Exploring the Dimensionality of “Religiosity”

and “Spirituality” in the Fetzer

JAMES ALAN NEFF

Reliable and valid measures of religiosity and spiritual dimensions appropriate for ethnically diverse groups are
needed. I examine the factor structure of the relatively new Fetzer multidimensional measure (Fetzer Institute
1999). Confirmatory factor analyses (CFA) of national data collected in the 1998 General Social Survey are
conducted to (1) test the fit of the proposed dimensionality of the Fetzer measure, specifically comparing the
Fetzer structure with a simpler single-factor model suggested by preliminary exploratory factor analyses; and (2)
evaluate ethnic differences in factor structure of the Fetzer items. Data from 1,201 adults, 1,019 whites and 182
African Americans, are examined to compare the fit of the multidimensional Fetzer model against an alternative
single-factor (“unitary spirituality”) model. CFA analyses indicated that only the Fetzer models had acceptable fit
indices (NFI and/or NNFI approaching or reaching 0.90). Ethnic differences in loadings, while significant, were
not substantively different. These findings suggest that a multidimensional measure like the Fetzer is useful for
multiethnic research and that combining the daily spirituality and values/beliefs dimensions into a single factor
provides a simpler model overall (compared to the Fetzer model separating these dimensions out) while providing
acceptable fit.

INTRODUCTION

Scholars acknowledge that religious and spiritual influences are poorly understood (cf. Ellison
1993; Armstrong 1996) and efforts to clarify both theoretical and methodological issues surround-
ing these dimensions and their important implications for health and well-being are increasing
(Fetzer Institute 1999). Fundamental to systematic research in this area is an acknowledgment of
the multidimensional nature of these concepts. As well, reliability and validity of measurement
are critical concerns—particularly in research with minority populations, where measures must
be comprehensible as well as culturally competent.
The conceptualization of “religiosity” and “spirituality” involves complex issues (Hill and
Hood 1999) well beyond the scope of the current examination. The present study is more modest,
involving an empirical analysis of the dimensionality of a relatively new measure, referred to
here as the Fetzer multidimensional measure (Fetzer Institute 1999) designed to tap a variety
of dimensions of spirituality and religiosity as defined by an expert panel (Idler et al. 2003).
The purpose of this research note is decidedly empirical—I seek to clarify the psychometric
properties of the Fetzer measure rather than to advance the conceptualization of spirituality and
religiosity concepts. While the Fetzer measure was not developed to provide a theory-based,
all-encompassing measure of religiosity or spirituality (Idler et al. 2003), the measure has been
widely cited. A Google Scholar search for “Fetzer Multidimensional Spirituality Measure” yielded
over 140 citations, thus attesting to its general acceptance. The Fetzer measure was developed to
provide relatively brief measures for a number of dimensions that would hopefully prove useful in
research on relationships between religiosity/spirituality and health and well-being dimensions.
The premise of the present analyses is that it is important to assess how well the Fetzer items fit

James Alan Neff is Professor of Community and Environmental Health in the College of Health Sciences, 3136 Technology
Building, Old Dominion University, Norfolk, VA 23529. E-mail: [email protected]

Journal for the Scientific Study of Religion (2006) 45(3):449–459

the original conceptualization in order to establish the potential utility of the measures for future
research.
I conduct confirmatory factor analyses (CFA) of national data collected in the 1998 General
Social Survey (GSS) to examine the dimensionality of the Fetzer items to (1) test the fit of the
proposed dimensionality of the Fetzer measure, specifically comparing the Fetzer structure with
simpler models (i.e., single “spirituality” factor vs. more complex models), and (2) evaluate ethnic
differences in the factor structure of the Fetzer measure (i.e., comparing spiritual dimensions for
non-Hispanic whites and African Americans).

