Copyright 1996 by
The Cerontological Society of America
The Cerontologist
Vol.36, No. 4,454-463
This study examined the effects of religious attendance on three dimensions of psychological
well-being using panel data from a three-generations study of Mexican Americans from Texas
(N = 624). Well-being dimensions included life satisfaction (the 13-item LSIA), and respective
seven- and four-item depressed and positive affect subscales of the CES-D. Two-wave path
analyses revealed a cross-sectional association between religious attendance and life
satisfaction in the two oldest generations, and a salutary longitudinal effect of religious
attendance on subsequent depressed affect in the youngest generation. Findings for life
satisfaction and depressed affect withstood controlling for health and five sociodemographic
correlates of religious attendance and well-being.
Key Words: Religious attendance, Psychological well-being, Mexican Americans,
Panel analysis, Intergenerational research
Religious Attendance and Psychological
Well-Being in Mexican Americans: A Panel
Analysis of Three-Generations Data1
Jeffrey S. Levin, PhD, MPH, 2 Kyriakos S. Markides, PhD,3 and Laura A. Ray, MPA<
Recently, there has been considerable growth in
gerontological research on the impact of religious
involvement on mental health and psychological
well-being. Although religion has never been a mainstream topic in gerontology, gerontologists have
shown longstanding if sporadic and unsystematic
interest in its influence on well-being (see Levin,
1989). From the 1950s through the early 1970s, the
establishment and growth of research in this area
was principally due to the pioneering work of Moberg (e.g., 1953,1965,1971). The further emergence
of religion and aging as a field of study is documented in the first edition of The Encyclopedia of
Aging (Markides, 1987). The years since have seen
the publication of key literature reviews (Levin, 1989;
Moberg, 1990; Witter, Stock, Okun, & Haring, 1985)
and books (e.g., Koenig, 1994a, 1995; Koenig,
Smiley, & Gonzales, 1988; Thomas & Eisenhandler,
1994), inclusion of a chapter on religion in the fourth
edition of the Handbook of the Psychology of Aging
(McFadden, 1996), and publication of Aging, Spirituality, and Religion: A Handbook (Kimble, McFadden,
Ellor, & Seeber, 1995), which includes a chapter reviewing and critiquing research, methods, and the-
1
The work of Dr. Levin was supported by the National Institute on Aging
through NIH FIRST Award (R29) Grant No. AC-09462 and NIH Research (ROD
Grant No. AG-10135. The work of Dr. Markides and Ms. Ray were supported
by NIH Research (ROD Grant No. AG-08633.
2
Address correspondence to Dr. Levin, Associate Professor, Department
of Family and Community Medicine, Eastern Virginia Medical School, P.O.
Box 1980, Norfolk, VA 23501.
3
Professor, Department of Preventive Medicine and Community Health,
The University of Texas Medical Branch, Galveston.
"Faculty Associate, Department of Preventive Medicine and Community
Health, The University of Texas Medical Branch, Galveston.
454
ory linking religion and well-being (Levin & Tobin,
1995).
Despite these recent developments, research on
religion, aging, and well-being has been largely unfocused, and efforts at synthesis and at developing
programmatic emphases have been few. With the
notable exception of several widely cited articles
(e.g., Blazer & Palmore, 1976; Koenig, Moberg, &
Kvale, 1988; Krause & Tran, 1989; Markides, Levin, &
Ray, 1987; Mindel & Vaughan, 1978), gerontological
research published in this area is largely isolated
from conceptual work on both religion and wellbeing and from theoretical and methodological developments in religion and aging. This has led some
to ask, "Is religion taboo in gerontology?" (Sherrill,
Larson, & Creenwold, 1993) and has created incentive for systematic efforts to identify limitations of
prior studies in order to redirect research (Levin,
1989, 1995a, 1995b; Levin & Tobin, 1995). Identified
needs include more precise conceptualization and
measurement of both religious involvement and psychological well-being, and a greater emphasis on
studies that are longitudinal, focus on ethnic minority populations, and are theory- or hypothesis-driven.
First, researchers must meaningfully account for
the different ways of being religious and practicing
religion. Cerontologists have begun to differentiate
among types of religious involvement and to validate
multidimensional measures (e.g., Ainlay & Smith,
1984; Chatters, Levin, & Taylor, 1992; Krause, 1993;
Krause & Tran, 1989; Levin, Taylor, & Chatters, 1995).
This work was stimulated by Mindel and Vaughan
(1978), who distinguished between formal, institutional, public ("organizational") forms of religious
involvement (e.g., religious attendance, church or
The Gerontologist
synagogue membership) and more informal, noninstitutional, private ("nonorganizational") forms of
involvement (e.g., bible study, private prayer, watching religious television). Subsequent research has
confirmed the importance of this distinction, as different religious dimensions exert different effects on
various psychosocial and health-related outcomes
(e.g., Koenig, Moberg, & Kvale, 1988; Levin, Chatters, & Taylor, 1995). Research questions and study
hypotheses thus ought to be based more closely on
the particular religious construct(s) under consideration in order to avoid the common misperception
that all measures containing variations of the word
"religion" are interchangeable and equally valid indicators of an imaginary "singular something called
religiosity" (Levin & Tobin, 1995, p. 38).
Second, religious gerontologists must recognize
that psychological well-being is also best characterized as multidimensional. Within gerontology and
geriatrics, numerous measurement instruments have
been developed to assess the various dimensions or
domains of well-being, such as life satisfaction, depressed affect, positive affect, congruence, and symptomatology (see Levin & Tobin, 1995). In religious
research, however, single items or unidimensional
indices are too often used, and results are generalized to "well-being" or "mental health" as a whole. It
is important for research questions and hypotheses
to be framed explicitly in terms of the particular
aspect(s) of well-being under study, certain of which
(especially depression and affect or mood disorder)
have been the subject of considerable clinical research among geriatricians (Koenig & Blazer, 1992).
