Existential security and the gender gap in religious values

Transcription

1 [1 st Draft: Total length 11,606 words] Existential security and the gender gap in religious values Pippa Norris (Harvard University) and Ronald Inglehart (University of Michigan) Pippa Norris Ronald Inglehart McGuire Lecturer in Comparative Politics Institute for Social Research John F. Kennedy School of Government University of Michigan Harvard University Ann Arbor, Cambridge, MA Michigan, Synopsis: A pervasive generalization in the social sciences suggests that women are universally more religious in their attitudes, values and behavior than men around the world and across the centuries, leading Stark (2002) to propose a physiological or biological root cause. Yet a recent comparison of a wide range of world faiths suggests that the standard view needs revision. Far from a universal phenomenon, Sullins (2006) reports that the size of the gender gap in religiosity varied by type of indicator and by type of culture. In particular, among those of Jewish and Muslim faith, compared with women, men were significantly more active in collective religious rituals, not less. Moreover Miller and Stark (2002) found that, contrary to expectations, the gender gap in religiosity was weaker in societies where traditional sex roles predominated. Reframed in this light, the puzzle becomes how to explain variations in the gender gap in religiosity. This study develops the argument that disparities in existential security can help to explain this phenomenon, and then analyzes structural and attitudinal indicators to test this account. Data is derived from five waves of the World Values Survey from , covering almost half the world s countries, and most major types of religious cultures. The conclusion draws together the main findings and considers their implications for understanding religious practices and behaviors. Draft chapter for the SSRC conference on Religion & International Affairs, New York, February 15-16, 2008 and the edited volume by Timothy Shah, Alfred Stepan, and Monica Toft 1

2 Are women more religious than men? For many decades the standard perspective in the literature has asserted that they are, whether gauged by religious identities, values, or practices. Over the years a wealth of empirical studies in the scientific study of religion have demonstrated that, at least in predominately Protestant and Catholic cultures, church congregations are disproportionately female, while women were also more likely to express faith in God and to hold religious beliefs (Argyle and Beit-Hallahmi 1975; Walter and Davie 1998; Miller and Hoffman 1995; Francis 1997). A number of alternative explanations have been proposed to account for this long-standing pattern, based on structural and cultural theories of sex-role differentiation. The apparent ubiquity of this phenomenon led Stark (2002) to speculate about a physiological or biological root cause, with testosterone encouraging men to take greater risks, including discarding the threat of eternal damnation in religions promising an afterlife. Much of the systematic empirical evidence supporting the standard claim has been gathered from historical censuses and social surveys, mainly conducted in predominately Protestant and Catholic cultures in post-industrial nations. Until recently, however, little systematic cross-national survey data has been available to test whether similar patterns hold across a wider range of world faiths and in developing societies. The conventional wisdom was recently challenged by Sullins (2006) who found that gendered patterns of religiosity varied both by type of culture and by type of indicator, suggesting a more multifaceted reality. The study expanded the comparative framework by analyzing 71 societies included in the 2 nd and 3 rd waves of the World Values Survey conducted in the early to mid-1990s. In particular, Sullins confirmed that women were indeed usually more religious than men in many religious cultures, especially in the affective forms of religious expression reflecting personal spirituality, exemplified by feelings of religious identity and the importance of religious values. By contrast, in many societies the traditional gender gap was found to be far smaller, or even negligible, in collective forms of religious rituals, such as church attendance and church membership in Protestant and Catholic societies. Among Jews and Muslims, by contrast, the gap even reversed, with men being significantly more active in religious practices than women. In addition, Miller and Stark (2002) examined the gender gap in religiosity against several societal-level indicators of traditional and egalitarian sex roles, such as the proportion of women in the paid workforce and the fertility rate, and found that the gap was weaker in societies where traditional sex roles predominated. Reframed in this light, if these findings are indeed confirmed as robust in a broad range of contexts and time-periods, universal physiological explanations become implausible, and the 2

