Aging,Social Developmental,and Cultural Factors in Changing Patterns of Religious Involvement Over a 32-Year Period

Abstract

Religious change has been studied from the perspectives of both individual development and societal progress, but a lack of long-term longitudinal data has limited the capacity to examine them simultaneously. This study uses repeated cross-sectional data from the World Values Survey/European Values Study, covering 80 nations over the period from 1981 to 2013 to estimate age, period, and cohort effects on mean changes in the subjective importance of God and attendance at religious services. A cross-classified mixed model approach was used, examining both random effects indicating between-country differences in these changes and fixed factors unifying them in a broader framework. Older age was associated with greater personal and organizational religious involvement in a large majority of societies, but the strength of this association differed by culture, with the largest mean effects occurring in Western nations. Period effects were detected in many cultural areas but were very heterogeneous in direction and magnitude. Period changes were related to national wealth, with increase in per capita gross domestic product being related to declines in mean religious involvement. Cohort effects were in evidence in relatively few societies. These results indicate that both individual aging processes and changes in the material environment may influence changes in religious involvement, in keeping with both psychological and sociological theory, but that culture also plays a role in the nature and speed of these changes.

Keywords

religion culture aging cohort effects societal change wealth

Religious involvement is a subject that has been of intense interest to scholars of both society and the mind since the earliest days of these disciplines (e.g., Freud, 1989; James, 1902; Marx, 1970). Religion is a complex phenomenon, involving a range of beliefs, behaviors, and identities (Fetzer Institute & National Institute on Aging, 1999; Hill et al., 2000). Moreover, it is highly dynamic, often shifting in its usage, meaning, and salience both within individuals across the life span and within societies through time (Finke & Iannaccone, 1993; Hayward & Krause, 2013a; Inglehart & Baker, 2000; McCullough, Enders, Brion, & Jain, 2005). It is this dynamic quality that has often been one of the most controversial aspects regarding the study of religion but also one of the most difficult to study empirically. Some researchers, particularly those working from a developmental or cognitive perspective, have tended to see religious involvement of some kind as a relatively universal human phenomenon (Atran & Norenzayan, 2005; Barrett, 2000) and one that changes largely within individuals in response to psychological needs (Norenzayan, Dar-Nimrod, Hansen, & Proulx, 2009; Uecker, Regnerus, & Vaaler, 2007). Others, particularly those operating from the perspective of social theory, have often viewed involvement in religion as a learned response to particular conditions of the social environment and, thus, something that is prone to change across societies in response to factors such as material development (Inglehart, 1990, 1997; McCleary & Barro, 2006). Although recent years have seen a rapid increase in the number of longitudinal and repeated cross-sectional studies mapping the nature of change over time in religious involvement (Hayward & Krause, 2013a, 2013b; Schwadel, 2010a, 2011), it remains difficult to distinguish between changes that may stem from these two quite different sources.

In this study, we use one of the largest and most comprehensive datasets regarding change in religious involvement available, the World Values Survey/European Values Study (WVS/EVS), covering more than 700,000 individuals from 80 countries and a span of more than 30 years to begin to illuminate the intersection between social and developmental elements of these trends. Each nation examined has been included in the WVS/EVS in between 2 and 6 waves. The advantage of having repeated cross-sectional data is that we can apply a cross-classified random effects modeling (CCREMs) approach to age–period–cohort analysis (Raudenbush, 1993; Yang & Land, 2006, 2008). Two elements of this approach are key in addressing the problems of linear dependency between age, period, and cohort effects inherent in using cross-sectional data. First, birth cohorts are defined by grouping participants into broader categories (e.g., by decade) which are analyzed categorically, sidestepping the dependency between age and birth year. Second, and more importantly, repeated data collection means that—for example—a 30-year-old participant in 1981 (born in 1951) can be compared with a 30-year-old participant in 2011 (born in 1981). Because the 2011 study also includes individuals from the 1951 birth cohort (by then aged 60), the CCREM method can be used to estimate the proportion of the difference in the outcome variable at each time attributable to being age 30 versus 60 (age effect), being born in 1951 versus 1981 (cohort effect), or being surveyed in 1981 versus 2011 (period effect). These two elements are combined and executed by using multilevel modeling methods to treat cohort (i.e., decade of birth) and period (i.e., time of study) as grouping elements within which individual observations are nested.

Religious Involvement and Individual Development

A long-standing issue in the psychology of religion is the question of whether and how religious involvement changes with age. Early researchers documented phenomena including rejection of religious upbringing in adolescence (Starbuck, 1901), religious conversion in adulthood, and increasing interest in religious affairs later in life (James, 1902). Although there is fairly clear evidence that older individuals tend to be more involved in religion across a range of dimensions in comparison with their younger contemporaries, the supposition that this cross-sectional correlation can be attributed to intraindividual change during adulthood remains inconclusively resolved (Krause, 2008). From a developmental perspective, a number of theorists have suggested that religious involvement becomes more important to individuals as they approach late life because it can be a means of creating a sense of meaning and coherence in the perception of one’s life that becomes increasingly important during the final stage of human development (Erikson, 1968; Tornstam, 1997).

Social-psychological theory places a more cognitive emphasis on the same basic notion, positing that religious involvement serves as means of reducing the anxiety associated with factors including uncertainty about the self in relation to others (Hogg, Adelman, & Blagg, 2010) and the knowledge of one’s own death (Vail et al., 2010). Because aging tends to make these concerns more salient, religious involvement is thought by this view to become more important as time goes on. At the same time, religious groups serve as important sources of both tangible and emotional social support (Krause, 2002, 2008), and thus, older adults’ increasing needs may serve as an impetus to increase their religious involvement to secure more of this support (Hayward & Krause, 2013b). Other possible explanations of the relationship between aging and religious involvement include the possibility that older adults simply have more free time to devote to religious activities (e.g., Shaver & Sosis, 2014).

Unfortunately, because the vast majority of this research is cross-sectional in nature, it is difficult to separate aging and cohort effects. That is, more recent generations may be less involved with religion than their parents and grandparents were, making its correlation with age an artifact of changes in socialization over the years. For example, a recent study by the Pew Forum (Pew Forum on Religion and Public Life, 2012) concluded that, in the United States, members of the Millennial generation (born after 1980) are much less likely to belong to a religious group than were members of previous cohorts. In addition, whereas psychological explanations of within-person religious change generally imply that these effects should be broadly universal, the vast majority of empirical research in this area is confined to a small spectrum of Western cultures, which may not be representative of the human condition as a whole (Henrich, Heine, & Norenzayan, 2010). Aging may not affect religious involvement in the same way in all cultural environments, either because it does not fulfill the same basic psychological needs in all contexts or because the same needs do develop across the life course in the same way in all contexts. For example, aging may not increase mortality salience and self-uncertainty in cultures that place greater value on social roles held by the elderly (Martens, Greenberg, Schimel, & Landau, 2004), or conversely, these sources of anxiety may arise in the same way but may be addressed using other nonreligious means.

