597066

research-article2015

JCCXXX10.1177/0022022115597066Hayward and KrauseJournal of Cross-Cultural Psychology

Article

Aging, Social Developmental,

and Cultural Factors in Changing
Patterns of Religious Involvement
Over a 32-Year Period: An
AgePeriodCohort Analysis of
80 Countries

Journal of Cross-Cultural Psychology

2015, Vol. 46(8) 979995
The Author(s) 2015
Reprints and permissions:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/0022022115597066
jccp.sagepub.com

R. David Hayward1 and Neal Krause1

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
1University

of Michigan, Ann Arbor, MI, USA

Corresponding Author:
R. David Hayward, Department of Health Behavior and Health Education, School of Public Health, University of
Michigan, 1415 Washington Heights, Ann Arbor, MI 48109-2029, USA.
Email: rdhaywar@umich.edu

Downloaded from jcc.sagepub.com at UNIV MASSACHUSETTS LOWELL on March 10, 2016

980

Journal of Cross-Cultural Psychology 46(8)

Institute & National Institute on Aging, 1999; Hill etal., 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, DarNimrod, 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 ageperiodcohort 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 thatfor examplea 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

Downloaded from jcc.sagepub.com at UNIV MASSACHUSETTS LOWELL on March 10, 2016

Hayward and Krause

981

life because it can be a means of creating a sense of meaning and coherence in the perception of
ones 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 ones own death (Vail etal., 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 withinperson 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,

Downloaded from jcc.sagepub.com at UNIV MASSACHUSETTS LOWELL on March 10, 2016

982

Journal of Cross-Cultural Psychology 46(8)

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.

Downloaded from jcc.sagepub.com at UNIV MASSACHUSETTS LOWELL on March 10, 2016

Hayward and Krause

983

The Present Study

This study aims to begin filling these gaps by charting changes in two dimensions of religious
involvementsubjective 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 ageperiodcohort
model. We focus on these dimensions because they provide good coverage of two contrasting
elements of religious involvement, the individuals 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 etal., 2010). As such, we would expect to find that the strength of this relationship should be constant across culturesthat 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 involvementthat 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).

Downloaded from jcc.sagepub.com at UNIV MASSACHUSETTS LOWELL on March 10, 2016

984

Journal of Cross-Cultural Psychology 46(8)

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

Downloaded from jcc.sagepub.com at UNIV MASSACHUSETTS LOWELL on March 10, 2016

Hayward and Krause

985

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 ageperiodcohort 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 zonesGDP 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.
Downloaded from jcc.sagepub.com at UNIV MASSACHUSETTS LOWELL on March 10, 2016

986

Downloaded from jcc.sagepub.com at UNIV MASSACHUSETTS LOWELL on March 10, 2016

Individual
Religious importance
Religious attendance
Gender (male)
Age
Cohort
1880
1890
1900
1910
1920
1930
1940
1950
1960
1970
1980
1990
Country
Year of survey
GDP (natural log)

23
1981.6 (0.8)
9.7 (0.6)

24 (0.1%)
266 (0.7%)
2,036 (6.0%)
3,882 (10.1%)
4,872 (12.7%)
5,214 (13.5%)
6,818 (17.7%)
9,360 (24.3%)
6,034 (15.7%)

42
1990.2 (0.6)
9.2 (1.2)

20 (<0.1%)
1,230 (1.0%)
5,642 (4.5%)
13,534 (10.8%)
17,876 (14.3%)
21,698 (17.3%)
27,894 (22.3%)
28,304 (22.6%)
9,064 (7.2%)

125,542
6.3 (3.4)
4.0 (2.6)
65,078 (51.8%)
42.2 (16.4)

M (SD) or
n (%)

M (SD) or
n (%)

56,760
6.5 (3.2)
4.3 (2.6)
24,900 (52.0%)
41.0 (17.9)

Wave 2

Wave 1

Table 1. Descriptive Statistics by Wave.

53
1996.3 (1.0)
8.3 (1.3)

208 (0.1%)
2,238 (1.5%)
9.514 (6.2%)
17,338 (11.3%)
21,512 (14.0%)
32,202 (21.0%)
36,890 (24.0%)
32,580 (21.2%)
1,204 (0.8%)

154,014
7.1 (3.2)
4.4 (2.5)
79,776 (51.8%)
41.0 (15.9)

M (SD) or
n (%)

Wave 3

64
2000.2 (1.2)
8.6 (1.4)

2 (<0.1%)
160 (0.1%)
1,778 (0.9%)
9,512 (5.1%)
18, 486 (9.9%)
23,532 (12.5%)
33,408 (17.8%)
40,592 (21.6%)
44,252 (23.6%)
15,814 (8.4%)