3 puzzle becomes how to explain more complex variations in the gender gap in religiosity around the world. To account for differences in the size and significance of the gender gap in religiosity, Part I of the present analysis builds upon a theory of human existential security, originally developed to explain cross-national patterns of religiosity and trends in secularization (Norris and Inglehart 2004). The concept of existential security is a broad and multifaceted notion but its core meaning reflects the degree to which survival is seen as secure or uncertain. The degree to which survival is seen as precarious or safe varies widely, which generates a series of testable propositions. This argument is compared with accounts based on the psychological propensity to take risks, as well as the standard sociological theories based on common structural differences in men and women s lives, such as patterns of child-rearing, marital status, and engagement in the paid work-force, and cultural explanations based on sex-role socialization and attitudes towards gender equality. The key research questions are whether the gender gap in religious values is universally evident across most nations, religious cultures, and types of society, and whether the effect of gender on religious values weakens once models control for indicators of existential security, social structure and cultural attitudes? Part II describes the survey evidence, comparative framework, and measures. Religiosity is understood here as a multi-dimensional concept open to many different indicators, but nevertheless not all of these are equally suitable for comparative studies covering many different world religions. In particular, the comparison of specific spiritual beliefs and particular sacred rituals are problematic to interpret cross-culturally, as it remains difficult to establish indicators which are functionally-equivalent in diverse world religions. For this reason, in this study we decided to focus on the religious values gender gap, defined and measured as the difference in the perceived importance of religion in the lives of women and men. This is one of the most important indicators of spirituality, it is universal across all types of world faith and all forms of New Age spirituality, and religious values are also a strong predictor of other religious practices and attitudes (Norris and Inglehart 2004). To replicate, update, and expand many other comparative studies, empirical evidence is derived from the World Values Survey, including all five waves conducted from , covering over 85 percent of the world s population and most of the major world religions (with a total sample of 342,737 individual respondents). Part III compares the descriptive bivariate correlations for structural and cultural factors associated with religious values, estimating the size of the gender gap with each, and then uses multivariate 3

4 OLS regression models to test the factors driving the gender gap in religious values. The conclusion in Part IV draws together the major findings and considers their implications for understanding social differences in religious practices and behaviors. I: Theories explaining the gender gap in religiosity A recent literature review by Francis (1997) noted that many studies have observed the traditional gender gap in religiosity, but nevertheless the reasons underlying this phenomenon continue to prove puzzling. Sociological accounts emphasize structural differences in men s and women s lives which are thought to influence religiosity, such as patterns of child-rearing, psychological versions emphasize female personality traits, while socio-biological explanations propose physiological characteristics, such as levels of testosterone, which differentiate behavior in women and men. The core argument advanced in this study is that the gender gap in religious values should not be understood as sui generis; instead it can be explained most satisfactorily by patterns of existential security, which also illuminate broader cross-cultural differences in religiosity and secularization. Women often give higher priority to religion not because of their sex per se but because they usually experience less security in their lives, due to enduring and persistent disparities in physical strength, economic resources, and political power. Elsewhere we have developed the argument that economic and political changes go together with cultural developments in coherent and consistent ways (Norris and Inglehart 2004). We have demonstrated that the process of human development has significant consequences for religiosity; as societies transition from agrarian to industrial economies, and then develop into postindustrial societies, the growing economic and physical security that usually accompany this process tends to reduce the importance of religious values in people s lives. The main reason, we believe, is that the need for religious reassurance becomes less pressing under conditions of greater existential security. These effects operate at both the societal-level (socio-tropic) and at the personal level (ego-tropic), although the former seems to be more important. The concept of existential security is complex and multidimensional, as many different types of threats may engender perceived dangers and hardships in different contexts (Paris 2001; Ogata and Sen 2003). The 1994 UNDP Human Development Report argued that human security is shaped by multiple threats arising from economic, food, health, environmental, personal, community and political dimensions. 1 Most postindustrial nations are characterized by conditions of widespread social protection and control, high life expectancy, health and welfare services, a functioning and 4

5 stable state, as well as conditions of enduring peace and security from external invasion, which mean that although perceived dangers remain, existential security is maximized. Under these conditions, the thesis predicts that fewer people in these societies will regard traditional spiritual values, beliefs and practices as vital to their lives, or to the lives of their community. This does not imply that all forms of religion necessarily disappear as societies develop; residual and symbolic elements often remain, such as formal adherence to religious identities, just as church buildings and cathedrals remain long after congregations have shrunk away, even when their substantive meaning has faded away. But we expect that people living in advanced industrial societies will often grow increasingly indifferent to traditional religious leaders and institutions, and become less willing to engage in spiritual activities. Contrary to the assumptions of the supply-side religious markets school (Finke and Stark 1992, 2000), we assume that the demand for spirituality is far from constant; instead striking variations are evident in different societies and groups due to experience of the basic living conditions, poverty, inequality, and freedom from fear common in rich and poor nations. Virtually all of the world's major traditional religious cultures provide reassurance that, even though the individual alone can not understand or predict what lies ahead, a higher power will ensure that things work out. Both religion and secular ideologies assure people that the universe follows a plan, which guarantees that if you follow the rules, everything will turn out well, in this world or the next. The rains will come, life will prosper, and (for religions believing in an afterlife) the good will be rewarded and death will not end our contact with loved ones. This belief reduces stress and uncertainty, enabling people to shut out terrifying anxieties and focus on coping with their immediate problems. Without such a belief system, extreme stress and pain tends to produce withdrawal reactions. Under conditions of insecurity, people have a powerful need to see authority as both strong and benevolent-- even in the face of evidence to the contrary. The theory is not mechanically deterministic: perceptions of threats can generate increased fears even in affluent societies, as exemplified by the events of 9/11 and their aftermath in the US. The thesis advanced in this study therefore predicts that the people of poorer societies and more vulnerable sectors of society will usually regard religion as far more important than more those of affluent nations and people living in states with generous welfare safety-nets and with deeper reservoirs of household savings, private insurance, and investments, which usually experience greater freedom from fear in their lives. Levels of social and individual security are expected to lead towards predictable probabilistic variations in religious values 5