Religious Involvement and Societal Change

The other long-standing perspective on religious change focuses on sociocultural change rather than on individual development. The antecedents of this view can be found, for example, in Marx (1970) and Freud (1989), but it has more recently been developed in the form of secularization theory (Swatos & Christiano, 1999). According to this general view, being involved in religion fulfills certain social functions, such as providing explanations for important existential questions and acting as a mechanism of social control, that may also be better fulfilled by other more recent sociocultural developments, such as the scientific method and the modern state. Hence, social development is expected to lead the weakening of religious influence over time. Commensurate with this perspective, a number of studies have found that wealthier countries and those with more highly educated populations also tend to have lower levels of adherence to religious measures such as belief in God and attendance at religious worship services (Barber, 2011; Diener, Tay, & Myers, 2011; Ruiter & van Tubergen, 2009; Verweij, Ester, & Nauta, 1997). A related line of research attributes the relationship between rising wealth and decreasing religious involvement to the strengthening of secular institutions that fulfill certain social functions, for example, inculcating a sense of generalized trust by providing an external means of social control, that were previously underwritten by shared religious adherence (Norenzayan, 2013; Norenzayan & Gervais, 2013). However, the resilience of religious adherence even in the most highly developed economies, and the resurgence of conservative religious ideologies worldwide in recent decades, has cast doubt on the most stringent formulation of secularization theory (Berger, 1999; Stark, 1999; Swatos & Christiano, 1999). A more nuanced view is given by the postmaterialist value shift perspective (Inglehart, 1997), which holds that material development is associated with a change in cultural values away from those emphasizing survival and group order (including religious orthodoxy) and toward those emphasizing self-expression and rational inquiry (which may nevertheless include more individualized forms of religious involvement). Cross-cultural research provides some support for this view (Inglehart & Baker, 2000).

However, studies of within-country change in religious beliefs and practices give a mixed picture. For example, Finke and Stark (2005) used a variety of historical records of church attendance in the United States to piece together a picture of changes from colonial times to the end of the 20th century and found, in contrast to the secularization narrative, that religious involvement increased dramatically over the course of the 19th and early 20th centuries before remaining nearly stable or declining only marginally. Similarly, Gallup and Jones (1989) found that levels of belief in God remained highly stable in the latter half of the 20th century, and Presser and Chaves (2007) found a similar picture with respect to frequency of attendance at religious services. Using more advanced methods of analyzing repeated cross-sectional data from the General Social Survey, Schwadel (2011) found a mixed pattern depending on the measure of religious involvement examined, with worship attendance and belief in the literal truth of the Bible declining between 1972 and 2006, but belief in the afterlife and the frequency of prayer increasing slightly. Conversely, studies in some parts of Europe have tended to show more evidence of decreasing religious involvement over a similar period, including in Great Britain (Crockett & Voas, 2006) and the Netherlands (Need & de Graaf, 1996).

These apparent cultural differences may provide some indication that the relationship between development and religious involvement is not uniform. Supporting this general viewpoint is cross-cultural research from the postmaterialism perspective (Inglehart, 1997; Inglehart & Baker, 2000). Using WVS/EVS data, these researchers examined regional differences in basic values, particularly on axes-related survival versus self-expression (prioritizing physical safety over quality of life), and traditionalism versus rationalism (prioritizing obedience to traditional authority over reasoned judgment). They found that although there were global trends toward more emphasis on self-expression and rationalism over time (particularly as national development increased), mean national scores on these value indexes also tended to be clustered in ways that could be linked with broader shared historical cultures. For example, although they share many developmental similarities in terms of politics and economics, persistent differences in values can still be identified between historically Catholic (e.g., France and Italy) and Protestant (e.g., Germany and Switzerland) nations in Europe (Hayward & Kemmelmeier, 2011; Inglehart & Baker, 2000). The historically dominant religious group in a region appears to play a significant role in identifying these cultural zones (Catholic Europe, Protestant Europe, Orthodox nations, Confucian nations, Islamic nations), whereas geography (Africa, the Baltic, South Asia) and history of settlement (Latin America, English-speaking nations) define others (Inglehart & Baker, 2000). Trajectories of change in religious involvement may differ between cultural zones, because religion plays a different role in individuals’ lives or because certain values are more persistent in some social environments. Equally, life course trajectories of religious involvement may also vary between cultures; some value complexes may encourage lifelong involvement, whereas in others, it may play a more context-specific role and, thus, change more with life circumstances. However, there has been relatively little research applying the same measures to comparable samples both across nations and over time, making it difficult to assess the nature of these differences.

The Present Study

This study aims to begin filling these gaps by charting changes in two dimensions of religious involvement—subjective importance of God and frequency of attendance at religious services—over a period of decades and across a spectrum of nations and cultures in an age–period–cohort model. We focus on these dimensions because they provide good coverage of two contrasting elements of religious involvement, the individual’s psychological investment in religious belief (i.e., individual religiousness) and embeddedness in a religious group or community (i.e., organizational religiousness). Although it is debatable that the extent to which these facets of religious involvement carry the same importance among adherents of different religious traditions, these elements are common and pervasive throughout the world. These methods allow us to use repeated cross-sectional data to estimate components of change related to individual age, historical period, and generational differences. In doing so, we evaluate expectations derived from two major theoretical perspectives outlined above.

The first hypotheses regard age and cohort effects. We aim to examine the cross-cultural generality of the relationship between aging and religious involvement that has been found in research in a limited number of national settings (primarily in studies conducted in the United States). Such studies have largely indicated that elements of both individual religious involvement (e.g., the subjective importance of God) and organized religious involvement (e.g., attendance at religious services) tend to be more prevalent at older ages. Two potential explanations of this relationship can be contrasted using this dataset. First religious involvement may increase with age as a function of basic elements of human psychological development, such as increasing needs for meaning-making (Cacioppo, Hawkley, Rickett, & Masi, 2005) and self-uncertainty reduction (Hogg et al., 2010). As such, we would expect to find that the strength of this relationship should be constant across cultures—that is, the average age-related change should be about the same everywhere (Hypothesis 1a). Conversely, to the extent that age-related increases in religious involvement reflect an increasing reliance on the social benefits of religious participation in response to increasing needs that are unmet by other institutions, we would expect to see cultural differences, especially depending on the extent of other forms of social support (Hypothesis 1b). Related to both of these hypotheses is the question of whether age-related differences in religious involvement are due primarily to intraindividual change or to enduring generational differences between cohorts. Because the timing and meaning of generations vary widely between nations, and because their potential impact on religious involvement in each of these contexts remains a matter of conjecture, we do not propose any specific hypotheses with respect to the nature of these cohort effects, but we predict that the age effects examined above will persist in spite of controlling for differences between generations (Hypothesis 2).