187,890
7.6 (3.1)
4.5 (2.4)
97,778 (52.1%)
41.5 (16.4)

M (SD) or
n (%)

Wave 4

6 (<0.1%)
336 (0.2%)
3,618 (2.6%)
10,086 (7.1%)
16,422 (11.6%)
22,618 (16.0%)
27,908 (19.7%)
30,400 (21.5%)
29,314 (20.7%)
972 (0.7%)
48
2006.2 (0.9)
8.7 (1.4)

141,830
7.7 (3.0)
4.5 (2.6)
74,234 (52.3%)
42.4 (16.6)

M (SD) or
n (%)

Wave 5

22 (<0.1%)
632 (1.0%)
3,219 (5.3%)
6,305 (10.4%)
8,837 (14.5%)
10,398 (17.1%)
11,382 (18.7%)
13,772 (22.6%)
6,262 (10.3%)
42
2011.6 (0.8)
8.7 (1.4)

60,941
7.4 (3.1)
4.3 (2.6)
32,243 (52.9%)
43.3 (17.1)

M (SD) or
n (%)

Wave 6

24 (<0.1%)
288 (<0.1%)
3,640 (0.5%)
13,898 (2.0%)
41,682 (5.9%)
72,219 (10.2%)
96,287 (13.6%)
134,319 (19.0%)
150,126 (21.2%)
127,678 (18.0%)
60,104 (8.5%)
7,234 (1.0%)
81
1998.3 (7.9)
8.7 (1.4)

726,977
7.2 (3.2)
4.4 (2.6)
374,009 (52.1%)
41.8 (16.5)

M (SD) or
n (%)

Total

Hayward and Krause

987

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.
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.

Downloaded from jcc.sagepub.com at UNIV MASSACHUSETTS LOWELL on March 10, 2016

988

Downloaded from jcc.sagepub.com at UNIV MASSACHUSETTS LOWELL on March 10, 2016

Individual variables (Level 1)

Age
Gendera
Income
Country Wave variables (Level 2)
Year of survey
Wave GDPb,c
Country variables (Level 3)
M GDPb,d
Cultural zone
Africa
Age
Year
Baltic
Age
Year
Catholic Europe
Age
Year
Confucian
Age
Year
English-speaking
Age
Year
Islamic
Age
Year

(0.002)
(0.006)
(0.001)
(0.008)
(0.01)
(0.02)

0.03***
0.06***

0.07***

(SE)

0.02***
0.70***
0.04***

Model 1

Model 2

9.27***
0.002
0.02
4.77***
0.04***
0.006
6.15***
0.04***
0.02*
5.00***
0.01**
0.03*
6.74***
0.03***
0.02*
8.92***
0.005
0.02

0.70***
0.04***

Importance of God

Table 2. Fixed Effects Summary for Cross-Classified Mixed Effects Model.

(0.62)
(0.005)
(0.02)
(0.66)
(0.006)
(0.03)
(0.49)
(0.003)
(0.007)
(0.59)
(0.005)
(0.01)
(0.54)
(0.004)
(0.007)
(0.49)
(0.003)
(0.01)

(0.006)
(0.001)

(SE)

0.04***

0.004
0.04***

0.02***
0.31***
0.002

Model 1

(0.01)

(0.007)
(0.007)

(0.001)
(0.006)
(0.001)

(SE)

6.48***
0.0001
0.0007
3.04***
0.02***
0.006
3.26***
0.02***
0.02**
3.26***
0.01**
0.008
4.12***
0.02***
0.03***
4.94***
0.02***
0.02*

0.32***
0.002*

Religious attendance

(0.52)
(0.005)
(0.02)
(0.41)
(0.003)
(0.02)
(0.41)
(0.003)
(0.008)
(0.48)
(0.004)
(0.01)
(0.44)
(0.004)
(0.008)
(0.41)
(0.003)
(0.01)

(0.006)
(0.001)

(SE)

(continued)

Model 2

989

Downloaded from jcc.sagepub.com at UNIV MASSACHUSETTS LOWELL on March 10, 2016

.32
.004
.01
2,817,646

Model 1

.35
.004
.008
3,086,890

(SE)

Model 2

.15
.005
.02
2,855,483

8.61
0.01**
0.02
6.45***
0.008*
0.11***
4.67***
0.05***
0.03***
8.77***
0.004
0.007

Importance of God

(0.52)
(0.004)
(0.01)
(0.50)
(0.003)
(0.01)
(0.38)
(0.004)
(0.007)
(0.67)
(0.006)
(0.02)

(SE)

Note. GDP = gross domestic product; ICC = intraclass correlation coefficient; BIC = Bayesian information criteria.
aMale is the comparison category.
bPer capita GDP / 1,000.
cCentered on country mean.
dMean for country across all waves, centered on grand mean.
*p < .05. **p < .01. ***p < .001.