6 (reflecting the vital importance of religion in people s lives), although at the same time residual religious beliefs, ritualistic practices and institutions will continue to bear a strong imprint stamped by each specific type of faith. How does this theory help to explain to the gender gaps evident in religious values? The existence of enduring and deep-rooted differences in the way that women and men experience existential security are widely recognized, as illustrated by unequal levels of economic resources derived from subsistence farming, disparities in income, credit loans and debt, and the feminization of poverty; by disparities in access to clean water, nutrition, and health care in developing societies; by lack of power, representation in elected office, and voice in the public sphere; and by the physical and social vulnerabilities which women face against threats from aggression, the use of force, crime, and violence (including domestic violence). In short, feminist perspectives call attention to the way that girls and women experience existential security differently from boys and men, in large part because of enduring disparities in status and power (Lammers 1999; Bunch and Carillo 1998). This is most vividly demonstrated in some of the bloodiest civil wars, such as in contemporary Darfur, where reports by Amnesty International and others have documented how far rape and other forms of sexual violence are being used as a weapon of war in order to humiliate, punish, control, inflict fear and displace women and their communities. 2 But the contrasts in existential security experienced by most men and women go far beyond the most extreme cases, and the international community, bilateral aid agencies, and UN peace-building initiatives have increasingly come to recognize important contrasts in the way that women and men experience development, military interventions, and civil conflict, with important implications for effective development programs and policy interventions. The argument developed here therefore builds on and extends our previous work by suggesting that women generally give higher priority to religion in large part because of the disparities in safety, protection, and security which they experience in their lives. Is there convincing evidence that existential security is linked with religious values? And, if so, does this help to explain the gender gap in religiosity? Despite the complexity of the core concept of existential security, the thesis generates a series of clear hypotheses that can be tested empirically. One central prediction is that the salience of religious values will tend to be strongest in the least developed societies; the fact that this holds true has been demonstrated in considerable detail elsewhere (Norris and Inglehart 2004). The people of low-income countries are relatively vulnerable to life-threatening socio-tropic threats arising from sudden natural and 6

7 man-made disasters, such as floods, famines, and social instability, and the appeal of religious values is strongest in this context. The existential security thesis has attracted considerable interest, with subsequent studies replicating the findings (Zaidise, Canetti-Nisim, and Pedahzur 2007). Building upon this foundation, the thesis has a number of additional implications. Does the salience of religious values vary according to the degree of existential security that different groups within a given society experience? Of particular interest for the present analysis, does the salience of religious values reflect the relative economic status, power, and sense of safety of women and men living within given societies? This can be examined empirically by analyzing a series of indicators, comparing the strength of religiosity against gender disparities in household income, household savings, and the perceived importance of living in secure conditions. In particular, underprivileged females living in households with a minimal income are predicted to prove most religious in their outlook. Another test of the hypothesis can be made by examining contrasts in household savings and debt. Here again, we would expect the most vulnerable individuals with the least financial resources (a group that includes disproportionate numbers of females) to give the highest priority to religion. The poor are most vulnerable to sudden catastrophic economic risks, especially in the least developed societies, as they lack a safety-net of personal savings to cushion the shock of bad times, such as a health crisis, crop failure, or layoffs. Educational and age patterns are also expected to prove important, with the less educated, and the oldest (and thus most infirm) sectors of the population, giving the highest priority to religious values. Existential security can also be monitored more directly by examining the attitudes and priorities of those who say they give the greatest emphasis to living in safe conditions which are free from danger; we would expect this group to be disproportionately female and to place relatively strong emphasis on religion. Risk analysis and physiological differences The existential security thesis can be compared with a growing series of studies which link religiosity to more instrumental and individualistic notions of preferences for risk-taking. This proposition was first outlined by Miller and Hoffman (1995), and subsequently developed by Miller and Stark (2002) and Stark (2002). In religious cultures which emphasize the reward of life after death for the devout, especially in Christianity, Islam, and Orthodox Judaism, their argument suggests that to be irreligious is to risk divine punishment. It is well-established that 7