Third, according to Koenig and associates (Koenig,
Smiley, & Gonzales, 1988), there is "a dire need for
longitudinal studies that follow middle-aged persons
. . . over time into late old age" (p. 162), especially in
the area of religion and well-being. Relatively few
studies in this area have explored religious effects on
well-being longitudinally (e.g., Anson, Antonovsky,
& Sagy, 1990; Idler & Kasl, 1993; Koenig et al., 1992;
Markides, 1983; Markides, Levin, & Ray, 1987; Williams, Larson, Buckler, Heckmann, & Pyle, 1991), but
the presence of at least some positive findings suggests the need to pursue this more carefully. Evidence of both cohort and aging effects in religious
involvement (e.g., Guy, 1982; Haganaars, 1990), as
well as the often-noted presence of confounding
between behavioral measures of religious involvement and functional or somatic measures of health
and well-being in cross-sectional analyses of olderadult samples (see Levin, 1989), points to the value of
longitudinal research.
Fourth, unlike other areas within gerontology,
considerable research has focused on patterns and
outcomes of religious involvement within particular
racial and ethnic minority groups — notably the work
of Taylor and associates on African Americans (summarized in Chatters & Taylor, 1994) and of Markides
and associates on Mexican Americans (e.g., Levin &
Markides, 1986,1988; Markides, Levin, & Ray, 1987).
Additional work has drawn on samples of AfricanAmerican respondents (e.g., Brown & Gary, 1987,
Vol. 36, No. 4,1996
1988; Brown, Ndubuishi, & Gary, 1990; Ellison & Gay,
1990; Krause & Tran, 1989; Levin, Chatters, & Taylor,
1995) and has examined black-white differences in
dimensions of well-being (e.g., St. George & McNamara, 1984). Nonetheless, nearly all of these studies
are cross-sectional, empirical research on U.S. Hispanics has declined over the past decade, and particular analyses tend to focus on a respective single
dimension of well-being.
Finally, as in much of social gerontology, empirical
research has been mostly atheoretical, the justification being that little is known and thus purely "exploratory" research is required. On the contrary,
since 1980 alone, over 100 empirical studies reporting the effects of religious variables have been published in the aging literature (Levin, 1995a), many
focusing on mental health and well-being outcomes.
Further, within this area over the past 25 years, half a
dozen competing mid-range theoretical perspectives can be identified (Levin, 1989), although they
are often more implicit than explicit. Several gerontologists interested in religion and well-being recently have begun more explicitly to evaluate alternative theoretical perspectives (Krause & Tran, 1989),
test well-specified multifactorial structural models
(Krause, 1992; Krause & Tran, 1989; Levin, Chatters,
& Taylor, 1995), and develop comprehensive theoretical frameworks for examining religious effects on
well-being outcomes such as depression (Ellison,
1994). Research on religion, aging, and well-being is
well enough along that theory- or hypothesis-driven
designs and analyses are a reasonable minimum expectation for current and future studies in this area.
Research Issues and Hypotheses
This particular study redresses each of these limitations, while also revisiting and updating a program of
research on Mexican Americans conducted from the
late 1970s through the middle 1980s. In a series of
reports based on three samples from Texas — a small
interview study of low-income and mostly MexicanAmerican elderly respondents, a panel study of older
Mexican Americans and Anglos, and a three-generations study of Mexican Americans — Markides and
associates examined the impact of religious attendance on measures of health and well-being. Crosssectional path analysis findings revealed a positive
effect on life satisfaction of an index of activity,
including a measure of religious attendance, net of
the effects of health (Markides & Martin, 1979). This
finding for religious attendance and life satisfaction
was replicated in two-wave panel data on both Mexican Americans and Anglos (Markides, 1983;
Markides, Levin, & Ray, 1987). A third wave of data
showed a similar but nonsignificant trend, with interpretation complicated by health-related sample attrition and the confounding of religious attendance
and health status in older respondents (Markides,
Levin, & Ray, 1987). Similar cross-sectional findings
for religious attendance were also obtained with
respect to subjective health in older Mexican Americans (Markides & Martin, 1983). Three-generations
455
data also revealed consistent cross-sectional effects
for religious attendance: positive effects on subjective health in the oldest generation and protective
effects against physical symptoms in the youngest
generation (Levin & Markides, 1985); positive effects
on subjective health in older and younger women,
net of the effects of social support and social class
(Levin & Markides, 1986); and positive effects on life
satisfaction in older men and older and middle-aged
women, the latter two findings net of the effects of
subjective and functional health (Levin & Markides,
1988). In sum, this body of work suggests that the
frequency of religious attendance is a cross-sectional
correlate of life satisfaction in older Mexican Americans, and that this effect withstands controlling for
the effects of subjective health, long believed to be
the principal correlate of well-being in older adults
(George & Landerman, 1984).
This study seeks to extend this earlier work on the
effects of religious attendance on psychological wellbeing among Mexican Americans in several ways.
First, a second wave of data is now available on the
three-generations study respondents, so this association can be examined in the context of both intergenerational differences and temporal change simultaneously. Second, additional dimensions of
well-being will be examined, in order to better specify
how and understand why religious attendance influences well-being in this population. These dimensions include life satisfaction, depressed affect, and
positive affect. Third, path analysis will be used to
examine across two waves of data the associations
between religious attendance and each well-being
outcome in respective multifactorial models. Effects
will be estimated separately by generational cohort,
and analyses will also be rerun controlling for the
effects of subjective health, age, and other known
sociodemographic correlates or predictors of religious attendance and well-being. These two-wave
panel models will enable the estimation of longitudinal effects of religious attendance (at Time 1) on psychological well-being (at Time 2), adjusting for Time 1
well-being and Time 2 attendance (see Figure 1).
Based on earlier findings, several study hypotheses can be posed. Each hypothesis pertains to a
particular dimension of psychological well-being.
Hypothesis 7. — It is hypothesized that religious
attendance will be cross-sectionally associated with
life satisfaction at Time 2, as it was at Time 1 (Levin &
Markides, 1988). This finding is expected primarily
among older respondents, reflecting the role of formal, institutional religious involvement in establishing continuity across life-course stages, emphasizing
the intrinsic and enduring meaning of life, fostering a
sense of feeling blessed by God, and providing both
personal and community resources that enhance
coping with age-associated losses (Tobin, 1991). Patterns of religious attendance at a given time thus
should be related to feelings of life satisfaction,
which, more than depressed or positive affect, appear to embody reflection about one's current life
situation in general (George, 1981).
456
Figure 1. The proposed panel model of religious attendance and
psychological well-being.