Second, we examine the view that material development is related to a decline in religious involvement. The repeated nature of the data at the national level allows us to examine two important elements of this relationship. First, there should be a negative relationship in mean national wealth (i.e., between countries) and the extent of mean religious involvement, with wealthier nations having lower mean levels of both personal and organizational involvement among their populations in comparison with poorer nations (Hypothesis 3a). Second, and much more difficult to evaluate without this type of time series data, there should also be a negative relationship between the extent of within-country change in wealth and religious involvement—that is, religious involvement should decline most rapidly in the countries experiencing the greatest increases in wealth (Hypothesis 3b). Postmaterialism theory may suggest a somewhat different set of expectations in comparison with secularization theory. If it is assumed that value shift primarily affects religious traditionalism, then it is possible that development is related to declining religious attendance (an aspect of involvement in traditional organized religion) but not necessarily the subjective importance of God (a more individual aspect of involvement; Hypothesis 3c).

Method

Individual-level data for this study came from the WVS Waves 1 to 6 (WVS Association, 2014b) and EVS Waves 1 to 5 (EVS Group, 2011). The WVS and EVS are coordinated cross-national social surveys which, collectively, have been carried out in 108 nations over a period of 33 years from 1981 to 2014. Because the core of both surveys includes, by design, a broad range of overlapping items, it is possible to combine them for analytical purposes. Not all nations were included in all waves of data collection; the first wave of the WVS/EVS included 23 nations, whereas subsequent waves included as many as 64 (in Wave 4), and nations included in one wave were not necessarily repeated in the next. Thus, coverage varies widely, with the number of available survey waves for each nation ranging from 1 to 6. Because this article focuses on national change, only those nations that were surveyed at least twice were included in the present analyses. In addition, the survey instruments used in different nations occasionally excluded key variables analyzed in this study, and hence, surveys that did not measure either importance of God or religious attendance were eliminated from these analyses. This left 272 surveys from 80 separate nations (total individual-level N = 726,977).

The design of the WVS/EVS includes parallel repeated cross-sectional surveys conducted by investigators operating within each sampled country. In each case, survey items were translated into local languages and adapted when necessary to fit local cultural conditions (e.g., the standard response options for religious group affiliation would be very different in the United States and in India) and back-translated for verification. Sampling, using a mixture of probability and nonprobability methods depending upon country and wave of data collection, was conducted within each nation, and data were collected in person by trained interviewers (WVS Association, 2014a).

Measures

Religious involvement

This study examines two religious variables: the importance of God and frequency of religious attendance. Each of these is measured with a single item. The question “How important is God in your life?” had responses on a 10-point scale anchored with not at all important and very important. Responses for the question “Apart from weddings and funerals, how often do you attend religious services these days?” were given on a 7-point scale ranging from never or practically never to more than once a week.

Age and birth cohort

Respondents reported their current age and year of birth. Ages ranged from 12 to 104 (M = 41.8, SD = 16.5). The challenges involved in specifying appropriate birth cohorts has long been an issue of discussion, as it is difficult to specify meaningful cohorts in the absence of strong theoretical expectations (Mason, Mason, Winsborough, & Poole, 1973). In this case, the challenges are even greater, because generational influences and timing are likely to vary greatly between countries. Therefore, we adopted the strategy of defining cohorts in uniform 10-year intervals across all countries. Birth cohort was defined by decade (e.g., the 1980 cohort includes those born in the years 1980-1989) and ranged in the data from 1880 to 1990.

Demographics

Demographic covariates included in the model for reasons of control included gender and income. Participants’ gender was classified as male or female by self-report. Income was coded within country by decile (those in lower than the 10th percentile of incomes coded as 1, those in the 10th-20th percentiles as 2, and so forth). Previous research suggests that women tend to be more religiously involved than men (Miller & Stark, 2002). Individual income has also been found to be related to less religious involvement (Ruiter & van Tubergen, 2009).

National wealth

Historical per capita gross domestic product (GDP) figures in constant 2005 U.S. dollars were obtained from the World Bank (2013). These figures were matched with the individual-level data within nation by year of survey administration. Two versions of GDP are used in these analyses. Mean GDP (averaged within country across all waves included in the WVS/EVS) is used to gauge the impact of wealth differences between countries. The centered within-country deviation from the national mean was used to gauge within-country differences based on change in per capita GDP over time and was computed by subtracting the figure for the year corresponding to each wave from the across-wave mean.

National culture

Countries were classified on the basis of historical culture using the categories outlined by Inglehart and Baker (2000) in their analysis of value complexes in the WVS/EVS data. That study found that countries in each of these historical cultural zones tend to group together in terms of their orientations on a variety of values, in addition to having important elements of shared history. Validation of these zones was based partially on position on axes of traditionalism/rationalism and survival/self-expression. Regular religious attendance was one component of the traditionalism/rationalism scale, leading to some concern about conflation with the outcome measures in the current study. However, historical cultural zones also conform to regions defined by geography and shared history.

Analytical Approach

The data were analyzed using a CCREM approach (Yang & Land, 2006). Briefly, CCREM provides a strategy for attempting to partially disentangle age, period, and cohort effects in repeated cross-sectional data, by taking advantage of the fact that members of different cohorts have been observed at a range of different ages. This allows us to compare individuals from different cohorts when they were the same age. For example, individuals born in 1960 would be ages 30 and 50 in 1990 and 2010, respectively, whereas individuals born in 1970 would be 20 and 40 when measured at the same time. The model serves to partial out differences between, for example, the 1960 and 1970 cohorts in 1990 and 2000, when they were both, respectively, 30 years old (age effect), from differences shared between these cohorts in 1990 and 2000, irrespective of age (period effect), from differences between the two that remain stable between 1990 and 2000 (cohort effects). Although it is of course logically impossible to separate these effects fully in the absence of longitudinal data (Bell & Jones, 2014), it is a valuable method for estimating effects that would otherwise be impossible to separate and has been applied to a number of outcomes in the recent literature (Dassonneville, 2013; Schwadel, 2010b; Yang, 2008).