BACKGROUND

While the explicit focus of the present analyses is upon the dimensionality of the Fetzer mea-
sure, some background with regard to the issue of parsimony versus complexity of measurement
of religiosity and spirituality is instructive. Historically, much research on religion and spiritual-
ity has involved rudimentary unidimensional and, often, single-item measures (Williams 1993)
such as religious affiliation (church membership and/or denominational affiliation) and religious
participation (e.g., frequency of church attendance; see the reviews by Gartner et al. 1991 and
Witter et al. 1985).
Acknowledging limitations of single-item measures, more complex measures often follow
the classic conceptual distinction between religiosity and spirituality, where religiosity involves
formal or informal religious practice (public or private) and spirituality involves the individual’s
relationship to some transcendent force (God or higher power; cf. Armstrong 1996). Spirituality
is said to represent an integrative force in the individual’s life (Ellison 1983), providing meaning,
and core values and principles for organizing one’s life.
Among such religiosity-spirituality measures, there is variation in the extent to which mea-
sures differentiate between subdimensions of religiosity and spirituality. As an example of an
intermediate model, Chatters, Levin, and Taylor (1992) propose a three-dimensional model of
religiosity (their terminology), which distinguishes between two religiosity dimensions, orga-
nizational involvement (formal involvement such as church attendance) and nonorganizational
involvement (informal involvement such as prayer and Bible studies at home), and one spirituality
dimension, which they refer to as subjective religiosity (beliefs, knowledge, attitudes, and the
perceived importance of religion in the individual’s life).
While analyses support the posited three-factor model (Chatters et al. 1992; Levin, Taylor,
and Chatters 1995), the Chatters’s measure differentiates between subdimensions of religiosity,
but not spirituality. Specifically, spirituality is treated as a single dimension, captured by their
“subjective religiosity” factor, and measured with only three items. The authors acknowledge
that their measure does not follow from an explicit conceptual framework—rather, it is based
upon items available in their data—and give no consideration to the question of what spirituality
subdimensions might look like. Finally, in terms of measuring religiosity and spirituality in diverse
populations, Levin et al. (1995) note that the three dimensions, though developed specifically using
an African-American sample, may not reflect the diversity or complexity of religious experience
for African-American populations.
Both religiosity and spirituality dimensions are measured with greater complexity by the
Fetzer/NIA Brief Multidimensional Measure of Religiousness/Spirituality (Fetzer Institute 1999).
This measure taps religiosity subdimensions (Private Religious Practices, Organizational Reli-
giosity, and Self-Rated Religiosity) as well as multiple spirituality subdimensions (Daily Spiritual
Experiences, Spiritual Values/Beliefs, Forgiveness, Positive and Negative Spiritual Coping, and
Religious Support). A subset of Fetzer items in each dimension were included in the 1998 GSS
of over 1,300 adults, including roughly 200 African-American adults. Internal consistency reli-
ability estimates for the GSS measures in the total sample were generally acceptable (minimum
Cronbach’s alpha of 0.64; most reliabilities >0.70).
FETZER MULTIDIMENSIONAL MEASURE 451