8 men have a greater propensity to engage in risky behavior such as crime and delinquency, drinking and drug use, whether due to male socialization processes or to socio-biological drives linked with different levels of testosterone. This tendency, they argue, means that men are also more likely to engage in risky behavior by being less religious. By contrast, women are generally more risk-averse and therefore also more religious. The reasons why men are usually more willing to take risks were not addressed by Miller and Hoffman (1995), but later Stark (2002) proposed the socio-biological explanation, where higher testosterone levels in men encouraged riskier behavior. This risk preference thesis has received mixed support from the available empirical survey evidence, however, in part because of alternative ways that risk preferences can be operationalized and modeled. Miller and Stark (2002) observed significant correlations, while later studies by Freese (2004), and by Roth and Kroll (2007), reported no relationship between risk preferences and the gender gap in religiosity. The risk preference theory suggests an instrumental notion, where individual men and women consciously weigh the costs and benefits of alternative choices; for example, in Pentecostal churches, whether sins are worth the risk of eternal damnation. By contrast, the existential security thesis offers a far broader framework which relates to the generic threats and dangers facing societies, communities, and individuals. The theoretical framework is also preferable because it offers a more comprehensive explanation; it does not just seek to account for the gender gap in religiosity but it also provides insights into many observable contrasts in patterns of secularization and religiosity. The results of the analysis of existential security can also be directly compared empirically with the risk aversion thesis: if the risk aversion thesis is correct, those who express a propensity for adventure and taking risks, and for living an exciting life, should prove less religious in their values, as well as being disproportionately male. Structural differences in gender roles and cultural attitudes towards gender equality In addition to testing these hypotheses, we can also test the implications of the existential security thesis against those of other standard factors emphasized in the literature. One of the most common structural theories in sociology emphasizes differences in the life-styles of women and men, particularly the traditional roles of women as the primary child-rearer in the home and the predominance of men in the paid labor-force and the public sphere (Becker and Hofmeister 2001). A strict division of gender roles and responsibilities in the home, patterns of 8

9 marriage and divorce, and the size and composition of the typical family unit, have changed substantially over time, especially in post-industrial societies, but nevertheless women continue to bear the primary responsibility for care of children and dependents (Mason and Jenson 1995; Scott et al 1993, 1998). In particular, Azzi and Ehrenberg (1975) hypothesize that church attendance may be higher for women if this represents an extension of the division of labor in the home. Previous attempts to test these relationships, however, have reported mixed results; De Vaus and McAllister (1987) examined Australian public opinion and found that mothers with dependent children attended church more frequently than women without dependent children. Similar patterns were also evident, however, for men with and without children. If the structural proposition is true, however, it implies that the size of the gender gap in religiosity should shrink among women who are in (full or part-time) paid employment, while it should expand among women who are full time home-makers, especially if they care for dependent children. These propositions can be tested empirically by examining whether the size of the gender gap in religious values varies systematically according to marital status, the presence of children in the household, and the respondent s participation in the paid workforce, after controlling for other important factors, including age and education. Lastly, the cultural attitudes of women and men towards gender equality could also play a role in determining patterns of religiosity (Walter and Davie 1998). If socialization patterns encourage traditional roles for women and men, in which women accept that their primary reward comes from being a wife and mother, then we would expect that these attitudes would also contribute towards a gender gap in religion. Religious authorities often advocate that women should remain in traditional roles in the domestic sphere and female fertility is encouraged, as exemplified by Papal encyclicals in the Catholic Church banning abortion and the use of artificial contraceptives. One of the central injunctions of virtually all traditional religions is to strengthen the family, to encourage people to have children, to encourage women to stay home and raise children, and to forbid abortion, divorce, or anything that interferes with high rates of reproduction. In this respect, greater adherence to religious beliefs and practices may encourage women to adopt traditional roles as full-time mothers and careers of dependents. Moreover those who accept traditional gender roles may also feel that this includes engaging in religious values and practices. This hypothesis can be tested by examining whether the size of the gender gap in religious values expands among those women who accept the idea of traditional gender roles, 9

