This item is only available as the following downloads:
FOLLOW THE RAINBOW BRICK ROAD: A CROSS NATIONAL ANALYSIS OF LGBT RIGHTS BY DANA ZIEGLER A Thesis Submitted to the Division of Social Sciences New College of Florida in partial fulfillment of the requirements for the degree Bachelor of Arts Under the sponsorship of Dr. Barbara Hicks Sarasota, Florida May 2012
i Acknowledgements I would like to tha nk my sponsor, Dr. Hicks, for going above and beyond for me Thanks also to my parents and my sister Kacey for their unwavering love and support. Finally, m any thanks to Liz Beth Susanna, Shelby, and the ladies of the Strictly common room for their friendship and emotional support through out my thesis process.
ii Table of Contents Chapter 1 Introduction, Literature Review, Theory, and 1 Introduction 1 5 8 Hypotheses 15 Chapter 2 18 Case Selection 18 Dependent Variables: LGBT Rights Legislation 19 Hypothesis 1: Sequencing 2 3 Hypothesis 1: Methods Hypothesis 1: Results and Discussion ....27 Hypoth esis 2: Neighbor .....39 Hypothesis 2: Methods ..42 Hypothesis 2: Results and Discussion ...42 Hypothesis 3: Religion ..53 Hypothesis 3: Methods ..56 Hypothesis 3: Results and Discussion ...58 Hypothesis 4: Demographic and Attitudinal Variable s Hypothesis 4: Methods ..79 Hypothesis 4: Results and Discussion ....80 Hypothesis 5: LGBT Rights Organizations ...101 Hypothesis 5: Methods Hypothesis 5: Results and Discussion ..103 Conclusion 114 122 Appendix 133
iii Tables Dependent Variables Table 2.1 Dependent Variables: LGBT Rights Hypothesis 1 Table 2. 2 Hypothesis 1.1: Most Common World Patterns 28 Table 2.3 Hypothesis 1.1: Regional and Federal State Patterns 30 Ta b le 2.4 Hypothesis 1.2 World Patterns ..... 31 Table 2.5 Hypothesis 1.2: Regional and Federal State Patterns 32 Table 2.6 Hypothesis 1.3: Most Common World Patterns with PCU 33 Table 2.7 Hypothesis 1.4: Most Common World Patterns with PCU 35 Table 2.8 Hypothesis 1.5: Most Common World Sequences Regarding Partnership and A doption 36 Table 2.9 Hypothesis 1.6: World Sequences 36 Table 2.10 Hypothesis 1: Frequency of Hypothes ized Sequence Ranks 38 Hypothesis 2 Table 2.11 Hypothesis 2: Explanatory Variables ..... 41 Table 2.12 Hypothesis 2: BorderBeforePer Mann Whitney Test 44 Table 2.13 Hypothesis 2: BorderBeforePer Means .... 44 Tabl e 2.14 Hyp .... 45 Table 2.15 Hypothesis 2: BorderTotalPer Mann Whitney Test 46 Table 2.16 Hypothesis 2: BorderTotalPer Means ... 47 Table 2 17 Hypothesis 2: NeighborTotalPer Mann Whitne y Test 48 Table 2.18 Hypothesis 2: NeighborTotalPer Means .... 48 T able 2 19 Hypoth sis 2: RegionBeforePer Mann Whitney Test 49 Table 2.20 Hypothesis 2: RegionBeforePer Means 50 Table 2 21 Hypothesis 2: RegAfterPer Means for Sex, Con, and Emp By Regio n 51 Table 2 22 Hypothesis 2: R egAfterPer Means for PCU and M b y Region 52 Table 2.23 Hypothesis 2: RegA fterPer Means for AnyP and Fam by Region ... ..... 53 Hypothesis 3 Table 2 24 Hypothesis 3: Broad Religion (BroadRelig) Frequenc ies Table 2 25 Hypothesis 3: Sig. Results for BroadRelig Chi square Tes t Table 2 26 Hypothesis 3: Sig. Results for BroadRelig Fisher's Exact Test Table 2 27 Hypothesis 3: Broad Religion 1 (BroadRelig1) Table 2.28 Hypothesis 3: Sig. Results for BroadRelig1 Chi square Test T able 2 29 Hypothesis 3: Sig. Results for BroadRelig1 Fisher's Exact Test Table 2.30 Hypothesis 3: General R eligion (GenRelig) Freq uencies
iv Table 2 31 Hypothesis 3: Sig Results for GenRelig Chi square Test Table 2.32 Hypothesis 3: Sig. Results for GenRelig Fis her's Exact Test Table 2.33 Hypothesis 3: General Religion 1 (GenRelig1) Frequencies Table 2 .34 Hypothesis 3: Sig. Results for Gen Relig 1 Chi square Test Table 2.35 Hypothesis 3: Christian vs. Muslim Frequencies Table 2.36 Hypothesis 3: Christian vs. Muslim Statistics Table 2.37 Hypothesis 3: C hristian vs. Other Frequencies Table 2.38 Hypothesis 3: Christian vs. Other Statistics Hypothesis 4 Table 2.39 Hypothesis 4. Homosexuality Justifiable Means Table 2 40 Hypothesis 4: Average for Homosexuality Justified (HJ) Frequencies Table 2 41 Hypothesis 4: Age Structure Means Table 2 42 Hypothesis 4: Average Age Frequencies Table 2.43 Hyp othesis 4: Age 15 24 Frequencies Table 2 44 Hypothesis 4: Postmaterialism Means Table 2 45 Hypothesis 4: Materialist (Mat) Frequencies Table 2 46 Hypothesis 4: Mixed Frequencies Table 2 47 Hypothesis 4: Postmaterialist (Post) Frequencies 90 Table 2 48 Hyp othesis 4: Religiosity Means .. 91 Tab le 2 49 Hypothesis 4: Religiosity Frequencies 92 Table 2 50 Hypothesis 4: Education Index (EI) Means .. 93 Table 2.51 Hyp othesis 4: Education Index (EI) Frequencies 94 T able 2 52 Hypothesis 4: Gender Inequality Index (GII) Means 95 T able 2.53 Hypothesis 4 : Gender Inequality Index (GII) Frequencies ... 95 Ta ble 2 54 Hypothesis 4: Freedom House Freedom Status Frequencies 96 Table 2 55 Hypothesis 4 : Freedom House Civil Liberties and Political Rights Means ....... 97 Table 2 56 Hypothesis 4: Freedom House Political Rights (PR) and C ivil Liberties (CL) Frequencies 98 Hypothesis 5 Table 2 57 Hypothesis 5: ILGA All LGBTI Organizations Freq s 105 Table 2 58 Hypothesis 5: All LGBTI Organizations Freq s by Legislation .. 106 Table 2 59 Hypothesis 5: Al l ILGA Directory Orgs Fre q s .. 108 Table 2 60 Hypothe sis 5: All ILGA Directory Orgs b y Population Freq s ... 108 Table 2 61 Hypothesis 5: All ILGA Directory Orgs and All ILGA Directory Orgs By Population Means .. 109 Table 2 62 Hypothesis 5: All ILGA Directory Orgs Means by Legislation 109 Table 2 63 Hypothe sis 5: All ILGA Directory Orgs b y Population Means By Legislation 111 Table 2.64 Hypothesis 5: All ILGA Orgs Wilcoxon Mann Whitney Test ... 112 Table 2 65 Hypothesis 5: All ILGA Orgs by Population Wilcoxon Mann Whitne y Test 112
v FOLLOW THE RAINBOW BRICK ROAD: A CROSS NATIONAL ANALYSIS OF LGBT RIGHTS Dana Ziegler New College of Florida, 2012 ABSTRACT The world has seen many changes in the area of LGBT rights, specifically decriminalization of homosexuality, equalization of the age of consent for sexual activity, anti discrimination legislation on the basis of sexual identity, same sex partnership and marriage, and adoption rights. This study seeks to understand what factors influence the passage of LGBT rights legislatio n, and what common characteristics are shared by countries with broad LGBT rights. Through cross national statistical analys e s of as many national and sub national cases as possible, five hypotheses are investigated. The order of legislation was examined to determine whether there are standard sequences in the passage of LGBT rights legislation. The influences of n eighboring countr ies and regional relationships, as well as differences between predominant religious heritages, were also explored. Demograp justifiability of homosexuality, postmaterialism levels, religiosity, age structure, education, democratic freedoms, and gender equality were next investigated. Finally, t he p ossible effe ct of LGBT advocacy organizations was assessed. Evidence was found in
vi support of the hypotheses about both sequencing and the impact of the independent variables studied but the outcomes of analyses stress the need for further study Overall, despite so me robust patterns the results supported a variety of paths to LGBT rights rather than one predominant path Dr. Barbara Hicks Division of Social Sciences
1 CHAPTER 1 : INTRODUCTION, LITERATURE REVIEW THEORY, AND HYPOTHESES In the past decade, same sex marriage has been legalized in ten diverse countries spanning four continents. Argentina, Belgium, Canada, Iceland, the Netherlands, Norway, Portugal, Spain, South Africa, and Sweden in addition to Mexico City and some states in the U.S., have extended marriage equality to gays and lesbians by legally validating their relationships and families within society. How can we explain this transformation fro m a relatively recent global past in which private same sex sexual activity was criminalized, to a present in which same sex unions are sanctioned by the state? The diversity of religious, geographical, and cultural heritages in these countries is striking. What do these heterogeneous societies have in common? For most of history, homosexuality has been either completely ignored or vilified as socially unacceptable. In accordance with religious and mora l traditions, most nations criminalized homosexual acts at some point in their history. It is only since the late 18 th century that countries began to repeal criminal legal codes targeted at homosexual acts some decriminalization has been very recent, eve n in liberal democratic societies such as
2 the U.S. where criminal codes in several states were finally struck down in 2003. The large scale mobilization of movements has gained momentum in national and international contexts in recent years, manifesting i n demands for increased rights. These rights include decriminalization, equal ages of consent for sexual activity, non discrimination legislation, partnership rights for gay couples, marriage rights, and adoption rights. However, while some states are mo ving toward broader lesbian, gay, bisexual, and trans (LGBT) rights, others staunchly oppose the legitimac y of LGBT rights on traditional or religious grounds. Most of the countries of the world have decriminalized homosexuality, and many are slowly but s urely extending rights. Some societies have become tolerant of LGBT lifestyles, while others ignore, restrict, or persecute them. accompanied the historical trajectory of greater conc ern for human rights that has orientation, gender expression, and gender identity, have been framed as fundamental human rights, highlighting concerns for the right to privacy, equal treatment, and protection. The movement also has connections to gender equality because of the significance of making partnerships and marriage gender neutral and less grounded in a tradition of patriarchy, especially through rights for lesbian cou ples and single women such as in vitro fertilization and single parent adoption. The human rights frame has been gaining traction at the domestic and international level to counter violence and discrimination against LGBT persons. However, the advances argued by the human rights frame are countered by moral and
3 religious arguments about the sanctity of the heterosexual family. LGBT rights have entered the political arena for example with broader rights becoming part of Leftist party platforms and campa ign promises and Rightist parties arguing that the family and children must be protected. Gay marriage, partnership, and adoption have become salient and often polarizing issues in national political discourse. Changes in legislation regarding sexual or ientation have practical and symbolic significance. Although the criminal status of homosexuality is often ignored in practice, in some places homosexual acts face capital punishment and reinforce the cultural acceptability of violence against LGBT people Decriminalization of homosexuality can lead to greater social acceptance of LGBT lifestyles, and non discrimination legislation can give LGBT persons legal recourse and protection in employment. Even t hough it is often hard to monitor, non discriminati on legislation explicitly spells out protection for gender identity and/or sexual preference, which has great symbolic value. In many countries the age s of consent for specific sexual acts differ (often with homosexual acts having a higher age of consent), which implies that such acts are graver or more delinquent. Equalizing ages of consent for heterosexual and homosexual activities is a largely symboli c measure that can foster a sense of equality in society. The recognition of gay families as legitimate and valued elements of social and political life is incredibly important. Extending partnership rights to same sex couples, through cohabitation, civ il union, partnership, or marriage normalizes their relationships and provides increased quality of life. Cohabitation rights are rights that are less than marriage for couples who live together (often applying to both heterosexual and homosexual couples) Civil unions and domestic partnerships refer to arrangements that
4 confer benefits to same sex couples that are similar to marriage benefits. The rights provided by such unions differ from country to country some may be functionally equivalent to marria ge while others remain a far cry from it. The legalization of same sex marriage opens marriage to two persons regardless of gender identity, often including language that makes marriage a gender neutral institution. Same sex marriage may or may not be ac companied by all of the rights given to heterosexual couples such as joint adoption. In many countries, only heterosexual married couples may adopt, or gay individuals or couples are explicitly banned from adoption. Allowing single parent, joint parent, and second parent adoption allows all persons to provide needy children with loving families. The dynamics that have resulted in a wide array of LGBT rights innovations cross nationally are not fully understood. A variety of literatures have addressed L GBT rights, but only a handful of studies have directly sought to explain cross national variation. Understanding domestic dynamics is vital in order to make sense of global change. Why have claims for gay rights gained so much momentum in recent years? Can we explain changes in LGBT rights legislation through cultural changes that are positive changes in LGBT rights? One can gain some insight into the answers to these br oad questions by asking: What factors influence the passage of LGBT rights legislation at the state and subnational level? By consulting relevant literature, I created several hypotheses t o address this question. First is the argument that t here is a stan dard pattern for the passage of LGBT rights legislation: decriminalization, followed by equalization of the age of consent, anti
5 discrimination legislation, recognition of same sex partnership/marriage, and recognition of same sex parental rights. I expec t that neighbor and regional effects are at work that states are influenced by legislation in neighboring states, resulting in policy diffusion. I also anticipate that the predominant historical religious context will have an effect on LGBT rights recogni tion. Several demographic and attitu dinal factors may account for variations in the existence of legislation, such as tolerance toward homosexuality, low religiosity, democratic freedoms, gender equality, higher levels of education, and postmaterialist va lues. Finally LGBT organizations advocating for rights at the international and domestic levels should have a positive effect on LGBT rights. LITERATURE REVIEW Very few studies have attempted to tease out explanatory variables for the legal transformation of same sex policy on a cross national scale. Most discussion of factors that led to legislation has been concentrated in in depth country case studies that focus on domestic LGBT social movement activity and the political, economic, and so cial context for the framing and passage of legislation (see Waaldijk 2001; Van Meerendonk and Sche epers 2004; Pichardo Galn 2010; Platero 2007 & 2008 ; Martinez and Dodge 2010; Stefnsson and Eydal 2004; Corijn 2004; Matthews 2005; Manzo 2010). The key factors studied in the case literature are a ttention to political party involvement or lack of involvement, public attitudes toward homosexuality, contribution of scientific literature, political arguments about human rights, equality, and the family, and legal innovations in LGBT rights arrangements. In addition, some comparative studies have been done, including small N studies of two or more countries or regional studies, such as of the U.S. states or the European Union (see Waaldijk 2005, 2007;
6 Rydstro m 2004 ; Maxwell 2000 ). Often these studies seek to compare LGBT rights arrangements in a region (such as Scandinavia and the EU) or use an exemplar case to suggest a course of action for a state with less liberal policies (as in comparisons between the U. S. and European nations with more liberal rights). A variety of academic studies and reports relate to LGBT rights; most of these studies focus on normative goals, such as advocating for gay marriage and decriminalization of homosexual activity, as well as providing and analyzing data on legislation. While there is some discussion of the process of that leads to greater LGBT rights, such as LGBT social movement activism and interaction with government elites, as well as why coun tries choose marriage rights or partnership rights, most of the discussion revolves around the implications of different partnership arrangements for LGBT individuals, especially marriage. Queer studies and feminist literatures often address the tension between recognition of same sex unions as rights of citizenship, and regulation of sexuality within a heteronormative framework (Weeks 2008, Josephson 2005). The International Lesbian, Gay, Bisexual and Transgender Association (ILGA) publishes yearly reports on the state of LGBT human rights, focusing their efforts on the decriminalization of homosexuality but also accounting for same sex partnership, anti discrimination legislation, same se x couple adoption, and equal age s of consent (ILGA 2010 2011). ILGA provides a wealth of information about existing laws and advocacy opportunities, but has not focused on trying to identify facto rs behind legislative changes. Within the field of legal analysis, the study of comparative sexual orientation law is nascent but growing. Most work has focused on analyzing cross national differences in
7 LGBT legislation and judicial rulings, or it analyzes legislation with an eye to international human rights law especially in the areas of nondiscrimination, equality, privacy, and liberty (see Waaldijk 2005; International Commissio n of Jurists 2009, Lau 2008). While this literature is rich in data and analysis of the legislation itself, it lacks efforts to explain legal transformations. What is largely missing from these literatures and reports is a discussion of factors that lead to innovations in legislation. Understan ding what influences changes in sexual orientation law is central to understanding legal innovations, how and why LGBT movements influence state policy, and ultimately how we can learn from the past in order to attain LGBT rights in the future. Issues of LGBT rights are also recently being addressed by the study of the global politics of LGBT human rights, an interdisciplinary approach that brings together politics, sociology, socio legal studies, social policy, and gender and queer studies (see Kollman a nd Waites 2009; Kollman 2007; Swiebel 2009). LGBT rights have been framed by LGBT social movements as fundamental human rights (Kollman and Waites 2009). Kollman and Waites (2009) hypothesize that the political processes involved in translating new, glo balized sexual categories into national and international policy change are contingent on several factors including institutional structure, ruling party coalitions, elite attitudes, economic climate, cultural understandings of welfare, and historic timing International bodies such as the UN and the European Court of Human Rights have begun to pay attention to LGBT issues in the frame of human rights abuses, right to
8 privacy, and equality of treatment under the law. Recent international declarations, su ch as the 2007 Yogyakarta Principles on the Application of Human Rights Law in Relation to Sexual Orientation and Gender Identity (the Yo gyakarta Principles), identify fundamental human rights that states are obligated to respect without prejudice based on sexual orientation or gender 2008). The Declaration of Montreal presented at the International Conference on LGBT Human Rights in 2006 proposed the formation of a UN convention for eliminating discrimination targeted towar d sexual orientation and gender identity (Kollman and Waites 2009, 5). These treaties and activities at the international level speak to a growing global consensus on the responsibility of states to recognize the fundamental human rights of LGBT individua ls. In order to gain a better understanding of transnational dynamics and the flurry of activity at the global level, one must analyze the dynamics occurring at the national and subnational level. THEORY Three authors have attempted to distill patterns and factors that explain the passage of LGBT rights legislation from the lens of large N cross national study. 2004) economics groundwork for the current study. Waaldijk (1992) outlines a sequencing argument in the passage of legislation for the European Union. He hypothesizes that all countries go through a sequence of LGBT rights legislation, beginning with decriminalization of homosexual acts and followed by equalization of the age of consent, anti discrimination
9 legislation, same sex partnership, and finally recognition of same se x p arental rights (Waaldijk 1992 53 54). He revises this argument in a 2007 article that ranks the EU countries and Council of Europe non EU member countries on their LGBT rights. Reinforcing the trend of standard sequences, he adds the trend of steady pr ogress, 638) in an attempt to predict law developments in EU countries in this area. The trend of re progressively gaining increased rights in Europe as a result of very small steps toward states will only transition from a regime of oppression to a regime that gr ants rights to LGBT persons after it has passed symbolic legislation, such as decriminalization of homosexuality or anti discrimination legislation (Waaldijk 2007, 638). Waaldijk implies that the argument applies internationally in his chronology of LGBT rights legislation around the world, presenting a chronology of legal recognition of homosexuals that 2009, 3). I hope to test the validity of izable on an international level. Badgett (2004) performs statistical analysis to explain the adoption of same sex partnership laws in OECD states, hypothesizing the importance of the perceived efficiency of gay marriage and institutional change factors. For her efficiency component, Badgett proposes that the visibility of gay couples will correlate with same sex unions (SSU) and the declining social and economic value of marriage will negatively correlate with SSU (103). Regarding institutional variables of shifts in
10 political power and collective resources and shifts in political interests and norms about sexuality, gender, and homosexuality, Badgett hypothesizes that high levels of religiosity will correlate negatively with SSU, and that having a leftis t government, union membership a strong presence of LGBT organizations, and more liberal social norms will correlate positively with SSU (Ibid). She concludes that high levels of cohabitation, tolerance toward homosexuality, and low religiosity predict p assage of SSU legislation variables on a larger scale. Alt hough Frank and McEneaney (1999) specifically examine policies that regulate same sex sexual activity and the g rowth of l esbian and gay social movements, their cross national study of eighty four countries provides insight into causal linkages related to LGBT policy more generally. y greater cultural individualism, gender equality, and international linkages, exerts influence on the formation of LGBT social movements and the passage of liberal policies on same sex relations, which in turn mutually reinforce one another. The existenc e of dense ties to international society and commitment to gender equality was strongly positively correlated with more liberal policies within states (Frank and McEneaeney 1999, 930). These cultural effects may play a role in the proliferation of other L GBT targeted policies such as partnership rights. To create an index of individualism, they provide a score for Protestant tradition, human rights, and psychology authorships (924). Individualism was cor related with the existence of lesbian and gay movem ents, so it could be useful to disaggregate the index and test religious rights. Given
11 the framing of LGBT issues as human rights issues and the interconnectedness of LGBT to gender rig hts, testing the effects of human rights and gender equality commitments may shed some light on causality. In order to engage with alternative explanations for policy change, such as econom ic and political explanations, Frank and McEneaney incorporate l ogged per capita on LGBT movement formation, but much less effect compared to cultural explanations on state policies (Frank and McEneaeney 1999, 926 931). They menti on but do not incorporate measures of postmaterialist attitudes in a country, instead using per capita GDP, which may not capture the full range of cultural beliefs that is brought on by economic growth Postmaterialism theory can provide insight into br oad cultural changes, including attitudes toward homosexuality, which are sweeping the globe. Inglehart and Abramsom (1999) hypothesize that economic growth in advanced industrialist nations leads to express ion and quality of life (Inglehart and Abramsom 1999). An empirical study of gender and postmaterialist values found that having postmaterialist values and being a woman was correlated with feminism ( Hayes, McAllister, and Studler 2000). Concern for LGBT f the industrialized countries, citing the example of the proliferation of gay and lesbian g roups
12 (Inglehart 1990, 12). When looking at trends with regards to traditional sexual and social norms, Inglehart explains variation in cultural values by the degree to wh harmonize with Judeo Christian values (Inglehart 1990, 182). He finds that post materialists are less likely than materialists to hold traditional beliefs, that the young are more likely to hold socially liberal beliefs than the old, and that the shift to postmaterial publics is coupled with a decrease in traditional values (Inglehart 1990, 191 193). When asked to rate whether homosexuality was ever justified on a ten point scale, 63% of Materialists believed homosexuality is never justif ied, compared to only 36% of Postmaterialists (World Values Survey 1981 1982, in Inglehart 1990, 195). Younger, less religious, postmaterialist individuals were least likely to feel that homosexuality is never justified. While it is useful to look at age differences, religiosity, and postmaterialist inclinations, the restriction of analysis to Western industrialized nations with historically Christian populations can only go so far when trying to understand global trends. Religions have historically bee n opposed to family models or sexual activities that differ from the heterosexual man and wife model. With less religiosity in society, it is hypothesized that there will be a greater commitment to LGBT rights. It is also important to measure the amount of power that religious institutions have in a society whether religious institutions have direct or indirect political power. The religious culture of advanced industrialist countries played a part in the likelihood of individuals being feminists (Hayes McAllister, and Studler 2000). Minkenberg (2002) analyzed the
13 relationship between religion and the abortion policy outcomes in liberal democracies, finding that confessional heritage (Catholicism or Protestantism) and cultural values (levels of relig io sity) had the greatest effect; whereas institutional differentiation (separation of church and state) and political mobilization (Christian democratic political parties) had lesser effects. As a moral issue, gay rights policy may be influenced by similar factors to that of abortion policy. Norris and Inglehart incorporate ten major world religions while examining between the West and Islam, arguing that cultural di fferences are more likely related to gender equality and sexual liberalization (2003, 11). Using data from the World Values Surveys, they classify seventy two countries based on predominant religious heritage, including Sinic/Confucian, Sub Saharan Africa Latin America, Orthodox, Central Europe, Islamic, Hindu, Japanese, and the West (Protestant and Catholic countries) in order to assess patterns of cultural attitudes with regards to political and liberal social values. Some of these categorizations lump regions together as sharing a common religious heritage. Not only was there a gap between Western Christian and Islamic nations in social values, but there was much variance across the religious society groups, especially with regards to approval of homo sexuality and gender equality. Weeks suggests that variations in demands for gay marriage versus partnership rights are a result based cultural traditions that either emphasize or deemphasize the importance of marriage in society ( 2008, 791). It would be useful to examine whether religious heritage (beyond simply Christian beliefs) correlated with particular outcomes of LGBT rights recognition.
14 With a domestic focus, a study by Van Meerendonk and Scheepers (2004) examines public op inion toward equal marriage rights for gays and lesbians in the Netherlands. They find evidence to support their hypotheses that people who attend church more often, affiliate with Christian political parties, are members of Christian denominations, are l ess educated, and are older are more likely to be against equal marriage rights. These factors religiosity, political affiliation, education, and age and other demographic factors should be taken into account when looking at the passage of LGBT rights leg islation in a state. Creation of gay rights legislation, such as partnership and cohabitation rights, could be conceived of as policy innovations. One can look at the literature on theories of the policy process to hypothesize about how and why LGBT policy innovations have come about. Berry and Berry (2007) create a model of government innovation that takes into account diffusion of nearby, regional, or national policies and internal factors as part of the theories of the policy process literature. While designed for innovation across U.S. states, this model may be generalizable on a cross national basis. In combination with internal domestic factors, such as demographics and government structure, what kind of neighbor, regional or internati onal eff ects can be discerned?