The Fetzer measure has the advantage of having been designed by a consensus panel of experts
and offers an advance over simpler measures in tapping multiple subdimensions of both religiosity
and spirituality. As the authors of the Fetzer measures have recently noted (Idler et al. 2003), Fetzer
dimensions were selected to emphasize possible “pathways” of spiritual influences upon health,
though the scales were not intended to reflect an overarching conceptual model of religiosity or
spirituality. Rather, individual panel members developed measures of concepts in their specific
areas of interest. Idler et al. (2003) present further reliability analyses to support the internal
consistency of the measures and also provide gender-age analyses that they argue demonstrate the
potential utility of the measures. Thus, while the individual scales as defined by the authors have
some empirical psychometric support, and the structure of the Daily Spiritual Experience Scale
has been examined with exploratory factor analysis techniques (Underwood and Teresi 2002),
there have not been confirmatory analyses to test the fit of the posited multidimensional structure
of the Fetzer measures.
A related empirical issue with regard to the Fetzer measures involves their utility for cross-
cultural or multicultural research. Given the diversity and complexity of religious experience in
the African-American population (Levin et al. 1995), one might ask about the extent to which
the structure of religiosity and spirituality among African Americans might differ from that of
other ethnic groups, specifically, Anglo Americans (or non-Hispanic whites). The historical cen-
trality of religion and church among African Americans (Lincoln and Mamiya 1990; Ellison and
Sherkat 1995), and spiritual tenets of the Black Church that emphasize relationships with others
and community involvement rather than simply emphasizing relationships with a transcendent
force or a focus upon the afterlife (Armstrong 1996; Fletcher 1998; Mattis 2000), all suggest
the value of empirically examining ethnic differences in the dimensionality of religiosity and
spirituality.
The Fetzer multidimensional measure has been proposed to be useful in health research.
Accepting that the measure was not theoretically derived—and was not intended as an all-
encompassing measure—I examine the dimensionality of the measure empirically to see if CFA
techniques support the proposed dimensions of the measure. Specifically, these analyses seek to
help clarify the utility of the Fetzer measures by addressing: (1) the fit of the proposed dimensional-
ity of the Fetzer measures, and (2) whether similar factor structures are obtained for non-Hispanic
whites and African Americans. Confirmatory techniques cannot prove that the Fetzer items truly
measure religiosity and spirituality, but they can help to assess whether the data fit the dimensions
as proposed by the Fetzer authors.

METHODOLOGY

Sample Characteristics

The data presented here were collected as part of the 1998 GSS, a national probability sample
of 2,832 respondents. Data on the Fetzer spirituality measure were available on approximately
1,400 of the total GSS sample. Based upon the minimum number of cases for which complete
Fetzer data are available, the racial/ethnic composition of the sample was 75.8 percent non-
Hispanic white (N = 1,012), 14.1 percent African American (N = 188), 6.3 percent Hispanic
(N = 84), and 3.8 percent “Other” (N = 51). To allow for an adequate number of cases for analysis,
I use the data for non-Hispanic whites and African Americans.
Demographically, GSS sample respondents were predominantly female (45.7 percent male)
with an average age of 46.5, an average of 13.6 years of education, and an approximate av-
erage household income of $22,500. Approximately one-half of respondents (48.6 percent)
were married. Roughly 34.2 percent of respondents were not working either full or part time.
African-American respondents were significantly (p < 0.05) more likely than whites to be female
(39 percent vs. 46 percent, respectively), unmarried (63 percent vs. 51 percent), younger (average
452 JOURNAL FOR THE SCIENTIFIC STUDY OF RELIGION

age 43 vs. 46 years), and less educated (average years of education 12 vs. 14). There were no
significant racial/ethnic differences in terms of likelihood of being unemployed or household
income.

Description of the Fetzer Items

A subset of 23 of the Fetzer items was included in the GSS. Brief versions of the following
scales were included (see Fetzer Institute 1999): Perceived Religiosity (2 items), Daily Spirituality
(6 items), Positive Coping (3 items), Negative Coping (3 items), Forgiveness (3 items), Religious
Involvement (3 items), and Spiritual Values and Beliefs (3 items). All items were scored on a
5-point strongly agree through strongly disagree Likert format.