10 while the gap diminishes among women who adhere to more liberal notions of gender equality in the home, workplace and school. II: Evidence, data and measures Despite considerable debate, many of the theoretical arguments that have been proposed over the years have not yet been tested empirically through multivariate analysis that reveal whether given generalizations prove robust when controlling for different cultural contexts, socioeconomic indicators, and time-periods. Moreover even where reliable empirical correlations have been established, the direction of causality proposed in many accounts remains difficult to interpret, in recursive relationships. Evidence for religious beliefs, values and behavior in many different societies is available in the World Values Survey (WVS-5) which covers a wide range of countries from all world regions, as well as many different types of faiths. The World Values Surveys is a global investigation of socio-cultural and political change. This project has carried out representative national surveys of the basic values and beliefs of the publics in more than 90 independent countries, containing in total almost 5.5 billion people or over 85% of the world s population and covering all six inhabited continents. It builds on the European Values Surveys, first carried out in 22 countries in A second wave of surveys, in 41 countries, was completed in ; a third wave was carried out in 55 nations in ; and a fourth wave, with 59 countries, took place in The fifth wave took place in over forty countries in The total sample from all combined waves contains 342,737 individual respondents. The WVS survey includes some of the most affluent market economies in the world, such as the U.S., Japan and Switzerland, with per capita annual incomes as high as $40,000; together with middle-level industrializing countries including Taiwan, Brazil, and Turkey, as well as poorer agrarian societies, exemplified by Uganda, Nigeria, and Viet Nam, with per capita annual incomes of $300 or less. Some smaller nations have populations below one million, such as Malta, Luxembourg and Iceland, while at the other extreme a billion or more people live in both India and China. The survey contains predominately Protestant societies such as Australia, Sweden, and the United States, Catholic societies including Belgium, the Philippines and Mexico, as well as Orthodox countries such as Russia and Greece, and a few with diverse Eastern religions, such as Japan, Viet Nam, and China. The survey also includes some of the first systematic data on public opinion in many Muslim states, including Arab countries such as Jordan, Iran, Egypt, and 10

11 Morocco, as well as Indonesia, Iran, Turkey, Bangladesh and Pakistan. The most comprehensive coverage of countries in the surveys is available in Western Europe, North America and Scandinavia, where public opinion surveys have the longest tradition, but countries are included from all world regions, including several Sub Saharan African nations. Measures of religiosity Since religiosity is a multidimensional phenomenon, studies have focused on many different aspects (Shiner 1966; Dobbelaere 1981; Swatos and Christiano 1999). Many sociologists have emphasized trends in religious participation, monitored through church and census historical records of congregations and through social surveys of reported churchgoing. It can be argued that behavior provides a concrete indicator of the importance of religion for social norms and habitual practices. Yet each major religion emphasizes its own distinct practices in spiritual rituals, ceremonies, and observances, often associated with the life-changes of birth, marriage and death, as well as celebration of certain holy days, and there are multiple variations within each religion s sects, denominations, and communities. Christian religious practices are exemplified by habitual church attendance on Sundays and special holidays, as well as by the role of prayer, charitable giving, the significance of communion, and the rituals of baptism, confirmation, and marriage. But within this common repertoire, Anglicans, Methodists, and Baptists each emphasize their own specific rituals. Elsewhere meditation rituals and ceremonies are central to Buddhism, along with the observation of festivals, blessings and initiations, and the role of monastic communities. For Muslims, the Qur-an specifies the five Pillars of Islam, including public profession of faith by recitation of the shahada, daily performance of the salat prayer ritual, annual giving of obligatory alms (zakat), fasting during Ramadan, and performance once in a lifetime in the rituals of the Great Pilgrimage to Mecca (the Hajj). Alternative New Age forms of spirituality involve an even wider range of diverse activities, including psychic, pagan, metaphysical, personal growth and holistic healthcare, with practices exemplified by yoga, meditation, aroma therapies, channeling, divination, and astrology. Given these differences, it does not seem to be appropriate to use measures of participation in collective religious services in cross-national studies covering many different types of faith and religious cultures. The frequency of attendance at collective church services, for example, is widely used to compare the size of Christian congregations in the United States, but this concept does not travel well as an indicator of the strength of religious involvement in other faiths. 11