15 H YPOTHESES Hypothesis 1: Standard Sequencing Countries follow a standard sequence in LGBT rights legislation, beginning with decriminalization of homosexual acts and followed by equal ages of consent for sexual activity, anti discrimination legislation, same sex partnership (of a variety of forms, such as domestic partnership or marriage), and recognition of same sex parental rights (adoption by LGBT individuals and families). Waaldijk suggests in several papers that there is a standard order in which rights are legislated for LGBT individuals. This hypothesis has been examined with regards to the EU (Waaldijk 2007), but not on an international scale. By examining the chronological order in which rights were extended in a country, one can gain a better unders tanding of the entire process of the attainment of LGBT rights. Hypothesis 2 : Neighbor Effects Regional and neighbor to neighbor policy diffusion has an effect on LGBT rights. There is often diffusion of policy innovations across the U.S. states, and in an increasingly global world nations can look to their neighbors and abr oad for guidance on policy prec edent. With the rise of international advocacy for LGBT rights and greater communication on human rights in international organizations such as the United Nations, one expects that states will look to the examples of their neighbors. Additionally, shared cultures in world regions may account for consensus on social issues across borders. Comparing state polici es to those of other states in the region is v aluable
16 in order to determine whether there is a regional diffusion effect of sexual orientation law. Hypothesis 3 : Religion Historically predominant religious culture of a nation has an effect on LGBT rights Minkenberg (2002) and Norris and Inglehart (2003) stressed the importance of religious institutional heritage in predicting social values. Alt hough many studies have focused on the importance of Protestantism as a predictor of individualist values that privilege sexual freedom, it would be useful to account for the effect of predominant religious culture in an international comparison beyond mer e consideration of Christianity or lack thereof. Hypothesis 4 : Demographic and Attitudinal Variables Tolerance toward homosexuality, education, postmaterialist values, commitment to gender equality, democracy, younger age cohorts, and lower levels of religiosity predict broader LGBT rights. Much of the literature has pointed to demographic factors and state commitments to socia l rights (such as gender rights and human rights ) that correlate with liberal social values. Commitment to democratic ideals ex pressed through political rights and civil liberties should reflect greater concern for safeguarding the rights of LGBT individuals. Younger individuals tend to have more liberal attitudes, so perhaps larger youth cohorts translate into more legislation f or LGBT rights. Changing public attitudes account for increasing social acceptance of homosexual families and lifestyles. One expects that
17 societies friendly and sympathetic to LGBT lifestyles would legislate greater advances in rights. Similarly, lower religiosity has been associated with greater acceptance of LGBT lifestyles and less commitment to the traditional heterosexual family model. Higher levels of education typically correlate with more liberal social values, and postmaterialist values have been associated with greater acceptance of homosexuality. By testing the effects of these factors we can try to discern whether the existence of certain demographic and value characteristics in a population translate into broader LGBT rights. Hypothesis 5 : LGBT Organizations Greater numbers of int ernationally connected LGBT organizations result in greater LGBT rights. Groups mobilizing for LGBT rights play a key role in effecting policy change. One predicts that the presence of LGBT groups mobilizing at t he domestic level, as well as connecting on the international level, will have an effect on the sexual orientation policy in a nation. The international involvement of LGBT social movements also accounts for linkages between the state and the internationa l community that subjects the state to emerging global norms regarding the human rights of LGBT individuals.
18 CHAPTER 2 : VARIABLES, METHODS, RESULTS, AND DISCUSSION CASE SELECTION In order to carry out a broad scale, cross national analysis of LGBT rights, data was drawn from every country possible in addition to some subnational units Data was g athered for all regions including the Americas, Africa, Asia, Oceania, and Europe, as well as the subnational units of the United States, Brazil, United Kingdom, Mexico, Canada, and Australia. For Mexico, data was on ly available for five states. For the sequencing, religion, regional and LGBT rights organizations hypotheses most countr ies were included in analysis In some tests, countries were excluded for lack of data. For the demographic variables tested in Hypothesis 4 the case selection only extends to those countries included in the World Values Survey Gender Inequality Index, Education Index, or Freedom House Freedom in the World Report respectively for each portion of the hypothesis Countries were classified into regional and sub regional categories according to the UN Statistical Division Classification of Regions.
19 DEPENDENT VARIABLES: LGBT RIGHTS LEGISLATION The hypotheses seek to explore the effect of certain explanatory variables on LGBT rights legislation Sex, Con, AnyDiscrim, Emp, GS, AnyP, PCU, M, Marriage, A, and Fam Defined in Table 2. 1, t he legislation v ariables were coded with the year in which the relevant law was passed to the greatest extent possible In some cases, only the year of enactment was known or the year of passage was unknown Still other cases of legislation exist because of court ruling s, so the year for the court decision is recorded. Multiple references are made in the following analy ses it should be understood that not all legislation is the result of legislative action. In the c ase of variables any anti discrimination legislation ( AnyDiscrim ) any partnership ( AnyP ) and family recognition ( Fam ) the first instance of each broad type of legislation is the year recorded. State Sponsored Homophobia 2011 report and Legal Recognition of Homosexual Orientation in the Countries of the W orld a Chronological Overview with F ootnotes (2009). When data was missing or in congruous between the two primary sources, a wide variety of online news sources, legislation documents, maps and case studies were consulted for country specific information. Although most information was found, information is still missing. For this r eason, information is given for each analysis on the number of cases studied.
20 Decriminalization of homosexuality (Se x ) is the removal of criminal sanctions for homosexual sexual acts, often codified in sodomy laws. In Africa, some states have nev er criminalized homosexuality, but this is not to say that homosexuality is acceptable in those societies. In some countries, h omosexual sexual activities require a higher age of consent than for heterosexual activity, so the variable equal ages of consent (Con) represents the removal of unequal ages of consent. In many cases the age of consent was Table 2.1 Dependent Variables: LGBT Rights Dep. Variable Description of legislation Coded Sex Decriminalization of homosexual sexual activity 1 (Yes), 2 (No), 0 (Never criminalized) Con Equalization of the age of consent for heterosexual and homosexual sexual activity 1 (Yes), 2 (No) AnyDiscrim Any anti discrimination legislation, includes Con, Emp, and constitutional protections 1 (Yes), 2 (No) Emp Anti discrimination legislation in employment 1 (Yes), 2 (No) GS Anti discrimination legislation in goods and services 1 (Yes), 2 (No) AnyP Any partnership legislation, includes cohabitation rights, PCU, M, and Marriage 1 (Yes), 2 (No) PCU Partnership and/or civil unions available for same sex couples 1 (Yes), 2 (No) M Same sex marriage or marriage equivalent partnership (equal or near equal rights of marriage) 1 (Yes), 2 (No) Marriage Same sex marriage 1 (Yes), 2 (No) A Joint parent and/or second parent adoption available to same sex couples 1 (Yes), 2 (No) Fam Any family legislation, includes all forms of partnership and A 1 (Yes), 2 (No)
21 equalized in tandem with the de criminalization of homosexuality. Several countries have equalized the age of consent, but it is unclear in which year this was done, so there is missing data for Con. Three categorizations of anti discrimination legislation are studied. The first two are l egislation in employment (Emp), such as that which prevents employers from discriminating on the basis of sexual orientation, and legislation in goods and services (GS), such as that which prevents the denial of government services on the basis of sex ual orientation. A third variable, any type of anti discrimination legislation (AnyDiscrim), includes Emp, GS, and other anti discrimination legislation such as constitutional protections. The recorded year for AnyDiscrim is the year for the first instan ce of such legislation. Thus, AnyDiscrim serves as an indicator for comparison between countries for all types of anti discrimination legislation with regards to sexual orientation. Four categorizations for partnership are studied. Any type of partners hip (AnyP) is the broadest partnership variable, which includes any cohabitation rights recognized by the state for same sex partners, civil union or partnership rights of various configurations, and same sex marriage. The year recorded for AnyP is the fi rst instance of such legislation in order to represent the first in itiation of partnership legislation in a state. While allowing for comparison across states with all different types of same sex partnership arrangements, the disadvantage to AnyP is that the year of passage may not be reflective of the substance of the legislation. For example, if the only type of partnership passed in country X were some minimal cohabitation rights in 2000, but the first instance of partnership in country Y was marriage equivalent partnership in 2010, it
22 woul d appear that country X was more progressive, when in fact country Y has the more substantive partnership. To disaggregate AnyP, types of partnership are broken down into partnership or civil union (PCU), marriage equivalent partnership (M), and same sex marriage (Marriage). Partnership or civil union (PCU) is the next step up from cohabitation rights, and can encompass a multitude of different benefits con ferred onto partners that vary by state, from tax and health insurance benefits to near equal marriage rights. The terms there is no broad definition or agreed upon configuration of rights for either. Because of the broad spectrum of same sex pa rtnership arrangements observed in the world, partnership is further specified in a variable for m arriage equivalent partnership ( M ) Cases with M have nearly all or all of the rights and benefits that marriage has for same sex couples, without the title The last specification for partnership rights is same sex marriage (Marriage ). Applying to the fewest cases worldwide same sex marriage is partnership that is equivalent to heterosexual marriage. The relative value for each type of partner ship is outside the scope of this thesis; the categorizations instead reflect the number of rights conferred by each type of legislation rather than any intrinsic value of the legislation. Throughout analysis, terms escribe the instance of all of the LGBT rights covered by the set of hypotheses examined in this thesis This term is not meant to imply that equal rights exist for homosexual persons and same sex partners in a s ociety. The goal of this study is to explo re patterns with regards to LGBT rights legislation, not to evaluate the symbolic or practical effects of such legislation.
23 Joint parent and/or second parent adoption (A) is also included in analysis. The earliest year for availability of either is r ecorded. Joint parent adoption means that two people of the same sex can jointly adopt a ch ild, while second parent ado ption means that one can adopt A multitude of adoption and family related laws for same sex couples are observed across the globe, including single parent adoption by homosexuals, various joint and second parent adoption arrang e ments and in vitro fertili zation for same sex couples Because of the wide variability of laws from sta te to state, two standard arrangement s for whic h information was available were chosen for analysis Unfortunately, given the availability of data, the two type s of adoption could not be disaggregated. The variable A should be understood as an indicator of steps toward equality in family recognition. T he presence of joint parent adoption and/or second parent adoption by no means indicates that there is full equality for all types of adoption or family a rran gements for same sex couples. The variable Fam stands for all types of family legislation, which includes joint and second parent adoption (A) and all types of partnership. Since AnyP aggregates the partnership form PCU, M, and Marriage by recordin g the first instance of any of the three partnership types, it is used for Fam. The year recorded for Fa m is the earlier of A and AnyP. Fam indicates the presence of some, even minimal, recognition by states of family rights for same sex couples.
24 HYPOTHESIS 1: SEQUENCING In order to look at patterns of passage of LGBT rights legislation, several sub hypotheses were developed that reflect the spirit of W t hat all countries first decriminaliz e homosexual sexual activit y, then equaliz e the age of consent, pass anti discrimination legislation, allow same sex partnership, and recognize same sex parental rights (1992, 53 54). This section also tests the hypothesis that partnership and joint parent adoption before those rights are passed (Waaldijk 2007, 636 638). The hypothesis is broken down into sub hypotheses in order to test different possible confi gurations of rights that are in line with the sequence. Hypothesis 1.1: The order of passage of legislation is decriminalization (Sex), equalization of the age of consent (Con), anti discrimination legislation (AnyDiscrim), partnership legislation (AnyP) joint parent adoption (A). Hypothesis 1. 1 equalization of the age of consent, and joint parent adoption are s pecific legislative (or in some cases judicial) acts, whereas anti discrimination and partnership appear in many forms. In thi s analysis and for Hypothesis 1.2 the most far ranging terms possible were used for most types of legislation. For example, AnyP reflects the first instance of any type of partnership enacted, ranging from some recognition of cohabitation by the state to same sex marriage. AnyDiscrim refers to the first instance of any type of anti discrimination legislation, whether in employment, goods and services, or constitutional protections. B ecause of the wide ranging combinations of adoption rights and difficul ty
25 of locating legislation, the variable A represents the presence of joint and or second parent adoption that is not disaggregated. Hypothesis 1. 2 The order of passage of legisl ation is decriminalization (Sex ), anti discrim ination legislation (AnyDiscr im ), pa rtnership legislation (AnyP) joint parent adoption (A ). In this hypothesis, consent was excluded. Many cases were missing data for the variable consent, so this version of the hypothesis allows for the inclusion of more cases Hypothesis 1.3: The order of passage of legislation is decriminalization (Sex ), equaliza tion of the age of consent (Con ), anti discrim ination legislation (AnyDiscrim ), pa rtnership or civil union legislation (PCU) joint parent adoption (A ). Hypothesis 1 .3 is the same as Hypothesis 1.1 except AnyP is replaced with PCU, in order to test the hypothesis when instances of cohabitation are excluded in favor of more developed partnership rights This difference is important because cohabitation rights represent a very limited recognition of partnership rights by the state, whereas partnership and civil union rights carry more weight both substantively and symbolically. Hypothesis 1.4: The order of passage of legislation is decriminalization (Sex ), anti discrim ination l egislation (AnyDiscrim ), partnership or civil union legislation (P CU) joint parent adoption (A ). This hypothesis excludes consent (Con) so as to include more cases, and replaces AnyP with PCU with the same logic as Hypothesis 1.3 Hypothesis 1.5 : Partnership or civil union legislation (PCU) is passed before joint parent adoption rights (A). whether partnership rights precede adoption rights. It tests the argument that recognition
26 of a family unit would come before broader parental rights for a dopting non biological children. Hypothesis 1. 6 : Symbolic legislation, either decriminalization of homosexuality, equal age of consent, or anti discrimination legislation is passed before substantive legislation, either any partnership recognition or adoption rights. (2007) Passage of any of the three types of symbolic legislation before either of the two types of substantive legislation supports the hypothesis. hat criminal and anti discrimination provisions do (2007, 23). Hypothesis 1: Methods Legislation was ranked in the order of passage by year for each sub hypothesis under revi ew For example, the data set for Hypothesis 1.1 includes every combination of legislation variables Sex, AnyDiscrim, AnyP, and A that was observed in the cases with four possible ranks from 1 to 4 including combinations in which legislation was not observed, for example, if only Sex and AnyDiscrim w ere passed for Hypothesis 1.1 L egislation types that were passed in the same year were ranked in the order favorable to the hypothesis, since co ncurrent passage reflects the spirit of the hypothesis rather than providing evidence against it. The frequencies of the combinations of sequences for legislation were examined Frequencies were also found for each legislation variable for each sub hypothesis in order to se e the frequencies for which each was ranked. For example, in Hypothesis 1.1, Sex occur red first in the order of legislation 232 times second 3 times, and third 1 time
27 This analysis is helpful for observing genera l trends in rank, especially since some countries did not pass all of the le gislation in a given hypothesis. Reference to cases means that the order of the passage of legislation fits the hypothesis even tho ugh the country may not have passed all legislation in the pattern. For example, for the hypothesized pattern Sex_AnyDiscrim_AnyP_A, a country would be on track if its pattern was Sex_AnyDiscrim or Sex_AnyDiscrim_AnyP States with no legislation are excluded from analysis. Additionally, in cases where data was missing for any of the legislation, the political unit was excluded from the analysis for that hypothesis. For example, if the sub hypothesis examined four pie ces of legislation, and country Z had data for three types of legislation but was missing the year of passage for one type of legislation, country Z would be excluded for that sub hypothesis completion of the hypothes ized sequence at hand. S refers specifically to the to any observed order. Hypothesis 1: Results and Discussion Hypothesis 1.1: Sex, Con, AnyDiscrim, AnyP, A Table 2 .2 illustrates the most common patterns found around the globe with at least 6 cases. The sequence, Sex_Con_AnyDiscrim_AnyP_A, occurs 11 times. However, the pattern Sex_Con_AnyDiscrim_A_AnyP, in which joint parent adoption was legalized before any partnership legislation, was more common than the hypothesized sequence by 6 cases. T his result is somewhat misleading because all 17 cases for the Sex_Con_AnyDiscrim_A_ AnyP hypothesis were subnational units of
28 Brazil (15 cases) and the United States (2 cases), whereas only 3 of the 11 cases for the Hypothesis 1.1 sequence are accounted for by subnational units. Another distortion of frequencies occurs in the next most c ommon patterns after the sequence Sex_ AnyDiscrim_AnyP_A and Sex_AnyDiscrim_A_AnyP, which are accounted for almost entirely by multiple units in Canada and Brazil. Table 2. 2 Hypothesis 1.1: Most Common World Patterns Freq % of World Numerical Pattern Legislative Pattern 63 18.16 1_2_N_N Sex_Con 23 6.63 1_2_3_4_N Sex_Con_AnyDiscrim_AnyP 21 6.05 1_2_3_N_N Sex_Con_AnyDiscrim 17 4.9 1_2_3_5_4 Sex_Con_AnyDiscrim_A_AnyP 12 3.46 1_N_N_N_N Sex 11 3.17 1_2_3_4_5 Sex_Con_AnyDiscrim_AnyP_A 9 2.59 1_N_2_3_4 Sex_AnyDiscrim_AnyP_A 9 2.59 1_2_N_4_3 Sex_AnyDiscrim_A_AnyP 6 1.73 1_2_4_3_N Sex_Con_AnyP_AnyDiscrim 171 49.28 The numerical pattern is a comparison to the hypothesized sequence : decriminalization of homosexual sexual activity (Sex), equal ages of consent for sexual activity (Con), any anti discrimination legislation (AnyDiscrim), any partnership legislation (AnyP), and joint parent adoption (A), or 1_2_3_4_5. The first slot in the numerical pattern represents decriminalization (Sex), the second slot equal age of consent (Con), the third slot anti discrimination legislation (AnyDiscrim), the fourth slot any partnership (AnyP), and the last slot joint parent adoption (A). The number corresponding to each variable space represents the o rder in which legislation was passed, from 1 to 5. N means no legislation for that variable space. Another 107 cases, 30.84% of global cases, were on track for the hypothesis, in that they completed part of the sequence. Evidence of many on track cases, the top three most common patterns, also offers support to the hypothesis in the sense that political units are on th e predicted path toward greater rights. Looking at cases that have passed all of the legislation is especially useful for attempting to p redict further innovations around the world. When looking specifically at cases that passed all five types of
29 legislation, Hypothesis 1.1 has the highest frequency on the national level (with the exclusion of Brazilian subnational units). However, there is much variation in the 39 cases with all five rights in the 11 cases besides the hypothesis and the Brazilian pattern, there were 9 observed variations. So, while many countries appear on track or follow the sequence, there is evidence of several differ ent paths to LGBT rights. Observation of regional patterns is useful in order to detect whether patterns are concentrated in certain areas of the world. Table 2. 3 describes regional and federal state observations of the Hypothesis 1.1 sequence as well as a combination of sequence and on track cases for the hypothesis. The majority of instances of Hypothesis 1.1 were observed in Europe. Of the 18 Asian cases that were on track for the sequence 13 of the cases only had the first two types of le gislation; likewise in Oceania the on track cases had only 1 or 2 types of legislation I n Europe which had a majority of on track cases, the bulk of the on track cases had either 3 types of legislation in the sequence (13 cases) or 4 types (8 cases) In the Americas the majority of cases were also on track ; the highest frequencies were for 2 types of legislation in the sequence (6 cases) and 4 types (7 cases) In federal states, most of the states in the U nited States were on track, the majority o f which were cases with 2 types of legislation (31 cases); and in Australia the on track cases either had 3 types of legislation or 4 types It appears that on the state level, Europe and the Americas account for many of the cases that are close to compl eting the sequence. These regions accounted for 15 of the 23 cases that had 4/5 of the Hypothesis 1.1 sequence. For this analysis and much of Hypothesis 1, Africa is excluded from detailed analysis, because it has very few countries
30 with any LGBT rights legislation, and the focus is on understanding patterns for countries that have completed or near completed sequences. Table 2.3 Hypothesis 1.1: Regional and Federal State Patterns Region Total States Total w/Righ ts Freq 1.1 Sequence % 1.1 Sequence Freq 1.1 + OT % 1.1 + OT Asia 53 20 0 0.00% 18 33.96% Africa 58 11 0 0.00% 3 5.17% Europe 53 53 7 13.21% 35 66.04% The Americas 52 32 1 1.92% 24 46.15% Oceania 26 10 0 0.00% 6 23.07% Federal States Total Subunits Australia 9 9 2 22.22% 4 44.44% Brazil 27 27 0 0.00% 0 0.00% Canada 13 13 0 0.00% 0 0.00% Mexico 5 1 0 0.00% 1 20.00% United States 51 51 1 1.96% 39 76.46% Total number of states or subunits that have passed at least one piece of legislation 1.1 + OT refers to states or subunits that follow the Hypothesis 1.1 sequence or are on track Hypothesis 1.2 : Sex_AnyDiscrim_AnyP_ A Table 2.4 reports all of the observed sequences when looking at combinations of Sex, AnyDiscrim, AnyP, and A. H ypothesis 1.2, sequence Sex_AnyDiscrim_AnyP_A occurs in 24 c ases including states and subnational units Subnational units represent 16 of the 24 cases. Roughly half, 47.55% of cases, are on track for the pattern or fit the sequence, and comprise the top four frequencies. About a fifth of cases that have a t least one type of legislation do not follow the pattern. The fact that almost 75% cases with at least one type of legislation are on track for the pat tern or deviate slightly from it, with A before AnyP, perhaps indicates that pursuing such patterns of legislation will result in greater success in gaining LGBT rights. There were seven paths to full rights for the 55 units that passed all four pieces of legislation. The next most observed pattern after th e hypothesis, Sex_AnyDiscrim_AnyP
31 _A, occurred entirely in states in Brazil and the U.S. Europe had 5 cases for sequence Sex_AnyP_AnyDiscrim_A, compared to 6 cases for Hypothesis 1.2. Testing Hypothesis 1.2 has strengthened the findings in Hypothesis 1.1 that there are a vari ety of paths to full rights but that many countries are on track for the hypothesis. Ta ble 2.4 Hypothesis 1.2 World Patterns Frequency % of World Numerical Pattern Legislative Pattern 82 23.63 1_N_N_N Sex 30 8.65 1_2_N_N Sex_AnyDiscrim 29 8.36 1_2_3_N Sex_AnyDiscrim_AnyP 24 6.92 1_2_3_4 Sex_AnyDiscrim_AnyP_A 18 5.19 1_2_4_3 Sex_AnyDiscrim_A_AnyP 13 3.75 1_N_3_2 Sex_A_AnyP 11 3.17 1_3_2_N Sex_AnyP_AnyDiscrim 6 1.73 1_3_2_4 Sex_AnyP_AnyDiscrim_A 6 1.73 1_N_2_N Sex_AnyP 4 1.15 1_4_2_3 Sex_AnyP_A_AnyDiscrim 4 1.15 N_1_N_N AnyDiscrim 3 0.86 1_N_2_3 Sex_AnyP_A 3 0.86 1_N_N_2 Sex_A 1 0.29 1_4_3_2 Sex_A_AnyP_AnyDiscrim 1 0.29 2_1_3_4 AnyDiscrim_Sex_AnyP_A 1 0.29 3_1_2_4 AnyDiscrim_AnyP_Sex_A 1 0.29 2_1_N_N AnyDiscrim_Sex 1 0.29 2_1_3_N AnyDiscrim_Sex_AnyP 238 100.01 The numerical pattern is a comparison to the hypothesized sequence : decriminalization of homosexual sexual activity (Sex), any anti discrimination legislation (AnyDiscrim), any partnership legislation (AnyP), and joint parent adoption (A), or 1_2_3_4. The first slot in the numerical pattern represents decriminalization (Sex), the second slot anti discrimination legislation (AnyDiscrim), the third slot any partnership (AnyP), and the last slot joint parent adoption (A). The number corresponding to each variable space represents the order in which legislation was passed, from 1 to 4. N means no legislation for that variable space. Over 100% due to rounding. Table 2. 5 describes regional and federal state observations for the Hypothesis 1.2 sequence, Sex_AnyDiscrim_AnyP_A and on track patterns Over half of the political units that followed the Hypothesis 1.2 sequence were located in Europe and Canada. The sequence seems more common on the subnational unit level than the state level. Well
32 over the majority of states in Europe an d the Americas are on track for the sequence. In Asia, all 21 of the states that passed legislation are on track for the sequence 17 of which only had1 type of legislation. S imilarly in the Americas 34 cases were on track but the highest frequency was for 1 type (20 cases), and likewise in Oceania 8 of the 10 on track cases had only 1 type. Unlike in Hypothesis 1.1, the Americ as do not account for a large portion of on track patterns with multiple pieces of legislation. This difference results partly from data missing on the consent ( Con) variable in this region, which meant several cases were excluded from the analysis Europe also had a majority of on track cases; the highest frequencies were for on track cases with 2 types of legislation in the sequence (17 cases ) or 3 types (9 cases). In federal states, most of the states in the United States Aust r alia, and Canada were on track The most observed frequency for the U.S. was 26 for 1 type of legislation. Table 2.5 Hypothesis 1.2: Regional and Federal State Patterns Region Total States Total w/Rights Freq 1.2 Sequence % 1.2 Sequence Freq 1.2 + OT % 1.2 +OT Asia 53 21 0 0 21 40.00% Africa 58 8 1 1.89 3 5.17% Europe 53 52 6 11.32 39 73.59% The Americas 52 40 1 1.92 34 65.38% Oceania 26 12 0 0 10 38.47% Federal States Total Subunits Australia 9 9 3 33.33 8 88.89% Brazil 27 27 0 0 0 0.00% Canada 13 13 8 61.54 9 69.23% Mexico 5 5 1 20 5 100.00% United States 51 51 4 7.84 36 70.58% Total number of states or subunits that have passed at least one piece of LGBT rights legislation. States missing years for any legislation are not included. 1.2 + OT refers to states or subunits that follow the Hypothesis 1.2 sequence or are on track
33 Hypothesis 1.3 : Sex, Con, AnyDiscrim, PCU, A Table 2. 6 describes the most common world patterns for Hypothesis 1.3, Sex_Con_AnyDiscrim_PCU_A which has 9 cases Included are only those patterns that contain legislation on partnership or civil union (PCU), because Hypothese s 1.3 and 1.4 focus on the sequence where there is a higher threshold for partnership law above cohabitation rights There is an overrepresentation of the most common full rights pattern Sex_Con_AnyDiscrim_A_PCU, on the national level because 15 of 22 cases were in Brazi l, likewise for pattern Sex _Con_A_PCU 12 of 13 cases were in Brazil. Hypothesis 1.3 was found in Europe (7 cases), the Americas (1 case), and the U.S. (1 case). T able 2.6 Hypothesis 1.3: Most Common World Patterns with PCU Freq % of World Numerical Pattern Legislative Pattern 24 6.92 1_2_3_4_N Sex_Con_AnyDiscrim_PCU 22 6.32 1_2_3_5_4 Sex_Con_AnyDiscrim_A_PCU 13 3.75 1_2_N_4_3 Sex_Con_A_PCU 9 2.59 1_2_3_4_5 Sex_Con_AnyDiscrim_PCU_A 5 1.44 1_N_2_3_4 Sex_AnyDiscrim_PCU_A 5 1.44 1_N_2_4_3 Sex_AnyDiscrim_A_PCU 3 0.86 1_N_N_2_N Sex_PCU 3 0.86 1_N_2_3_N Sex_AnyDiscrim_PCU 3 0.86 1_2_4_5_3 Sex_Con_A_AnyDiscrim_PCU 3 0.86 1_3_2_4_5 Sex_AnyDiscrim_Con_PCU_A 3 0.86 1_3_2_4_N Sex_AnyDiscrim_Con_PCU 90 25.9 The numerical pattern is a comparison to the hypothesized order: decriminalization of homosexual sexual activity (Sex), equal ages of consent for sexual activity (Con), any anti discrimination legislation (AnyDiscrim), partnership or civil union legislation (PCU), and joint parent adoption (A), or 1_2_3_4_5. The first slot in the numerical pattern represents decriminalization (Sex), the second slot equal age of consent (Con), the third slot anti discrimination legislation (AnyDiscrim), the fourth slot partnership/civil union (PCU), and the last slot joint parent adoption (A). The number corresponding to each variable space represents the order in which legislation was passed, from 1 to 5. N means no legis lation for that variable space.