ANALYSIS AND RESULTS

Analysis Procedure

Analyses began with exploratory principal components analyses of the Fetzer item set to as-
sess the feasibility of a single factor solution as a reference point for further analyses and to explore
possible differences in dimensionality between non-Hispanic whites and African Americans.
To explore more systematically the dimensionality of the Fetzer items, CFA techniques
were used, following Jöreskog and Sörbom‘s (1993) “alternative models” (AM) approach. The
Jöreskog and Sörbom (1993, 2000) approach allows goodness-of-fit tests of specific models in
reproducing the covariance or correlation matrix via χ 2 statistics. The improvement in model fit
in comparisons of a hierarchically ordered series of models is assessed with incremental χ 2 tests
and fit indices (Bentler and Bonett 1980). The Jöreskog approach has the additional advantage of
fitting models simultaneously in multiple subgroups and testing the invariance of factor loadings
or factor correlations across groups (Byrne 1989).
While the multidimensional Fetzer model could be used for a test of exact fit in a “strictly
confirmatory” analysis (i.e., does the Fetzer model fit?), for purposes of parsimony, I also consider
a simpler reference model (a single-factor model (M1 ) incorporating all 23 items in a unitary
dimension) suggested by preliminary exploratory factor analyses. The full Fetzer model is included
as model M2 The theoretic point is to test whether spirituality, as measured by the Fetzer items,
can most simply and parsimoniously be modeled and, more specifically, how well the posited
Fetzer multidimensional model fits the data.
Preliminary principal components analyses of the total item sets yielded four components
(eigenvalues >1.0) for whites and six for African Americans, both suggesting global spirituality
dimensions in the first component (percentage of variance explained of 43 percent for whites
and 37 percent for African Americans), consisting of all items other than negative coping. For
both racial groups, the three negative coping items had either negligible or negative loadings on
the first factor relative to other items and these items did not load consistently on another single
factor. Given these findings (combined with the inability of LISREL to fit CFA models including
these negative coping items, as discussed below), the negative coping items were dropped from
the analysis.
Subsequent principal components analyses of the remaining 20 items (Table 1) again sug-
gested a first global spirituality factor (percentage of variance explained of 48 percent for whites
and 42 percent for African Americans) as well as a second factor (Forgiveness) for both whites
and African Americans. Loadings for the first principal component ranged from approximately
0.39 to 0.85 for both whites and African Americans (average values of 0.67 and 0.63, respec-
tively), indicating relatively high and consistent loadings in both groups. Use of several extrac-
tion and rotational methods yielded little consistent evidence of further dimensionality or racial
differences.
FETZER MULTIDIMENSIONAL MEASURE 453

TABLE 1
EXPLORATORY UNROTATED PRINCIPAL COMPONENTS ANALYSES: WHITES AND
AFRICAN AMERICANS
Whites African Americans
Component Component

I II III I II III IV
Daily spirituality/values/beliefs
I believe in a God who watches over 0.716 0.053 −0.438 0.556 −0.236 0.200 −0.169
me (V1 )
I feel a deep sense of responsibility for 0.471 −0.154 0.404 0.380 −0.220 0.429 0.354
reducing pain and suffering in the
world (V2 )
I try hard to carry my religious beliefs over 0.720 −0.092 0.051 0.648 0.031 0.514 0.066
into all my other dealings in life (V3 )
I desire to be closer or in union 0.827 −0.113 −0.080 0.756 −0.062 −0.238 0.260
with God (D1 )
I feel God’s love, directly or through 0.846 −0.027 −0.027 0.760 −0.029 −0.086 0.282
others (D2 )
I am spiritually touched by the beauty of 0.667 −0.013 0.280 0.641 0.139 −0.094 0.461
creation (D3 )
I feel God’s presence (D4 ) 0.825 −0.051 −0.079 0.587 −0.281 −0.366 0.180
I find strength and comfort in religion (D5 ) 0.850 −0.137 −0.025 0.847 −0.097 −0.229 −0.064
I feel deep inner peace or harmony (D6 ) 0.680 0.062 0.324 0.639 0.117 −0.504 0.132
Forgiveness
I have forgiven myself for things that I 0.388 0.725 0.169 0.395 0.660 0.275 0.051
have done wrong (F1 )
I have forgiven those who hurt me (F2 ) 0.433 0.667 0.136 0.525 0.568 0.041 0.047
I know that God forgives me (F3 ) 0.577 0.379 −0.351 0.470 0.323 −0.017 −0.009
Positive religious coping
I think about how my life is part of a larger 0.716 0.053 −0.438 0.581 0.049 0.242 −0.06
spiritual force (PC1 )
I work together with God as partners (PC2 ) 0.471 −0.154 0.404 0.663 −0.269 0.215 0.22
I look to God for strength, support, and 0.720 −0.092 0.051 0.711 −0.286 0.245 −0.07
guidance (PC3 )
Private religiosity
How often do you pray privately in places 0.827 −0.113 −0.080 0.727 −0.014 0.024 −0.25
other than at church or synagogue (PR1 )
Within your religious or spiritual tradition, 0.846 −0.027 −0.027 0.574 0.256 −0.140 −0.21
how often do you meditate (PR2 )
Public religiosity
How often do you go to religious 0.667 −0.013 0.280 0.722 −0.002 −0.140 −0.21
services (P1 )
Self-rated religiosity
To what extent do you consider yourself 0.825 −0.051 −0.079 0.741 −0.170 −0.030 −0.28
a religious person (SR1 )
To what extent do you consider yourself 0.850 −0.137 −0.025 0.745 −0.028 −0.113 −0.36
a spiritual person (SR2 )
Eigenvalue 9.68 1.26 1.15 8.32 1.35 1.28 1.06
Percent of variance 48.42 6.30 5.75 41.62 6.73 6.39 5.32
454 JOURNAL FOR THE SCIENTIFIC STUDY OF RELIGION