Hypothesis 2. — It is hypothesized that religious
attendance will exert an epidemiologically protective
effect on depressed affect. This expectation is based
on longitudinal findings that suggest a salutary impact of religious attendance on subsequent depression (Idler & Kasl, 1992), perhaps due to buffering the
deleterious cumulative effects of life stress on mental health (Williams et al., 1991). Prior findings from a
sample of respondents drawn from the full adult age
range indicate that this effect may not be limited just
to older adults (see Williams et al., 1991).
Hypothesis 3. — It is hypothesized that religious
attendance will have a significant positive relationship with positive affect, although whether this manifests cross-sectionally or longitudinally is uncertain.
Prior gerontological studies of religion and well-being have emphasized measures of either life satisfaction or depressed affect and related "negative" mental health outcomes. Relatively little of this research
since 1980 has investigated the effects of religious attendance on "positively" defined mood-related constructs (see Levin, 1995a), and findings are mixed.
Religious attendance has been found to be associated cross-sectionally with measures of happiness
(Heisel & Faulkner, 1982; Steinitz, 1980) and morale
(Lee & Ishi-Kuntz, 1987), but nonsignificant findings
have been reported as well (Ellison, 1991). In addition, this research has not been limited to the olderadult age range, so there is no pressing reason to
expect a salutary effect to appear solely among the
oldest cohort. Prior findings also do not indicate
whether a relationship, if it emerges, would withstand controlling for the effects of known sociodemographic correlates of religious attendance and
positive affect.
In sum, religious attendance is hypothesized to
have (a) a positive cross-sectional association with
The Gerontologist
life satisfaction, especially among older respondents, (b) a salutary longitudinal effect on depressed
affect, and (c) a significant positive relationship of
some type with positive affect. Whether religious
attendance also exerts longitudinal effects on life
satisfaction or is cross-sectionally associated with
depressed affect is uncertain. Panel data from older
Israeli Jews suggest not (Anson et al., 1990).
Methods
The Study
A three-generations study of Mexican Americans
was originally conducted in the San Antonio area
during 1981 to 1982. Multi-stage area-probability
sampling procedures were employed to select older
Mexican Americans (aged 65 to 80) who had children
and adult grandchildren (18 years old and over) living
within 50 miles of San Antonio, Texas. An eligible
grandchild was then selected randomly, a process
that also identified the middle-aged parent who was
the son or daughter of the oldest-generation respondent. The total sample consisted of 375 such threegeneration lineages, amounting to a total of 1,125
respondents.
Respondents were followed up approximately 11
years later. Of the 1,125 original respondents, 624
(56%) were reinterviewed. Dropouts were as follows: 227 were deceased (199 in the oldest generation), 75 refused to be reinterviewed, 24 had moved
out of town or were in prison, 37 were too ill to be
interviewed, and 138 were lost to follow-up. Of the
624 reinterviewed subjects, 111 were from the oldest
generation (24 males, 87 females), 263 were from the
middle generation (80 males, 183 females), and 250
were from the youngest generation (81 males, 169
females). The findings reported in this study are
based on respondents interviewed at both times.
The sample used in this study has the following
sociodemographic characteristics at Time 1, from
oldest to youngest generation: respondents have a
median age of 72.5, 49.8, and 27.0 years; 81%, 68%,
and 67% are female; 47%, 79%, and 79% are married;
8%, 56%, and 69% are employed; and respondents
average 3.38, 9.04, and 12.09 years of education.
Approximately 84% of respondents are Roman Catholic, with little variation by generation.
Measures
This study includes a single-item measure of religious attendance, three summary scales measuring
dimensions of psychological well-being (life satisfaction, depressed affect, and positive affect), and six
exogenous constructs, including subjective health.
The well-being measures are derived from existing
instruments that have been used extensively in studies
of Hispanic and ethnic-minority populations, including research conducted by the present investigators.
Frequency of attendance at religious services is
measured by a single ordinal item, "How often do
you attend religious services?" (coded: 4 = once a
Vol. 36, No. 4,1996
week or more often, 3 = two or three times a month,
2 = once a month, 1 = less frequently). High scores
on this indicator therefore signify more frequent
religious attendance.
Life satisfaction is measured by the 13-item version
of the Life Satisfaction Index (LSI) originally proposed by Neugarten, Havighurst, and Tobin (1961).
The LSI assesses agreement with general statements
about life, and subsequent research has identified
dimensions of mood tone, zest, and congruence
(Liang, 1984). The present study uses the trichotomous scoring system proposed by Wood, Wylie, and
Sheafor (1969). This revision, known as the LSIA, is
coded such that a score of 2 = agree, 1 = uncertain,
and 0 = disagree. Reliability analysis conducted in
SAS software (using the CORR procedure with the
ALPHA option) led to the deletion of three LSI items,
and the resulting 10-item scale had acceptable areliabilities across the three generations and two
waves of observation (from oldest to youngest generation, at Time 1: .60, .67, .60; at Time 2: .67, .75,
.74). In this study, the overall score (range: 0-2) was
calculated by averaging responses across the scale's
10 items.
Depressed affect and positive affect are measured
by the respective seven-item depressed affect and
four-item positive affect subscales (Liang, Tran,
Krause, & Markides, 1989) of the 20-item Center for
Epidemiologic Studies Depression (CES-D) scale
(Radloff, 1977). The CES-D measures the frequency
of particular feelings within the past week (coded: 0
= rarely or none of the time [less than one day], 1 =
some or a little of the time [1-2 days], 2 = occasionally or a moderate amount of time [3—4 days], 3 =
most or all of the time [5-7 days]). Reliabilities for
depressed affect (range: 0-21) were very high at both
Time 1 (.87, .87, .81) and Time 2 (.87, .88, .89) and
were stable across waves. Reliabilities for positive
affect (range: 0-12) were somewhat lower but still
quite acceptable at both waves of observation (Time
1: .77, .76, .63; Time 2: .69, .69, .80).
Exogenous constructs used in these analyses consist of five sociodemographic variables and a singleitem self-rating of health status. These include age (at
Time 1), gender (1 = female, 0 = male), education
(years of school completed), marital status (1 = married, 0 = not married), employment status (1 = yes,
0 = no), and subjective health (4 = excellent, 3 =
good, 2 = fair, 1 = poor). In this study, from oldest
to youngest generation, respondents' mean ratings
of subjective health at Time 1 are 2.39,1.96, and 1.62,
and at Time 2 are 2.78, 2.37, and 2.03.