34 The 24 cases with 4 of the 5 rights in hypothesized sequence were found most ly in Europe, the Americas, and the United States. The near complete patterns indicate a future trajectory toward the hypothesized sequence. In the exploration of Hypothesis 1.3, one finds, as with previous sub hypotheses, that there are diverse paths to full rights. Of the 43 political units with full rights, 11 different orders of the passage of rights were observed. Instead of a strict order of rights that guarantees completion of a sequence of rights, the picture that emerges from testing Hypothesis 1 provides evidence for a multiplicity of paths to full rights. Hypothesis 1.4 : Sex, AnyDiscrim, PCU, A Table 2.7 describes the most common world p atterns for Hypothesis 1.4 Sex _AnyDiscrim_PCU_A observed in 22 cases Included are only those patterns that contain legislation on partnership or civil union (PCU) For the most common full rights pattern, Sex_AnyDiscrim_A_PCU 20 of 28 cases were found in subnational units in Brazil, Canada, and the U.S. On the national level, the sequence was the most common pattern of full rights, found in 14 states. The pattern with the highest frequency was for three types of legislation in the sequence, with 26 of 28 units found in Europe, the Americas, and the U. S. It may be that these units are headed towa rd completion of the sequence. There were five different paths to full sequence, around 90% of which were the sequence Sex_AnyDiscrim_PCU_A or the pattern Sex_AnyDiscrim_A_PCU. With the exclusion of equal ages of consent (Con) from the sequence, there is less variation in the paths to full sequence that include partnership/civil union (PCU). This perhaps indicates that the difference is in the order of passage for consent (Con). It may not be vital to
35 equalize the age of consent before moving on to other rights legislation. Of the 22 cases in which Hypothesis 1.4 was observed, 1 3 are states and 9 are subnational units. Europe had the highest frequency for the observation, followed by Canada. Table 2.7 Hypothesis 1.4: Most Common World Patterns with PCU Freq % of World Numerical Pattern Legislative Pattern 32 9.22 1_2_3_N Sex_AnyDiscrim_PCU 28 8.07 1_2_4_3 Sex_AnyDiscrim_A_PCU 22 6.34 1_2_3_4 Sex_AnyDiscrim_PCU_A 15 4.32 1_N_3_2 Sex_A_PCU 5 1.44 1_N_2_N Sex_PCU 2 0.58 1_3_2_4 Sex_PCU_AnyDiscrim_A 2 0.58 1_3_4_2 Sex_A_AnyDiscrim_PCU 106 30.55 The numerical pattern is a comparison to the hypothesized order: decriminalization of homosexual sexual activity (Sex), any anti discrimination legislation (AnyDiscrim), partnership or civil union legislation (PCU), and joint parent adoption (A), or 1_2_3_4. The first slot in the numerical pattern represents decriminalization (Sex), the second slot anti discrimination legi slation (AnyDiscrim), the third slot partnership/civil union (PCU), and the last slot joint parent adoption (A). The number corresponding to each variable space represents the order in which legislation was passed, from 1 to 4. N means no legislation for that variable spa ce. Hypothesis 1.5 : PCU, A Hypothesis 1.5, that partnership or civil union precedes joint parent adoption rights, or PCU_A, occurred in 19 cases shown in Table 2.8 Given the 46 observations in which A came before PCU, Hypothesis 1. 5 does not seem to be an accurate description of the most common sequence found in the world currently. However, when the 27 Brazilian subnational units are remo ved, 19 cases of A_PCU remain. This suggests that the frequency of both order s of passage when observed on the national level may be similar for both types of legislation. Still there are 42 cases that have passed partnership but not adoption, so the future trend may sway in favor of the hypothesis. One can also see that there are a handful of c ases that passed both types of legislation concurrently.
36 Table 2.8 Hypothesis 1.5: Most Common World Sequences Regarding Partnership and Adoption Freq % of World Numerical Pattern Sequence Order 46 13.26 2_1 A_PCU 42 12.1 1_N PCU 19 5.48 1_2 PCU_A 6 1.73 3_3* Concurrent 113 32.57 Hypothesized order is partnership or civil union (PCU) followed by joint parent adoption (A), or 1_2. The first slot represents PCU and the second slot A. N means no legislation. **Adoption of legislation within the same year. Hypothesis 1.6 : Symbolic, Substantive The sequence hypothesis Symbolic_Substantive, that symbolic legislation such as decriminalization, equalization of the age of consent, or anti discrimination legislation occurs before substantive legislation, such as same sex partnership or adoption rights, o ccurred in 123 cases. The frequencies for patterns are shown in Table 2.9. There were no cases in which substantive legislation occurred before symbolic legislation Another 138 cases were on track for the sequence by taking the first step toward grea ter rights by passing symbolic legislation. in all of the cases studied that had both types of legislation Table 2.9 Hypothesis 1.6: World Sequences Frequency % of World Numerical Pattern Legislative Pattern 138 39.77 1_N Symbolic 123 35.45 1_2 Symbolic_Substantive 261 75.22 Hypothesized order is symbolic legislation (Symbolic) followed by substantive legislation (Substantive). N means no legislation.
37 Hypothesis 1: General Trends for Legislation Table 2.10 describes general trends in the order of legislation hypotheses 1.1 1.4 offering a more comprehensive lens for exploring Hypothesis 1 and separating observations by legislation type The descriptive statistics encompass all cases, including those with incomplete patterns. The strength of support for sequence rank declines with each successive piece of legislation within each hypothesis 1.1 1.4 from Sex through A, the frequency and p ercent of cases fitting the hypothesized rank falls. D ecriminalization ( Sex ) was passed first in roughly 98% of cases. This was the most observed portion of the hypothesis, that homosexuality is decriminalized before other legislation is passed. In the two hypotheses that included age of consent (Con), Hypotheses 1.1 and 1.3, Con was found in the hypothesized second slot almost 90% of the time. A large proportion of these cases may be explained because Sex and Con are very often passed at the same time, and concurrent legislation was ranked favorably to the hypothesis. The order of passage for any anti discrimination legislation (AnyDiscrim) was more varied. In the sequences that placed AnyDiscrim third, for Hypothesis 1.1 and 1.3, roughly 60% of the time that was the case. However when Con was removed from observation and hypothesized order for AnyDiscrim was second, 77% of the cases fol lowed Hypothesis 1.1, and 91% aligned with Hypothesis 1.4. A ny type of partnership (AnyP) was ranked in the hypo thesized rank the majority of the time it appeared in Hypotheses 1.1 and 1.2 Likewise, PCU in Hypotheses 1.3 and 1.4 was found in the sequence rank for the majority of cases with PCU. Joint parent adoption (A) was the least predicted across every hypoth esis. For Hypotheses 1.3 and 1.4, that raised the
38 threshold for partnership rights from AnyP to PCU, the hypothesized rank for A was not th e most common rank. All hypothesized ranks were the most common rank unless otherwise indicated. For Hypothesis 1.3, the most common rank for A was 4th (41.1%). For Hypothesis 1.4, the most common rank for A was 3rd (39.73%). What does this tell us about sequencing of LGBT rights legislation? Almost always homosexuality is decriminalized (Sex) and/or ages of conse nt are equalized (Con) first. The majority of the time, it appears that anti discrimination legislation is next followed by some type of partnership (AnyP or PCU) The order of j oint parent or second parent adoption rights (A) is the most varied; usually it occurs after Sex and/or Con, but it may precede partnership or anti discrimination in many cases. Again, there is Table 2.10 Hypothesis 1: Frequency of Hypothesized Sequence Ranks Legislation + Sequence Rank Freq in Sequence Rank % in Sequence Rank Total Cases w/Legislation Hypothesis 1.1 1. Sex 216 98.18 220 2. Con 165 87.77 188 3. AnyDiscrim 78 62.4 0 125 4. AnyP 52 46.02 113 5. A 16 25 .00 64 Hypothesis 1.2 1. Sex 230 98.29 234 2. AnyDiscrim 101 77.1 0 131 3. AnyP 69 58.47 118 4. A 32 4.24 74 Hypothesis 1.3 1. Sex 217 98.19 221 2. Con 166 88.77 187 3. AnyDiscrim 83 64.34 129 4. PCU 62 56.88 109 5. A 15 20.55 73 Hypothesis 1.4 1. Sex 232 98.72 235 2. AnyDiscrim 122 91.04 134 3. PCU 71 65.14 109 4. A 25 34.25 73
39 evidence for a variety of paths toward broader LGBT rights, with some patterns that appear more common than others What is most important is that rights are being granted, so getting caught up in the minutia e of sequencing should not distract from the goal of observing overall trends in legislation. While there is not j ust one formula for LGBT rights, t he findings of Hypothesis 1 provide some valuable insight about the progression of LGBT rights legislation. on a global scale is that many countries are still early on the path toward granting broader LGBT rights. In particular, the lack of cases of joint parent adoption (A) made analysis of full sequences difficult. Additionally, the lack of data on when ages of consent were equalized (Con) in many countries complicat ed analyzing sequences. It is also im portant to stress that t here were outliers in every region such as South Afri ca and Israel in Africa and Asia, respectively, which stand out in their regions with multiple LGBT rights Additionally, just because states appear on track for a hypothesized sequence does not mean t hat they will ever extend greater LGBT right s I t is with optimism that the term on track anticipation of future trend s of broader LGBT rights around the world Future research could concentrate on smaller sequences, perhaps encompassing on e to three rights, in order to gain a fuller picture of the patterns occurring around the world. As more rights are passed around the world, further research can gain a better idea of common patterns.
40 HYPOTHESIS 2: NEIGHBORHOOD In Hypothesis 2 the dependent LGBT legislation variables Sex, Con, Emp, AnyP, PCU, M and Fam are analyzed in order to explore their relationship to region and neighborhood By testing observations of the passage of legislation by country in comparison to neighboring possible explanatory power of diffusion in the region and neighborhood can be explored Data for every country possible as well as some subnational units, was drawn upon for Hypothesis 2. Subnational units were i ncluded in analyses when they differed for the year of passage of legislation. Table 2 11 defin es the six neighbor variables chosen for analysis. For ease of countries that have the leg islation. Countries that have legislation are sometimes referred to as countries with a Yes value for legislation, while countries that do not have legislation are sometimes referred to as countries with No value for legislation. Variables BorderBeforePer, BorderAfterPer, and BorderTotalPer are based on the lengths (in kilometers ) of the borders shared between countries, which ar e then converted into a percent, in order to assess border state relationships. By using lengths of borders rather than simply the number of bordering countries, the assumed impact of each neighboring country is adjusted based on the size of the shared border. This approach is based on the assumption that countries that share a longer border experience a greater volu me of interchange culturally and otherwise, than countries with a comparatively shorter border. In addition, island countries were excluded from analysis by virtue of the fact that they border the sea.
41 Table 2.11 Hypothesis 2: Explanatory Variables Variable Description BorderBeforePer Percentage of border shared between country X and neighbors that passed legislation Z before country X BorderAfterPer Percentage of border shared between country X and neighbors that passed legislation Z after or in the same year as country X BorderTotalPer shared border that is occupied by countries that passed legislation Z. The sum of ZBorderBeforePer and ZBorderAfterPer. NeighborTotalPer Total percent of neighbors (out of total number of neighboring countries) that passed legislation Z. RegBeforePer Percent of region al neighbors that passed legislation Z before country X RegAfterPer Percent of regional neighbors that passed legislation Z after or in the same year as country X *Z stands for the type of LGBT rights legislation studied for Sex, Con, Emp, AnyP, PCU, M, and Fam. BorderBeforePer gives an idea of the percentage of the border that is occupied by states that innovated with legislation prior to a country X. As such, it reflects past innovations. One hypothesizes that these neighbors influence country X to adopt legislation. BorderAfterPer gives an idea of the number of countries (as a percentage of the border) that may be influenced by Country X after it has passed the legislation. Since not all countries have adopted all types of legislation, BorderAfterPer only applies to those countries with a Y es value for legislation. One must also note that countries with a Y es value for legislation can have values in both BorderBeforePer and BorderAfterPer, whereas c ountries with a N o value for legislation only can have values for BorderBeforePer.
42 A summary of BorderBeforePer and BorderAfterPer, the variab le BorderTotalPer shows the total percent of the border that is occupied by countries with legislation, which provides the full bordering neighbor context for each country for the present time. Again, the totals for countries with a N o for value legislat ion will reflect only BorderBeforePer. NeighborTotalPer is similar to BorderTotalPer, but not based on border length, it is rather the percentage of countries out of the total number of neighbors that have the legislation. RegBeforePer and RegAfterPer look at regional relationships, rather than border state relationships. Regional groupings are taken from the UN Classification of Regions, and include the Americas, Asia, Oceania, Europe, and Africa. The se variables are based on a percentage of regiona l neighbors as part of the total number of regional neighbors that have passed the legislation, either before or after Country X. R egional neighbors do not necessarily share borders. These variables are similar to BorderBeforePer and BorderAfterPer, but instead describe regional neighborhood effects. RegAfterPer only contains those cases that h ave passed legislation. Island countries ar e included in regional data. Hypothesis 2: Methods In order to see whether there were differences along neighbor variables between countries that did and did not pass the legislation, the Wilcoxon Mann Whitney test was used. The Wilcoxon Mann Whitney test is a non parametric test that is appropriate for two independent samples of non normally distributed data. It is a conservative measurement that t ests whether two sets of observations have d ifferences in the locations of the ir populations (like the one way analysis of variance for normally distributed data). For BorderBeforePer, BorderTotalPer, NeighborTotalPer, and RegBeforePer this test
43 was appropriate for determining whether there were difference s between the values for Yes and No on each type of legislation. The means for Yes and No on each type of legislati on were found for all variables, BorderBeforePer, BorderTotalPer, NeighborTotalPer, RegBeforePer, BorderAfterPer (on Yes), and RegionAfterPer (on Yes). Conclusions for these analyses about neighbo r and regional effects are necessarily suggestive rather th an decisive, because the analyse s look at patterns of regional obser vations on legislation, but do not tease out specific explanatory variables that may be at work, such as interaction levels between neighbor countries. Further testing of interaction levels would prov ide more substantial evidence for neighborhood effects. Hypothesis 2: Results and Discussion Percentage of border with legislation before state: BorderBeforePer The Wilcoxon Mann Whitney tes t values are reported in Table 2 12 for BorderBef orePer for each type of legislation. The differences between Y and N on legislation for BorderBeforePer for Sex and M were not significant, while the differences for Con, Emp, AnyP, PCU, and Fam were significant. So for decriminalization and marriage equivalent partnership the effect of neighboring countries with legislati on prior to country X seemed negligible.
44 Table 2.12 Hypothesis 2: BorderBeforePer Mann Whitney Test Legislation Chi square DF P value # of cases Y # of cases N Total Cases Sex 0.5387 1 0.463 139 35 174 Con 11.21 1 0.0008 129 84 213 Emp 10.22 1 0.0014 80 137 217 AnyP 16.79 1 <.0001 ** 58 157 215 PCU 20.67 1 <.0001 ** 53 163 216 M 2.88 1 .0897 34 181 215 Fam 21.2 1 <.0001 ** 61 154 215 *Results Table 2.13 displays the breakdown for the means for BorderBeforePer by legislation. For Sex, which did not have a significant difference, the means were very similar for Y es and N o Likewise for M, the difference between means was only about 8%. Overall, the means for innovating countries range from around 20% of the border to 40% of the border, which is fairly low. The means for countries without legislation range from 12% of th e border to 35% of the border, which is very low. Table 2.13 Hypothesis 2: BorderBeforePer Means Legislation Yes (Y) No (N) Total Cases # of cases Mean Std Dev # of cases Mean Std Dev Sex 139 41.61 37.73 35 35.93 33.65 174 Con 129 40.11 36.88 84 22.89 29.01 213 Emp 80 29.56 32.82 137 17.63 30.26 217 AnyP 58 36.65 37.35 157 15.52 31.96 215 PCU 53 31.52 35.89 163 12.63 26.91 216 M 34 20.60 31.39 181 12.22 25.40 215 Fam 61 34.50 35.87 154 14.82 27.26 215
45 P ercentage of border with legislation after state: BorderAfterPer The means for Yes on legislation for BorderAfterPer, t he average amount of the border that passed legislation after country X, are presented in Table 2 14. In general, the mean was highest for types of legislation passed early in the standard sequence (examined in Hypothesis 1), and declined for each successive piece of legislation. This makes sense because there are fewer and fewer cases of countries that pass each successive piece of legislation hence potential border lengths decline Fam is the exception, because it combines partnership and adoption so it has no place in the standard sequence T he standard deviations are quite large for each piece of legislation. The means are also substantial, ranging from about 20% to 50% of the border. Of the countries that decriminalized homosexuality, almost fifty percent of their borders passed legislatio n after they di d. For Emp, this was the case for about a third of the border For AnyP, PC U and Fam, this value was around one fourth of the border. Finally, after M was passed, on average one fourth of the border followed suit in the cases observed. Table 2.14 Hypothesis 2: BorderAfterPer Means Legislation Yes (Y) # of cases Mean Std Dev Sex 140 49.53 38.67 Con 128 42.56 38.51 Emp 80 33.71 38.33 AnyP 53 24.78 37.06 PCU 53 26.32 36.09 M 34 19.89 30.84 Fam 61 26.73 37.65
46 P ercentage of border with legislation: BorderTotalPer Table 2.15 shows the statistics for BorderTotalPer by legislation. For all types of legislation, the difference between groups with Y and N on legislation was significant. This perhaps provides evid ence toward the hypothesis, indicating that countries with legislation larger shares of their borders with countries that also have the legislation than do countries that do not have the legislation. To tilt in favor of the hypothesis, the neighbors of Y cases would represent higher percentages of the total border than the N cases. Table 2.15 Hypothesis 2: BorderTotalPer Mann Whitney Test Legislation Chi square DF P value # of cases Y # of cases N Total Cases Sex 100.63 1 <.0001 ** 146 56 202 Con 105.95 1 <.0001* ** 142 82 224 Emp 58.12 1 <.0001 ** 80 137 217 AnyP 50.32 1 <.0001 ** 58 157 215 PCU 57.36 1 <.0001 ** 53 163 216 M 19.8 1 <.0001 ** 34 172 206 Fam 61.91 1 <.0001 ** 61 154 215 Observation of the differences in means for BorderTotalPer can help determine whether hypothesized values charact e rize the data. From Table 2.16 i t appears that the values for means on Y legislation are indeed higher than those for N on legislation. The means for Y range from about 40% to 90%, while the means for N range from about 15% to 35%. These values perhaps indicate that the presence of legislation varies by region in the world, because there appear to be pockets of countries that share the same value for legislati on, evidenced by the high means for Y. In contrast, the N cases have a lower percent of the border that is taken up by neighbors with legislation, so it would make
47 sense, if one is assuming that regional effects are at work, that those countries would not have the same amount of pressure or influence toward passing legislation as Y cases have. Percentage of total neighbors with legislation: NeighborTotalPer The results for the Wilcoxon Mann Whitney test for NeighborTotalPer are displayed in Table 2.17 As mentioned previously, NeighborTotalPer differs from BorderTotalPer by looking at the total percentage of neighbor countries, rather than the amount of shared border between country X and its neighbo rs, essentially giving equal weight to the influence of ea ch neighboring country with legislation. Like BorderTotalPer, the NeighborTotalPer is also significant for every type of legislation. T able 2.16 Hypothesis 2: BorderTotalPer Means Legislation Yes (Y) No (N) Total Cases # of cases Mean Std Dev # of cases Mean Std Dev Sex 146 91.69 21.55 56 36.46 33.56 202 Con 142 86.43 25.68 82 26.21 33.29 224 Emp 80 63.27 37.76 137 20.57 32.46 217 AnyP 58 60.44 38.04 157 18.56 34.68 215 PCU 53 57.83 38.62 163 15.39 29.50 216 M 34 40.50 37.04 172 15.49 28.93 206 Fam 61 62.03 37.86 154 17.32 30.26 215
48 The means for NeighborTotalPer are listed in Table 2.18. In comparison to BorderTotalPer, the means on Y for Sex, Con, and Emp are almost identical, while the means for N for all types of legislation were very similar. For Y in NeighborTotalPer, the means were about 15% higher on AnyP, PCU, M, and Fam than for BorderTotalPer. This change in values is in the direction of the hyp othesis, indicating that a large number of the neighboring countries for Yes cases also have the legislation. Values for Y were very high percentages, at least 50% for every type of legislation. Table 2.18 Hypothesis 2: NeighborTotalPer Means Legislation Yes (Y) No (N) Total Cases # of cases Mean Std Dev # of cases Mean Std Dev Sex 152 91.42 21.10 53 36.29 30.10 205 Con 148 86.20 24.71 71 31.87 31.51 219 Emp 150 64.49 33.50 110 24.95 33.36 260 AnyP 97 75.33 35.38 164 18.38 30.19 261 PCU 58 69.12 35.79 147 15.48 28.22 205 M 49 53.40 39.64 182 13.96 24.98 231 Fam 108 77.49 33.19 154 16.35 27.09 262 Countries with Sex and Con had, on average, nearly 90% of their border shared with innovating countries. For AnyP and Fam the values were around three fourths of Table 2 17 Hypothesis 2: NeighborTotalPer Mann Whitney Test Legislation Chi square DF P value # of cases Y # of cases N Total Cases Sex 105.31 1 <.0001 *** 152 53 205 Con 94.16 1 <.0001 *** 148 71 219 Emp 68.18 1 <.0001 *** 110 150 260 AnyP 106.26 1 <.0001 *** 104 157 261 PCU 72.81 1 <.0001 *** 58 147 205 M 48.77 1 <.0001 *** 49 182 231 Fam 129.41 1 <.0001 *** 108 154 262
49 neighboring countries. The lowest value was for marriage equivalent partnership (M), but the mean value was still quite high at 50% innovating border countries. These variables w ere included in order to see wheth er bordering countries passed similar types of legislation, comparing legislation types in a more general sense. These findings suggest clustering of countries geographically for LGBT rights legislation. Percentage of region with legislation before stat e: RegBeforePer In Table 2.19 the results for the Wilcoxon Mann Whitney test by legislation are reported for RegBeforePer. The differences between Y and N cases by legislation were significant for Sex, AnyP, PCU, and M, and nearing significance for Con. Differences were not signifi cant for anti discrimination legislation, Emp or family legislation, Fam which combines partnership and adoption legislation Significant differences on LGBT rights legislation suggest that the number of innovating regional neighbors for innovating countries is different than the number for countries without legislation. A look at the means will allow elaboration on the details of this difference and provide a clue as to why Emp and Fam did not have significant differences as expected. Table 2 19 Hypothesis 2: RegionBeforePer Mann Whitney Test Legislation Chi square DF P value # of cases Y # of cases N Total Cases Sex 11.93 1 0.0006 ** 179 81 260 Con 3.18 1 0.0744 149 106 255 Emp 91 1 0 .713 6 91 198 289 AnyP 5.39 1 0.0202 68 212 280 PCU 6.06 1 0.0138 62 228 290 M 6.7 1 0.0096 ** 37 253 290 Fam 2.55 1 .1102 71 218 289 *Re ** Results significant,
50 In Table 2.20 one can observe the descriptive statistics for RegionBeforePer. The cases for Y and N for Emp were very similar in both mean and standard deviation. These findings are at odds with the hypothesis, and suggest that for this type of anti discrimination legislation, region al effects are not at play. Likewise, the means and standard deviations for Fam were less than 3 percentage points of difference. Since significant differences were observed for partnership variables, it may be that patterns of adoption explain the lack of difference. Overall, the differences between means were not great on any of the legislation variables, despite findings of significance. This suggests that regional explanations may not be as useful for explaining variations in legislation around the world as neighbor to neighbor explanations. Table 2.20 Hypothesis 2: RegionBeforePer Means Legislation Yes (Y) No (N) Total Cases # of cases Mean Std Dev # of cases Mean Std Dev Sex 179 38.33 25.68 81 47.76 17.08 260 Con 149 36.20 16.89 106 34.29 26.39 255 Emp 91 25.04 20.52 198 20.83 18.48 289 AnyP 68 20.84 15.75 212 17.50 17.86 280 PCU 62 19.76 14.41 228 15.68 16.42 290 M 37 15.28 10.58 253 10.19 11.69 290 Fam 71 21.09 16.55 218 18.35 18.39 289 Percentage of region with legislation after state: RegAfterPer RegAfterPer looks at regional relationships, reporting the percentage of the region that passed legislation afte r innovating country X. Table 2.21 provides the means for Sex, Emp, and Con, divided by region, including the subnational region of the U.S. Only the Y es values on legislation were studied. When broken into regional observations, interesting patterns emerge. Europe is the stand out region, with on
51 average almost 50% of the region for Sex, Con, and Emp passing legislation after an innovating country This tells us that much of the region passed legislation. For the U.S., the percentage mean observed for Sex was high, but this is somewhat misleading because all of the U.S. states were mandated to decriminalize after the 2003 Supreme Court ru ling in Lawrence v. Texas. The Americas had the next highest means after Europe and the U.S. Asia then Oceania followed, with Africa consistently having the lowest means, because very few countries have p assed LGBT rights legislation in that region. Ta ble 2 21 Hypothesis 2: RegAfterPer Means for Sex, Con, and Emp By Region Legislation Sex Con Emp # of cases Mean Std Dev # of cases Mean Std Dev # of cases Mean Std Dev Africa 6 2.59 2.23 7 5.17 3.72 6 4.31 3.23 The Americas 40 38.94 22.87 20 20.19 11.55 15 16.92 11.17 Asia 21 19.50 11.60 18 15.62 10.23 4 2.83 2.44 Europe 50 49.88 28.66 49 49.02 28.05 39 42.05 21.41 Oceania 12 22.15 12.82 6 11.54 6.44 3 3.85 3.85 U.S. 51 53.83 25.42 20 20.19 11.55 13 24.28 14.49 The means for Sex and Con across regions on RegAfterPer ranged from a low of 3% to a high of about 50%, whereas the means for Emp ranged from about 3% to about 40%. The values RegAfterPer declined for each successive piece of legislation for most regions, which makes sense because fewer cases on the whole have passed Emp in comparison with Sex, for example. The levels for RegAfterPer were similar for Asia and Ocean ia for all three pieces of legislation. Europe and the U.S. were closest on Sex, with around 50%. Table 2.22 groups the means for PCU and M on RegAfterPer. Africa and Asia only have 1 case each of the legislation, and Oceania had only on e case for M, so no regional effects can be observed. For the U.S. Europe, and the Americas after passage
52 of PCU in country X, about 15 20% of the region observed the p assage of legislation. These values are very similar. Oceania had a small number of innovating st ates, so its mean was about 7%. For M, Europe had the highest mean at about 18% and the highest number of states with M by far. Comparing with t he U.S. and the Americas does not make sense because there are so few cases, so means of states passing legis lation after the fact will necessarily be low. Table 2 22 Hypothesis 2: R egAfterPer Means for PCU and M b y Region Legislation PCU M # of cases Mean Std Dev # of cases Mean Std Dev Africa 1 0 1 0.00 The Americas 12 14.42 8.96 4 4.81 2.48 Asia 1 0 1 0.00 Europe 22 21.73 12.38 17 17.41 9.45 Oceania 4 6.73 3.68 1 0.00 U.S. 8 20.34 10.1 2 11.70 11.09 Table 2.23 illustrates the last two legislation variable, AnyP and Fam. These variables are grouped together because they are both combination legislation variables and broader in scope than the other legislation variables. Africa and Asia have no observations of either type of legislation. Europe and the U.S. have very similar means for both legislation groups. The Americas have means of about 15% for both types, while Oceania is 10% for both types. Again, comparison between regions is difficult with variation in frequencies of legislation across regions.