Using this approach, we evaluate two models of age–period–cohort change in the importance of God and attendance at religious services. Age is an individual-level (Level 1) predictor, while respondents are cross-classified at Level 2 both by birth decade (cohort) and year of data collection (period). Cohort and survey period are both nested in turn within country (Level 3), with random effects for age and year specified at Level 3. Model 1 focuses on the role of absolute and relative national wealth in changes in these dimensions of religious involvement over time. Individual-level predictors include age, gender, and income. Mean GDP is a predictor at the national level, and change in GDP (relative to the national mean) and year are predictors at the survey-wave level. Model 2 provides a contrast by instead examining differences in age and period effects between historical cultural zones—GDP variables are removed, and cross-level culture by age and culture by year effects are added. Additional computational details and specifications for this model are given in the supplementary file.

Results

Descriptive statistics by wave are presented in Table 1. Deletion due to item nonresponse led to 7.3% missing data in the CCREM models for the subjective importance of God and 6.3% missing in the models for frequency of religious attendance.

Table 1.

Descriptive Statistics by Wave.

	Wave 1	Wave 2	Wave 3	Wave 4	Wave 5	Wave 6	Total
	M (SD) or n (%)	M (SD) or n (%)	M (SD) or n (%)	M (SD) or n (%)	M (SD) or n (%)	M (SD) or n (%)	M (SD) or n (%)
Individual	56,760	125,542	154,014	187,890	141,830	60,941	726,977
Religious importance	6.5 (3.2)	6.3 (3.4)	7.1 (3.2)	7.6 (3.1)	7.7 (3.0)	7.4 (3.1)	7.2 (3.2)
Religious attendance	4.3 (2.6)	4.0 (2.6)	4.4 (2.5)	4.5 (2.4)	4.5 (2.6)	4.3 (2.6)	4.4 (2.6)
Gender (male)	24,900 (52.0%)	65,078 (51.8%)	79,776 (51.8%)	97,778 (52.1%)	74,234 (52.3%)	32,243 (52.9%)	374,009 (52.1%)
Age	41.0 (17.9)	42.2 (16.4)	41.0 (15.9)	41.5 (16.4)	42.4 (16.6)	43.3 (17.1)	41.8 (16.5)
Cohort
1880	24 (0.1%)						24 (<0.1%)
1890	266 (0.7%)	20 (<0.1%)		2 (<0.1%)			288 (<0.1%)
1900	2,036 (6.0%)	1,230 (1.0%)	208 (0.1%)	160 (0.1%)	6 (<0.1%)		3,640 (0.5%)
1910	3,882 (10.1%)	5,642 (4.5%)	2,238 (1.5%)	1,778 (0.9%)	336 (0.2%)	22 (<0.1%)	13,898 (2.0%)
1920	4,872 (12.7%)	13,534 (10.8%)	9.514 (6.2%)	9,512 (5.1%)	3,618 (2.6%)	632 (1.0%)	41,682 (5.9%)
1930	5,214 (13.5%)	17,876 (14.3%)	17,338 (11.3%)	18, 486 (9.9%)	10,086 (7.1%)	3,219 (5.3%)	72,219 (10.2%)
1940	6,818 (17.7%)	21,698 (17.3%)	21,512 (14.0%)	23,532 (12.5%)	16,422 (11.6%)	6,305 (10.4%)	96,287 (13.6%)
1950	9,360 (24.3%)	27,894 (22.3%)	32,202 (21.0%)	33,408 (17.8%)	22,618 (16.0%)	8,837 (14.5%)	134,319 (19.0%)
1960	6,034 (15.7%)	28,304 (22.6%)	36,890 (24.0%)	40,592 (21.6%)	27,908 (19.7%)	10,398 (17.1%)	150,126 (21.2%)
1970		9,064 (7.2%)	32,580 (21.2%)	44,252 (23.6%)	30,400 (21.5%)	11,382 (18.7%)	127,678 (18.0%)
1980			1,204 (0.8%)	15,814 (8.4%)	29,314 (20.7%)	13,772 (22.6%)	60,104 (8.5%)
1990					972 (0.7%)	6,262 (10.3%)	7,234 (1.0%)
Country	23	42	53	64	48	42	81
Year of survey	1981.6 (0.8)	1990.2 (0.6)	1996.3 (1.0)	2000.2 (1.2)	2006.2 (0.9)	2011.6 (0.8)	1998.3 (7.9)
GDP (natural log)	9.7 (0.6)	9.2 (1.2)	8.3 (1.3)	8.6 (1.4)	8.7 (1.4)	8.7 (1.4)	8.7 (1.4)

Table 2 presents the fixed effects for the two CCREM models described above. Consistent with previous findings, the coefficients for the demographic controls indicate that women tend to have greater average subjective religiousness and more frequent religious attendance than men and that higher income is related to lower levels of religious involvement on both measures. Model 1 indicates that there is a positive association between age and the importance of God (even after accounting for period and cohort differences). In addition, net of development effects, there was a mean increase in the country-level mean for the importance of God over the course of the period studied in the WVS/EVS. Furthermore, there were both between and within-country effects for development. Wealthier countries had lower mean importance of God, as indicated by the coefficient for mean GDP. The same pattern held within the same country as well, with lower levels of mean importance of God observed in years when GDP was higher. The same basic pattern was found for frequency of religious service attendance, with the exception that Model 1 for that outcome indicated that there was no mean effect for year (i.e., no period effect on the global scale, net of development effects). Thus, these models offer support for the notion that economic development is linked with declining levels of religious involvement on the national scale.

Table 2.

Fixed Effects Summary for Cross-Classified Mixed Effects Model.