Data Analysis
This study has two objectives: (a) to identify generational differences and temporal changes in religious
attendance and three dimensions of psychological
well-being, and (b) to examine multifactorial panel
models of the effects of religious attendance on
these dimensions of psychological well-being and
test respective hypotheses. Each study objective has
a respective data analysis strategy.
457
To address the first objective, generational differences and temporal changes in each of the four study
constructs are examined through two-way analyses
of variance (ANOVAs) using a factorial design that
tests for main effects of generation and time and for
their multiplicative interaction. In these analyses, the
"Type I I I " solution is presented, whereby both main
effects and the interaction term are estimated simultaneously (Freund & Littell, 1981). This type of analysis has been used successfully in prior gerontological
research on religion (Levin & Taylor, 1993; Levin,
Taylor, & Chatters, 1994).
To address the second objective, two-wave path
models are tested separately by generation and dimension of psychological well-being. These models
are fully recursive, with constructs "causally" ordered, as in Figure 1, from Time 1 religious attendance to Time 1 psychological well-being to Time 2
religious attendance to Time 2 psychological wellbeing. The models are estimated twice: first, specifying the endogenous constructs only (Model I); second, controlling for the effects of the six exogenous
constructs — age, gender, education, marital status,
employment status, and subjective health (Model II).
Separate path models are estimated for each generation for each of the three identified dimensions of
psychological well-being.
All analyses were conducted using the REG and
GLM procedures in PC versions 6.08 and 6.10 of SAS.
Results
Descriptive Results
As shown in Table 1, there are statistically significant generational differences and/or temporal
changes in all three of the psychological well-being
measures and in religious attendance. First, life satisfaction is lowest among the oldest generation and
increases from Time 1 to Time 2. Although these
differences and changes are statistically significant,
they are not substantively large. Second, depressed
affect is highest among the oldest generation at both
times, and this generational difference is relatively
pronounced. In terms of the hypothetical range of
this subscale (0-21), however, this cohort, on average, is largely unafflicted with depressed affect at
both Time 1 and Time 2. Third, positive affect increases from Time 1 to Time 2, and there is a visible
but nonsignificant trend toward less positive affect in
the oldest generation at both times. In light of the
hypothetical range of this subscale (0-12), these respondents, on average, show moderately high levels
of positive affect at both Time 1 and Time 2.
For religious attendance, there are statistically
significant generational differences and temporal
changes, as well as a significant interaction. At Time
1, there is a gradient such that religious attendance is
less frequent in successively younger generational
Table 1. Generational Differences and Temporal Changes in Religious Attendance and Dimensions of Psychological Weil-Being
Generation3
Variable
Religious attendance
Life satisfaction
Depressed affect
Positive affect
Effects"
Time
Oldest
Middle
Youngest
1
3.21C
(1.10)'
2.93
(1.20)
2.43
(1.30)
2
2.38
(1.43)
3.03
(1.21)
2.50
(1.33)
1
1.30
(0.37)
1.39
(0.38)
1.41
(0.34)
1.31
(0.43)
1.50
(0.43)
1.48
(0.44)
2.96
(3.77)
2.17
(3.45)
2.03
(2.90)
3.25
(4.56)
2.01
(3.65)
2.56
(3.94)
9.06
(2.84)
9.41
(2.58)
9.38
(2.55)
9.63
(3.01)
10.21
(2.37)
10.16
(2.63)
Generation
Time
Gen. x Time
21.46*
8.03*
11.88*
11.78*
6.46*
1.25
6.17*
0.97
1.14
2.64
19.70*
0.16
a
/Vs for these analyses: oldest generation = 108; middle generation = 260; youngest generation = 249.
F-scores.
c
Means (and standard deviations).
*p < .05 (2-tailed); **p < .01; ***p < .001.
b
458
The Gerontologist
tion, there are no longitudinal effects of Time 1
religious attendance on Time 2 life satisfaction, but
none were necessarily expected. For measures of
affect, findings are just the opposite. Time 1 religious
attendance is associated with less depressed affect at
Time 2 within the youngest generation (see Table 3).
This longitudinal effect confirms Hypothesis 2. Time
1 religious attendance is also associated with greater
positive affect at Time 2 in the youngest generation
(see Table 4). This confirms the more uncertain expectations contained in Hypothesis 3. Finally, religious attendance and both affect measures are unassociated cross-sectionally, at both waves, but
cross-sectional associations, at least for depressed
affect, were not specifically hypothesized. It should
be kept in mind that estimation of these longitudinal
effects on Time 2 affect controls for the effects of
Time 1 scores on these respective indices.
Model II findings present net effects for religious
attendance on each of the three well-being dimensions. These results represent the Model I analyses
rerun controlling for the structural effects of the six
exogenous constructs (age, gender, education, marital status, employment status, and subjective health).
In these analyses, the principal findings from the
Model I analyses remain: (a) a cross-sectional association between religious attendance and life satisfaction in the two oldest generations, although only at
Time 1, and (b) a longitudinal effect of Time 1 religious attendance on Time 2 depressed and positive
affect in the youngest generation. All of these effects
are notable in that they are independent of both
Time 1 affect scores and the effects of health, age,
and other known sociodemographic correlates or
predictors of both religious attendance and wellbeing.
Although upholding Model I results for life satisfaction and depressed affect, and thus reaffirming
Hypotheses 1 and 2, Model II findings add some
complicated twists with respect to positive affect (see
Table 4). The longitudinal effect of Time 1 religious
cohorts. From Time 1 to Time 2, there is only a slight
increase in attendance in the middle and youngest
generations, but a considerable decline in attendance in the oldest generation. At Time 1, the oldest
generation attends religious services, on average,
between two or three times a month and weekly or
more (mean = 3.21), the highest level of any of the
six generation-time combinations. At Time 2, the
oldest generation attends religious services, on average, about once a month (mean = 2.38), the lowest
level of any of the generation-time combinations.
This decline in religious attendance over time among
the oldest generation is the most dramatic finding in
Table 1.