53 Table 2.23 Hypothesis 2: RegA fterPer Means for AnyP and Fam by Region Legislation AnyP Fam # of cases Mean Std Dev # of cases Mean Std Dev Africa 1 0.00 1 0.00 The Americas 12 15.86 9.53 13 15.83 8.70 Asia 1 0.00 2 0.94 0.01 Europe 24 23.83 13.64 26 25.77 14.93 Oceania 5 10.00 3.44 5 10 3.44 U.S. 10 22.75 14.33 14 22.69 14.75 G eneral Observations for Hypothesis 2 Most legislation when comparing BorderBeforePer on the Y es and N o values had significant difference s, except for Sex and M Decriminalization legislation is the most common type of LGBT rights legislation in the world, and M is the least common type of le gislation in the world. Decriminalization may not have a border effect compared to other legislation types, because it is not a recent innovation. M is a recent innovation in legislation, but countries may still be early in the sequence of LGBT rights le gislation and not ready to take that step. For both measurements of the bordering states that had legislation, BorderTotalPer and NeighborTotalPer, there were significant differences between the number of bordering states or shared borders with legislatio n for Y es and N o cases on all types of legislation For region, RegionBeforePer had significant or nearing significant results for all legislation besides Emp and Fam. When looking at the means for RegionBeforePer, the differences on Y es and N o for legislation were not very large in comparison with the differences observed in the neighborhood variables. One would expect with the levels of means for No almost as high as for Yes that the innovating countries would be exerting influence on the reg ion
54 toward passing legislati on. In comparison to the border state variable BorderBeforePer, for which the means for No were much lower than for Yes, the regional innovators prior to country X did not appear to put as much pressure on countries that had not innovated. This may indi cate that geographical proximity or perhaps smaller regions have more of an effect on th e diffusion of legislation than broad region As noted above no definitive conclusions about effects can be made, only conjectures about the reasons for the patterns observed. It would be useful to look at before and after passage as a percentage of the total number of neighbors, as a supplement to looking at before and after passage adjusted by border lengths. When looking at the number of countries that passed legislation following an innovating country on BorderAfterPer and RegAfterPer, it is difficult to make conclusions because only the Y es value on legislation was applicable in this case, and only means could be discussed. Further t esting that traced patterns on the state level could provide better insight into the meaning of the descriptive data observed. One question that could be tested is: w as passage of legislation in co untry X associated with a high percentage of neighbor or regional countries that subsequently passed the legislation? Testing this question would require ranking BorderAfterPer and RegAfterPer into ranges and observing the frequencies for each range. On the whole, it seems that neighborhood and regional effect s are at play. Specific country by country analysis is beyond the scope of this thesis, but dissection by regions would also be helpful in order to understand the patterns that are happening at the national level. Teasing out regional differences can he lp explain significant differences found in the Wilcoxon Mann Whitney tests, for example. It may be that similarities in
55 other dimensions among countries that are clustered geographically, such as culture, religion, and development, explain some differenc es rather than or in addition to geographical proximity and exchange across borders. HYPOTHESIS 3: RELIGION Hypothesis 3 looks at the effect of religious heritage on the dependent variables decriminalization of homosexuality ( Sex ) equalized age of consent ( Con ) anti discrimination legislation in employment ( Emp ) and goods and services ( GS ) any type of anti discrimination legislation ( AnyDiscrim ) same sex marriage ( Marriage ) same sex marriage or marriage equivalent partnership ( M ) and joint parent adoption ( A ) Several religion variables were created in order to examine this hypothesis from multiple angles. Countries were categorized by predo minant religious heritage, from religious demographic data in the CIA World Factbook for the year 2012 No subnational units were analyzed. R eligious data was taken at a certain point in time even though religious adherence is not unchanging. However, it is assumed that the religions represented in the present should be fairly represen tative of religious heritage. To count as the predominant religion, a religion had to have fifty percent or more adherents. In the case that there was no single religion that fit this rule, the top two religions were each counted as predominant religions The variable BroadRelig or broad religion, describes the majority religion for each country. BroadRelig includes the religion with fifty percent or more adherents, named in the broadest terms possible. For example, a country with a Coptic Christian In the case that there was more than one predominant religio n, neither of which constituted fifty percent of the population, a
56 co mbination variable was created that links the two religions together with no space in between. For example, a country with large percentage s of Muslim and Christian adhere nts would be co ded MuslimChristian. In order to create more cases for the statistical viability of determining differences between religions the variable BroadRelig1 was created to separate out countri es with dual heritages or more than one predominant religion effectively removing combination variables such as MuslimChristian Thus, this variable only represent s one part of the dual heritage the heritage less represented in the data (not necessarily less represented in the country) was chosen for analysis. Tho ugh this alteration does not entirely reflect the weights of heritage in a country, it is helpful in order to get another perspective on religious effects. If paired with Christian heritage, the non Christian heritage was coded as variable BroadRelig1 si nce there are plenty of Christian cases for analysis For example, a country coded MuslimChristian for variable BroadRelig would be coded Muslim for variable BroadRelig1 When Muslim was paired with Indigenous Beliefs, Indigenous Beliefs was coded as Bro adRelig1 in order to create a statistically viable number of cases for the Indigenous Beliefs category Along the same logic, w hen Buddhist was in combination with another faith, Bu ddhist was coded as BroadRelig1 A total of twenty four cases were altere d from BroadRelig to BroadRelig1. Gen eral Relig, or general religion, is also the majority religion, but more specific in that it describes the denomination of the broader religious category into which the country falls For example, S T his variable was created in order to see whether there were significant differences based on a more specific definition of the
57 predominant religion Again, multiple predomin ant denominations we re coded as such. The variable Gen eral Relig1 was created to separate out dual denominational countries for statistical viability and followed a similar rationale as in BroadRelig1 For example, for a country that had dual heritage that included Protestant or Roman Catholic heritage, the non Protestant and non Roman Catholic cases were chosen for analysis, because there were plenty of cases for those two faiths. ProtRC, the category for dual Protest a nt and Roman Catholic heritage was not separated out because it has a large number of cases. After observing many significant differences between Muslim and Christian countries, the variable XMBroad was created to categorize countries as either Muslim or Christian Countries that did not fit these categorizations were excluded. Additionally, the variable XtianOther separates traditions into the dichotomous categories of Christian (if there was a predominant or shared predominant Christian faith) and Other (if there was no p redominant Christian faith). All countries were included in the XtianOther variable. By looking at country cases from several different angles, including the broadest categorization of religion and more specific denominational diff erences, as well as separating out dual heritage countries and re classifying them into a single heritage, one can gain a better overall understanding of patterns on LGBT rights legislation, as well as patterns of differences between religious heritages. Clearly, separating out dual heritage countries is not truly representing the heritage for such countries, but doing so can provide greater understanding of the relationship betwee n religion and LGBT rights and improve clarity of analysis, since the inclus ion of many dual heritage combinations is somewhat overwhelming.
58 Hypothesis 3: Methods Each of the religion variables BroadRelig, BroadRelig1, GeneralRelig, GeneralReli g1, XMBroad, and XtianOther was tested on each aforementioned legislation variable Sex, Con, AnyDiscrim, Emp, GS, AnyP, PCU, M, Marriage, and A T o determine whether there were significant differences between religions with regards to LGBT rights legislation n on parametric tests were performed, including the c hi square test of independence and Fish est. Pair wise comparisons between religious categories were performed to see whether there were differences on LGBT rights for each religion variable. The chi square test provides a chi square value, a probability val ue and the phi coefficient. The phi coefficient is a measure of effect size, much like a correlation coefficient, that indicates the strength of the relationship between the two religious categories tested. If the cell sizes were too smal l for a valid chi square test, was used s Exact Test can be used for two by two table comparisons when the sample size and cell sizes are small The frequencies for each piece of legislati on were compared to type of religion in order to provide descriptive statistics F or the chi square test, the degree of freedom was always 1, because the religious categories were compared on t wo possible values for legislation, Y es or No. Throughout reporting of results, most attention is paid to variables Sex, Con, AnyP, AnyDiscrim, and A. Since AnyP is not mutually exclusive of PCU, M, and Marriage, and AnyDiscrim is not mutually exclusive of GS and Emp, an overview of these two variable s AnyP and AnyDiscrim, is many times sufficient for looking at differences between religious categories. Attention is also focused on the religious
59 categories with the highest total frequencies, because the goal is to look at differences among major worl d religions. Hypothesis 3: Results and Discussion BroadRelig Broad Religion Tab le 2.24 describes the categories of religious adherence possible for broad religion (BroadRelig) as well the frequencies for each type of legislation by religious category Religious categories in this and following tables are first ordered by the highest total frequency of observations and then by highest frequency for LGBT rights legislation for clarity of analysis. The highest total observations were for Christian and Muslim majority countries, with all other observations by religious category below ten. The of non affiliated persons. Dual heritage countries had eight categories Table 2 24 Hypothesis 3: Broad Religion (BroadRelig) Frequencies BroadRelig Freq Sex Con AnyD Emp GS AnyP PCU M Mar A Christian 142 114 94 60 57 40 41 37 22 9 15 Muslim 51 22 7 2 2 1 0 0 0 0 0 Buddhist 9 0 0 0 0 0 0 0 0 0 0 IndigenousChristian 7 7 1 0 0 0 1 1 0 0 0 MuslimChristian 6 6 3 1 1 1 0 0 0 0 0 None 5 1 1 1 1 0 0 0 0 0 0 NoneChristian 4 4 4 4 4 4 4 3 1 1 2 Indigenous Beliefs 4 3 1 0 0 0 0 0 0 0 0 Hindu Christian 3 3 2 1 1 0 0 0 0 0 0 IndigeneousMuslim 2 2 1 0 0 0 0 0 0 0 0 Hindu 2 0 0 0 0 0 0 0 0 0 0 Jewish 1 1 1 1 1 0 1 1 1 0 1 TaoistBuddhist 1 0 0 1 1 1 0 0 0 0 0 NoneBuddhist 1 0 0 0 0 0 0 0 0 0 0 ShintoistBuddhist 1 0 0 0 0 0 0 0 0 0 0 Total: 239 163 119 71 68 47 47 42 24 10 18
60 What is striking is that for Buddhist, Hindu, NoneBuddhist, and ShintoistBuddhist countries there is no LGBT rights legislation. In Indigenous Beliefs and IndigenousMuslim countries, only the legislation types Sex and Con were observed. HinduChr istian and NoneChristian countries observed Sex, Con, and some kind of anti discrimination legislation. By far, Christian majority countries made up the bulk of countries with LGBT rights legislation across all types of legislation studied. Christian and NoneChristian countries were the only categories to pass all types of legislation. Christian and Muslim countries constituted most of the state s that passed legislation variables Sex and Con decriminalization and equalization of consent However, only a portion of the total number of countries in these religious categories passed the legislation. For variable Sex, all of the HinduChristian, IndigenousChristian, IndigenousMuslim, Jewish, MuslimChristian, and NoneChristian countries had t he legislation, while most of the Indigenous Beliefs countries had the legisla tion. Of these categories, fewer countries overall had Con ; but at least one did in each category. All of the countries in categories NoneChristian, Jewish and HinduChristian passed Con. Only seven of the fifteen religious categories passed any type of anti discrimination legislation. These categories included Christian, Jewish, Muslim, MuslimChristian, None, NoneChristian, and TaoistBuddhist. E ven fewer categories, four, ob served any type of partnership legislation Christian, IndigenousChristian, NoneChristian, and Jewish. These same four categories represented those countries with PCU. Only Christian, Jewish, and NoneChristian had either marriage equivalent partnership ( M ) or joint parent adoption ( A ) Two categories passed same sex marriage (Marriage) Christian with 9 cases and NoneChristian with 1 case.
61 The overall frequencies for each piece of legislation are as expected when viewed as part of the sequencing trend. The highest frequencies for legislation are found for Sex and Con, the second highest group of frequencies for anti discrimination legislation, and the lowest group of frequencies for partnership legislation and adoption. The only Jewish majority country is Israel, which is quite liberal in its LGBT rights, with Sex, Con, Emp, and marriage equivalent partnership (M). As a singular case it will not be emphasized in analysis. Additionally, the TaoistBuddhist case (Taiwan) is unique in that it has not decriminalized homosexuality or equalized consent, but has anti discrimination legislation. For the chi square test, significant differences were found between Christian and Muslim countries on five legislation variables, expr essed in Table 2.25 No other pair of religions had significant differences for the chi square test. For PCU and M, the phi coefficient expresses a weak association while for Sex, Con, and AnyDiscrim the phi coefficient express es a moderate association. Since Christian countries have the highest frequencies of legislation and Muslim countries have lower frequencies of legislation, this difference makes sense. Table 2 25 Hypothesis 3: Significant Results for BroadRelig Chi square Test Legislation Comparison P value DF Phi Coefficient # of cases Sex Christian v. Muslim 23.86 <.0001 ** 1 0.3516 193 Con Christian v. Muslim 37.91 <.0001 ** 1 0.4551 183 AnyDiscrim Christian v. Muslim 28.8 <.0001 ** 1 0.3956 184 PCU Christian v. Muslim 16.31 <.0001 ** 1 0.2899 194 M Christian v. Muslim 8.85 0.0029 1 0.2136 194 Effective sample size.
62 Many cell sizes were too small to perform chi square tests, so was used for pair wise comparisons. Table 2.26 presents s ignificant differences found for several pairs of religious categories on the legislation variables Sex, Con, AnyDiscrim, Emp, PCU, and A. There were no significant differences for M or Marriage. Many of the observations are between religious categories with no legislation and those with legislation such as between Buddhist and C hristian heritage countries All of the significant pair wise differences were between a state with Christian heritage (including dual heritage) and another state, either also of Christian heritage or otherwise. Table 2 26 Hypothesis 3: Significant Results for BroadRelig Fisher's Exact Test Legislation Comparison P value Effective Sample Size Sex Buddhist v. Christian <.0001 ** 152 Con Buddhist v. Christian <.0001 ** 147 Emp Buddhist v. Christian 0.0134 152 AnyDiscrim Buddhist v. Christian 0.0103 142 Con Muslim v. NoneChristian 0.0016 ** 49 PCU Muslim v. NoneChristian <.0001 ** 55 Emp Christian v. Muslim <.0001 ** 194 A Christian vs. Muslim 0.0120 194 AnyDiscrim Christian v. NoneChristian 0.0482 137 Emp Christian v. NoneChristian 0.0299 147 Sex Buddhist v. IndigenousChristian <.0001 ** 16 Sex Buddhist v. Muslim 0.0199 59 Sex Christian v. None 0.0079 ** 148 Con Muslim v. IndigeneousChristian 0.0049 ** 52 Emp Christian v. IndigenousChristian 0.0435 150 Emp Buddhist v. NoneChristian 0.0014 ** 13
63 BroadRelig1 Broad Religion 1 BroadRelig1 separates out dual heritage countries and puts them into broader categories by coding the less common heritage as the main heritage, effectively removing the combination variables and leaving seven major religion variables. Table 2.27 displays the frequencies of religious categories for BroadRelig1 as well as the frequencies for each for all of the legislation variables in analysis. Over 80% of countries fall into either the Christian or Muslim categories. Again, Christian majority countries o verwhelmingly represent the total passage of legislation. All of the categories passed at least one piece of legislation. Table 2 27 Hypothesis 3: Broad Religion 1 (BroadRelig1) Frequencies BroadRelig1 Freq Sex Con AnyD Emp GS AnyP PCU M Mar A Christian 142 114 94 61 58 41 42 38 22 9 16 Muslim 57 28 10 3 3 2 0 0 0 0 0 Indigenous Beliefs 13 12 7 0 0 0 1 1 0 0 0 Buddhist 12 0 0 1 1 1 0 0 0 0 0 None 9 5 5 4 4 3 3 2 1 1 1 Hindu 5 3 2 1 1 0 0 0 0 0 0 Jewish 1 1 1 1 1 0 1 1 1 0 1 Total: 239 163 119 71 68 47 47 42 24 10 18 Buddhist Indigenous Beliefs, and Hindu countries had the lowest proportions of legislati on, with Buddhist nations only passing anti discrimination legislation. The Indigenous Beliefs category had many observations for Sex and Con, and one observation for partnership legislation. Hindu countries had a moderate proportion of cases for Sex, Con, and anti discrimination legislation. Muslim countries had many observations for Sex and Con, with a few observations for anti discrimination legislation. For None countries, there was at least one observation for every piece of legislation.
64 The highest frequencies for Sex were observed i n the Christian, Muslim, and Indigenous Beliefs categories. Of the 13 Indigenous Beliefs states, 12 of them have decriminalized homosexuality. About half of the Muslim countries did so, and about 80% of Christian countries followed suit. Nearly h alf of the None and Hi ndu countries decriminalized. About 65% of Christian countries passed Con, compared to 18% of Musli m countries, and about half of the Hindu and None countries. For types of anti discrimination legislation, most countries were Christian, with a small nu mber of observations in Muslim and Buddhist countries The None category had the next highest frequency, with about half passing anti discrimination legislation. For partnership legislation, most c ountries were Christian, in addition to low frequency observations in None, Jewish, and Indigenous Beliefs. For joint parent adoption (A), most countries were Christian with a singular observation in each of the Jewish and None categories. Table 2.28 describes the significant results for the chi square test, found mainly between Christian and Muslim countries, but also between Christian and Indigenous Beliefs as well as Indigenous Beliefs and Muslim. Weak association was found for AnyDiscrim between C hristian and Indigenous Beliefs as well as M and Sex between Christian and Muslim. There was a moderately strong association for Con, AnyDiscrim, and Emp between Christian and Muslim countries, as well as for Sex between Indigenous Beliefs and Muslim coun tries. On almost every piece of LGBT rights legislation there was a significant difference between Muslim and Christian countries, because Muslim countries hav e far fewer pieces of legislation.
65 Table 2.28 Hypothesis 3: Significant Results for BroadRel ig1 Chi square Test Legislation Comparison P value DF Phi Coefficient # of cases Sex Christian v. Muslim 18.4 <.0001 1 0.3042 199 Con Christian v. Muslim 35.4 <.0001 1 0.4328 189 AnyDiscrim Christian v. Muslim 30.3 <.0001 1 0.3995 190 Emp Christian v. Muslim 24.7 <.0001 1 0.3513 200 GS Christian v. Muslim 17.4 <.0001 1 0.3041 188 AnyP Christian v. Muslim 21.2 <.0001 1 0.3255 200 PCU Christian v. Muslim 18.7 <.0001 1 0.3058 200 M Christian v. Muslim 9.85 <.0001 1 0.222 200 Sex Indigenous Beliefs v. Muslim 7.75 .0054 1 .3352 69 AnyDiscrim Christian v. Indigenous Beliefs 9.77 .0018 1 .2596 145 Effective sample size. Table 2.29 describes significant results for the pair wise comparisons with the except M and Marriage, and they were mainly found between states that have legislation and states that have no legislation. For example, Christian and Indigenous Beliefs differed because there was no legislation for Indigeneous Beliefs on GS and Emp. There were no significant differen ces between Christian and None countries, which both have legislation of every type. Many differences on legislation occurred between Christian and Buddhist countries. This makes sense given the frequencies observed Buddhist countries have very little ri ghts legislation whereas Christian countries represent most of the legislation passed. There were also differences between Muslim and None, since None cases tended to have more legislation passed than Muslim cases.