	Importance of God				Religious attendance
	Model 1		Model 2		Model 1		Model 2
	b	(SE)	b	(SE)	b	(SE)	b	(SE)
Individual variables (Level 1)
Age	0.02***	(0.002)			0.02***	(0.001)
Gender^a	0.70***	(0.006)	0.70***	(0.006)	0.31***	(0.006)	0.32***	(0.006)
Income	−0.04***	(0.001)	−0.04***	(0.001)	−0.002	(0.001)	−0.002*	(0.001)
Country × Wave variables (Level 2)
Year of survey	0.03***	(0.008)			0.004	(0.007)
Wave GDP^b,c	−0.06***	(0.01)			−0.04***	(0.007)
Country variables (Level 3)
M GDP^b,d	−0.07***	(0.02)			−0.04***	(0.01)
Cultural zone
Africa			9.27***	(0.62)			6.48***	(0.52)
Age			0.002	(0.005)			−0.0001	(0.005)
Year			−0.02	(0.02)			0.0007	(0.02)
Baltic			4.77***	(0.66)			3.04***	(0.41)
Age			0.04***	(0.006)			0.02***	(0.003)
Year			−0.006	(0.03)			−0.006	(0.02)
Catholic Europe			6.15***	(0.49)			3.26***	(0.41)
Age			0.04***	(0.003)			0.02***	(0.003)
Year			−0.02*	(0.007)			−0.02**	(0.008)
Confucian			5.00***	(0.59)			3.26***	(0.48)
Age			0.01**	(0.005)			0.01**	(0.004)
Year			0.03*	(0.01)			0.008	(0.01)
English-speaking			6.74***	(0.54)			4.12***	(0.44)
Age			0.03***	(0.004)			0.02***	(0.004)
Year			−0.02*	(0.007)			−0.03***	(0.008)
Islamic			8.92***	(0.49)			4.94***	(0.41)
Age			0.005	(0.003)			0.02***	(0.003)
Year			0.02	(0.01)			−0.02*	(0.01)
Latin America			8.61	(0.52)			4.74***	(0.43)
Age			0.01**	(0.004)			0.02***	(0.004)
Year			0.02	(0.01)			−0.02*	(0.01)
Orthodox			6.45***	(0.50)			3.65***	(0.41)
Age			0.008*	(0.003)			0.006	(0.003)
Year			0.11***	(0.01)			0.05***	(0.01)
Protestant Europe			4.67***	(0.38)			2.79***	(0.31)
Age			0.05***	(0.004)			0.02***	(0.004)
Year			−0.03***	(0.007)			−0.03***	(0.007)
Southeast Asia			8.77***	(0.67)			5.96***	(0.57)
Age			0.004	(0.006)			0.01*	(0.006)
Year			0.007	(0.02)			0.0007	(0.02)
Model Covariance and Fit
Country ICC	.32		.15		.21		.09
Cohort ICC	.004		.005		.005		.005
Period ICC	.01		.02		.02		.02
Fit (BIC)	2,817,646		2,855,483		2,718,670		2,755,446
Null Model Covariance Parameters
Country ICC	.35				.24
Cohort ICC	.004				.005
Period ICC	.008				.02
Fit (BIC)	3,086,890				2,965,949

Note. GDP = gross domestic product; ICC = intraclass correlation coefficient; BIC = Bayesian information criteria.

Male is the comparison category.

Per capita GDP / 1,000.

Centered on country mean.

Mean for country across all waves, centered on grand mean.

p < .05. **p < .01. ***p < .001.

Model 2 was applied to test the cultural alternative hypothesis that, rather than differences in economic development, it is differences in cultural history that are most important in shaping the trajectory of national changes in religious involvement over time. Regarding the importance of God, we see large mean differences among historical cultural zones (indicated by differing intercepts between zones). The greatest extent of mean religious involvement was observed in Africa, with the least involvement reported in Protestant Europe. Positive individual-level age effects were detected in the Baltic, Catholic Europe, Confucian, English-speaking, Latin America, Orthodox, and Protestant Europe cultural zones; no age effects emerged in the Africa, Islamic, and Southeast Asia zones. With respect to period effects, negative change by year was found in the Catholic Europe, English-speaking, and Protestant Europe zones; conversely, increasing importance of God was found in the Confucian and Orthodox zones; no significant change was found in the Africa, Baltic, Islamic, Latin America, or Southeast Asia cultural zones. Results were substantially similar for attendance, with the following exceptions: There was no significant age effect in the Orthodox zone, whereas there was evidence of a significant positive effect in the Southeast Asia zone; there was no significant period effect in the Confucian zone, but there were significant negative time trajectories in the Islamic and Latin America cultural zones.

Finally, the model covariance and fit statistics (see Table 2) allow for the two hypothetical models to be compared, as well as for the relative effect sizes of variables at each level of analysis to be estimated. A comparison of Bayesian information criteria (BIC) addresses the first of these questions. The BIC is a measure of fit between model and data that corrects for the number of parameters in the model, with a smaller BIC indicating better fit after penalizing models with larger numbers of variables (Kuha, 2004). For both outcomes, Model 1 fit substantially better than Model 2, indicating that differences in development (as represented by national GDP) may do a better job of explaining patterns of religious involvement than differences based on national historical culture. An analysis of the intraclass correlation coefficients (ICC) at each level helps to provide an estimate of the effects of the model variables at each level, indicating the extent of between-individual variation explained by within-group similarity within country, cohort, and survey year. The largest of these group effects are evidently between nations (i.e., due to shared national characteristics), with small but nonnegligible proportions also shared at the period and cohort levels. The cohort-level ICC figures provide the best method of evaluating the extent of cohort differences in these religious factors and their relative magnitude in comparison with age and period effects. Although not insignificant, cohort effects appear to account for the smallest proportion of variance by far for both outcomes.

Discussion

These results afford some important insights into some of the social and psychological factors that may contribute to changes in the way that individuals and populations are involved in religion over time. First, the use of an age–period–cohort model allowed us to test the developmental hypothesis that religious involvement increases with age, net of generational differences related to birth cohort (Hypothesis 1a). In support of this view, there was a positive relationship between age and both personal and organizational forms of religious involvement on average throughout the world. However, there was also significant variation between countries in terms of the magnitude of the age effect (Hypothesis 1b). Age appeared to be especially important in Western cultural zones (the English-speaking world and much of Europe), as well as in Latin America and the Confucian countries, but had no mean relationship in Africa. Results were mixed in Southeast Asia and the Islamic world. These cultural differences are inconsistent with the view that increasing religious concern is a universal aspect of adult development, although they may also reflect cross-cultural differences in the conception of God that may affect the interpretation of the measure used in this study. That is, Hypothesis 1a appeared to be better supported than Hypothesis 1b. Hypothesis 2 was also supported, given that the magnitude of variance explained by cohort effects was very small. Thus, there was minimal evidence that cohort replacement could account for apparent age differences in religious involvement. However, it is important to note that cohorts were defined atheoretically by birth decade for the purposes of this study, due to lack of guidance from either theory or previous research in this area; other methods of defining cohort by more meaningful generations might yield stronger results in this regard, highlighting the value of further research.