12 Other commentators, such as Greeley (2003), give greater attention to the strength of common religious beliefs, such as faith in an afterlife or in metaphysical beings, on the grounds that people can continue to adhere to these beliefs even if they no longer participate regularly in services of worship. But beliefs about specific theological doctrines, creeds, and injunctions vary even more substantially across the major world faiths, as well as differing among specific sects, denominations, and communities. The predominant religious cultural traditions in any society, such as the legacy of Protestantism and Catholicism in Western Europe, are expected to leave a distinct imprint upon the contemporary moral beliefs and social attitudes that are widespread among the public in these nations. Predominant religious cultures are understood here as pathdependent, adapting and evolving in response to developments in the contemporary world, and yet also strongly reflecting the legacy of the past centuries. Yet the major faiths of the world express divergent teachings and doctrines on many moral values and normative beliefs, such as those surrounding the roles of women and men, the sanctity of life, and the importance of marriage and the family. As a result, comparing the strength of religious beliefs is also problematic for comparing like-with-like in cross-national studies. The expression of subjective religious identities can also be compared across different societies although their social significance remains unclear. Belonging to a particular sect, such as the Shi a or Sunni in Baghdad, or being part of Protestant or Catholic neighborhoods in Belfast, may express genuine deeply-held membership of a community with a common shared culture. But religious identities may also function as formal labels devoid of spiritual meaning; in Britain, for example, a plurality of the population habitually report on official census forms that they are nominal Anglicans, but beyond baptisms, weddings and funerals, this does not mean that they are active, practicing members who regularly fill the pews in the Church of England (Furlong 1999). 4 Yet others emphasize religion as a societal institution. Dobbelaere (1999), for example, regards secularization as a broad process reducing the societal significance and meaning of religion, notably how far the public regards spiritual values as important to their lives and how far they listen to religious leaders as an important source of moral authority and spiritual guidance. In this study, we do not examine direct evidence of the power and status of religious institutions and authorities, such as the role of evangelical churches in the bible-belt American South, ideological divisions within the Anglican synod, the influence of radical Islamic parties in the Middle East, or the structure, resources and leadership of the Roman Catholic Church in Italy. 12

13 Nor do we focus primarily upon the relative organizational strength of different denominations and sects within specific countries, and the way that this may affect women and men, such as attempts to build churches and expand congregations by US evangelicals in South Korea, or rivalry for hearts and souls between Catholic clergy and Pentecostal missionaries in Guatemala and El Salvador. These are important matters, but it remains difficult to establish functionallyequivalent survey items that would compare these issues among types of faith which encourage different types of spiritual expression and rituals of behavior. While recognizing that religiosity is multidimensional, in cross-national analysis, values are most easily treated as functionally-equivalent indicators of this phenomenon. In general, the study of values refers to the goals that people believe to be important for themselves, their families, their community or the nation, such as the value of wealth and material prosperity, of collective solidarity and social equality, of security and freedom from fear, and of individual competition, hard-work, and enterprise. Religious values can be gauged by how far respondents believe in the importance of God in monotheistic religions, or in the importance of the gods in pantheistic religions. This is an important indicator since it has been established elsewhere (Norris and Inglehart 2004) that religious values are strongly related to active participation in collective religious services and to the frequency of individualized prayer or meditation. Not surprisingly, where people thought that God was important in their lives, they were more likely to engage in religious rituals and practices. Moreover religious values are also an important proxy for societal patterns of secularization; we established that the strongest decline in religious values has occurred in affluent and secure nations, where the importance of religion has faded most. Secularization involves the lessening importance of religion in people s lives, and growing indifference to spiritual matters among the public. By contrast, where religious values continue to remain a vital part of people s everyday lives, in poor agrarian societies, where people are also most active in worship and prayer. The 10-point importance of God scale was carried in all five waves of the World Values Survey, facilitating the broadest comparison over time and across societies. To double-check on the robustness of this indicator, since 1990 the survey has also monitored the importance of religion, using a 4-point scale. Models were replicated using this indicator but this did not substantially alter the outcome of the analysis, lending greater confidence to the results. Identifying the predominant religious culture in each country is important because the values and beliefs of Catholicism, Protestantism, Orthodoxy, Muslim and Eastern religions are 13