66 T able 2 29 Hypothesis 3: Significa nt Results for BroadRelig1 Fisher's Exact Test Legislation BroadRelig1 Comparison P value Effective Sample Size Sex Buddhist v. Christian <.0001 ** 155 Con Buddhist v. Christian <.0001 ** 150 AnyDiscrim Buddhist v. Christian 0.0128 145 Emp Buddhist v. Christian 0.0295 155 AnyP Buddhist v. Christian 0.037 155 PCU Buddhist v. Christian 0.0393 155 Con Muslim v. None 0.0355 60 AnyDiscrim Muslim v. None 0.005 ** 66 Emp Muslim v. None 0.005 ** 66 Emp Christian v. Indigenous Beliefs 0.002 ** 156 GS Christian v. Indigenous Beliefs 0.0192 143 Sex Buddhist v. Muslim <.0001 ** 68 Con Buddhist v. Hindu 0.05 16 Con Indigenous Beliefs v. Muslim 0.0298 64 A Christian v. Muslim 0.0069 ** 200 Results sign ** Results sign General Religion GenRelig GenRelig breaks countries down into more specified religious categories in order to observe differences on the denominational level. Combination denominational heritages are also included. For ease of coding, the variable Prot refers to Protestant, RC re fers to Roman Catholic, OO refers to Oriental Orthodox, Orthodox refers to Eastern Orthodox. The largest religious categories for GenRelig were Roman Catholic, Protestant, and Sunni Muslim, followed by Protestant and Roman Catholic (ProtRC) and Eastern Or thodox. With the overview of frequencies and LGBT rights le gi slation for GenRelig in Table 2.30 one observes that rights are mainly concentrated in Protestant and Roman Catholic countries, with legislation also found in moderate frequencies in dual heritage Protestant and Roman Catholic (ProtRC) and None and Roman Catholic
67 (NoneRC) countries, as well as None and Eastern Orthodox countries, respectively. About half of Sunni Muslim countries decriminalized homosexuality, and these countries h ad low frequencies of legislation for anti discrimination legislation. Table 2.30 Hypothesis 3: General R eligion (GenRelig) Frequencies GenPrimRelig Freq Sex Con AnyD Emp GS AnyP PCU M Ma r A Roman Catholic 66 57 52 33 32 24 23 22 10 4 5 Protestant 47 32 23 12 11 9 11 11 9 4 8 Sunni Muslim 45 22 7 2 2 1 0 0 0 0 0 ProtRC 16 15 9 7 6 5 5 4 3 1 2 Eastern Orthodox 12 10 10 8 8 2 2 0 0 0 0 None 7 3 3 3 3 2 2 1 0 0 0 Buddhist 7 0 0 0 0 0 0 0 0 0 0 Indig eneousProt 4 4 3 0 0 0 1 1 0 0 0 Indigenous Beliefs 4 3 1 0 0 0 0 0 0 0 0 Shia Muslim 4 0 0 0 0 0 0 0 0 0 0 NoneRC 2 2 2 2 2 2 2 2 1 1 2 HinduRC 2 2 1 1 1 0 0 0 0 0 0 IndigenousRC 2 2 2 0 0 0 0 0 0 0 0 Indig enous Muslim 2 2 1 0 0 0 0 0 0 0 0 SunniRC 2 2 1 0 0 0 0 0 0 0 0 Musli m00 2 2 0 0 0 0 0 0 0 0 0 Hindu 2 0 0 0 0 0 0 0 0 0 0 Tibetan Buddhist 2 0 0 0 0 0 0 0 0 0 0 Jewish 1 1 1 1 1 0 1 1 1 0 1 SunniOrthodox 1 1 1 1 1 1 0 0 0 0 0 TaoistBu ddhist 1 0 0 1 1 1 0 0 0 0 0 ProtSunni 1 1 1 0 0 0 0 0 0 0 0 ProtHind u 1 1 1 0 0 0 0 0 0 0 0 IndigenousChristian 1 1 0 0 0 0 0 0 0 0 0 NoneB uddhist 1 0 0 0 0 0 0 0 0 0 0 ShiaSunni 1 0 0 0 0 0 0 0 0 0 0 ShintoistBuddhist 1 0 0 0 0 0 0 0 0 0 0 Ibadhi Muslim 1 0 0 0 0 0 0 0 0 0 0 Oriental Orthodox 1 0 0 0 0 0 0 0 0 0 0 Total : 239 163 119 71 68 47 47 42 24 10 18 The categories HinduRC, Indigenous Beliefs, IndigenousMuslim, Indigenous RC, ProtHindu, ProtSunni, and SunniRC only had legislation observed for Sex and Con,
68 while MuslimOO and IndigenousChristian only had legislation observed for Sex. HinduRC, SunniOrthodox and IndigenousProt had Sex and Con in combination with another type of legislation in low frequencies. The TaoistBuddhist country only had anti discrimination legislation. The categories Buddhist, Ibadhi Muslim, Oriental Orthodox, Shia Muslim, ShiaSunni, Sh intoistBuddhist, and Tibetan Buddhist had no LGBT rights legislation. Tab le 2.31 describes significant results for the pair wise chi square tests. Significant differences with weak associations were found between Protestant and Roman Catholic countries for Sex, Con, AnyDiscrim and Emp. Interestingly, there were no significant differences between these categories for partnership legislation or anti discrimination legislation in goods and services (GS). There were differences betwe en Protestant and Sunn i Muslim with moderately strong effect size s for Con, AnyDiscrim, AnyP, and PCU and weak effect size for Emp. However, Protestant and Sunni Muslim countries did not differ significantly on Sex. Between Roman Catholic and Sunni Muslim, moderately strong effect size was found for Sex, Con, AnyDiscrim, Emp, and AnyP. There were a few singular pairs of religious categories that had significance for the chi square test listed in the bottom portion of the table. Worth mentioning is the moderately strong eff ect size for Sex between Protestant and Roman Catholic (ProtRC) and Sunni Muslim, as well as between None and Sunni Muslim. No significant differences were found between the Muslim denominations Shia and Sunni.
69 Table 2 31 Hypothesis 3: Significant Results for GenRelig Chi square Test Legislation Comparison P value DF Phi Coefficient # of cases Sex Protestant v. Roman Catholic 5.48 0.0192 1 0.2203 113 Con Protestant v. Roman Catholic 10.23 0.0014 ** 1 0.305 110 AnyDiscrim Protestant v. Roman Catholic 5.41 0.0201 1 0.228 104 Emp Protestant v. Roman Catholic 7.33 0.0068 ** 1 0.2546 113 Con Protestant v. Sunni Muslim 10.48 0.0012 ** 1 0.3511 85 AnyDiscrim Protestant v. Sunni Muslim 9.7 0.0018 ** 1 0.3358 86 Emp Protestant v. Sunni Muslim 6.81 0.0091 ** 1 0.2721 92 AnyP Protestant v. Sunni Muslim 11.96 0.0005 ** 1 0.3606 92 PCU Protestant v. Sunni Muslim 11.96 0.0005 ** 1 0.3606 92 Sex Roman Catholic v. Sunni Muslim 17.25 <.0001 ** 1 0.396 110 Con Roman Catholic v. Sunni Muslim 39.29 <.0001 ** 1 0.6117 105 AnyDiscrim Roman Catholic v. Sunni Muslim 27.54 <.0001 ** 1 0.5049 108 Emp Roman Catholic v. Sunni Muslim 24.42 <.0001 ** 1 0.4691 111 AnyP Roman Catholic v. Sunni Muslim 19.78 <.0001 ** 1 0.4221 111 Sex ProtRC v. Sunni Muslim 9.5 0.0021 ** 1 0.3979 60 Sex Eastern Orthodox v. Sunni Muslim 4.28 0.0386 1 0.2764 56 Con Eastern Orthodox v. Protestant 4.04 0.0446 1 0.2661 57 AnyDiscrim None v. Sunni Muslim 10.28 0.0013 ** 1 0.4447 52 Effective sample size. Results ** Ta ble 2.32 reports significant pair wise diff of which there were many. Eastern Orthodox and Sunni Muslim differed on several types of legislation, with Eastern Orthodox passing more legislation. Likewise, Protestant and Sunni Muslim countries differed on legislation because Muslim countries had eithe r no legislation or one piece of legislation for each of GS, A, and M. All differences between Buddhist countries and non Buddhist countries make sense because Buddhist countries have no LGBT rights legislation for GenRelig. Eastern Orthodox and Protesta nt countries, which both passed legislation for AnyDiscrim and Emp, differed significantly on those variables. Otherwise, the general pattern is that Christian heritage countries,
70 such as Eastern Orthodox, Protestant, and ProtRC, NoneRC, have significant differences with other non Christian heritages, especially Sunni Muslim and Buddhist. Table 2.32 Hypothesis 3: Significant Results for GenRelig Fis her's Exact Test Legislation GenRelig Comparison P value Effective Sample Size Con Eastern Orthodox v. Sunni Muslim <.0001 ** 52 AnyDiscrim Eastern Orthodox v. Sunni Muslim <.0001 ** 57 Emp Eastern Orthodox v. Sunni Muslim <.0001 ** 57 GS Protestant v. Sunni Muslim 0.0052 ** 85 A Protestant v. Sunni Muslim 0.0057 ** 92 M Protestant v. Sunni Muslim 0.0026 ** 92 Con Buddhist v. Protestant 0.0135 52 Sex Buddhist v. Protestant 0.0009 ** 54 AnyDiscrim Eastern Orthodox v. Protestant 0.039 53 Emp Eastern Orthodox v. Protestant 0.012 59 AnyP ProtRC v. Sunni Muslim <.0001 ** 61 Con ProtRC v. Sunni Muslim 0.002 ** 54 Sex Buddhist v. Roman Catholic <.0001 ** 73 Sex Buddhist v. Sunni Muslim 0.015 51 Sex Hindu v. Roman Catholic 0.064 69 Sex ProtRC v. Protestant .0502 63 Con None v. Roman Catholic 0.0491 72 GS None v. Sunni Muslim 0.0443 52 AnyDiscrim NoneRC v. Sunni Muslim 0.0056 ** 47 General Religion 1 GenRelig1 GenRelig1 differs from GenRelig by separating out countries with dual heritage into one of the main religious denominati ons because the frequencies for such countries were generally low, which makes statistical comparison difficult. Frequencies of religious c ategories are listed in Table 2.33 The religious categ ory ProtRC was not separated because it contained 16 cases. There are 14 main denomi national groups for GenRelig1. Roman Catholic, Protestant, ProtRC, and None categories had legislation of
71 every type studied. The Sunni Muslim, Eastern Orthodox, Indigen ous Beliefs, and Hindu categories had at least four types of legislation each. Oriental Orthodox countries only had decriminalization, and only one Buddhist country had anti discrimination legislation. The denominations Shia Muslim, Tibetan Buddhist, and Ibadhi Muslim had no LGBT rights legislation. Roman Catholic countries had the highest frequencies for legislation for every variable except for Marriage, for which it tied with Protestant, and joint parent adoption (A), for which it was surpassed by Pro testant. Table 2.33 Hypothesis 3: General Religion 1 (GenRelig1) Frequencies GenPrimRelig1 Freq Sex Con AnyD Emp GS AnyP PCU M Mar A Roman Catholic 66 57 52 33 32 24 23 22 10 4 5 Sunni Muslim 48 25 9 2 2 1 0 0 0 0 0 Protestant 47 32 23 12 11 9 11 11 9 4 8 ProtRC 16 15 9 7 6 5 5 4 3 1 2 Eastern Orthodox 13 11 11 9 9 3 2 0 0 0 0 Indigenous Beliefs 13 12 7 0 0 0 1 1 0 0 0 Buddhist 10 0 0 1 1 1 0 0 0 0 0 None 9 5 5 5 5 4 4 3 1 1 2 Hindu 5 3 2 1 1 0 0 0 0 0 0 Shia Muslim 5 0 0 0 0 0 0 0 0 0 0 Oriental Orthodox 3 2 0 0 0 0 0 0 0 0 0 Tibetan Buddhist 2 0 0 0 0 0 0 0 0 0 0 Jewish 1 1 1 1 1 0 1 1 1 0 1 Ibadhi Muslim 1 0 0 0 0 0 0 0 0 0 0 Total: 239 163 119 71 68 47 47 42 24 10 18 Table 2.34 reports the significant results for GenRelig1 chi square tests. There were significant differences with moderately strong effect size between Roman Catholic and Sunni Muslim countries for almost every type of legislation sans M, Marriage, and A, which have low overall frequencies. There were also moderate effect siz e differences for Protestant and Sunni Muslim countries for Con, AnyDiscrim, and AnyP. Protestant and Roman Catholic countries differed for AnyDiscrim and Con, with a larger proportion
72 of Roman Catholic countries having both types of legislation. Eastern Orthodox and Protestant nations differed on Con and AnyDiscrim, for which a larger proportion of Eastern Orthodox countries passed the legislation. Roman Catholic and Indigenous Beliefs differed on anti discrimination legislation because there is no legi slation for Indigenous Beliefs. Sunni Muslim and Eastern Orthodox differed on Sex, for which Eastern Orthodox had a higher proportion of frequencies. Table 2.34 Hypothesis 3: Significant Results for Gen Relig 1 Chi square Test Legislation Comparison P value DF Phi Coefficient # of cases Sex Roman Catholic v. Sunni Muslim 15.2 <.0001 ** 1 0.3960 113 Con Roman Catholic v. Sunni Muslim 36.7 <.0001 ** 1 0.5832 108 AnyDiscrim Roman Catholic v. Sunni Muslim 29.3 <.0001 ** 1 0.5141 111 Emp Roman Catholic v. Sunni Muslim 26.1 <.0001 ** 1 0.4783 114 GS Roman Catholic v. Sunni Muslim 20.3 <.0001 ** 1 0.4271 111 AnyP Roman Catholic v. Sunni Muslim 21 <.0001 ** 1 0.4287 114 PCU Roman Catholic v. Sunni Muslim 19.8 <.0001 ** 1 0.417 114 Con Protestant v. Sunni Muslim 8.66 0.0033 ** 1 0.3136 88 AnyDiscrim Protestant v. Sunni Muslim 10.5 0.0012 ** 1 0.3437 89 AnyP Protestant v. Sunni Muslim 12.7 0.0004 ** 1 0.3657 95 Con Eastern Orthodox v. Protestant 4.67 0.0307 1 0.2837 58 AnyDiscrim Eastern Orthodox v. Protestant 6.63 0.01 ** 1 0.3505 54 AnyDiscrim Indigenous Beliefs v. Roman Catholic 11.2 0.0008 ** 1 0.3869 75 Emp Indigenous Beliefs v. Roman Catholic 10.6 <.0011 ** 1 0.3662 79 AnyDiscrim Protestant v. Roman Catholic 5.41 0.0201 1 0.228 104 Con Protestant v. Roman Catholic 10.2 0.0014 ** 1 0.305 110 Sex Eastern Orthodox v. Sunni Muslim 4.19 0.0407 1 0.2643 60 Effective sample size. Many significant differences were found between denominational GenRelig1 en the magnitude of findings, Table 38 can be referenced in the Appendix on page 147 ; only general findings will be discussed here. There were differences between ProtRC and S unni Muslim, as well as None and Sunni
73 Muslim, on almost every piece of legislation. There were also significant differences on many types of legislation between Buddhist and Roman Catholic, as well as Eastern Orthodox and Sunni Muslim. Buddhist nations, with very little legislation, differed significantly from most of the other denominations, those with higher frequencies in legislation. The only difference between Shia and Sunni nations was on variable Sex, which approached significance. A great numbe r of differences were not observed between Christian denominations of Protestant, ProtRC, Roman Catholic, Eastern Orthodox, and Oriental Orthodox. In fact, most differences were between Christian heritage nations and non Christian nations. There were a handful of differences between None and Christian countries, but dif ferences were most often observed between None and non Christian countries. About a third of the differences were between Sunni Muslim countries and other countries. Christian v s. Muslim Because of the multiple observations of differences between Chris tian and Muslim countries, the religion variable Christian vs. Muslim was formed to look specifically at the dichotomy between countries with major heritage for either religion. Table 2.35 shows the total frequencies for the two categories and frequencies for each LGBT rights legislation variable. For partnership and adoption rights, Muslim countries did not pass any legislation For Sex, Con, and anti discrimination legislation, Muslim cases are observed in low frequencies compared to Christian countrie s.
74 Table 2.35 Hypothesis 3: Christian vs. Muslim Frequencies Religion Freq % of Cases Sex Con AnyD Emp GS AnyP PCU M Mar A Christian 158 74.88 130 104 64 61 43 43 39 22 9 17 Muslim 53 25.12 24 8 2 2 1 0 0 0 0 0 Total 211 100 154 112 66 63 44 43 39 22 9 17 Table 2.36 reveals the significant differences for the chi Exact Test between Christian and Muslim countries. All differences were significant except for Marriage. Alt hough there was no statistical significance, the fact that Christian countries have 9 and Muslim countries have 0 instances of same sex marriage is worthy of note. For Sex, Con, AnyDiscrim, and Emp, the phi coefficient indicates a moderately strong effect size. For GS, AnyP, PCU, and M, the phi coefficient indicates a weak effect size. For A, a significant difference was Table 2 36 Hypothesis 3: Christian vs. Muslim Statistics Legislation P value DF Cramer's Phi Effective Sample Size Chi square Test Sex 26.11 <.0001* 1 0.3526 210 Con 37.88 <.0001* 1 0.4352 200 AnyDiscrim 27.57 <.0001* 1 0.3703 201 Emp 22.99 <.0001* 1 0.3301 211 GS 17.16 <.0001* 1 0.2936 199 AnyP 18.16 <.0001* 1 0.293 211 PCU 16.05 <.0001* 1 0.2758 211 M 8.24 0.0041* 1 0.1976 211 Fisher's Exact Test A NA 0.031* NA NA 211 Marriage NA 0.1159 NA NA 211 5. ** 1. ** 01.
75 Christian vs Other Because of the multiple observations of differences between Christian countries and countries with different heritages, the religion variable Christian vs. Other was formed to look specifically at the dichotomy between countries with Christian heritage versus countries with no Christian heritage. Table 2.37 shows the total frequencies for the two religion categories and frequencies for each LGBT rights legislation variable. Both Christian and Other countries observed legislation for every legislation variable, except for Marriage, for which the only observat ions are in Christian countries. There were lower frequencies on every legislation variable for Other. Table 2.37 Hypothesis 3: C hristian vs. Other Frequencies Religion Freq % of Cases Sex Con AnyD Emp GS AnyP PCU M Mar A Christian 159 66.53 131 105 63 60 42 43 39 22 9 16 Other 80 33.47 31 13 8 8 5 4 3 1 0 2 Total 239 100 162 118 71 68 47 47 42 23 9 18 Table 2.38 displays significant differences for the chi Exact Test between Christian and Other countries. There were significant differences for the chi square on every legislation variable, except for Marriage which was significantly moderate effect size, while for Emp, GS, AnyP, PCU, M and A the effect size was weaker.
76 Table 2.38 Hypothesis 3: Christian vs. Other Statistics Legislation P value DF Phi coefficient Effective Sample Size Chi square Test Sex 45.21 <.0001* 1 0.4358 238 Con 51.28 <.0001* 1 0.4743 228 AnyDiscrim 24.89 <.0001* 1 0.3304 228 Emp 20.11 <.0001* 1 0.2901 239 GS 15.43 <.0001* 1 0.2613 226 AnyP 16.37 <.0001* 1 0.2617 239 PCU 15.86 <.0001* 1 0.2576 239 M 9.69 0.0018* 1 0.2014 239 A 4.37 0.0366* 1 0.1352 239 Fisher's Exact T est Marriage NA 0.031* NA NA 239 General Observations for Hypothesis 3 A summary discussion with regards to frequencies of LGBT rights legislation by religious heritage is helpful for an overview of patterns. Countries with Buddhist and Hindu heritage including dual heritage countries, tended to have the lowest observations of LGBT rights legislation. Countr ies with heritage None had low frequencies, but combination NoneChristian countries had a high proportion of states with legislation. When countries had Indigenous Beliefs heritage, either as a primary or combination heritage, a relatively high proportion of legislation was observed for Sex and Con, but n one or only a singular case was observed in any of the other legislation types. Christian combination heritage countries at least had high proportions of total states with Sex and Con, and many had low fr equency observations of anti discrimination legislation or partnership law. Muslim heritage countries observed moderate frequencies of Sex and
77 Con, and only a handful of cases of anti discrimination l egislation, with no instances of partnership or family legislation. On the denominational level of GenRelig and GenRelig1 the Christian denominations Roman Catholic, Protestant, and combination Protestant and Roman Catholic (ProtRC) had the highest frequencies and proportions of legislation, followed by Eas tern Orthodox, None and NoneRC countries. Shia Muslim and Shia and Sunni Muslim combination (ShiaSunni) countries had no LGBT rights legislation, while Sunni Muslim countries accounted for LGBT rights legislation in Sex, Con, and anti discrimination legislation. Statistical analysis on the pair wise level allowed more specific examination of religious differences th an a broad analysis of differences Some patterns of differences on the pair wise level emerge from analysis. Across BroadRelig and BroadRelig1 there were multiple significant results for the difference between Christian and Muslim countries. Along the same vein f or the denominational religion variables GenRelig and GenRelig1, there were many observed differences between Christian heritage countries such as Protestant, Roman Catholic, and Eastern Orthodox and Sunni Muslim countries. For these variables and in the specific Christian v. Muslim comparison there are definite differences in Muslim and Christian majority nations. There were some differences between Christian denominations Roman Catholic countries had the largest proportion of states across every type of legislation except A, for which Protestant countries exceeded them. Additionally, some differences with weak to low moderate effect sizes were observed between Protestant and Roman Catholic and Eastern Orth odox and Protestant countries For the most part, the Protestant and ProtRC
78 categories had similar proportions of countries with LGBT rights legislation, following Roman Catholic countries in overall highest proportion of legislation. Given this overview of patterns of LGBT rights le gislation and these statistical findings, it appears that Christian countries are the stand out cases for greater LGBT rights legislation. Across every religion variable and type of legislation, there were higher proportions of Christian heritage states ( including combination variables) with legislation than any other religious heritage. This was evident in the Christian vs. Muslim and Christian vs. Other religion variables, which illustrated differences in frequencies as well as statistical differences. Some of the limitations of these analyses include the categorization of countries and other factors. Since countries were categorized by 2012 data, instead of by the religious data at a time in close proximity to the passage of legislation, religious ca tegorizations may not be completely accurate. For example, a country that decriminalized homosexuality in the late 19 th century may have had a different religious majority than is observed today. Additionally, study ing differences across religions is com plicated by the fact that many religions are concentrated geographically. Thus, differences between religions could be due to shared geographical proximity, perhaps the influence of a common regional culture, level of development, or regional interactions rather than inherent differences between religions. Specifically since Christian nations have the highest proportions of legislation, and many Christian countries such as the European cases have high levels of development, these factors may explain so me of the differences.
79 HYPOTHESIS 4: DEMOGRAPHIC AND ATTITIDUNAL VARIABLES Demographic and attitudinal factors were compared to the passage of LGBT rights specifically decriminalization (Sex), equal ages of consent (Con), anti discrimination legislation in employment (Emp) and goods and services (GS), any type of partnership (AnyP), partnership or civil union (PCU), marriage equivalent partnership (M), and joint parent adoption (A) Only states with legislation were analyzed for this hypothes is. Because the argument is that the pattern of change in years prior contributes to the outcome of legislation, data for demographic variables was taken five years before the year in which legislation was passed. Another reason for building in this lag is that it takes several years for change to be translated into legislative action through legislative processes. For example, for a partnership law passed in 1995, the variable for Gender Inequality would be taken for the year 1990. In some cases, there was no data available five years before legislation was passed. In that case data was taken as close to five years as possible between zero and nine years before the legislation in question. Da ta from the World Values Survey (1981, 1990, 1995, 2000, 2005 2008), Human Development Index, and Freedom in the World Reports (1975 2006) were used. The World Values Survey (WVS) was t he source of several variables. The survey question from the WVS about the justifiability of homosexuality asked respondents to rate their attitudes homosexuality from 1 (never justifiable) to 10 (always justifiable). LGBT rights were co mpared to responde always n ever and the mean for the sample. Postmaterialism was likewise included, with percentages of Additionally, the age of respondents in the WVS was used, grouped 15 24, 25 34, 35 44,
80 55 64, 65 and above, as well as the mean age. ( 2002) method was used: t he religiosity variable represents the percentage of people who attend religious services at least once a month, take n from World Values Surveys data on average attendance at church services. Indices related to the Human Development I ndex were used in order to judge commitment to gender equality and levels of education for each country. The Educational Inde x (EI) part of the Human Development Index reflected the mean years of schooling (of adults) and expected years of schooling (of children). The Gender Inequality Index (GII) measures the inequality of women in the dimensions of labor market, empowerment, and reproductive health. These dimensions are further broken down into maternal mortality, adolescent fertility, parliamentary representation, educational attainment, and participation in the labor force (UN Development Program). The Freedom House Fre edom i n the World country ratings were used to capture the democratic features of states (Country Ratings and Status, FIW, 1973 2012) Stat us of each country was recorded as either free (F), partially free (PF), or not free (NF) Ratings for civil liberties (CL ) and political right s (PR ) were recorded on a scale from 1 to 7, 1 representing the freest status for rights and liberties. Hypothesis 4: Methods Since this hypothesis is not comparing differences on the legislation variables because all countries in the data set have passed the legislation, descriptive data is the most useful for analysis. Frequencies and means were used in order to observe trends in the data demographic and attitudinal variables compared to LGBT rights legislation variables When examining means, the mean, standard deviation, and total number of
81 cases are reported. The standard deviation is important for determining whether the m ean accurately characterizes the data across country cases. For frequencies, values for demographic and attitudinal variable are grouped into ranges for ease of analysis, with the frequency, percent, and total number of cases reported for each legislation variable. The ranges of values with the highest frequency are bolded for each type of legislation in the frequency tables. Legislation is (2007) standard sequence, which also tends to proceed from the highest number of cases with legislation (for lower order rights, such as decriminalization) t o the lowest number of cases (for the highest order rights such as adoption). Hypothesis 4: Results and Discussion Justifiability of Homosexuality For the World Values Survey question about the justifiability of homosexuality, homosexuality was never justified (score of 1) and the percent of respondents that said homosexuality was always justi fied (score of 10) were chosen for analysis. The average rating for the justifiability of homosexuality is helpful for gaining an understanding of where respondent attitudes fall and for comparing ratings across countries. The latter two variables provid e a snapshot of the extreme ends of the spectrum of responses. One would expect that countries that passed LGBT rights legislation would have higher average acceptability of homosexuality in society than countries without the legislation. However, one ca n only speculate about the values observed for res homosexuality for societies with legislation for Hypothesis 4.