Several potential explanations have been put forth to account for the relationship between aging and religious involvement, including psychological factors such as an increasing need for meaning-making (Martens et al., 2004), needs for self-uncertainty reduction (Hogg et al., 2010), and an increasing need to draw upon social support and other resources mediated by religious involvement (Hayward & Krause, 2013b). The presence of substantial cultural differences would seem to point toward the importance of social factors, as changes in psychological needs are likely to be a relatively universal facet of adult development, whereas the extent of support drawn from the religious group is likely to be highly dependent on broader social conditions such as the availability of secular alternatives. However, it is also possible that different cultures also differ in some way in their propensity to provide alternative methods of satisfying psychological needs such as meaning-making and self-uncertainty reduction. The present study has provided valuable context by expanding the literature on religion and aging into a cross-cultural framework, shedding important new light by suggesting that the relationship between religion and aging is widespread, but not universal, and that it differs in magnitude even within the scope of places where it exists. Future research should build on this foundation by examining potential psychological and social causes of this relationship.

Second, the use of repeated cross-sectional data collection allows us to track changes in religious involvement at the national level, net of the effects of cohort replacement. Throughout the years, a number of theorists have argued that increasing material development should lead to diminishing religious involvement, as the growth of scientific literacy and the increasing personal security granted by general societal wealth and the presence of state institutions reduces the need for religious beliefs and institutions to fill these gaps (Barber, 2011; Freud, 1989; Inglehart, 1997; Marx, 1970). Our results provide support for that hypothesis. Wealthier nations had lower mean levels of religious involvement in comparison with poorer ones (in support of Hypothesis 3a), and most importantly, the rate of GDP change within each country was linked negatively with the rate of change in religious involvement over the study period (in support of Hypothesis 3b). Contrary to Hypothesis 3c, the same basic pattern held for both individual and organized involvement in religion. We interpret this as more supportive of the secularization view than the postmaterialism view of shifting religious involvement, as it seems to suggest that rising wealth may affect not only aspects of adherence to traditional norms but also more subjective and psychological aspects of religious involvement.

Other societal factors influencing the trajectories of these changes over time may include changes in the structure of the national market for religion (Finke & Iannaccone, 1993). The availability of new religious groups may serve as a catalyst for increased religious involvement both by those who join the new groups and by members of old groups that increase their engagement with their members to compete. Similarly, changes in the way that governments regulate religious activity can serve to either increase or decrease religious involvement, not only by banning certain forms of involvement outright but also by enhancing or detracting from the beneficial impact of religious involvement on individual well-being (Elliott & Hayward, 2009; Hayward & Elliott, 2014). In addition, there is evidence that religious involvement has an impact on individual well-being only when it is socially normative (Diener et al., 2011; Hayward & Elliott, 2014; Okulicz-Kozaryn, 2010). Thus, another possibility is that highly religious nations tend to become more religious over time, because the benefits of being religiously involved are greater, whereas highly irreligious nations become even less so because of the costs of norm deviation. In this study, we focused on the most widely researched potential influences on period effects in religious involvement, and our findings provide useful support for these views, but future research could focus on delineating some of these other possible influences.

On the country-level, the results of the present study largely recapitulate some significant previous single-country studies. For example, the age and period effects found here for frequency of religious attendance in the Unites States are similar to those reported by Schwadel (2011) based on age–period–cohort analysis of repeated cross-sectional General Social Survey (GSS) data. Similar age effects have been observed in numerous studies with U.S. samples (Levin & Taylor, 1997; Taylor & Chatters, 2011). Other countries have been less-comprehensively researched in the past, but the results are largely consistent. We, therefore, have reasonable confidence that the WVS/EVS data are similar in important respects to other similar data sources, providing evidence of convergent validity for the present results.

The data used cover a very broad cross-section of cultures, making it plausible that similar results would be obtained in nations not covered by the WVS/EVS. However, as countries were not sampled randomly for inclusion in the WVS/EVS, generalizability cannot be definitely ascertained. At the individual level, random probability sampling was used in many but not all national studies, and thus, there is the potential for differences in the degree of within-country generalizability at the individual level on this basis.

Limitations of this study include the fact that the WVS/EVS did not follow the same individuals over time. Thus, although we can draw some inferences about age effects on the basis of repeated cross-sectional observation, it is not possible to come to any conclusions about individual trajectories of change over time. In addition, although the CCREM approach is very valuable for attempting to disaggregate age, period, and cohort effects, there is necessarily still a level of linear dependency among these effects which makes it impossible to definitively separate them (Bell & Jones, 2014). Specific criticisms have been made regarding the capacity of the CCREM to conduct age–period–cohort analyses (Bell & Jones, 2014; Luo, 2013), although it remains a useful tool (Yang & Land, 2006, 2008). Unfortunately, alternatives such as the intrinsic estimator method (Yang, Schulhofer-Wohl, Wenjiang, & Land, 2008) are not generalizable to the multilevel framework demanded by cross-cultural data. Relatedly, the use of atheoreticial birth cohorts may lead to some bias in estimation in the presence of linear trends (Bell & Jones, 2014) but is unavoidable in this instance due to the highly heterogeneous timing of generations across countries. Thus, interpretation should be made with care, and future research should seek to resolve the methodological difficulties associated with this type of research, ideally by implementing longitudinal designs. In addition, the question of cross-cultural validity of measurement is always present when comparisons are made across countries with widely varying sociocultural backgrounds (Hui & Triandis, 1985). Although we think that the two outcome measures used in this study are good indicators of personal and organizational religious involvement across a range of contexts, it is possible that they are less relevant in certain religious and cultural traditions than in others. Finally, this modeling approach cannot fully account for the possibility of reciprocal effects between changes in wealth and changes in religiousness—that is, positive or negative changes in the national religious climate may affect subsequent changes in GDP, whereas treating religiousness as the outcome allows only for modeling of causality in the opposite direction.

Religion involves a complex web of social and psychological phenomena, and thus, changes in an individual’s involvement in it cannot be fully understood without also understanding the dynamics of religious change in the broader culture, and vice versa. This study presents perhaps one of the fullest pictures to date of these patterns on a global scale and the factors that contribute to them. The insights they provide are valuable for our understanding of these processes in both the individual and the social contexts—the study of aging and religious involvement is contextualized by an understanding of the broader social and cultural forces simultaneously affecting behavior, and the study of societal change is incomplete without addressing the individual trajectories that make up the broader pattern. We hope that this study will serve as a catalyst for future research at the interface of individual and group change and its sociocultural context.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Supplementary Material

Supplementary materials are available on the Journal of Cross-Cultural Psychology website at .