14 expected to imprint themselves on each society, via the major channels of cultural transmission and socialization, irrespective of how far individuals actively participate in religion. The classification of the predominant religion in each society is drawn from the Encyclopedia Britannica Book of the Year 2001, using a dataset on religious pluralism derived from this source collected by Alesina and colleagues (2003). 5 The classification of all the countries contained in the combined WVS under this categorization is given in Table A1. Of all these, the categorization of eastern religions remains the most heterogenious, given the vast diversity of religions and sects contained within this group. In addition, the classification is less useful in pluralistic societies containing highly heterogeneous religious populations. The reliability and consistency of the classification was crosschecked against alternative reference sources that are widely cited in the research literature. 6 The type of society was also classified into three categories using the UNDP Human Development Index 2004, combining literacy and education, longevity and per capita income (UNDP 2007). Details about the measures and classifications used for all variables are given in the technical appendix. It should be noted that important differences in the countries and the questions included in each wave of the survey limit the multivariate analysis, for example the Schwartz value scales for the importance of security and risk were only contained in the most recent (2005-6) wave. The earliest surveys were conducted mainly in Protestant and Roman Catholic societies, so that time-series analysis over successive surveys is limited for other types of religion. III: Gender gaps in religious values As a first step, to get a sense of the cross-cultural contrasts, we can start by comparing the size of the gender gap in religious values by aggregating all five waves of the World Values Survey from 1981 to The gender gap in each society is measured as the mean position of women on the 10-point importance of God scale minus the mean position of men. A positive coefficient indicates that women are more religious in their priorities. An alternative comparison, using the 4-point importance of religion scale, generated similar contrasts, which increases confidence in the results of the comparison, although this measure was not included in the first wave of the survey. Cross-cultural comparisons Figure 1 illustrates the main cross-cultural contrasts, confirming the long-standing observation that women tend to attach greater importance to religion than men in almost all 14

15 societies under comparison, and the gender gap is often fairly substantial in size. Nevertheless despite this generalization, the societies vary in the size of the gender gap on the 10-point scale, with the largest gender gap evident in Orthodox societies such has Hungary, Russia, Ukraine, and Belarus, where the church had been suppressed for decades by the Communist state; in Italy, Portugal and Spain in Catholic Europe, as well as in Norway and Finland, two relatively secular Evangelical-Lutheran cultures. A wide range of faiths and levels of development are evident in the middle range of the scale while at the opposite pole there is a minimal gender gap in religious values, and even a modest reversal, found in some Muslim states such as Saudi Arabia, Pakistan, Jordan, Indonesia, Iraq, Morocco and Egypt. Many of those living in Muslim cultures express such a universally high level of religiosity that a ceiling effect may operate in the scale, flattening any social differences. [Figure 1 and Table 1 about here] To examine the societal-level factors linked with religious values more broadly we can compare the salience of religion at aggregate level by type of society, by the major types of world religion, and by world region, starting with bivariate descriptive statistics. Confirming the existential security thesis, Table 1 illustrates that religion was generally regarded as far more important in agrarian societies, classified by the UNDP as the lowest level of human development; on the 10-point scale of the importance of God, there was a 3-point gap between people living in the more affluent post-industrial nations (mean 6.35) and those living in agrarian societies (mean 9.45). This is consistent with what we have documented elsewhere (Norris and Inglehart 2004) where we have shown that similar disparities exist between rich and poor nations in many other indicators of religiosity, such as in patterns of participation in collective religious rituals and practices. Religion is almost universal in some of the poorest developing countries under comparison, such as Pakistan, Ghana and Indonesia, while by contrast, although there are a few notable exceptions such as Italy, Ireland, and the United States, many other affluent societies such as Sweden, Japan, Australia, and France are among the most secular around the globe. The comparison by the major type of world religious culture also shows significant differences, with the predominately Muslim cultures attaching by far the greatest importance to spiritual values, while Jewish and Catholic cultures proved moderate, while Protestant and Orthodox religious cultures generally proved more secular. 7 The comparison by world regions confirmed the patterns observed for development, with Scandinavia and Western Europe proving 15

16 the most secular, which the Middle East and North Africa, and Sub-Saharan Africa, were the most religious. [Table 2 about here] At the same time, when we examine the gender gap in religious values, the opposite patterns can be observed; as Miller and Stark (2002) noted, the gender gap is greatest in postindustrial societies while it diminishes to insignificance in poorer agrarian societies. The gap is also strongest in the more secular Orthodox and Protestant cultures, while shrinking in Muslim nations. And the gap is also largest in Scandinavia and Western Europe. All of these patterns suggest that processes of secularization and human development, diminishing the importance of religion in people s lives in many of the most affluent societies, have affected men more than women, widening the religious gender gap. Table 2 describes the bivariate correlations at societal-level between the mean importance of religion for women and for men, as well as the size of the gender gap, compared against a range of standard indicators of human development and existential security, including per capita GDP (in purchasing power parity), the Human Development Index, the Gender-related Development Index, and measures of infant mortality, life expectancy, fertility rates and female literacy rates. All the indicators prove to be strongly correlated in the expected direction; women are far more religious in developing societies with higher poverty, shorter life-spans, greater fertility and lower literacy for women. But at the same time these factors also proved to be equally or even more strongly correlated with male religiosity. As a result the gender gap is reversed; women remain more religious than men in more affluent societies that have experienced greater human development. Thus religiosity for both men and women tends to be far stronger in developing societies, but the religious gender gap seems to expand with processes of modernization. [Table 3] Within-society comparisons To explore these phenomena further, let us compare men and women s religiosity within given societies. Table 3 provides similar comparisons of the mean importance of religion scale among men, women, and the resultant gender gap, across a range of social sectors and demographic groups in the pooled sample across all waves of the World Values Survey. Household income was classified into deciles in each society. Consistent with the existential security thesis, the level of household income on the 10-point scale was strongly correlated with 16