82 Table 2.39 presents the descriptive data on the means for these three variables for each LGBT rights variable. Overall, means tend to grow larger as one moves down t he legislation column from Sex to A. This perhaps indicates that a higher threshold of acceptance for homosexuality is required for each progressive right. The standard deviations for the never justif ied and always justified ratings were quite large, so the mean is not very representative of the sample cases. Standard deviation was small for the average overall score for the justifiability of homosexuality from around 1 to 2, so the mean provides a b The means for PCU, M, and A are near the midpoint of scores for the justifiability of homosexuality. Emp, GS, and AnyP are around 4, while Con is close to 3 and Sex closer to 2. It is interesting that the means for all of the legislation are in the lower half of the spectrum of the justifiability scale (except for A, which is just above the midpoint). Table 2 39 Hypothesis 4. Homosexuality Justifiable Means Legislation Justified Average % Never Justified % Always Justifed # of cases Mean Std Dev Mean Std Dev Mean Std Dev Sex 2.26 0.983 70.89 18 3.56 3 9 17 Con 3.02 1.272 56.09 20 6 6.37 6 7 30 Emp 4.14 2.291 45.43 23. 4 12.87 12 7 43 GS 3 .99 1.48 41.81 20 7 12.88 9 4 29 AnyP 4.35 1.503 39.02 19 1 16.42 12 7 27 PCU 4.76 1.378 29.01 11 7 21.71 13 2 23 M 4.96 1.882 24.65 13 7 26.01 16 8 15 A 5.43 2.074 23.65 17. 6 28.39 17 1 15 Ba sed on respondent's answer to the question: "Is homosexuality ever justifiable?" rated from 1 (never justified) to 10 (always justified) in the World Values Surveys. Table 2.40 shows the frequencies for ranges of average scores on the justifiability of homosexuality (HJ Avg) by legislation, giving an idea of the distribution of scores. F our groups of ranges encompass the possible ratings in increments of 2 and 3 For Sex, all of the frequencies were either in Group 1 or Group 2, with the highest concentration in
83 the lowest acceptance range Group 1. Con had frequencies in Groups 1 through 3, with the highest concentration also in Group 1. Emp, GS, AnyP, PCU, M, and A had freq uencies in every group. The highest concentration of frequencies for Emp, GS, AnyP, PCU, and M were for Group 2, between 3 and less than or equal to 5. For A, the mode observed was for Group 4, the highest acceptance range, though frequencies were proxim ate in Group 1 and 3. On the whole, cases with anti discrimination legislation and family legislation had higher average acceptance levels for homosexuality than did decriminalization and age of consent. Table 2 40 Hypothesis 4: Average for Homosexual ity Justified (HJ) Frequencies Legislation Group 1 : 3 Group 2 : 5 Group3 : 7 Group 4 : 7 < HJ Avg Total C ases Freq % Freq % Freq % Freq % Sex 12 70.59 5 29.41 0 0 0 0 17 Con 15 50 12 40 3 10 0 0 30 Emp 15 34.88 16 37.21 9 20.93 3 6.98 43 GS 8 27.59 14 48.28 6 20.69 1 3.45 29 AnyP 5 18.52 12 44.44 9 33.33 1 3.7 27 PCU 2 8.7 12 52.17 7 30.43 2 8.7 23 M 2 13.33 7 46.67 3 20 3 20 15 A 4 26.67 2 13.33 4 26.67 5 33.33 15 With the whole spectrum in mind, the clustering of average attitudes toward homosexuality seems low, especially for Sex and Con There were few averages above 7 on the scale. However, since there is no comparison for countries that do not have the legislation, it is difficult to assess whether the averages on the justifiability of homosexuality are higher, as expected by the hypothesis. Future studies could use within country analysis to compare the change over time of attitudes to ward as
84 well as comparing to countries that do not have legislation The low number of total cases, especially for legislation variables Sex, A, and M, also limited analys is. Age Structure To examine age structure, the average age of respondents in the World Values Survey and the perc entage of respondents in the 15 24 age group were examined. One would expect that countries with larger percentages of youth cohorts would have more liberal LGBT rights, since younger people tend to have more liberal attitudes. The 15 24 cohort was c hosen for study because in the five years before the passage of legislation this group would have come of age and begu n to asse rt its political preferences. Table 2.41 reports the means for the age structure on each legislation variable. All of the mea ns for average age fell between 41 and 44, with standard deviations ranging from around 3 to 4.5. There are no striking observat ions, besides perhaps that Sex had the lowest mean for average age. Likewise for the mean percentages of the 15 24 age group, Sex stood out with the highest mean by about 2 percent. The age 15 24 means ranged from around 14 to 17 percent, with standard d eviations between 3 and 5 percent. F indings show no outstanding differences by legislation. Table 2 41 Hypothesis 4: Age Structure Means Legislation Average Age Mean Average Age Std Dev Age 15 24 Mean Age 15 24 Std Dev Total Cases Sex 41.33 2.92 16.91 4 18 Con 43.09 3.01 14.94 3 7 32 Emp 43.94 3.01 14.24 3 3 43 GS 44.31 3.08 13.72 3 4 30 AnyP 43.43 3.55 15 4 2 28 PCU 43.47 3.41 14.58 4 24 M 42.77 4.43 13.87 5 16 A 42.98 3.96 14.69 3 9 15 Mean for the average age of respondent. Mean for the percentage of respondents aged 15 24.
85 Table 2.42 displays the frequencies for the average age of respondents along LGBT rights variables, separated into three groups in increments of about 5 years For every legislation variable except Sex, the lowest frequency was observed in Group 1. The highest frequencies for Sex were tied for Group 1 and 2. The highest frequencies were in Group 2 for Con, Emp, and PCU, with A tying Group 2 and 3. GS and AnyP had the mode in Group 3. For Con, Emp GS, AnyP, PCU, M, and A, the bulk of average ages fell between 40 and 50 in Group 2 and 3. The frequencies for Sex were concentrated in the lower end of average age. Table 2 42 Hypothesis 4: Average Age Frequencies Legislation Group 1: Group 2: Group 3: # of cases Freq % Freq % Freq % Se x 8 44.44 8 44.44 2 11.11 18 Con 5 15. 63 17 53.13 10 31.25 32 Emp 5 11.63 20 46.51 18 41.86 43 GS 3 10 12 40 15 50 30 AnyP 5 17.86 10 35.71 13 46.3 28 PCU 3 16.67 11 45.83 9 37.5 24 M 3 18.75 5 31.25 8 50 16 A 3 20 6 40 6 40 15 Table 2.43 displays the frequencies for the percentage of respondents aged 15 24 along LGBT rights variables, separated into three groups in increments of 10% For every LGBT right, the highest frequencies for the youngest age c ohort fell between 10% and 20% in Group 2, and were all majorities upwards of 60%. In keeping with the hypothesis that there would be a large proportion of young people as a portion of the population there were cases in the largest cohort size, 20% and a bove for Group 3, but this group does not have the highest concentration of state obse rvations. But n either were
86 there a large proportion of cases in the smallest cohort size, bel ow 10% in Group 1, except for M, meaning that the cohort size did not provid e counterevidence for the hypothesis. Table 2.43 Hyp othesis 4: Age 15 24 Frequencies Legislation Group 1: 0%< Age 10% Group 2: 20% Group 3: 20% < Age # of cases Freq % Freq % Freq % Se x 0 0 13 72.22 5 27.78 18 Con 2 6.25 26 81.25 4 12.5 32 Emp 2 4.65 37 86.05 4 9.3 43 GS 3 10 25 83.33 2 6.67 30 AnyP 3 10.71 22 78.57 3 10.71 28 PCU 3 12.5 19 79.17 2 8.33 24 M 4 25 10 62.5 2 12.5 16 A 0 0 13 86.67 2 13.33 15 Much smaller frequencies were observed in Groups 1 and 3 than for Group 2 For Group 1, legislation variables had 0 to 4 observations, and for Group 3, legislation varia bles had 2 to 5 observations. Again, Sex stands out, as about a third of frequencies for the youngest age cohort are in Group 3 whereas for all other legislation in Group 3 the proportion is about less than 15% of total observations. This may relate to the fact that countries early in development have lower average ages, and countries may tend to decriminalize homosexuality earlier in development. In general however, there was no robust support for the hypothesis. Conclusions are d ifficult to make because there is no comparison for countries without the legislation, and the distribution of th e 15 24 age cohort for each country over time is unknown. Future research could study the change of age cohorts over time in order to determin e if legislation was passed when the youngest cohort was largest.
87 P ostmaterialist Attitudes Postmaterialist attitudes measured in the World Values Survey categorized respondents as Materialist, Mixed, or Postmaterialist. The means for each group are s hown in Table 2.44 One hypothesizes that countries with a higher proportion of Postmaterialist attitudes in their populations would be more likely to pass LGBT rights legislation than countries with large concentrations of Materialist attitudes Thus, study will focus on the proportion of Postmaterialist attitudes. Table X. shows the means by legislation for each of Materialist, Mixed, and Postmaterialist attitudes. The mode and highest proportion of observations fall into the Mixed category for every type of legislation. For Sex, Con, Emp, GS, AnyP, and A, the second highest proportion of cases are Materialist. For PCU and M, the second highest proportion is Postmaterialist. As expected under the hypothesis, the majority of respondents are not Materialist, but neither are Postmaterialist cases the most common. Standard deviations are lowest for the Mixed category, with more variation observed in the Postmaterialist category, and the most variation for the Materialist category, as shall be s een in the breakdown of frequencies for each category. Table 2 44 Hypothesis 4: Postmaterialism Means Legislation Materialist Mixed Postmaterialist Total Cases % Std Dev % Std Dev % Std Dev Sex 36.86 15 2 54.11 9 9 9.56 6 9 18 Con 36.25 15 8 53.06 8 6 11.05 9 4 32 Emp 28.46 13 6 57.33 6 9 14.2 9 3 44 GS 26.4 11 9 58.33 5 5 15.27 8 9 32 AnyP 23.97 1 0 58.76 4 9 17.28 7 8 28 PCU 19.9 7 8 59.79 3 8 20.3 7 5 24 M 17 .56 9 8 60.94 5 6 21.48 7 16 A 26.22 14 1 59.09 9 9 14.69 6 2 15
88 Comparing across legislation, Sex and Con have the highest percent of Materialists and lowest percent of Postmaterialists. The cases with anti discrimination legislation and joint parent adoption have the second highest percent of Materialists and second lowest percent of Postmaterialists. Cases with partnership represent the lowest proportions of Materialist respondents and highest proportion of Postmaterialist respondents. This pattern makes sense when viewed as reflective of the sequencing of rights. Countries with low development begin as mainly Materialist soci eties with concern concentrated in basic needs rather than higher order human rights. As they develop they begin to demand more non material goods, such as human rights, transitioning to Post materialist societies. One can conjecture that as postmaterial pref erences develop in society, these preferences proceed from less substantive to more substantive in the area of LGBT rights, beginning first with decriminalization of homosexuality, then an ti discrimination legislation, and family rights. Joint parent adoption seems to defy the pattern, fitting in better with trends for anti discrimination legislation. In Table 2.45 the frequencies for the Materialist (Mat) category with respect to LGBT rights legislation are illustrated, presented in four groups that represent possible frequencies in increments of 20%. For Sex and Con, the bulk of observations fall into Groups 2 and 3, from 20% to 60% Materialist. For Emp, GS, AnyP, PCU, M, and A, the bulk of observations fall into Groups 1 and 2, in the lower levels of Materialist ratings from 0% to 40%. For Emp down the legislation column through A, no cases were above 60% Materialist, while for PCU and M, no cases were above 40% Materialist. For more substantive partnership legislation, it make s sense that there would be fewer Materialist
89 attitudes in society compared to anti discrimin ation legislation, and even fewer Materialist attitudes when compared to the basic steps toward LGBT rights, Sex a nd Con. Table 2 45 Hypothesis 4: Materialist (Mat) Frequencies Legislation Group 1: 20% Group 2: 20% < Mat Group 3: 40% < Mat Group 4: 60% < Mat Total Cases Materialist Freq % Freq % Freq % Freq % Sex 2 11.11 8 44.44 6 33.33 2 11.11 18 Con 6 18.75 12 37.5 12 37.5 2 6.25 32 Emp 15 34.09 19 43.18 10 22.73 0 0 44 GS 12 37.5 14 43.75 6 18.75 0 0 32 AnyP 10 35.71 15 53.57 3 10.71 0 0 28 PCU 13 54.17 11 45.83 0 0 0 0 24 M 11 68.75 5 31.25 0 0 0 0 16 A 5 33.33 7 46.67 3 20 0 0 15 In Table 2.46 the frequencies for the distribution of the Mixed category acros s LGBT rights legislation are depicted in the same four grouped ranges that were used for the Materialist category. There were no observations that were less than 20% for Mixed. Almost all observations were between 40 and 80%, with the majority proportion for every type of legislati on observed in Group 3, from 40 to 60%. Table 2 46 Hypothesis 4: Mixed Frequencies Legislation Group 1: 20% Group 2: 20% < 40% Group 3: 40% < 60% Group 4: 60% < 80% Total Cases Freq % Freq % Freq % Freq % Sex 0 0 2 11.11 10 55.56 6 33.33 18 Con 0 0 3 9.38 23 71.88 6 18.75 32 Emp 0 0 1 2.27 29 65.91 14 31.82 44 GS 0 0 0 0 19 59.38 13 40.63 32 AnyP 0 0 0 0 16 57.14 12 42.86 28 PCU 0 0 0 0 13 54.17 11 45.83 24 M 0 0 0 0 9 56.25 7 43.75 16 A 0 0 0 0 9 60 6 40 15
90 Postmaterialist respondents by legislation are shown in Table 2.47 again grouped into ranges of 20%. For Groups 3 and 4, from 40 to 80%, there were no observations for any of the legislation. The modes for most legislation were in Group 1, with the modes for PCU and M in Group 2. Again, for PCU and M, which are more substantive types of same sex partn ership, there was the highest proportion of cases in the highest group, Group 2, compared to other types of legislation. One would expect that joint parent adoption would also have a higher proportion of cases with large groups of Postmaterialists, but thi s is not the case at least in the context of the small sample size As noted in previous analyses of Hypothesis 4, the study of postmaterialist attitudes would benefit from an overview with a country by country time analysis of changing attitudes compare d to legislative innovations The declining number of cases for each successive piece of legislation was also limiting factor for analysis. Table 2 47 Hypothesis 4: Postmaterialist (Post) Frequencies Legislation Group 1: Group 2: 40% Group 3: 40% < Group 4: 60% < Total Cases Freq % Freq % Freq % Freq % Sex 15 88.24 11.762 11.76 0 0 0 0 18 Con 25 80.65 6 19.35 0 0 0 0 32 Emp 33 75 11 25 0 0 0 0 44 GS 23 71.88 9 28.13 0 0 0 0 32 AnyP 18 64.29 10 35.71 0 0 0 0 28 PCU 11 45.83 13 54.17 0 0 0 0 24 M 6 37.5 10 62.5 0 0 0 0 16 A 11 73.33 4 26.67 0 0 0 0 15 R eligiosity Religiosity from the World Values survey is measured as the percentage of the population that attends religious services at least once a month. One hypothesizes that lower religiosity levels will be associated with broader LGBT rights legislation. Table
91 2.48 gives an overview of the mean religiosity percentages by legislation. The mean was largest for Sex at around 40% religiosity, following by Emp with approximately 35% religiosity and AnyP with around 33% religiosity. The lower means hovered around 30% religiosity for Con, GS, PCU, M, and A. However, it is difficult to make statements about the difference between means on legislation because t he standard deviation for each is quite large, between 18 and 27 percent. It appears that a closer look at the distribution of frequencies will be more telling than simply looking at the means by legislation. Table 2 48 Hypothesis 4: Religiosity Means Legislation Religiosity Mean Std Dev Total Cases Sex 39.69 27.12 16 Con 29.55 19.62 32 Emp 35.51 22.52 43 GS 31.63 18.24 31 AnyP 33.25 22.73 28 PCU 29.25 18.69 25 M 30.31 19.95 16 A 28.59 19.68 14 When broken down into groups of frequencies in 20% increments, one can get a better idea of the distribution of religiosity across legislation, outlined in Table 2 49. For Sex, Emp AnyP, PCU, and A, the mode for religiosity was between 0% and 20% in Group 1. Con had a mode in each of Group 1 and Group 2. GS and M had modes in Group 2. While the majority of cases across legislation fall into Groups 1 and 2, between 0 and 40% reli giosity, d istributions were quite spread out and varied by legislation. Examining the distribution of cases in Group 3 and above illustrates this. For Sex, half of cases had religio sity above 40%, while for Emp two fifths of cases, and for AnyP one third of cases were in Group 3 or higher. For Con and GS, about a fourth of cases had
92 religiosity higher than 40%, while for PCU, M, and A, about a fifth of cases had religiosity higher than 40%. Tab le 2 49 Hypothesis 4: Religiosity Frequencies Legislatio n Group 1: 20% Group 2: 40% Group 3: 60% Group 4: 80% Group 5: 80% < R Freq % Freq % Freq % Freq % Freq % Sex 5 31.3 3 18.75 3 18.75 4 25 1 6.25 Con 12 38.7 12 38.7 4 12.9 2 6.45 1 3.23 Emp 14 32.6 13 30.23 9 20.93 5 11.63 2 4.65 GS 10 32.26 13 41.9 4 12.9 4 12.9 0 0 AnyP 12 42.9 7 25 4 14.29 4 14.29 1 3.57 PCU 12 48 8 32 1 4 4 16 0 0 M 6 37.5 7 43.8 0 0 3 18.75 0 0 A 6 42.9 5 35.71 1 7.14 2 14.29 0 0 Alt hough proportions of cases differ across Groups 1 to 5 when compared by legislation, for the most part, on each legislation variable the number of cases declined for higher levels of religiosity This is characteristic of the assumptions of the hypothesis. The exceptions to this were for PCU, M, and A, for which there were more observations in Group 4 than Group 3. The low number of cases observed for these types of legislation may explain the difference in pattern. Additionally for these legislation variables, there were no religiosity levels above 80%, and on the whole, there were few observations above 80%. It would be helpful in further studies to harness in country comparisons in order to compare changing religiosity values over time with the introduction of legislation. Education Index The Education Index (EI) part of the Human Development Index (HDI ) combines adult and childhood education measurements to provide a score from 0 to 1. A l ower score means less education and lower human development, thus d eveloped countries
93 typically have an education index score of .8 or above. Table 2.50 show s the means for the Education Index. For the most part, from the top to the bottom of the le gislation column, mean EI grows larger, and standard deviation grows smaller. The increase in the value of the mean by successive legislation perhaps indicates that legislation granting higher order rights requires a higher education level for the populace, or that as countries reach higher levels education, and therefore development, they pass increasingly broad LGBT rights legislation. Since there is no comparison for the education levels of countries without legislation, one can only say that this is a possible explanation for countries with the legislation. Table 2 50 Hypothesis 4: Education Index (EI) Means Legislation EI Mean EI Std Dev Total Cases Sex 0.661 0.141 31 Con 0.669 0.138 42 Emp 0.704 0.165 56 GS 0.727 0.144 35 AnyP 0.748 0.095 31 PCU 0.788 0.084 28 M 0.82 0.099 22 A 0.837 0.098 14 The frequencies for the Education Index by legis lation are displayed in Table 2.51 with four groups in increments of .2 All countries had a score of at least .2, except for the outlier Mozambique, which had a score lower than .2 for Emp and is excluded from the table For Sex, Con, Emp, GS, AnyP, and PCU the modes and the majority proportion of observations were in Gro up 3. For M and A, the modes and majority proportion of obser vations were in Group 4. All of the observations for PCU, M, and A were in either Group 3 or 4. As substantive partnership rights, it makes sense that the cases exhibiting these rights would h ave education levels in excess of .6. The
94 observations of frequencies seem to fit the hypothesis well, that higher levels of education typify countries with LGBT rights, with cases of substantive family rights exhibiting even higher education levels. As aforementioned, analysis could be supplemented with temporal within country analysis to facilitate greater understanding of patterns. Table 2.51 Hyp othesis 4: Education Index (EI) Frequencies Legislation Group 1: .2 .4 Group 2: .4 .6 Group 3: .6 .8 Group 4: .8 < EI # of cases Freq % Freq % Freq % Freq % Sex 1 3.23 5 16.13 22 71 3 9.68 31 Con 2 4.76 5 11.9 32 76.2 3 7.14 42 Emp 3 5.36 5 8.93 34 60.7 13 23.21 56 GS 2 5.71 0 0 25 71.4 8 22.86 31 AnyP 0 0 3 9.68 20 64.5 8 25.81 28 PCU 0 0 0 0 15 53.6 13 46.43 14 M 0 0 0 0 8 36.36 14 63.6 22 A 0 0 0 0 4 28.57 10 71.4 35 Gender Inequality Index The Gender Inequality Index (GII) from the HDI reflects the percent loss in achievement due to gender inequality, and as such, a lower percent value r eflects more equality. Table 2 52 show s the means for GII by type of legislation. For the most part, fro m the top to the bottom of the legislation column, mean gender equality becomes larger, and standard deviation becomes larger. The range of difference in standard deviation is only about 4% from Sex through A Differences on means across legislation are quite considerable, for example with Sex and Con ranging from 40 to nearly 50% inequality, anti discrimination legislation around 30% inequality, and M and A around 20% inequality. With standard deviations in excess of 10%, a breakdown of the distribution for GII for all legislation is clarifying.