Latin America
Age
Year
Orthodox
Age
Year
Protestant Europe
Age
Year
Southeast Asia
Age
Year
Model Covariance and Fit
Country ICC
Cohort ICC
Period ICC
Fit (BIC)
Null Model Covariance Parameters
Country ICC
Cohort ICC
Period ICC
Fit (BIC)

Table 2. (continued)

.21
.005
.02
2,718,670

Model 1
(SE)

.24
.005
.02
2,965,949

Model 2

.09
.005
.02
2,755,446

4.74***
0.02***
0.02*
3.65***
0.006
0.05***
2.79***
0.02***
0.03***
5.96***
0.01*
0.0007

Religious attendance

(0.43)
(0.004)
(0.01)
(0.41)
(0.003)
(0.01)
(0.31)
(0.004)
(0.007)
(0.57)
(0.006)
(0.02)

(SE)

990

Journal of Cross-Cultural Psychology 46(8)

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 ageperiodcohort 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 etal., 2004), needs for self-uncertainty reduction (Hogg etal., 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

Downloaded from jcc.sagepub.com at UNIV MASSACHUSETTS LOWELL on March 10, 2016

Hayward and Krause

991

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 etal., 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 ageperiodcohort 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
ageperiodcohort 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

Downloaded from jcc.sagepub.com at UNIV MASSACHUSETTS LOWELL on March 10, 2016

992

Journal of Cross-Cultural Psychology 46(8)

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 religiousnessthat 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 individuals 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 contextsthe 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.
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 http://jcc.
sagepub.com/supplemental.

References
Atran, S., & Norenzayan, A. (2005). Religions evolutionary landscape: Counterintuition, commitment, compassion, communion. Behavioral and Brain Sciences, 27, 713-730. doi:10.1017/S0140525X04000172
Barber, N. (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, A., & Jones, K. (2014). Another futile quest? A simulation study of Yang and Lands 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, A., & Voas, D. (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, R. (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

Downloaded from jcc.sagepub.com at UNIV MASSACHUSETTS LOWELL on March 10, 2016

Hayward and Krause

993

Diener, E., Tay, L., & 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
Elliott, M., & 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
Erikson, E. H. (1968). Identity, youth, and crisis. New York, NY: W.W. Norton.
European Values Study Group. (2011). European Values Study 2008: Integrated dataset (EVS 2008).
Cologne, Germany: GESIS Data Archive.
Fetzer Institute & National Institute on Aging. (1999). Multidimensional measurement of religiousness/
spirituality for use in health research. Kalamazoo, MI: John E. Fetzer Institute.
Finke, R., & 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
Finke, R., & Stark, R. (2005). The churching of America, 1776-2005: Winners and losers in our religious
economy. New Brunswick, NJ: Rutgers University Press.
Freud, S. (1989). Civilization and its discontents (J. Strachey, Trans.). New York, NY: W.W. Norton.
Gallup, G., & Jones, S. (1989). 100 questions and answers: Religion in America. Princeton, NJ: Princeton
Religion Research Center.
Hayward, R. D., & Elliott, M. (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
Hayward, R. D., & Kemmelmeier, M. (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
Hayward, R. D., & Krause, N. (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
Hayward, R. D., & Krause, N. (2013b). Changes in church-based social support relationships during older
adulthood. The Journals of Gerontology Series B: Psychological Sciences & Social Sciences, 68, 8596. doi:10.1093/geronb/gbs100
Henrich, J., Heine, S. J., & Norenzayan, A. (2010). The weirdest people in the world? Behavioral and Brain
Sciences, 33, 61-83. doi:10.1017/S0140525X0999152X
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
Hogg, M. A., Adelman, J. R., & Blagg, R. D. (2010). Religion in the face of uncertainty: An uncertaintyidentity theory account of religiousness. Personality and Social Psychology Review, 14, 72-83.
doi:10.1177/1088868309349692
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
Inglehart, R. (1990). Culture shift in advanced industrial society. Princeton, NJ: Princeton University Press.
Inglehart, R. (1997). Modernization and postmodernization: Cultural, economic, and political change in 43
societies. Princeton, NJ: Princeton University Press.
Inglehart, R., & Baker, W. E. (2000). Modernization, cultural change, and the persistence of traditional
values. American Sociological Review, 65, 19-51. doi:10.2307/2657288
James, W. (1902). The varieties of religious experience: A study in human nature. New York, NY: The
Modern library.
Krause, N. (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
Krause, N. (2008). Aging in the church: How social relationships affect health. West Conshohocken, PA:
Templeton Foundation Press.
Kuha, J. (2004). AIC and BIC: Comparisons of assumptions and performance. Sociological Methods &
Research, 33, 188-229. doi:10.1177/0049124103262065
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