Panel Analysis Results
Tables 2 through 4 present results of panel analysis
for each of the three respective well-being outcomes: life satisfaction, depressed affect, and positive affect. In each table, Model I findings represent
gross effects estimated in analyses that specify structural parameters among endogenous constructs
only, while Model II findings represent net effects
derived from rerunning the same analyses controlling for exogenous effects.
Model I findings reveal that (a) religious attendance exerts statistically significant effects on psychological well-being, (b) these effects are different
for different dimensions of well-being, and (c) these
effects vary by generation. Most notable of these
findings are cross-sectional associations between religious attendance and life satisfaction, as hypothesized, and longitudinal effects of religious attendance on depressed and positive affect, also as
hypothesized. Specifically, Time 1 religious attendance is associated with Time 1 life satisfaction in the
two oldest generations, and respective Time 2 measures are also associated in the two oldest generations (see Table 2). This confirms Hypothesis 1,
which anticipated cross-sectional effects. In addi-
Table 2. Panel Analysis Results for Effects of Religious Attendance on Life Satisfaction, by Generation
Dependent Variables
Model I:
Endogenous Variables Only
Independent
Variables
T1
Life
Satisfaction
T2
Religious
Attendance
.28**
.17**
-.00
.25***
.36***
.26***
T1
Religious
attendance
T1
Life
satisfaction
T2
Religious
attendance
-.08
-.04
.04
Model II:
Controlling for Exogenous Variables
T2
Life
Satisfaction
.03
-.03
.05
.29***
.21***
.19***
.14*
.12*
.08
T1
Life
Satisfaction
T2
Religious
Attendance
T2
Life
Satisfaction
.19*
.16**
.01
.29**
.34***
.24***
.02
-.03
.06
-.14
-.11
.02
.25*
.11
.14*
.15
.12
.04
Note: Numbers are standardized ((3) regression coefficients. Findings are ordered from top to bottom: oldest (N = 108), middle (N =
260), and youngest (N = 249) generation.
*p < .05 (2-tailed); **p < .01; ***p < .001.
Vol. 36, No. 4,1996
459
Table 3. Panel Analysis Results for Effects of Religious Attendance on Depressed Affect, by Generation
Dependent Variables
Model I:
Endogenous Variables Only
Independent
Variables
T1
Religious
attendance
Model II:
Controlling for Exogenous Variables
T1
Depression
Affect
T2
Relgious
Attendance
T2
Depression
Affect
T1
Depression
Affect
-.05
.04
-.08
.24***
.36***
.26***
.04
-.01
-.15***
.03
.05
-.11
.26**
.31***
.23***
.24***
.27***
.30***
.03
.06
-.07
T1
Depressed
affect
T2
Relgious
Attendance
-.01
.01
-.16*
.07
.08
-.08
.15
.20**
.21**
.03
-.03
.04
T2
Religious
attendance
T2
Depression
Affect
.08
-.02
.08
Note: Numbers are standardized ((3) regression coefficients. Findings are ordered from top to bottom: oldest (N = 108), middle (N =
260), and youngest (N = 249) generation.
*p < .05 (2-tailed); **p < .01; ***p < .001.
Table 4. Panel Analysis Results for Effects of Religious Attendance on Positive Affect, by Generation
Dependent Variables
Model I:
Endogenous Variables Only
Independent
Variables
T1
Religious
attendance
T1
Positive
affect
Model II:
Controlling for Exogenous Variables
T1
Positive
Affect
T2
Religious
Attendance
T2
Positive
Affect
T1
Positive
Affect
T2
Religious
Attendance
T2
Positive
Affect
-.11
-.05
.08
.22***
.36***
.26***
.05
.02
.11*
-.05
-.01
.09
.25**
.32***
.24***
.16
-.03
-.13*
.10
.03
.05
-.01
.16***
.07
T2
Religious
attendance
-.13
.07
.04
-.20*
-.03
-.03
-.02
.08
-.07
-.26*
.10
-.01
Note: Numbers are standardized (P) regression coefficients. Findings are ordered from top to bottom: oldest (N = 108), middle (N =
260), and youngest (N = 249) generation.
*p < .05 (2-tailed); **p < .01; ***p < .001.
attendance on Time 2 positive affect in the youngest
generation remains statistically significant but
changes from a positive association (p = .11, p < .05)
in Model I to a negative association (p = -.13, p <
.05) in Model II after controlling for exogenous effects. In other words, for the generation of grandchildren, frequent religious attendance at Time 1 is
associated with less positive affect at Time 2 once the
effects of health and sociodemographic characteristics are taken into account. Supplementary analyses
reveal that two of these exogenous variables, being
female and years of education, exert respective positive (p = .19, p<.01) and negative (p = -.22,p<.01)
net effects on Time 2 positive affect in this generation. Considerable prior research has shown positive
bivariate effects of each of these variables on both
religious attendance and positive affect, so perhaps
the negative net effect of education on Time 2 posi460
tive affect has contributed to this anomalous finding.
Another possibility is the presence of multicollinearity related to the exogenous effects of education, not
uncommon in gerontological research on well-being
(see Levin, 1994a).
Another complicated twist in the association between religious attendance and positive affect
emerges in the oldest generation. In Model I, Time 2
religious attendance and Time 2 positive affect in this
generation are unrelated (p = -.13, n.s.), whereas in
Model 11 they are negatively associated (p = -.26, p <
.05). Perhaps contributing to this association is a
negative net effect (p = -.20, p < .05) of Time 1
positive affect on Time 2 religious attendance in
Model II, a counterintuitive association not present
in Model 1 (p = .10, n.s.). In addition, positive affect
is the only one of the three well-being dimensions
whose Time 1 score does not uniformly predict its
The Gerontologist
Time 2 score. Further, although it is well-documented that declining levels of well-being in older
age cohorts are reflective of age-related declines in
health that operate to restrict religious attendance
(Levin, 1989), positive affect in this study increases
over time, on average, and supplementary analyses
show that health is not a significant predictor of
subsequent positive affect ((3 = .10, n.s.). Additional
analyses in the oldest generation also confirm that
age exerts a strong negative effect ((3 = -.33, p <
.001) on Time 2 religious attendance net of the effects
of Time 1 measures of both religious attendance and
positive affect. It is suggested, then, that this rather
confusing situation for positive affect may be an
artifact of the well-known confounding that exists
among religious attendance, age, and well-being
outcomes among elderly adults (Levin, 1989). Alternatively, it may signify that positive affect in general,
or this particular measure, is simply not reflective of
the many more valid well-being measures typically
used in prior gerontological studies, such as use of
the LSIA for life satisfaction or the CES-D for depressed affect.