95 T able 2 52 Hypothesis 4: Gender Inequality Index (GII) Means Legislation GII Mean GII Std Dev Total Cases Sex 47.83 12 4 21 Con 40.01 16 8 27 Emp 30.44 16 8 46 GS 28.33 14 6 30 AnyP 26.8 14 5 23 PCU 23.88 14 7 23 M 20.91 14 5 20 A 20.93 15 8 14 Four groups, in 20% increments from greatest equali ty to lowest equality in Table 2.53 illustrate the frequencies for GII. The majority of cases for Sex had inequality levels in Group 3, while most cases for Con fell into Groups 2 and 3. For Emp, GS, AnyP, PCU, M, and A, most cases fell into Groups 1 and 2, at the lowest levels of inequality. For PCU, M, and A, the majority of cases were in the lowest gender inequality bracket, Group 1. T able 2.53 Hypothesis 4 : Gender Inequality Index (GII) Frequencies Legislation Group 1: 20% Group 2: 40% Group 3: 40% < GII Group 4: 60% < GII # of cases Freq % Freq % Freq % Freq % Sex 0 0 5 23.81 11 53.38 5 23.81 31 Con 4 14.81 9 33.33 10 37.04 4 14.81 42 Emp 18 39.13 16 34.78 9 19.57 3 6.52 56 GS 12 40 11 36.67 6 20 1 3.33 35 AnyP 8 34.78 11 47.83 4 17.39 0 0 31 PCU 12 52.17 7 30.43 4 17.39 0 0 28 M 14 70 3 15 3 15 0 0 22 A 10 71.43 2 14.29 2 14.29 0 0 14 It appears that for countries with decriminalization and equalization of the age of consent, high gender equality was not the norm. For anti discrimination legislation and any type of partnership, gender inequality was typically less than 40%, and for more substantive partnership and adopt ion rights gender inequality was typically less than
96 20%. These observations are in line with the hypothesis that greater gender equality is associated with greater LGBT rights. Freedom in the World: Status, Civil Liberties, and Political Rights From t he Freedom in the World Reports, the freedom status, political rights rating, and civil liberties rating s are examined for each state in comparison to legislation One hypothesizes that case s with broad LGBT rights legislation are free states, with overal l Free status bolstered by high levels of civil liberties and political rights. Table 2.54 shows frequencies for freedom status of countries by legislation of the three groups including Free, Partially Free, or Not Free For every variable except Sex, Free states made up the majority of cases for legislation. Of cases with Sex and Con, there were examples in every group. For Emp, GS, AnyP, PCU, and M, there were only Free and Partially Free case ratings. For A, all legi slation occurred in Free states. Ta ble 2 54 Hypothesis 4: Freedom House Freedom Status Frequencies Legislation Freq Free % Free Freq Partially Free % Partially Free Freq Not Free % Not Free Sex 20 45.45 17 38.64 7 15.91 Con 29 53.7 17 31.48 8 14.81 Emp 43 74.14 15 25.86 0 0 GS 32 82.05 7 17.95 0 0 AnyP 31 91.18 3 8.82 0 0 PCU 28 93.3 2 6.67 0 0 M 22 95.65 1 4.35 0 0 A 15 100 0 0 0 0 From these observations of states with legislation, countries with varying freedom status were able to decriminalize and equalize the age of consent. However, for anti discrimination legislation and family rights, countries were at least Partially Free an d
97 mostly Free. These reflections follow the hypothesis, since the most LGBT rights were concentrated in countries with concern for democratic principles. The means for civil liberties (CL) and political rights (PR) are co mbined in Table 2 55. The possible scores for each of CL and PR range from 1 to 7, with a score of 1 representing the broadest civil liberties and political rights. Thus, a lower score means greater rights and liberties. All of the means for CL and PR in countries that had legislation were below 3.5. As observed for the Education Index and Gender Inequality Index, the change in means down the legislation column reflected hypothesized sequencing. From Sex down to A average ratings for CL and PR ranged from the least free ratings to the most free ratings. The highest values were observed for Sex and Con for both CL and PR, with scores between about 3 and 3.5. Average ratings were for the most part below 2 for CL and PR on legislation variables GS, AnyP, PCU, M, and A. The standard deviations for all means were at maximum about 2 and at minimum a fraction of 1, which indicates fairly narrow distributions for the bulk of cases, especially for anti discrimination legislation and family rights. Table 2 55 Hypothesis 4: Freedom House Civil Liberties and Political Rights Means Legislation Civil Liberties Mean* Civil Liberties Std Dev Political Rights Mean** Political Rights Std Dev Total Cases Sex 3.43 1.7 3.34 1.98 44 Con 3.06 1.85 2.81 2.06 54 Emp 2.12 1.06 1.83 1.22 58 GS 1.9 1.02 1.59 1.09 39 AnyP 1.59 0.82 1.29 0.68 34 PCU 1.5 0.73 1.27 0.69 30 M 1.43 0.73 1.26 0.69 23 A 1.53 0.64 1.13 0.35 15
98 Frequencies for political rights (PR) and civil liberties (CL) are displayed jointly in T able 2.56 for each type of LGBT rights legislation. Of four total groups, the first group represents the countries with the greatest rights, proceeding to the fourth group with the least rights. The greatest rights score, 1, was given a separate group, G roup 1, in order to better discern the distribution of cases in the group with the highest threshold for rights and liberties. Political rights (PR) and civil liberties (CL) will be analyzed separately for this section. Table 2 56 Hypothesis 4: Freedo m House Political Rights (PR) and C ivil Liberties (CL) Frequencies Legislation Group 1: PR = 1 Group 2: PR = 2 or 3 Group 3: PR = 4 or 5 Group 4: PR = 6 or 7 # of cases Political Rights Freq % Freq % Freq % Freq % Sex 10 22.73 14 31.82 12 27.27 8 18.8 44 Con 24 44.44 11 20.37 10 18.52 9 16.67 54 Emp 36 62.07 14 24.14 8 13.79 0 0 58 GS 28 71.79 7 17.95 4 10.26 0 0 39 AnyP 27 79.41 6 17.65 1 2.94 0 0 34 PCU 25 83.33 4 13.33 1 3.33 0 0 30 M 19 82.61 3 13.04 1 4.35 0 0 23 A 13 86.67 2 13.33 0 0 0 0 15 Legislation CL = 1 CL = 2 or 3 CL = 4 or 5 CL = 6 or 7 # of cases Civil Liberties Freq % Freq % Freq % Freq % Sex 6 13.64 19 43.18 14 31.82 5 11.36 44 Con 14 25.93 20 37.04 14 25.93 6 11.11 54 Emp 20 34.48 31 53.45 7 12.07 0 0 5 8 GS 17 43.59 19 48.72 3 7.69 0 0 39 AnyP 19 55.88 13 38.24 2 5.88 0 0 34 PCU 18 60 11 36.67 1 3.33 0 0 30 M 15 65.22 7 30.43 1 4.35 0 0 23 A 8 53.33 7 46.67 0 0 0 0 15 For political rights (PR), the majority of cases for GS, Emp, AnyP, PCU, M, and A were in Group 1, with the greatest political rights. Sex and Con had wider distributions, with observations across all levels of PR, whereas there were no observations in Group 4, the lowest range of political rights, for Emp, GS, AnyP, PCU,
99 M, or A. Sex had the widest distribution of frequencies, with about two thirds of observations in Groups 2 and 3, while Con had almost 45% of cases for Con in Group 1. Civil liberties (CL) had AnyP, PCU, M, and A with majority proportions of cases in Group 1, with CL = 1. Near majorities of Emp and GS were in Group 2, CL = 2 or 3, an d the modes for Sex and Con were also located in Group 2. Cases for Sex and Con were spread out among all groups, while Emp, GS, AnyP, PCU, M, and A did not have any cases in Group 4, CL = 6 or 7. It makes sense that states with very low political right s would not have higher order LGBT rights legislation such as anti discrimination legislation and family law, whereas states with low political rights could still decriminalize and equalize ages of consent. An authoritarian regime might extend these most basic LGBT rights without going any further. Also, the concentrations of partnership and family rights in Group 1 for CL and PR is perhaps substantiated because it is only possible to advocate for these broad LGBT rights when political rights and civil li berties are highly respected. General Observations for Hypothesis 4 From analysis of Hypothesis 4, one can begin to construct a hypothetical picture of the demographic and attitudinal features of countries that have LGBT rights legislation For all of the features studied, one cannot draw definitive conclusions from the analysis of frequencies and means, but some patterns can be dissected. In general, t he a verage age of the population was between 40 and 50 years (except for Sex), and the a verage size of the 15 24 age cohort was between 10 and 20%. The majority of the populations with LGBT rights legislation had Mixed attitudes on the postmaterialism scale. The proportion of Materialist respondents was generally concentrated from 20 to
100 40% while the proportion of Postmaterialist respondents was exclusively concentrated from 0 to 40%. For the majority of cases by legislation, religiosity levels were lower than 40% and education levels were higher than .6. Most states with LGBT rights leg islation had Free freedom status, the freest rating for political rights, and a rating in the two freest groups for civil liberties. For anti discrimination and partnership legislation, mean attitudes on the justifiability of homosexuality were mainly between 3 and 5 on a scale of 10. Sex and Con had most averages below 5 on the 10 point scale. With regards to average age, Sex stood out with frequencies concentrated in the youngest gro ups, from age 34 to 45. Additionally, anti discrimination, partnership, and family types of legislation for the most part had l ess than 40% g ender inequality, while decriminalization and consent had a high proportion of cases with over 40% gender inequ ality. One of the most interesting findings was the reflection of the sequencing pattern that was found again and again when examining several of the demographic and attitudinal variable s Legislation was generally presented in the standard sequence ord er, beginning with decriminalization (Sex), then equal ages of consent (Con), anti discrimination legislation (Emp and GS), partnership legislation (from the least rights to the most rights, AnyP, PCU, M), and joint parent adoption (A). As one moved throu gh the sequence as presented in the summary tables, the values grew in the direction of the hypothesis. T he m ean for the average score on justifiability of homosexuality grew larger wi th successive legislation, the mean for level of religiosity grew small er with successive legislation, the mean for level of education grew larger with successive legislation, the means for civil liberties and political rights grew smaller with successive
101 legislation, and the mean for level of gender inequality grew smaller w ith successive legislation. One of the main limitations for Hypothesis 4 was in collecting data at a point five years before the passage of legislation. In many of the data sources, data was not available for every year, so data was collected as close to five years before as possible Many states decriminalized homosexuality and equalized the age of consent in a time period that was not available for analysis, so the demographic and attitudinal variables reflected in the data may not be completely repr esentative. The findings for variable Sex are for countries that have decriminalized for the most part within the last 25 years, as well as to a lesser extent those that equalized age of consent (Con) in that time period Many countries with more advance d LGBT rights regimes decriminalized much earlier, so demographic and attitudinal variables are not included for those countries. Evidenced from some of the findings in Hypothesis 4, the countries that have decriminalized homosexuality and equalized age of consent in recent years may be of a different breed than countries that decriminalized previously. Much of the analysis pointed out differences between conditions for these less substantive LGBT rights and for more substantive LGBT rights legislation, such as anti discrimination and family legislation. Many of the countries granting higher order rights such as anti discrimination legislation, partnership, and adoption are not represented in analysis of legislation on homosexuality and consent, whe reas those who legalize Sex and Con in recent years have not granted higher order rights yet. Only a small number of cases were under review, ranging from 14 to 60, and cases were limited to those that passed LGBT rights legislation. For many of the
102 demogra phic and attitudinal variables, more in depth analysis would be facilitated by within country comparisons across time compared to legislation. Perhaps states could also be paired with states that have similar characteristics on all but the LGBT rights leg islation in question, such that comparisons could be made between countries with and without legislation. The issue of endogeneity is especially relevant for the study of the justifiability of homosexuality. The direction of causality is unknown it is u ncertain whether more accepting attitudes toward homosexuality would facilitate greater LGBT rights, or whether more accepting attitudes are a symptom of greater rights. While the five year lag was an attempt to consider causal order, data problems and la ck of longer time series make the results inconclusive. HYPOTHESIS 5: LGBT ORGANIZATIONS For Hypothesis 5 data from the International Lesbian, Gay, Bisexual, Trans, and Intersex Association (ILGA) directory on LGBT organizations was used. Three variables refle ct LGBT organizations directory (including LGBT supportive organizations) and all organizations adjusted for population. The first variable includes numeric ranges (0, 1 5, 6 20, 21 50, etc. ) of organizations that are specifically devoted to LG BTI issues, designated by ILGA, AllLGBTIOrgs. Though I directory did not separate these organizations from LGBT organizations in their list However, since intersex issues are part of the transnational organization that is also advocating for sexual and gender identity issues, one hypothesizes that more intersex organizations would be beneficial for advancing LGBT rights. The total counts f or each
103 country also include privately listed organizations, since ILGA allows organizations to choose whether they want to be publicly or privately listed. Organizers in countries where homosexuality is still a crime may not wish to advertise their infor mation. Because of the fact that private organizations were not searchable, it was not possible to create a variable that reflected the exact number database, whereas information on the range of organization sizes was available The total number of organizations in the ILGA directory includes allied organizations, human rights organizations, commercial organizations, etc. in addition to LGBT organizations. This variable AllDirectoryOrgs, reflects the amount of suppo rt for L GBT communities and is expressed in specific numbers of organizations rather than ranges. The third variable AllOrgsbyPop, is the total number of directory organizations divided by the population of each country in order to estimate the density o f organizations across the population This variable attempts to make up for the fact that comparison of the total number of organizations is difficult between countries with varying populations. The analysis for Hypothesis 5 is necessarily exploratory and will not merit certainty with respect to its conclusions, since major limitations exi st. For example, the data used for organization representation i s for the present time period. Unlike for Hypothesis 4, where demographic and attitudinal measures were gathered for points in time five years prior to the passage of legislation, limitations prevented such data collection. It is highly likely that countries in the past do not have the same number of LGBT organizations as they do today Additionally, the ILGA database only contains organizations that subscribe to it, so organizations having impacts on policy at the
104 domestic level may not be included. For a v ariety of reasons, LGBT organizations may not be part of ILGA, including the danger for such organizations in regions of the world that still criminalize or even exact capital punishment for homosexuality. Despite these limitations, one can hypothesize th at the ranges of values for LGBT and allied organizations indicate the trajectory from which countries have com e as well as potential bases of support for promoting legislation. Hypothesis 5: Methods The methods for Hypothesis 5 will employ descriptive data analysis and the Wilcoxon Mann Whitney test. For all LGBTI organizations, or AllLGBTIOrgs, only descriptive data on overall frequencies and frequencies by legislation could be assessed because the data was in range format. For all ILGA directory o rg anizations (AllDirectoryOrgs) and all ILGA directory organizations by p opulation (AllOrgsbyPop), the overall frequencies and means for each are assessed, as well as the Mann Whitney test performed for each type of legislation. The Wilcoxon Mann Whitney te st is a rank sum test that is appropriate for two independent samples, Yes and No values on legislation, to test whether the samples differed in size across ILGA directory organizations.
105 Hypothesis 5: Results and Discussion LGBTI Organizations For the total number of LGBTI organizatio spells out the frequencies for ranges. The highest frequencies were observed for states with no LGBTI organizations, followed by countries with 1 to 5 organizations A fair proportion had 6 to 20 organizations, while few countries, only 24 or about 11%, had over 21 organizations in the database. The United States was the outlier, with over 800 organizations devoted to LGBTI issues. Table 2 57 Hypothesis 5: ILGA All LGBTI Organizations Frequencies # of LGBTI Orgs Freq % 0 89 39.91 1 to 5 80 35.87 6 to 20 30 13.45 21 to 50 7 3.14 51 to 100 9 4.04 101 to 300 7 3.14 8 00 + 1 0.45 Total: 223 100 In Table 2.58 the frequencies for all LGBTI organizations are expressed in comparison to LGBT rights legislation. For both Sex and Con, there were large frequencies of countries with legislation that had 0 or 1 to 5 organizations. This perhaps indicates that the num ber of LGBTI organizations has little impact on those types of legislation. It would make sense that little impact was seen for this analysis, especially because many countries decriminalized or equalized age of consent many years ago. Additionally, ther e may be other reasons besides pressure from LGBT organizations that countries pass Sex and Con. For all other types of legislation, the frequencies for
106 countries with 0 L BGTI organizations were below 10. This perhaps indicates that a greater presence of LGBTI organizations is necessary to promote policy change on substantive rights above the mere acceptability of the existence of homosexual citizens. The size of the presence of organizations may vary widely, but the support of LGBT advocacy groups appea r s to be necessary. Because of the flaws of using the ILGA database, it is also possible that countries with legislation have organizations, but for one reason or another these organizations do not subscribe to ILGA. Table 2 58 Hypothesis 5: All LGBTI Organizations Frequencies by Legislation Legislation Y/N 0 1 to 5 6 to 20 21 to 50 51 to 100 101 to 300 Over 800 Total Cases Sex Y 40 42 22 6 9 7 1 127 N 39 33 8 1 0 0 0 81 All 79 75 30 7 9 7 1 208 Con Y 35 38 22 5 8 5 0 113 N 50 41 8 2 1 2 1 105 All 85 79 30 7 9 7 1 218 Emp Y 5 27 5 5 6 5 0 61 N 57 51 2 2 3 2 1 133 All 62 78 7 7 9 7 1 194 GS Y 2 15 9 4 5 5 0 40 N 60 63 21 3 4 2 1 154 All 62 78 30 7 9 7 1 194 AnyP Y 9 10 8 3 5 7 0 42 N 80 70 22 4 4 0 1 181 All 89 80 30 7 9 7 1 223 PCU Y 9 6 8 3 5 6 0 37 N 80 74 22 4 4 1 1 186 All 89 80 30 7 9 7 1 223 M Y 1 3 6 3 5 5 0 23 N 87 77 24 4 4 2 1 199 All 88 80 30 7 9 7 1 222 A Y 3 2 4 2 3 3 0 17 N 86 78 26 5 6 4 1 206 All 89 80 30 7 9 7 1 223
107 A focus on countries without legislation is helpful for discerning patterns in the numbers of LGBT I organizations for countries that did not pass legislation because one would expect that lower numbers of organizations would characterize countries without legislation Across legislation, few N cases had organization frequencies in excess of 21. The b ulk of observations for Sex, Con, and Emp had between 0 and 5 organizations. For GS, AnyP, PCU, M, and A, organization counts were mainly from 0 to 20 about 90% of case s With few observations with large numbers of organizations, it appears that countries without legislation may not have the level of advocacy needed for policy change. A focus on countries with legislation is helpful for discerning patterns in the numbers of LGBT I organizations for countries that passed legislation because one wou ld expect that legislation would be facilitated with higher numbers of LGBT organizations For Con, most of the organization frequencies ranged from 0 to 20. For Emp the largest proportion of cases about half, was observed in the 1 to 5 organization ra nge, with quite a bit of variability across the other sizes. For GS cases were concentrated in countries with 1 to 20 organizations, for AnyP from 0 to 20 o rganizations. Both M and A had a wide degree of variability in range sizes. One also observes a w ide distribution of cases across the upper ranges of organizations. Al l Directory Organizations and All Directory Organizations by Population Table 2.59 shows the frequencies for all organizations in the ILGA directory. As was the case for total LGBTI organiz ations, the most frequent cases have 0 organizations, followed by the groups with 1 to 5 and 6 to 20 organizations. There are fewer countries with no organizations compared to AllLGBTIOrgs since the directory includes allied
108 organizations and busi ness. Nearly 90% of states had between 0 and 20 organizations, with the remaining 10% had above 20 organizations. Table 2 59 Hypothesis 5: All ILGA Directory Orgs Frequencies # of ILGA Directory Orgs Freq % 0 85 37.95 1 to 5 80 35.71 6 to 20 30 13.39 21 to 50 11 4.91 51 to 100 8 3.57 101 to 300 9 4.02 9 00 + 1 0.45 Total 224 100 Table 2.60 displays the ILGA directory organizations when adjusted for population. The mode was for 0 organizations. About a fourth of observations fell into the .0001 to .001 (1E 2 to 1E 3) and the .00001 to .0001 (1E 3 to 1E 4) groups, respectively. The remaining 10% of observations were in the lowest percentage group (besides 0 obse rvations). Table 2 60 Hypothe sis 5: All ILGA Directory Orgs b y Population Frequencies % Density of LGBT Orgs by Population Freq % 0 85 38.81 .000001 to .00001 23 10.5 .00001 to .0001 53 24.2 .0001 to .001 58 26.48 The means for both AllDirectoryOrgs and Al l OrgsbyPop are listed in Table 2.61 For AllDirectoryOrgs, the outlier, the United States with 961 organizations was excluded so that the mean and standard deviation were not disproportionately affected by one case. The mean for AllDirectoryOrgs falls around 13, but a standard deviation o f 34 renders
109 the mean not very useful for describing the cases. The mean for ILGAOrgsbyPop had a very low standard deviation, with percentage of the population served by ILGA directory organizations around one ten thousandth of a percent. Table 2 61 Hypothesis 5: All ILGA Directory Orgs and All ILGA Directory Orgs By Population Means Variable Mean Std. Dev Total Cases ILGA All Directory Orgs 13.126 34.837 224 ILGA Orgs by Pop 0.0001% 0.0000029 224 Excludes the outlier 961 for the U.S. In T able 2.62 one can see the breakdown of means for AllDirectoryOrgs by LGBT rights legislation. There is a clear difference across all legislation types between means for Yes and No. However, standard deviations for the Yes cases were huge which indicates a lot of variability across cases. Standard deviation was lower but still substantial for No cases, so this must be kept in mind. This could indicate that there are cases with large number of directory organizations that are pulling up t he means. To avoid distortion f or this analysis, the extreme outlier of 961 cases for the U.S. was excluded. The next highest case, included in the analysis, was Brazil with 221 organizations. Table 2 62 Hypothesis 5: All ILGA Directory Orgs Means by L egislation Legislation Yes No Total Cases # of cases Mean Std Dev # of cases Mean Std Dev Sex 127 21.41 44.30 81 2.46 4.87 208 Con 114 19.36 40.88 104 6.92 26.31 218 Emp 61 30.43 50.01 133 7.95 26.51 194 GS 40 41.53 57.78 154 8.14 25.36 194 AnyP 42 45.43 65.54 181 5.63 14.78 223 PCU 37 47.60 67.28 186 6.27 17.04 223 M 23 65.74 70.40 199 7.11 21.24 222 A 17 51.65 66.21 206 9.95 28.96 223
110 For Yes on Sex and Con, the mean was around 20 organizations. The comparative mean values for No were much lower at around 3 and 7, respectively. For Yes on Emp, the mean was around 30, while the mean for GS was about ten points higher at around 41. Th e comparative values for No were around 8 for each piece of anti discrimination legislation. AnyP and PCU had means of nearly 50 for Yes compared to about 6 for No. The highest mean for Yes on organization was for marriage equivalent partnership (M) at about 66 organizations, compared to a No value of 7 organizations. Likewise, A diverged on Yes and No, with means of about 52 and 10, respectively. With progressive legislation down the column, the divergence of mean organization frequencies be tween Yes and No grew greater, for the most part. When looking at the values for No on legislation, one observes that the means range from a low of about 3 for Sex to a high of about 10 for A. However, the mean values for AnyP, PCU, and M are lower than those for Con, Emp, and GS. The growth of mean number of organizations (for the most part) with successive legislation makes sense, because one surmises that more LGBT supportive organizations would be required for higher order rights progressing from dec riminalization to anti discrimination to partnership and family rights. Also, since there are lower frequencies of cases with higher order legislation, a higher proportion of cases are in the No group that have other types of legislation facilitated by th e work of LGBT supportive organizations. Looking at the Yes values for legislation, the pattern of means is typical of many of the patterns for demographic variables in that the mean grows larger with successive legislation. The means by legislation for the percentage of the population served by all ILGA directory organi zations are exhibited in Table 2.63 Because of the small values for
111 ILGAOrgsbyPop, values for means and standard deviations are in scientific notation for ease of reading. The density of LGBT supportive organizations is very small for every country, for the most part ten thousandths of a percent. The differences between Yes and No for Sex and Con were not considerable. The differences between Yes and No for Emp, GS, AnyP, PCU, M, and A were more substantial than for Sex and Con. All of the means for No on legislation hovered around .0001, or 1.00E 04. For Yes on Sex and Con, the means were about .0001, or 1.00E 04 apart from the next lowest mean for PCU. Emp, AnyP, and PCU had means around 2.75E 04, while A was close to 3.00E 04, and GS and M were around 3.50E 4. Statistical tests can help aid analysis to determine whether there were significant differences between the Yes and No groups on legislation. T able 2 63 Hypothe sis 5: All ILGA Directory Orgs b y Population Means By Legislation Legislation Yes No Total Cases # of cases Mean Std Dev # of cases Mean Std Dev Sex 128 1.34E 04 3.26E 04 81 1.01E 04 2.50E 04 209 Con 114 1.56E 04 3.34E 04 105 1.09E 04 2.43E 04 219 Emp 61 2.81E 04 4.27E 04 134 .87E 04 2.10E 04 195 GS 40 3.59E 04 5.04E 04 155 .93E 04 2.02E 04 195 AnyP 42 2.75E 04 3.33E 04 182 .97E 04 2.71E 04 224 PCU 37 2.70E 04 3.44E 04 187 1.07E 04 2.73E 04 224 M 23 3.64E 04 3.85E 04 200 1.04E 04 2.68E 04 223 A 17 2.94E 04 3.60E 04 207 1.17E 04 2.82E 04 224 The means for organizations by population for cases with and without legislation are compared on the LGBT rights variables with the W ilcoxon Mann Whitney test the results of which are reported in Table 2.64 For AllILGAOrgs, differences between Y and N were significant for every legislation variable. These findings bolster the prediction that the presence of LGBTI supportive organizations are salient for legislation.
112 Table 2.64 Hypothesis 5: All ILGA O rgs Wilcoxon Mann Whitney Test Legislation Chi square DF P value # of cases Y # of cases N Total Cases Sex 21.97 1 <.0001 ** 128 81 209 Con 17.01 1 <.0001 ** 114 105 219 Emp 37.03 1 <.0001 ** 61 134 195 GS 37.56 1 <.0001 ** 40 155 195 AnyP 24.65 1 <.0001 ** 42 182 224 PCU 19.78 1 <.0001 ** 37 187 224 M 38.34 1 <.0001 ** 23 200 223 A 14.29 1 0.0002 ** 17 207 224 ** 0 1. Likewise Table 2.65 displays the results for the Wilcoxon Mann Whitney test for the density of LGBT rights supportive organizations by population For every LGBT rights legislation variable, the differences between cases in the Y and N groups were significant. These findings support the observations from the analysis of the mean s by legislation. Table 2 65 Hypothesis 5: All ILGA Orgs by Population Wilcoxon Mann Whitney Test Legislation Chi square DF P value # of cases Y # of cases N Total Cases Sex 11.93 1 0.006 128 81 209 Con 8.7 1 0.0032 114 105 219 Emp 47.87 1 < .0001 ** 61 134 195 GS 38.92 1 <.0001 ** 40 155 195 AnyP 22.73 1 <.0001 ** 182 42 224 PCU 16.34 1 <.0001 ** 37 187 224 M 29.65 1 <.0001 ** 23 200 223 A 11.03 1 0.0009 ** 17 207 224 * 0 1.
11 3 General Observations for Hypothesis 5 The legislation type, supports the argument that organizations are needed to support greater LGBT rights legislation. For decriminalization and con sent, there was not a great difference between the numbers of organizations in countries that did and did not have legislation. However, for the other types of legislation pertaining to anti discrimination protections, partnership, and family law, there w as considerable difference, especially in the upper echelons of the grouped ranges such as groups of 20 and above For states with these LGBT rights, a smaller proportion had no org anizations and cases were observed more often across the upper levels of organization than for states without rights. LGBTI organizations, civil rights groups, and other supportive organizations, was examined by legislation, it became clear that there were differences on the organization frequencies for each type of legislation when cases with and without rights were compared. While the means for cases without legislation were under 10 for every type of legislation, the means for cases with legislatio n were much higher, from around 20 to 65. However, the large standard deviations indicate that there is a wide variance in the numbers of LGBT supportive groups. Future research could focus on breaking analysis down into cluster s of states, such as comparison by region, in order to decipher whether certain classifications of countries pull up the mean. The measurement of ILGA directory organizations adjusted for population likewise had differences in density of organizations f or countries with and without legislation. Alt hough differences in means were not as striking as for ILGA supportive
114 and specifically LGBTI directory organizations, the presence of denser networks of ILGA directory groups was associated with cases that ha d legislation. With regards to differences in organization by types of legislation when legislation was present the presence of either LGBTI organizations, ILGA directory organizations, or denser ILGA organizations by population seemed to grow larger, o n the main, as legislation types became more substantive. Decriminalization and age of consent typically had the lowest levels of organization types while anti discrimination legislation had higher levels. In the hierarchy, so to speak, of partnership a nd family legislation, numbers of organizations were higher for marriage equivalent partnership and partnership/civil union than for any type of partnership. The exception to this rule was joint parent and/or second parent adoption, which had wide variabi lity in the numbers of organizations and seemed to have fewer organizations than did marriage equivalent partnership. The small sample size for same sex couple adoption may be part of the issue. For the most part, the measurement of organizations reflect ed assumptions about the standard sequence of rights. Ideally, one w ould draw upon data for all international organizations dedicated to LGBT rights as well as organizations working at the domestic level to study the impact of LGBT organizations on polic y. Unfortunately such data was too difficult to come by for this study. As it stands, one can only muse about the effect of such organizations, since only descriptive data about organizations was compared to legislation, and the quality of the data rende rs more certain conclusions dubious. For this and all hypotheses studied focus on particular regions or sets of countries would aid understand ing of broader global patterns.