References

Atran

Norenzayan

(2005). Religion’s evolutionary landscape: Counterintuition, commitment, compassion, communion. Behavioral and Brain Sciences, 27, 713-730. doi:10.1017/S0140525X04000172

Barber

(2011). A cross-national test of the uncertainty hypothesis of religious belief. Cross-Cultural Research, 45, 318-333. doi:10.1177/1069397111402465

Barrett

J. L.

(2000). Exploring the natural foundations of religion. Trends in Cognitive Sciences, 4, 29-34. doi:10.1016/S1364-6613(99)01419-9

Bell

Jones

(2014). Another “futile quest”? A simulation study of Yang and Land’s Hierarchical Age-Period-Cohort model. Demographic Research, 30, 333-360. doi:10.4054/DemRes.2014.30.11

Berger

P. L.

(1999). The desecularization of the world: Resurgent religion and world politics. Grand Rapids, MI: Wm. B. Eerdmans.

Cacioppo

J. T.

Hawkley

L. C.

Rickett

E. M.

Masi

C. M.

(2005). Sociality, spirituality, and meaning making: Chicago health, aging, and social relations study. Review of General Psychology, 9, 143-155.

Crockett

Voas

(2006). Generations of decline: Religious change in 20th-century Britain. Journal for the Scientific Study of Religion, 45, 567-584. doi:10.1111/j.1468-5906.2006.00328.x

Dassonneville

(2013). Questioning generational replacement: An age, period and cohort analysis of electoral volatility in the Netherlands, 1971-2010. Electoral Studies, 32, 37-47. doi:10.1016/j.electstud.2012.09.004

Diener

Tay

Myers

D. G.

(2011). The religion paradox: If religion makes people happy, why are so many dropping out? Journal of Personality and Social Psychology, 101, 1278-1290. doi:10.1037/a0024402

10.

Elliott

Hayward

R. D.

(2009). Religion and life satisfaction worldwide: The role of government regulation. Sociology of Religion, 70, 285-310. doi:10.1093/socrel/srp028

11.

Erikson

E. H.

(1968). Identity, youth, and crisis. New York, NY: W.W. Norton.

12.

European Values Study Group. (2011). European Values Study 2008: Integrated dataset (EVS 2008). Cologne, Germany: GESIS Data Archive.

13.

Fetzer Institute & National Institute on Aging. (1999). Multidimensional measurement of religiousness/spirituality for use in health research. Kalamazoo, MI: John E. Fetzer Institute.

14.

Finke

Iannaccone

L. R.

(1993). Supply-side explanations for religious change. Annals of the American Academy of Political and Social Science, 527, 27-39. doi:10.2307/1048674

15.

Finke

Stark

(2005). The churching of America, 1776-2005: Winners and losers in our religious economy. New Brunswick, NJ: Rutgers University Press.

16.

Freud

(1989). Civilization and its discontents ( Strachey

, Trans.). New York, NY: W.W. Norton.

17.

Gallup

Jones

(1989). 100 questions and answers: Religion in America. Princeton, NJ: Princeton Religion Research Center.

18.

Hayward

R. D.

Elliott

(2014). Cross-national analysis of the influence of cultural norms and government restrictions on the relationship between religion and well-being. Review of Religious Research, 56, 23-43. doi:10.1007/s13644-013-0135-0

19.

Hayward

R. D.

Kemmelmeier

(2011). Weber revisited: A cross-national analysis of religiosity, religious culture, and economic attitudes. Journal of Cross-Cultural Psychology, 42, 1406-1420. doi:10.1177/0022022111412527

20.

Hayward

R. D.

Krause

(2013a). Patterns of change in religious service attendance across the life course: Evidence from a 34-year longitudinal study. Social Science Research, 42, 1480-1489. doi:10.1016/j.ssresearch.2013.06.010

21.

Hayward

R. D.

Krause

(2013b). Changes in church-based social support relationships during older adulthood. The Journals of Gerontology Series B: Psychological Sciences & Social Sciences, 68, 85-96. doi:10.1093/geronb/gbs100

22.

Henrich

Heine

S. J.

Norenzayan

(2010). The weirdest people in the world? Behavioral and Brain Sciences, 33, 61-83. doi:10.1017/S0140525X0999152X

23.

Hill

P. C.

Pargament

K. I.

Hood

R. W.

McCullough

M. E.

Swyers

J. P.

Larson

D. B.

Zinnbauer

B. J.

(2000). Conceptualizing religion and spirituality: Points of commonality, points of departure. Journal for the Theory of Social Behaviour, 30, 51-77. doi:10.1111/1468-5914.00119

24.

Hogg

M. A.

Adelman

J. R.

Blagg

R. D.

(2010). Religion in the face of uncertainty: An uncertainty-identity theory account of religiousness. Personality and Social Psychology Review, 14, 72-83. doi:10.1177/1088868309349692

25.

Hui

C. H.

Triandis

H. C.

(1985). Measurement in cross-cultural psychology: A review and comparison of strategies. Journal of Cross-Cultural Psychology, 16, 131-152. doi:10.1177/0022002185016002001

26.

Inglehart

(1990). Culture shift in advanced industrial society. Princeton, NJ: Princeton University Press.

27.

Inglehart

(1997). Modernization and postmodernization: Cultural, economic, and political change in 43 societies. Princeton, NJ: Princeton University Press.

28.

Inglehart

Baker

W. E.

(2000). Modernization, cultural change, and the persistence of traditional values. American Sociological Review, 65, 19-51. doi:10.2307/2657288

29.

James

(1902). The varieties of religious experience: A study in human nature. New York, NY: The Modern library.

30.

Krause

(2002). Church-based social support and health in old age. The Journals of Gerontology Series B: Psychological Sciences & Social Sciences, 57, S332-S347. doi:10.1093/geronb/57.6.S332

31.

Krause

(2008). Aging in the church: How social relationships affect health. West Conshohocken, PA: Templeton Foundation Press.

32.

Kuha

(2004). AIC and BIC: Comparisons of assumptions and performance. Sociological Methods & Research, 33, 188-229. doi:10.1177/0049124103262065

33.

Levin

J. S.

Taylor

R. J.

(1997). Age differences in patterns and correlates of the frequency of prayer. The Gerontologist, 37, 75-89. doi:10.1093/geront/37.1.75

34.

Luo

(2013). Assessing validity and application scope of the intrinsic estimator approach to the age-period-cohort problem. Demography, 50, 1945-1967. doi:10.1007/s13524-013-0243-z

35.

Martens

Greenberg

Schimel

Landau

M. J.

(2004). Ageism and death: Effects of mortality salience and perceived similarity to elders on reactions to elderly people. Personality and Social Psychology Bulletin, 30, 1524-1536. doi:10.1177/0146167204271185

36.