Discussion
In summary, this study of Mexican Americans
has found that (a) religious attendance is cross-sectionally associated with life satisfaction among members of the oldest and middle generations, and (b)
religious attendance has salutary longitudinal effects
on depressed affect among members of the youngest generation. These cross-sectional findings for life
satisfaction at Time 1 and these longitudinal findings
for depressed affect both withstand controlling for
the effects of health, age, and other known sociodemographic correlates or predictors of both religious
attendance and psychological well-being. The net
findings of a positive cross-sectional association between religious attendance and life satisfaction in
older respondents and a salutary longitudinal effect
of religious attendance on subsequent depressed
affect each confirm hypothesized expectations. The
results of this study have important implications for
research on (a) Hispanic families, (b) minority aging,
and (c) religious factors in mental health and wellbeing.
First, these findings confirm earlier work pointing
to the importance of regular, formal religious involvement, such as religious attendance, for the
well-being of Mexican-American families. Findings
from the 1960s (Grebler, Moore, & Guzman, 1970)
and 1970s (Greeley, 1979) show that Mexican Americans, who are overwhelmingly Roman Catholic in
affiliation, attend religious services Jess frequently
than U.S. Catholics in general Further, the frequency of religious attendance has been found to
decline with successive generations of three-generational Mexican-American families (Markides & Cole,
1984), despite high levels of intergenerational consistency in religious affiliation (Aldous & Hill, 1965;
Markides & Cole, 1984). Nevertheless, religious attendance is a significant correlate or predictor of
Vol. 36, No. 4,1996
subjective health (Markides & Martin, 1983) and life
satisfaction (Markides, Levin, & Ray, 1987) in older
Mexican Americans, and, according to the present
study, a correlate of life satisfaction in middle-aged
Mexican Americans and a protective factor with respect to depressed affect in the youngest generation,
as well. Religious attendance serves as a source of
connectedness to one's cultural traditions and, thus,
for older and middle-aged adults, represents a salient correlate of life satisfaction. For members of the
youngest generation, more acculturated than their
elders and less regular in their religious participation, religious attendance may not be as direct a
source of contemporaneous life satisfaction, but
rather may serve a protective function for mood
disorders over time, perhaps reflecting the increasing importance of religion for this cohort as it ages
into the child-rearing years.
Second, religious involvement is widely held to be
a crucial resource in the lives of older adults, a radix
of meaning (Post, 1992) that is especially valuable as
individuals begin to disengage from other roles and
formal institutional involvements due to retirement,
declining health, or other reasons. Religious attendance benefits older adults both through providing
access to church- and synagogue-based social services of various types and through the "personal
satisfactions" enhanced by worship and religiosity
(Moberg, 1990). Among these personal benefits of
religion are the provision of hope (Koenig, 1994b)
and a means of "preserving the self" (Tobin, 1991), as
well as the offer of forgiveness and reconciliation
with God (Kaplan, Munroe-Blum, & Blazer, 1994). For
minority elders, religion and churchgoing provide a
sense of continuity and help to maintain ties to the
church, a repository of culture and ethos and often
the principal autonomous social institution in respective ethnic-minority communities. According to
Ellison (1994), formal religious involvement also benefits the well-being of older adults through more
tangible means, such as by reducing the risk of
chronic and acute stressors, offering cognitive and
institutional frameworks that buffer stress and facilitate coping, and providing both internal psychological resources (such as self-efficacy) and concrete social resources. The present study's findings linking
religious attendance to satisfaction with life reflect
the salutary impact of these themes — meaning,
continuity, and control — in the lives of older Mexican Americans.
Finally, these results provide solid evidence of
salutary effects for religious involvement on at least
two different domains of mental health and psychological well-being. As in recently published work on
older African Americans (Levin, Chatters, & Taylor,
1995), religious attendance is positively associated
with life satisfaction even after controlling for the
effects of health. The present study thus provides
replication of this result in a different ethnic population and represents further evidence that the beneficial impact of organizational religious activities on
the well-being of older adults is not solely the result
of the confounding often found to exist between
461
Koenig, H. G. (1994a). AgingandGod: Spiritual pathways to mental health in
midlife and later years. New York: Haworth Press.
Koenig, H.G. (1994b). Religion and hope for the disabled elder. In J.S. Levin
(Ed.), Religion in aging and health: Theoretical foundations and methodological frontiers (pp. 18-51). Thousand Oaks, CA: Sage.
Koenig, H. G. (1995). Research on religion and aging: An annotated bibliography. Westport, CT: Greenwood Press.
Koenig, H. G., & Blazer, D. G. (1992). Mood disorders and suicide. In J. E.
Birren, R. B. Sloane, & G. D. Cohen (Eds.), Handbook of mental health
and aging (2nd ed., pp. 379-407). San Diego: Academic Press.
Koenig, H. G., Cohen, H. J., Blazer, D. G., Pieper, C , Meador, K. G., Shelp,
F., Goli, V., & DiPasquale, R. (1992). Religious coping and depression in
elderly hospitalized medically ill men. American Journal of Psychiatry,
149, 1693-1700.
Koenig, H. G., Moberg, D. O., & Kvale,). N. (1988). Religious activities and
attitudes of older adults in a geriatric assessment clinic. Journal of the
American Geriatrics Society, 36, 362-374.
Koenig, H. G., Smiley, M., & Gonzales, J. A. P. (1988). Religion, health, and
aging: A review and theoretical integration. New York: Greenwood
Press.
Krause, N. (1992). Stress, religiosity, and psychological well-being among
older blacks. Journal of Aging and Health, 4, 412-439.
Krause, N. (1993). Measuring religiosity in later life. Research on Aging, 15,
170-197.
Krause, N., & Tran, T. V. (1989). Stress and religious involvement among
older blacks. Journal of Gerontology: Social Sciences, 44, S4-S13.