115 C HAPTER 3: CONCLUSION Global diversity in LGBT rights sparked my interest in learning whether certain factors lead to the adoption of LGBT rights legislation. This question also encompassed my quest for a greater understanding of the commonalities among countries that have leg islation including tha t which reflects basic societal acceptance of homosexual lifestyles, protects individuals from discrimination on the basis of sexual orientation, and recognizes a diverse set of family arrangements. The forms of legislation encompass ed in this study were the decriminalization of homosexuality, equalized ages of consent for sexual activity, anti discrimination legislation in goods and services as well as employment, same sex cohabitation rights, same sex partnership and civil unions, marriage equivalent same sex partnership, same sex marriage, joint and second parent adoption, and family and partnership rights on a general level. My analysis incorporated five hypotheses about explanatory factors for LGBT rights legislation. Firs t, I examined the order of legi slation in order to determine whether there were standard sequences. Then, I explored neighboring country and regional relationships, followed by differences between predominant religious heritages. Attitudes related to the justifiability of homosexuality and postmaterialism and demographic variables such as religiosity, age structure, education, democratic freedoms,
116 and gender equality were next investigated. Finally I assessed the presence of LGBT advocacy organizations. characteristic of global patterns of LGBT rights passage. My analysis, through explorations of six configurations of the standard sequence, supported this hypothesis. I found that in countries with all types of rights, decriminalization of homosexual sexual activity and equalization of ages of consent tended to occur first, followed by anti discrimination legislation an d finally family recognition. Alt hough I hypothesized tha t partnership would occur before adoption rights, in many cases joint parent and/or second parent adoption was made available prior to the extension of partnership. Additionally, some type of discrimination protection was always present prior to the passage of some type of Throughout my test of Hypoth esis 1 I found a variety of paths, so my conclusion is not that there is only one way for rights to be granted. Most cases had not passed all of the legislation relevant to the current study, so this was a challenge for analysis. I found many on track c ases for legislation, in the sense that the order of legislation followed the standard sequence, even if some legi slation was yet to be passed. Even t hough these cases are illustrative of a portion of the pattern, they may never pass subsequent legislatio n. In Hypothesis 2 I sought to distill whether the neighborhood and region of a state was associated with legislation. I analyzed the effect of border state innovation on neighboring states, for the time periods before legislation was p assed in theoret ical
117 country X and after legislation was passed (or the innovation occurred ) in country X. For these before and after representations of border state effect, I used the percentage of shared borders that had legislation. I also looked at the present con figuration of legislation on the borders of country X, both as the total percent of shared borders occupied by countries with legislation, and as the total percent of neighbors with legislation. To study regional effects, I looked at the percentages of co untries in the region that had legislation before country X, and if country X also innovated, for regional neighbors that passed legislation at a point in time after country X. From analysis of the aforementioned variables, I found evidence for both nei ghboring country and regional effects by leg islation. With statistical testing of differences between countries with and without legislation, I found many significant differences for the percent of the border that was occupied by countries with legislatio n before country X, the total percent of the border that was occupied by countries with legislation, and the total percent of neighboring countries that had legislation. These statistical findings suggest that the presence of a larger shared border with i nnovating countries, either before the passage of legislation or generally, is associated with the passage of legislation. This result supports the hypothesis that neighbor effects impact legislation. The statistical findings were bolstered by a compariso n to the mean values. For these three neighbor variables, the differences in means between countries with and without the legislation were sub stantial and significant The exceptions to the rule were decriminalization and marriage equivalent partnership variable. This may be because decriminalization of homosexuality often happens with
118 regime change or when a new constitution is written and there are very few marriage equivalent partnership c ases from which to find significa nce. I also found evidence for regional effects, to a lesser degree than for border state effects, because evidence is only provided by one variable. The percent of the region that passed legislation before country X was higher on average for states tha t also passed the legislation than for states that have yet to pass the legislation. Additionally, the differences in means by legislation were not as striking as were the differences observed in border state relationships. My attempt at analyzing the r elationship between innovating states and the percentages of borders and regions after legislation was passed in country X failed to produce conclusive results. I had hoped to find that the passage of legislation in country X was correlated with a higher percentage of neighboring and regional countries that passed legislation, postulating that innovating countries would affect their neighbors. However, my broad scale look at this question was not specific enough to warrant conclusions. It would be neces sary to compare countries within regions, rather than remain at the level of cross regional analysis, and specify down to the country level in order to discern the patterns at work. For Hypothesis 3, I posed the question of whether predominant religious heritage factored in to the passage of legislation. I studied several configurations of predominant religious heritage, including broad and denominational classifications that were also adjusted to separate out religious dual heritages. By studying frequencies of legislation with regards to religious heritage and statistically testing pair wise differences between religious heritages, I found evidence for relationships between religion and LGBT rights
119 legislation. These relat ionships supported commonalities of legislation among states with certain types of religious affiliations as well as differences between religions on legislation. One of the most compelling findings was that Christian nations, on the broad and denominati onal level, tended to have the highest observations of legislation across all LGBT rights types. Additionally, significant differences were observed between Muslim co untries and Christian countries; Christian countries had the highest proportions of legis lation across all types, while Muslim countries had some instances of decriminalization, equal age of consent, and anti discrimination protections, but no family rights. After Christian heritage countries, countries with a combination None and Christian h eritage had the highest frequencies of legislation. Hindu, Buddhist, and Indigenous Beliefs heritages had low proportions of legislation, with almost no recognition of family rights. Out of the Christian majority states on the denominational level, Roman Catholic states had the highest frequencies of legislation, followed by Protestant and combination Protestant Roman Catholic states. However, I am hesitant to state that these observed differences between and commonly found characterizations of religiou s heritage are causal factors for LGBT rights legislation. There may be an intervening factor besides religion, because religions tend to be associated with certain development levels and in geographically proximate and culturally homogenous areas. Devel opment especially may be an explanatory factor for LGBT rights legislation, because Christian majority countries are associated with higher levels of development.
120 For Hypothesis 4, I sought to test various demographic and attitudinal variables with res pect to different types of legislation. I found evidence for some commonalities among countries with legislation along the studied variables when I used data lagged by five years prior to the passage of legislation. The distribution of attitudes toward t he justifiability of homosexuality was lower than I expected across legislation. Countries with decriminal ization and age of consent had low values on attitudes, and anti discrimination legislation had higher average values, all in the lower half of the r ating scale. These findings suggest that high acceptability of homosexuality in society is not necessary for the passage of legislation, but that more substantive rights require a moderate degree of acceptability. In general, the populations with legisl ation were composed of a majority of mixed attitudes on the postmaterialist scale. This suggests that societies must move somewhat beyond basic material concerns before passing legislation, especially for more substantive LGBT rights legislation, which is in line with the hypothesis. However, for non family legislation, proportions of M aterialists were higher than those of P ost materialists. Age structure was not especially relevant for explaining patterns, except that decriminalization had the youngest a verage age range, and average age seemed to increase with more substantive legislation. Since countries early in development have lower average ages, it is possible that development differences account for the differences observed for age structure. Lo w religiosity levels, high education levels, and high democratic freedom levels were associated with every type of legislation. The association of these variables with more liberal social attitudes validates these findings in favor of the hypothesis. Les s
121 gender equality was observed in countries with legislation such as decriminalization and equalized age of consent than for anti discrimination and family legislation, which had mainly levels of inequality that were less than 40%. Perhaps concern for gen der equality is not required for lower rights legislation, whereas for more practical legislation that substantiates the lifestyles of LGBT persons, concern for gender equality is a larger factor. Hypothesis 5 sought to look at the effect of LGBT organi zations, whose presence and activism should bolster passage of LGBT rights legislation. I found evidence in support of this hypothesis, with the caveat that major data problems impinge on my analysis. For measurements of LGBTI advocacy groups, LGBT suppo rtive organizations, and the density of LGBT supportive organizations by population, in general I found differences between countries with and without legislation for all types of rights in that cases with legislation had higher numbers of organizations t han did cases without legislation. However, the fact that for specifically LGBTI groups, decriminalization and age of consent were not substantially different for cases with and without legislation, suggests that differences became important for more subs tantive legislation related to societal acceptance of LGBT persons in the family and workplace. The findings for Hypothesis 5 are reported with hesitation, without certainty in conclusions. Since data collected was only from o ne directory of internat ionally connected LGBT organizations, many organizations that are influencing policy may not be included, such as those working specifically at the domestic level. More accurate and varied sources of data are necessary for more conclusive findings.
122 Man y roadblocks were encountered in analysis, many of which were discussed in detail in the general observations sections, and some of which suggest future directions for study. The direction of causation was an issue of concern, and correlation does not alw ays imply causation. There were also possibilities that omitted variables were actually correlated with LGBT rights legislation, making it seem as though other proxy or component variables were the cause. For example, high education levels and Christiani ty may be proxies for development levels. Future research should certainly look to development to see if this is the case. More regional and country level dissection of the variables studied would bolster understanding of their relationship to LGBT righ ts legislation. Additionally, future research could evaluate the impact of government variables, such as c ountries with Leftist parties in power and democratic consensus oriented governments (with coalition built cabinets, multipartism, and proportional r epresentation) since such characterizations of government seem likely to have broad LGBT rights In sum, I found support for every hypot hesis under review for my large scale cross nati onal analysis of LGBT rights. Alt hough the world is characterized by multiple variations in paths to LGBT rights, some generalizations can be made about the commonalities among countries with legislation and about the division between countries with and without legislation. My findings make some headway toward making sense of the multitude of global incidences of LGBT ri ghts. Hopefully this contribution to the literature on global analyses of LGBT rights will spark further research that can expand upon my tentative conclusions.
123 R eferences Adopting in Nunav May 3, 2012. http://www.adoptiveparents.ca/nun_issues.shtml Adoption Act. 1984. Australian Public Law SECT 10A, Vic torian Consolidated Legislation. http://www.austlii.edu.au/ au/legis/vic/ consol_ac t/ aa1984107/s10a.html Adoption and Children Act. 2002. U.K. National Archives. http://www.legislation. gov .uk/ukpga/2002/38/contents American Civil Liberties Union (ACLU) of Nevada. 2012. Nevada: A Practical Guid e for Same Sex and Opposite American Civil Liberties Union of Nevada April 5. http://aclunv.org/node/239. Uso Estratgico del Derecho: Reconocimiento del Oati Socio Legal Series 1(1): 1 35. Badgett, M. V. Lee. 2004. Variations on an Equitable Theme: Explaining International Same Sex Partner Recognition Laws. In Same Sex Couples, Same Sex Partnerships, and Homosexual Marriages: A Focus on Cross National D ifferentials Documents de travail n124 edited by Marie Digoix and Patrick Festy, 95 114. Theories of the Policy Process Second Edition, edited by Paul Sabatier, 20 35. Boudler: W estview Press.
124 Borrillo, Daniel, Eric Fassin, Wilfried Raut, Ian Sumner, Kees Waaldijk, M. V. Lee Badgett, Martine Corijn, et al. 2004. Same Sex Couples, Same Sex Partnerships, and Same Sex Marriages: A Focus on Cross National Differentials. Paper pres ented at Stockholm Conference, Stockholm University. Bruce Jones, Eddie, and Lucas Paoli Itaborahy. 2011. State Sponsored Homophobia: A World Survey of Laws Criminalising Same Sex Sexual Acts between Consenting Adults. International Lesbian, Gay, Bisexual Trans, and Intersex Association (ILGA) Canning, Paul. 2011. Northern Cyprus to Decriminalize Homosexuality. LGBTQ Nation. Central Intelligence Agency (CIA). 2012. The World Factbook https://www.cia.gov/ library/publications/the world factbook/ Cibotti, Andrea. 2011. Chilean Senate Approves Extensive Anti Discrimination Law. I Love Chile November 9 http://ilovechile.cl/2011/11/09/chilean senate approves extensive antidiscrimination law/38603 Danish Institute for Human Rights. 2010. Study on Homophobia, Transphobia and Discrimination on Grounds of Sexual Orientation and Gender Identity Danish Institute for Human Rights Treatment Act Diversity Enriches, June 10. http://www.erinevusrikastab.ee/en/ articles/paragraph/47.
125 and Liberalization of State Policies on Same Sex Sexual Relations, 1984 Social Forces 77(3): 911 943. Freedom House. 2012. Country Rat ings and Status, Freedom in the World 1973 2012. Accessed April 20, 2012 http://www.freedomhouse.org/report types/freedom world Anti Discrimination Law in http://www.glad.org/rights/ massachusetts/c/anti discrimination law in massachusetts/ Gay and Lesbian Advocates and Defenders (GLAD). 2009. Discrimination Law in discrimination law in vermont/ Hayes, Bernadette C., Ian Mc Allister, and Donley T. Studlar. 2000. Gender, Postmaterialism, and Feminism in Comparative Perspective. International Political Science Review 21 (4): 425 39. Holmes, Thomas J. The State Border Data Set. http://www.econ.umn.edu/ ~holmes/data/BorderData.html uman Rights HRC. http://www.norway geneva.org/unitednations/humanrights/hrc011206/ Human Right s Watch HRW, March 11. http://www.hrw.org/news/2009/03/11/cayman islands ensure equality all.
126 ILGA Europe, the European Region of the International Lesbian, Gay, Bisexual, Trans, and Intersex Association in cooperation with Transgender Europe. 2011. Rainbow Europe Map and Index 2011: Legal Situation for Lesbian, Gay, Bisexual, and Trans People in Europe. ILGA Europe. Inglehart, Ronald. 1990. Culture Shift in Advanced Industrial Society Princeton, NJ: Prince ton University Press. Inglehart, Ronald and Paul R. Abramsom. 1999. The American Political Science Review 93(3): 665 677. http:// www.jstor.org /stable/2585581 International Commission of Jurists. 2009. Sexual Orientation, Gender Identity and Justice: A Comparative Law Casebook Geneva: Atar Roto Presse SA. International Lesbian, Gay, Bisexual, Trans and Intersex Association (ILGA) 2012. irectory of LGBTI and Allied Organizations. ILGA. Accessed April 12. http://ilga.org/directory/en/ ILGA. Accessed May 3, 2012. http://ilga.org/ilga/en/index.html 2 ILGA April 4. 2011. Lesbian and Gay Rights in the World. ILGA. Lambda Legal. n.d. In Your State http://www.lambdalegal.org/ states regions
127 Jalal, Imrana, P. 2008. Legal Protection against Discrimination in Pacific Island Countries. T he Equal Rights Review 2: 32 39. BBC News July 23, 2011, sec Europe. http://www.bbc.co.uk/news/world europe jersey 14262401 BBC News January 13. http://www.bbc.co.uk/news/world europe jersey 16536572 Josephson, Jyl. 2005. Citizenship, Same Sex Marriage, and Feminist Crit iques of Marriage. Perspectives on Politics 3(2): 269 284. Kollman, Kelly. 2007. dea. International Studies Quarterly 51: 329 57. and Matthew Waites. 2009. The Global Politics of Lesbian, Gay, Bisexual and Transgender Human Rights: An Introduction. Contemporary Politics 15 (1): 1 17. Harvard Journal of Law and Gender 31: 67 100. Lucha Contra El Racismo, La Xenofobia y La Discriminacin, Public Law 17.817. 2004. http://www0.parlamento.gub.uy/leyes/ AccesoTextoLey.asp?Ley= 17817& Anchor=. ca of Madrid, Spain: An Journal of Homosexuality 57(2): 226 248. doi: 10.1080/00918360903488913.
128 Matthews, J. Scott. 2005. The Political Foundations of Support for Same Sex Marriage in Canada. Canadian Journal of Political Science / Revue Canadienne De Science Politique 38(4): 841 866. Gender Couples: A Netherlands Electronic Journal of Comparative Law 4(3). Mill bank, Jenni. 2005. Same Sex Families: Gay and Lesbian Families in Australia. Hot Topics: Legal Issues in Plain Language 53:7. http://www.legalanswers.sl.nsw .gov.au/hot_topics/pdf/same_sex_fam_53.pdf Minken Religion and Public Policy: Institutional, Cultural, and Comparative Political Studies 35(221): 221 247. National Democratic Institute Signs Anti 2011. December 20. http://www.ndi.org/Colombian president signs antidiscrimination law National Gay and Lesbian Task Force. 2011. State Nondiscrimination Laws in the U.S http://www.thetaskforce.org/downloads/reports/issue_maps/non_discrimination_6 _11_color.pdf 2011. BBC News April, 15. http://www.bbc.co.uk/news/world europe isle of man 13093113 Norris, Pippa, and Ronald Inglehart. 2003. Islamic Culture and Democracy: Testing the Clash of Civilizations' Thesis. In Human Values and Social Change: Findings
129 from the World Values Surveys. edited by Ronald Inglehart, 5 33. Leiden, Boston: Brill. Human Rights Law Review 8(2): 207 248. Olivera, Yanina. 2009. Uruguay Approves Latin America's First Gay Adoption Law. AFP September, 9. http://www.google.com/hostednews/ afp/article/ALeqM5gu1QYorSnG_WrGpbQ ic2fMpxObg Ottosson, Daniel. 2006. LGBT World Legal Wrap up Survey. ILGA http://typo3.lsvd.de/ fileadmin/pics/Dokumente/Homosexualitaet/World_legal_wrap_up_survey._Nov ember2006.pdf 2010. State Sponsored Homophobia: A World Survey of Laws Prohibiting Same Sex Activity between Consenting Adults ILGA. http://old.ilga.org/ Statehomophobia/ILGA_State_Sponsored_Homophobia_2010.pdf. The Pew Forum on Religion and Public Life. 2011. The Future of the Global Muslim Population by Country. http://features.pewforum.org/muslim population/. 2009. Mapping the Global Muslim Population: A Report on the Size and ion http://www.pewforum.org/ newassets /images/reports/Muslimpopulation/Muslimpopulation.pdf 2011. Global Christ ianity: An Interactive Feature. http://features.pewforum.org/global christianity/population number.php
130 Pichardo Galn, Jos Ignacio In Greenwood Encyclopedia of LGBT Issues Worldwide edited by Chuck Stewart, 393 412. Santa Barbara: Greenwood Press. Pink News Pink News, March 25. http://www.pinknews.co.uk/2010/03/25/guernsey equalises gay age of consent/. Pink News December 14. http://www.pinknews.co.uk/2006/12/14/h omophobia illegal in faeroe islands/ Feminist Legal Studies 15: 329 340. doi: 10.1007/s10691 007 9064 z. Challenges in a Post equality E ra: The Same se x International Journal of Iberian Studies: Open Forum 21(1): 41 49. doi: 10.1386/ijis.21.1.41/3. 2004. http://www.guernseylegalresources.gg/ccm/legal resources/laws/discrimination/prevention of discrimination enabling provisions bailiwick of guernsey law 2004.en. Reyes, Brian. 2011. ules: Age of Consent is 16 for All. Gibraltar Chronicle. Reyes, Jose. 2011. osible en Coahuila. Vanguardia November, 28 http://www.vanguardia.com.mx/adopciongayseraposibleencoahuila 1158988.html
131 Senate OKs 2 Bills: Anti discrimination Act and Kindergarten act. 2011. Tempo November 23 sec Met ro. http://www.tempo.com.ph/2011/senate oks 2 bills anti discrimination act and kindergarten act/#.T2OCt3JEO4Y Sexual Orientation Non Discrimination A New York. http://www.prideagenda.org/Issues Explained/Sexual Orientation Non Discrimination Act.aspx Shah, Dayna. 2002. Sexual Orientation Based Emplo yment Discrimination: States' Ex perienc e with Statutory Prohibitions. Washington, DC: United S tates General Accounting Office. http://www.gao.gov/new.items/d02878r.pdf Shoffman, Marc. 2006. Scotland Passes Gay Adoption Law. Pink News December, 8. http://www.pinknews.co.uk/2006/12/08/scotlan d passes gay adoption law/ Swiebel, Joke. 2009. Lesbian, Gay, Bisexual and Transgender Human Rights: The search for an International Strategy. Contemporary Politics 15 (1) (March 09): 19 35. Timeline: Same sex Marriage around the World. 2009. CBC News, sec World. http://www.cbc.ca/news/world/story/2009/05/26/f same sex timeline.html Human Development Reports. http://hdr.undp.org/en/statistics/gii/ Human Development Indicators. http://hdrstats.undp.org/en/indicators/ 103706.html.
132 United Nations Statistics. n.d. Division Classifications of Regions. http://unstats.un.org/unsd/methods/m49/m49regin.htm#europe. Composition of Macro Geographical (Continental) Regions, Geographical Sub Regions, and Selected Economic and other Groupings. http://unstats.un.org/unsd/methods/m49/m49regin.htm United S tates Department of State n.d. Country Profiles U.S. Department of State http://www.state.gov/countries/ 2009. Sexual Orientation / Gender Identity References: 2008 Country Reports on Human Rights Practices. U.S. Department of State. The Virgin Islands Constituti on Order, Public Law 1678. 2007. The National Archive s http://www.legislation.gov.uk/uksi/2007/1678/article/26/made Waaldijk, Kees. 1992 Standard Sequences in the Legal Recognition of Homosexuality: Europe's Fast, Present and Future. Australas ian Gay and Lesbian Law Journal 4 : 50 72 2001. Small Change: How the Road to Same Sex Marriage got paved in the Netherlands. In Legal Recognit ion of Same Sex Partnership: A S tudy of National, European and International Law edited by Robert Wintemute and Mads Andens, 43 7 464. Oxford: Ha rt Publishing. 2001. Towards the Recognition of Same Sex Partners in European Union Law: Expectations based on Trends in National Law. In Legal Recognition of Same Sex Partnerships: A Study of National, European and International Law ed ited by Robe rt Win temute and Mads Andenaes, 635 651. Oxford : Hart Publishing.
133 2005. More or Less Together: Levels of Legal Consequences of Marriage, Cohabitation and Registered Partnership for Different Sex and Same Sex Partners: A Comparative Study of Nine Europ ean Countries. Paris: Institut Natio 2007. Eight Major Steps in the Legal Recognition of Homosexual Orientation: A Chronological Overview of National Legislation in the Member States of the European Union. Choices: T he Sexual and Reproductive Health and Rights of Lesbian, Gay, Bisexual and Transgender People in Europe 40 : 1 1. 2009. Legal Recognition of Sexual Orientation in the Countries of the World: A Chronological Overview with Footnotes. Paper presented at The Global Arc of Justice Sexual Orientation Law around the World, Los Angeles. Weeks, Jeffrey. 2008. ecognition. Sexualities 11: 788 93. doi : 10.1177/1363460708096920. Worldwide Ages of Consent. 2012. http: //www.avert.org/age of consent.htm. World Values Survey Association. Online Data Analysis for the W orld V al ues Surveys 2005 2008 and Four Wave Aggregate of the Values Studies: (V202) J ust ifiable: Homosexuality, (V237) Age, (Y002) Post materialist Index 4 i tem, and (V186) How O ften do you Attend Religious S ervices http://www.wvsevsdb.com/wvs/ WVSAnalizeStudy.jsp 1981 2008 Official Aggregate v.20090901, 20 09. Madrid: Aggregate File Prod ucer: ASEP/JDS http://www.worldvaluessurvey.org/.
134 Appendix GenRelig1 Fisher's Exact Test: Signif i cant Results Legislation GenRelig1 Comparison P value Effective Sample Size Sex ProtRC v. Sunni Muslim 0.0003 ** 21 Con ProtRC v. Sunni Muslim 0.0061 ** 57 AnyDiscrim ProtRC v. Sunni Muslim 0.0003 ** 63 Emp ProtRC v. Sunni Muslim 0.0022 ** 64 GS ProtRC v. Sunni Muslim 0.0016 ** 62 AnyP ProtRC v. Sunni Muslim 0.0006 ** 64 PCU ProtRC v. Sunni Muslim 0.0029 ** 64 M ProtRC v. Sunni Muslim 0.0134 64 Con None v. Sunni Muslim 0.0477 52 AnyDiscrim None v. Sunni Muslim 0.0005 ** 57 Emp None v. Sunni Muslim 0.0006 ** 57 GS None v. Sunni Muslim 0.0015 ** 57 AnyP None v. Sunni Muslim 0.0003 ** 57 PCU None v. Sunni Muslim 0.0029 ** 57 A None v. Sunni Muslim 0.0226 57 Sex Buddhist v. Roman Catholic <.0001 ** 76 Con Buddhist v. Roman Catholic <.0001 ** 75 AnyDiscrim Buddhist v. Roman Catholic 0.0161 73 Emp Buddhist v. Roman Catholic 0.0365 76 AnyP Buddhist v. Roman Catholic 0.0272 76 PCU Buddhist v. Roman Catholic .0547 76 Con Eastern Orthodox v. Sunni Muslim <.0001 ** 56 AnyDiscrim Eastern Orthodox v. Sunni Muslim <.0001 ** 61 Emp Eastern Orthodox v. Sunni Muslim <.0001 ** 61 GS Eastern Orthodox v. Sunni Muslim 0.0277 61 AnyP Eastern Orthodox v. Sunni Muslim 0.0426 61 Sex Buddhist v. Eastern Orthodox <.0001 ** 23 Con Buddhist v. Eastern Orthodox <.0001 ** 23 AnyDiscrim Buddhist v. Eastern Orthodox 0.0097 ** 23
135 Legislation GenRelig1 Comparison P value Effective Sample Size Emp Buddhist v. Eastern Orthodox 0.0097 ** 23 Sex Buddhist v. None 0.0108 19 Con Buddhist v. None 0.0108 19 Emp Buddhist v. None .0573 19 AnyP Buddhist v. None 0.0325 19 Con Eastern Orthodox v. Oriental Orthodox 0.0179 16 AnyDiscrim Eastern Orthodox v. Oriental Orthodox 0.0625 16 Emp Eastern Orthodox v. Oriental Orthodox .0625 16 AnyDiscrim Indigenous Beliefs v. None 0.0062 ** 21 GS Indigenous Beliefs v. None 0.0211 21 Emp Indigenous Beliefs v. None 0.0048 ** 22 Emp Jewish v. Sunni Muslim .0612 49 PCU Jewish v. Sunni Muslim 0.0204 49 M Jewish v. Sunni Muslim 0.0204 49 Sex Roman Catholic v. Shia Muslim 0.0002 ** 71 Con Roman Catholic v. Shia Muslim 0.0028 ** 69 AnyDiscrim Roman Catholic v. Shia Muslim .0539 68 Emp Roman Catholic v. Sunni Muslim 0.0597 71 M Roman Catholic v. Sunni Muslim 0.0048 ** 114 A Roman Catholic v. Sunni Muslim .0725 114 Sex Buddhist v. Indigenous Beliefs <.0001 ** 23 Con Buddhist v. Indigenous Beliefs 0.0075 ** 23 Sex Buddhist v. Protestant <.0001 ** 57 Con Buddhist v. Protestant 0.0031 ** 55 Sex Buddhist v. ProtRC <.0001 ** 26 Con Buddhist v. ProtRC 0.002 ** 24 Sex Eastern Orthodox v. Shia Muslim 0.0025 ** 18 Con Eastern Orthodox v. Shia Muslim 0.0063 ** 17 Sex Indigenous Beliefs v. Sunni Muslim 0.0108 60 Con Indigenous Beliefs v. Sunni Muslim 0.0347 56 AnyDiscrim Eastern Orthodox v. Indigenous Beliefs <.0001 ** 25 Emp Eastern Orthodox v. Indigenous Beliefs 0.0005 ** 26 GS Indigenous Beliefs v. ProtRC 0.0425 26 Emp Indigenous Beliefs v. ProtRC 0.0205 29 Con Indigenous Beliefs v. Roman Catholic .0726 78 GS Indigenous Beliefs v. Roman Catholic 0.0072 ** 75 Sex Buddhist v. Hindu 0.022 15 Sex Buddhist v. Sunni Muslim 0.0016 ** 57 Sex Indigenous Beliefs v. Shia Muslim 0.0007 ** 18 Sex None v. ProtRC 0.0403 25
136 Legislation GenRelig1 Comparison P value Effective Sample Size Sex None v. Roman Catholic 0.0431 75 Sex ProtRC v. Protestant 0.0502 63 Sex ProtRC v. Shia Muslim 0.0055 ** 63 Sex Shia Muslim v. Sunni Muslim .0515 52 Con Hindu v. Roman Catholic .0747 70 AnyDiscrim Indigenous Beliefs v. Protestant 0.0476 53 Emp Eastern Orthodox v. Shia Muslim 0.0294 18 Emp Eastern Orthodox v. Protestant 0.0057 ** 60 PCU Eastern Orthodox v. None .0545 22 PCU Eastern Orthodox v. Roman Catholic 0.0152 79 A Protestant v. Sunni Muslim 0.0026 ** 95 * 0 1.