17 the distribution of religious values, with a 1.5 point gap between rich and poor. This means that the across-nation contrasts between agrarian and post-industrial societies was therefore roughly twice as strong as the within-nation comparison of rich and poor social sectors, but both point consistently in the direction expected by the existential security thesis. Religion was seen as more important for poor women than for rich women and similar contrasts by household income group were also observed among men. Nevertheless women were under-represented (46%) in the more affluent group, while they constituted 56% of the lowest income group. We also find a relationship with reported household savings, where those who reported saving money were less religious than those who experienced debt. Health showed similar patterns, with those reporting that they experienced poor health proving more religious than those saying that they were in very good condition. Similarly with age, the elderly were more religious than the young, as were the less educated compared with those with greater education. Those looking after dependent children were also more religious than those without children at home, as well those engaged in home-care compared with those in the paid workforce. By contrast, marital status did not prove to be significantly related to religiosity. In short, across most of the standard social and demographic variables, relatively strong religiosity was consistently found among the less well-off sectors of society, who tend to be more vulnerable to threats and dangers, including the poor and indebted, the elderly, the less educated, and those in poor health. Moreover there are more women than men in each of these populations; women are roughly 51.5% of the total population in the database as a whole, but they constitute 56% of the elderly, 56% of the poorest income group, and 59% of those with ill-health, and 98% of the home-makers in the world. [Table 4 about here] Cultural attitudes and religiosity All the social and demographic indicators which we have compared, and which are associated with the underlying concept of existential security, are therefore consistent predictors of the salience of religious values, but is there more direct attitudinal evidence? If we examine the cultural indicators, in Table 4, we can monitor attitudes towards security, risk and the gender equality scale. The Schwartz values scales contained in the fifth wave of the WVS asked respondents how well given values described themselves. As Table 4 shows, people who expressed a strong propensity to seek secure living conditions away from danger were also highly 17

18 religious, while those who did not, proved more secular. This provides direct evidence in support of the existential security thesis. We can also compare these results with the Miller and Stark (2002) thesis which predicts that in religions promising an afterlife, the propensity to take risks will be associated with more secular orientations, as well as being a more male characteristic. The Schwartz value scale asked respondents about the value of adventure, taking risks, and having an exciting life. Contrary to the Miller and Stark thesis, the results in table 4 show that those who were attracted to risks and adventure were more religious on average than those who were not. Therefore although women were disproportionately more likely to give high priority to security, while men were far more likely to value taking risks, emphasis on security proved a better predictor of the gender gap in religiosity than did one s emphasis on risk-taking. Lastly, the gender equality scale is also strongly correlated with religiosity; those expressing the most traditional views of women s roles in education, employment and political leadership also give the highest priority to spiritual values. This relationship, however, is probably recursive; many religious faiths and spiritual authorities emphasize a traditional division of sex roles in the home, family and public sphere, so that participation in religious activities is likely to reinforce these values, while at the same time those holding more egalitarian values may be less likely to be attracted to participate in religious organizations. To summarize the bivariate analysis in the pooled WVS across all waves, Figure 2 illustrates the size of the gender gap in religious values (the 10-point importance of God scale) across the range of societal-level and individual-level factors discussed here. As the bar chart shows, the gender gap was strongest in post-industrial societies, especially in relatively secular Scandinavia and Western Europe, as well as among the lowest income sectors of society, those with the worst state of health, the elderly, those who actively prioritize security, and those with egalitarian values towards gender equality. Women predominate in all these groups, associated with long-standing and well-established patterns of female poverty, inequality, and status, greater longevity, and men and women s cultural attitudes. By contrast, it is worth noting the dog that did not bark: the religious gender gap is relatively small or inconsistent according to marital status and the propensity to take risk. The Schwartz value items were only included in the fifth wave of the WVS survey so the multivariate analysis is limited to this wave, currently containing 44 societies and 61,618 18