Lee, G. R., & Ishi-Kuntz, M. (1987). Social interaction, loneliness, and
emotional well-being among the elderly. Research on Aging, 9, 459-482.
Levin, J. S. (1989). Religious factors in aging, adjustment, and health: A
theoretical overview. In W. M. Clements (Ed.), Religion, aging and
health: A global perspective (pp. 133-146). (Compiled by the World
Health Organization). New York: Haworth.
Levin, J. S. (1994a). Dimensions and determinants of general well-being
among older adults. Journal of Aging and Health, 6, 489-506.
Levin, J. S. (Ed.). (1994b). Religion in aging and health: Theoretical foundations methodological frontiers. Thousand Oaks, CA: Sage.
Levin, J. S. (1995a). Bibliography of gerontological research on religion.
Bethesda, M D : National Institute on Aging.
Levin, J. S. (1995b). Religion. In G. L. Maddox (Ed.), The encyclopedia of
aging (2nd ed., pp. 799-802). New York: Springer.
Levin, J. S., Chatters, L. M., & Taylor, R. J. (1995). Religious effects on health
status and life satisfaction among black Americans. Journal of Gerontology: Social Sciences, 50B, S154-S163.
Levin, J. S., & Markides, K. S. (1985). Religion and health in Mexican
Americans. Journal of Religion and Health, 24, 60-69.
Levin, J. S., & Markides, K. S. (1986). Religious attendance and subjective
health. Journal for the Scientific Study of Religion, 25, 31-40.
Levin, J. S., & Markides, K. S. (1988). Religious attendance and psychological well-being in middle-aged and older Mexican Americans. Sociological Analysis, 49, 66-72.
Levin, J. S., & Taylor, R. J. (1993). Gender and age differences in religiosity
among black Americans. The Gerontologist, 33,16-23.
Levin, J. S., Taylor, R. J., & Chatters, L. M. (1994). Race and gender
differences in religiosity among older adults: Findings from four national surveys. Journal of Gerontology: Social Sciences, 49, S137-S145.
Levin, J. S., Taylor, R. J., & Chatters, L. M. (1995). A multidimensional
measure of religious involvement for African Americans. The Sociological Quarterly, 36, 901-919.
Levin, J. S., & Tobin, S. S. (1995). Religion and psychological well-being. In
M. A. Kimble, S. H. McFadden, J. W. Ellor, & J. J. Seeber (Eds.), Aging,
spirituality, and religion: A handbook (pp. 30-46). Minneapolis: Fortress Press.
Liang, J. (1984). Dimensions of the Life Satisfaction Index A: A structural
formulation. Journal of Gerontology, 39, 613-622.
Liang, J., Tran, T. V., Krause, N., & Markides, K. S. (1989). Generational
differences in the structure of the CES-D Scale in Mexican Americans.
Journal of Gerontology: Social Sciences, 44, S110-S120.
Markides, K. S. (1983). Aging, religiosity, and adjustment: A longitudinal
analysis. Journal of Gerontology, 38, 621-625.
Markides, K. S. (1987). Religion. In G. L. Maddox (Ed.), The encyclopedia of
aging (pp. 559-561). New York: Springer.
Markides, K. S., & Cole, T. (1984). Change and continuity in Mexican
American religious behavior: A three-generation study. Social Science
Quarterly, 65, 618-625.
Markides, K. S., Levin, J. S., & Ray, L. A. (1987). Religion, aging, and life
satisfaction: An eight-year, three-wave longitudinal study. The Gerontologist, 27, 660-665.
Markides, K. S., & Martin, H. W. (1979). A causal model of life satisfaction
among the elderly. Journal of Gerontology, 34, 86-93.
Markides, K. S., & Martin, H. W. (1983). Religious behavior. In K. S.
Markides & H. W. Martin with E. Gomez, Older Mexican Americans: A
study in an urban barrio (pp. 64-76). Austin, TX: University of Texas
Press.
McFadden, S. H. (1996). Religion, spirituality, and aging. In J. E. Birren & K.
W. Schaie (Eds.), Handbook of the psychology of aging (4th ed., pp.
162-177). San Diego: Academic Press.
religious attendance and health status (see Ainlay,
Singleton, & Swigert, 1992). More importantly, this
study provides strong evidence of a salutary longitudinal effect of religious attendance on depressed
affect. This is especially noteworthy in that a protective effect on depressed affect persists across the 11
years between waves of data collection and withstands controlling for the effects of both Time 1 affect
and Time 2 religious attendance. These findings echo
the suggestion (Ellison, 1994; Koenig, Smiley, & Gonzales, 1988; Williams et al., 1991) that infrequent
religious involvement deserves further scrutiny as a
possible risk factor in psychiatric epidemiology.
References
Ainlay, S. C , Singleton, R., Jr., & Swigert, V. L. (1992). Aging and religious
participation: Reconsidering the effects of health. Journal for the Scientific Study of Religion, 31, 175-188.
Ainlay, S. C , & Smith, D. R. (1984). Aging and religious participation.
Journal of Gerontology, 39, 357-363.
Aldous, J. W., & Hill, R. (1965). Social cohesion, lineage type, and intergenerational transmission. Social Forces, 43, 471-482.
Anson, O., Antonovsky, A., & Sagy, S. (1990). Religiosity and well-being
among retirees: A question of causality. Behavior, Health, and Aging, 1,
85-97.
Blazer, D., & Palmore, E. (1976). Religion and aging in a longitudinal panel.
The Cerontologist, 16, 82-85.
Brown, D. R., & Gary, L. E. (1987). Stressful life events, social support
networks, and physical and mental health of urban black adults. Journal
of Human Stress, 13,165-174.
Brown, D. R., & Gary, L. E. (1988). Unemployment and psychological
distress among black American women. Sociological Focus, 21, 209220.
Brown, D. R., Ndubuishi, S. C , & Gary, L. E. (1990). Religiosity and psychological distress among blacks. Journal of Religion and Health, 29, 55-68.
Chatters, L. M., Levin, J. S., & Taylor, R. J. (1992). Antecedents and dimensions of religious involvement among older black adults. Journal of
Gerontology: Social Sciences, 47, S269-S278.
