypothesis 3 lacks support is precisely that bodily integrity rights enjoy greater levels of international agreement and support when compared to civil and political rights, giving rise to a ‘consensus paradox’. Whereas civil and political rights are often contested on cultural grounds as reflecting Western biases (Cole and Ramirez, 2013), bodily integrity rights are nearly universally accepted as legitimate, which ironically makes their violation more likely to be detected and reported. Indeed, these abuses are closely monitored and scrutinized by groups like Amnesty International, whose reports provide the source material for most of the rights indices analyzed here. Consequently, the basic premise underlying Hypotheses 3 seems valid, even if the specific prediction derived from it erred.
To conclude, this study challenges key tenets of Samuel Huntington’s controversial clash-of-civilizations perspective, at least with respect to cultural and especially religious differences in countries’ human rights practices. Huntington contends that religion is a defining characteristic of the world’s major civilizations, but religious composition is only weakly associated with human rights outcomes. Moreover, and perhaps more importantly, a country’s cultural or religious affinity to the West does not insulate its citizens from gross violations of their rights, but neither does cultural ‘distance’ from the West guarantee that their rights will be abused. There is little evidence of a ‘Confucian-Islamic connection’ in opposition to human rights, of the sort Huntington (1996a: 185; see also pp. 195–198) posited. If anything, countries with relatively close cultural ties to the West – Catholic, Orthodox, and Jewish, for example – are often identified as the worst abusers of human rights. Thus, whereas Huntington’s civilizational framework appeared at first blush to explain systematic variation in countries’ human rights practices, a more nuanced investigation calls the very foundations of that framework into question.
Footnotes
Appendix A
Funding
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.