Downloaded from jcc.sagepub.com at UNIV MASSACHUSETTS LOWELL on March 10, 2016

994

Journal of Cross-Cultural Psychology 46(8)

Luo, L. (2013). Assessing validity and application scope of the intrinsic estimator approach to the ageperiod-cohort problem. Demography, 50, 1945-1967. doi:10.1007/s13524-013-0243-z
Martens, A., Greenberg, J., Schimel, J., & 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
Marx, K. (1970). Cambridge Studies in the History and Theory of Politics: Critique of Hegels philosophy
of right (J. OMalley, Trans.). Cambridge, UK: Cambridge University Press.
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.
McCleary, R. M., & Barro, R. J. (2006). Religion and economy. Journal of Economic Perspectives, 20,
49-72. doi:10.1257/jep.20.2.49
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
Miller, A. S., & Stark, R. (2002). Gender and religiousness: Can socialization explanations be saved?
American Journal of Sociology, 107, 1399-1423. doi:10.1086/342557
Need, A., & 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
Norenzayan, A. (2013). Big gods: How religion transformed cooperation and conflict. Princeton, NJ:
Princeton University Press.
Norenzayan, A., Dar-Nimrod, I., Hansen, I. G., & Proulx, T. (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
Norenzayan, A., & Gervais, W. M. (2013). The origins of religious disbelief. Trends in Cognitive Sciences,
17, 20-25. doi:10.1016/j.tics.2012.11.006
Okulicz-Kozaryn, A. (2010). Religiosity and life satisfaction across nations. Mental Health, Religion &
Culture, 13, 155-169. doi:10.1080/13674670903273801
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.
Presser, S., & Chaves, M. (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
Raudenbush, S. W. (1993). A crossed random effects model for unbalanced data with applications in crosssectional and longitudinal research. Journal of Educational and Behavioral Statistics, 18, 321-349.
doi:10.3102/10769986018004321
Ruiter, S., & van Tubergen, F. (2009). Religious attendance in cross-national perspective: A multilevel
analysis of 60 countries. American Journal of Sociology, 115, 863-895. doi:10.1086/603536
Schwadel, P. (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.14685906.2010.01511.x
Schwadel, P. (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
Schwadel, P. (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
Shaver, J. H., & Sosis, R. (2014). How does male ritual behavior vary across the lifespan? Human Nature,
25, 136-160. doi:10.1007/s12110-014-9191-6
Starbuck, E. D. (1901). The psychology of religion: An empirical study of the growth of religious consciousness. London, England: W. Scott.
Stark, R. (1999). Secularization, R.I.P. Sociology of Religion, 60, 249-273. doi:10.2307/3711936
Swatos, W. H., & Christiano, K. J. (1999). Secularization theory: The course of a concept. Sociology of
Religion, 60, 209-228. doi:10.2307/3711934
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/s12111010-9144-z

Downloaded from jcc.sagepub.com at UNIV MASSACHUSETTS LOWELL on March 10, 2016

Hayward and Krause

995

Tornstam, L. (1997). Gerotranscendence: The contemplative dimension of aging. Journal of Aging Studies,
11, 143-154. doi:10.1016/S0890-4065(97)90018-9
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
Vail, K. E., Rothschild, Z. K., Weise, D. R., Solomon, S., Pyszczynski, T., & Greenberg, J. (2010). A terror
management analysis of the psychological functions of religion. Personality and Social Psychology
Review, 14, 84-94. doi:10.1177/1088868309351165
Verweij, J., Ester, P., & Nauta, R. (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
World Bank. (2013). World development indicators 2013. Available from www.worldbank.org
World Values Survey Association. (2014a). World Values Survey. Retrieved from http://www.world
valuessurvey.org/WVSContents.jsp
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
Yang, Y. (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
Yang, Y., & 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
Yang, Y., & Land, K. C. (2008). Ageperiodcohort analysis of repeated cross-section surveys: Fixed or
random effects? Sociological Methods & Research, 36, 297-326. doi:10.1177/0049124106292360
Yang, Y., Schulhofer-Wohl, S., Wenjiang, J. F., & Land, K. C. (2008). The intrinsic estimator for ageperiod-cohort analysis: What it is and how to use it. American Journal of Sociology, 113, 1697-1736.
doi:10.1086/587154

Downloaded from jcc.sagepub.com at UNIV MASSACHUSETTS LOWELL on March 10, 2016