Chatters, L. M., & Taylor, R. J. (1994). Religious involvement among older
African-Americans. In J. S. Levin (Ed.), Religion in aging and health:
Theoretical foundations and methodological frontiers (pp. 196-230).
Thousand Oaks, CA: Sage.
Ellison, C. G. (1991). Religious involvement and subjective well-being.
Journal of Health and Social Behavior, 32, 80-99.
Ellison, C. G. (1994). Religion, the life stress paradigm, and the study of
depression. In J. S. Levin (Ed.), Religion in aging and health: Theoretical
foundations and methodological frontiers (pp. 78-121). Thousand
Oaks, CA: Sage.
Ellison, C. G., & Gay, D. A. (1990). Region, religious commitment, and life
satisfaction among black Americans. The Sociological Quarterly, 31,
123-147.
Freund, R. J., & Littell, R. C. (1981). SAS for linear models: A guide to the
ANOVA and GLMprocedures. Cary, NC: SAS Institute.
George, L. K. (1981). Subjective well-being: Conceptual and methodological issues. Annual Review of Gerontology and Geriatrics, 2, 345-382.
George, L. K., & Landerman, R. (1984). Health and subjective well-being: A
replicated secondary data analysis. International Journal of Aging and
Human Development, 19, 133-156.
Grebler, L., Moore, J. W., & Guzman, R. C. (1970). The Mexican American
people: The nation's second largest minority. New York: Free Press.
Greeley, A. M. (1979). Ethnic variations in religious commitment. In R.
Wuthnow(Ed-), The religious dimension: New directions in quantitative
research (pp. 113-134). New York: Academic Press.
Guy, R. F. (1982). Religion, physical disabilities, and life satisfaction in older
age cohorts. International Journal of Aging and Human Development,
15, 225-232.
Haganaars, J. A. (1990). Categorical longitudinal data: Log-linear panel,
trend, and cohort analysis. Newbury Park, CA: Sage.
Heisel, M. A., & Faulkner, A. O. (1982). Religiosity in an older black
population. The Gerontologist, 22, 354-358.
Idler, E. L., & Kasl, S. V. (1992). Religion, disability, depression, and the
timing of death. American Journal of Sociology, 97, 1052-1079.
Kaplan, B. H., Munroe-Blum, H., & Blazer, D. G. (1994). Religion, health,
and forgiveness: Traditions and challenges. In J. S. Levin (Ed.), Religion
in aging and health: Theoretical foundations and methodological frontiers (pp. 52-77). Thousand Oaks, CA: Sage.
Kimble, M. A., McFadden, S. H., Ellor, J. W., & Seeber, J. J. (Eds.). (1995).
Aging, spirituality, and religion: A handbook. Minneapolis, M N : Fortress Press.
462
The Gerontologist
Mindel, C. H., & Vaughan, C. E. (1978). A multidimensional approach to
religiosity and disengagement. Journal of Gerontology, 33, 103-108.
Moberg, D. O. (1953). Church membership and personal adjustment in old
age. Journal of Gerontology, 8, 207-211.
Moberg, D. O. (1965). Religiosity in old age. The Gerontologist, 5(2), 78-87.
Moberg, D. O. (Ed.). (1971). Spiritual well-being: Background and issues.
Washington, DC: White House Conference on Aging.
Moberg, D. O. (1990). Religion and aging. In K. F. Ferraro (Ed.), Gerontology: Perspectives and issues (pp. 179-205). New York: Springer.
Neugarten, B. L , Havighurst, R. J., & Tobin, S. S. (1961). The measurement
of life satisfaction. Journal of Gerontology, 76,134-143.
Post, S. G. (1992). Aging and meaning: The Christian tradition. In T. R. Cole,
D. D. Van Tassel, & R. Kastenbaum (Eds.), Handbook of the humanities
and aging (pp. 127-146). New York: Springer.
Radloff, L. S. (1977). The CES-D Scale: A self-report depression scale for
research in the general population. Applied Psychological Measurement, 7,385-401.
Sherrill, K. A., Larson, D. B., & Creenwold, M. (1993). Is religion taboo in
gerontology?: Systematic review of research on religion in three major
gerontology journals, 1985-1991. American Journal of Geriatric Psychiatry, 1,109-117.
The 49th Annual
Scientific Meeting
Till; GlRONTOLOGICAL
SOCIETY or AMKRICA
November 17-20, 1996
Sheraton Washington Hotel
Washington, DC
St. George, A., & McNamara, P. H. (1984). Religion, race and psychological
well-being. Journal for the Scientific Study of Religion, 23, 351-363.
Steinitz, I . (1980). Religiosity, well-being, and Weltanschauung among the
elderly. Journal for the Scientific Study of Religion, 19, 60-67.
Thomas, L. E., & Eisenhandler, S. A. (Eds.). (1994). Aging and the religious
dimension. Westport, CT: Auburn House.
Tobin, S. S. (1991). Preserving the self through religion. In S. S. Tobin,
Personhood in advanced old age: Implications for practice (pp. 119133). New York: Springer.
Williams, D. R., Larson, D. R., Buckler, R. E., Heckmann, R. C , & Pyle, C. M.
(1991). Religion and psychological distress in a community sample.
Social Science and Medicine, 11, 1257-1262.
Witter, R. A., Stock, W. A., Okun, M. A., & Haring, M. J. (1985). Religion and
subjective well-being in adulthood: A quantitative synthesis. Review of
Religious Research, 26, 332-342.
Wood, V., Wylie, M., & Sheafor, B. (1969). An analysis of a short self-report
measure of life satisfaction: Correlation with rater judgments. Journal
of Gerontology, 24, 465-469:
Received June 29, 1995
Accepted January 26, 1996
100+ Exhibit Booths
Combined Book Exhibits
"Free Literature" Tables
Grand Opening Reception
Complimentary Coffee Breaks
Frequent Non-competitive Breaks
These are just some of the advantages of exhibiting to the 3,500
decision makers who comprise this audience of clinicians, educators,
administrators, researchers and other professionals. Price—we offer
low rates and substantial discounts. Size—you can still snap up a prime
spot. And don't forget location, location, location! Washington, DC!
GSA, 1275 K St NW #350, Washington, DC 20005-4006•Phone 202/842-1275 • FAX 202/842-1150
Vol. 36, No. 4, 1996
463