Marx

(1970). Cambridge Studies in the History and Theory of Politics: Critique of Hegel’s “philosophy of right” ( O’Malley

, Trans.). Cambridge, UK: Cambridge University Press.

37.

Mason

K. O.

Mason

W. M.

Winsborough

H. H.

Poole

W. K.

(1973). Some methodological issues in cohort analysis of archival data. American Sociological Review, 38, 242-258.

38.

McCleary

R. M.

Barro

R. J.

(2006). Religion and economy. Journal of Economic Perspectives, 20, 49-72. doi:10.1257/jep.20.2.49

39.

McCullough

M. E.

Enders

C. K.

Brion

S. L.

Jain

A. R.

(2005). The varieties of religious development in adulthood: A longitudinal investigation of religion and rational choice. Journal of Personality and Social Psychology, 89, 78-89. doi:10.1037/0022-3514.89.1.78

40.

Miller

A. S.

Stark

(2002). Gender and religiousness: Can socialization explanations be saved? American Journal of Sociology, 107, 1399-1423. doi:10.1086/342557

41.

Need

de Graaf

N. D.

(1996). “Losing my religion”: A dynamic analysis of leaving the church in the Netherlands. European Sociological Review, 12, 87-99. doi:10.2307/522776

42.

Norenzayan

(2013). Big gods: How religion transformed cooperation and conflict. Princeton, NJ: Princeton University Press.

43.

Norenzayan

Dar-Nimrod

Hansen

I. G.

Proulx

(2009). Mortality salience and religion: Divergent effects on the defense of cultural worldviews for the religious and the non-religious. European Journal of Social Psychology, 39, 101-113. doi:10.1002/ejsp.482

44.

Norenzayan

Gervais

W. M.

(2013). The origins of religious disbelief. Trends in Cognitive Sciences, 17, 20-25. doi:10.1016/j.tics.2012.11.006

45.

Okulicz-Kozaryn

(2010). Religiosity and life satisfaction across nations. Mental Health, Religion & Culture, 13, 155-169. doi:10.1080/13674670903273801

46.

Pew Forum on Religion and Public Life. (2012). “Nones” on the rise: One-in-five adults have no religious affiliation. Washington, DC: Pew Research Center.

47.

Presser

Chaves

(2007). Is religious service attendance declining? Journal for the Scientific Study of Religion, 46, 417-423. doi:10.1111/j.1468-5906.2007.00367.x

48.

Raudenbush

S. W.

(1993). A crossed random effects model for unbalanced data with applications in cross-sectional and longitudinal research. Journal of Educational and Behavioral Statistics, 18, 321-349. doi:10.3102/10769986018004321

49.

Ruiter

van Tubergen

(2009). Religious attendance in cross-national perspective: A multilevel analysis of 60 countries. American Journal of Sociology, 115, 863-895. doi:10.1086/603536

50.

Schwadel

(2010a). Period and cohort effects on religious non affiliation and religious disaffiliation: A research note. Journal for the Scientific Study of Religion, 49, 311-319. doi:10.1111/j.1468-5906.2010.01511.x

51.

Schwadel

(2010b). Age, period, and cohort effects on U.S. religious service attendance: The declining impact of sex, southern residence, and Catholic affiliation. Sociology of Religion, 71, 2-24. doi:10.1093/socrel/srq005

52.

Schwadel

(2011). Age, period, and cohort effects on religious activities and beliefs. Social Science Research, 40, 181-192. doi:10.1016/j.ssresearch.2010.09.006

53.

Shaver

J. H.

Sosis

(2014). How does male ritual behavior vary across the lifespan? Human Nature, 25, 136-160. doi:10.1007/s12110-014-9191-6

54.

Starbuck

E. D.

(1901). The psychology of religion: An empirical study of the growth of religious consciousness. London, England: W. Scott.

55.

Stark

(1999). Secularization, R.I.P. Sociology of Religion, 60, 249-273. doi:10.2307/3711936

56.

Swatos

W. H.

Christiano

K. J.

(1999). Secularization theory: The course of a concept. Sociology of Religion, 60, 209-228. doi:10.2307/3711934

57.

Taylor

R. J.

Chatters

L. M.

(2011). Religious media use among African Americans, Black Caribbeans, and Non-Hispanic Whites. Journal of African American Studies, 15, 433-454. doi:10.1007/s12111-010-9144-z

58.

Tornstam

(1997). Gerotranscendence: The contemplative dimension of aging. Journal of Aging Studies, 11, 143-154. doi:10.1016/S0890-4065(97)90018-9

59.

Uecker

J. E.

Regnerus

M. D.

Vaaler

M. L.

(2007). Losing my religion: The social sources of religious decline in early adulthood. Social Forces, 85, 1667-1692. doi:10.1353/sof.2007.0083

60.

Vail

K. E.

Rothschild

Z. K.

Weise

D. R.

Solomon

Pyszczynski

Greenberg

(2010). A terror management analysis of the psychological functions of religion. Personality and Social Psychology Review, 14, 84-94. doi:10.1177/1088868309351165

61.

Verweij

Ester

Nauta

(1997). Secularization as an economic and cultural phenomenon: A cross-national analysis. Journal for the Scientific Study of Religion, 36, 309-324. doi:10.2307/1387561

62.

World Bank. (2013). World development indicators 2013. Available from www.worldbank.org

63.

World Values Survey Association. (2014a). World Values Survey. Retrieved from http://www.worldvaluessurvey.org/WVSContents.jsp

64.

World Values Survey Association. (2014b). World Values Survey longitudinal data file 1981-2014 v.20140617 beta. Madrid, Spain: ASEP/JDS. Available from www.worldvaluessurvey.org

65.

Yang

(2008). Social inequalities in happiness in the United States, 1972 to 2004: An age-period-cohort analysis. American Sociological Review, 73, 204-226. doi:10.1177/000312240807300202

66.

Yang

Land

K. C.

(2006). A mixed models approach to the age-period-cohort analysis of repeated cross-section surveys, with an application to data on trends in verbal test scores. Sociological Methodology, 36, 75-97. doi:10.1111/j.1467-9531.2006.00175.x

67.

Yang

Land

K. C.

(2008). Age–period–cohort analysis of repeated cross-section surveys: Fixed or random effects? Sociological Methods & Research, 36, 297-326. doi:10.1177/0049124106292360

68.

Yang

Schulhofer-Wohl

Wenjiang

J. F.

Land

K. C.

(2008). The intrinsic estimator for age-period-cohort analysis: What it is and how to use it. American Journal of Sociology, 113, 1697-1736. doi:10.1086/587154

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.16 MB