Consistent with the exploratory analyses, initial CFA attempts to fit model M1 did not con-
verge; deletion of the three negative coping items resulted in convergence in all subsequent
analyses. Thus, models were fit with the remaining 20 items. Goodness-of-fit statistics (NFI,
NNFI, CFI, and RMSEA; Jöreskog and Sörbom 1993) for AM and for incremental χ 2 tests were
computed using covariance matrix input to LISREL 8.50 (Jöreskog and Sörbom 2000).
CFA analyses began by fitting the single-factor model (M1 ) with factor loadings and error
covariances constrained to be equal for whites and African Americans, examining the acceptability
of fit indices (fit values ≥0.90; RMSEA close to 0.05, cf. Jöreskog and Sörbom 1993). Non-
Hispanic whites served as the reference group in these analyses—i.e., initial parameter estimates
based upon this subgroup were then used for African Americans in constrained analyses. The
single-factor model (M1 ) yielded a significant lack of fit to the data (χ 2380df = 2592.77, p < 0.001).
However, significant lack of fit might be expected given a large sample size, and fit indices (e.g.,
NFI for model M1 = 0.83) indicated an improvement in model fit of 83 percent compared to a null
model (Bentler and Bonett 1980). M1 appears useful as a reference model for further comparisons,
despite lack of fit.
M2 , the multidimensional Fetzer model, yielded fit indices approaching acceptable
levels \ break (NFI = 0.88; NNFI = 0.89). Model fit was significantly improved relative to M1
(χ 2 for improvement in fit of M2 compared to M1 = 712.59 on 20 df, p < 0.001) and this model
yielded an improved RMSEA of 0.08. Finally, based upon high observed correlations between
the “daily spirituality” and “values/beliefs” latent variables (see below), an additional model (the
modified Fetzer model M3 ) was fit, which combined these two dimensions. This model yielded
a significant decrement in model fit relative to M2 (χ 2 difference between M2 and M3 = 201.81
on 6 df, p < 0.001), though the model retained an NFI of 0.87 (compared to 0.88 for the Fetzer
model). As the two dimensions had similar item content, combining them makes intuitive sense.
The two variants of the Fetzer model (M2 and M3 ) represent possibly acceptable solutions,
so additional models were fit (based upon model modification indices provided by LISREL) to
assess the effects of adding error covariances within latent variables as well as relaxing equality
constraints on factor loadings between whites and African Americans. These analyses indicate that
for both models M2 and M3 , the addition of error covariances and freeing of equality constraints
on factor loadings yield significant improvements in fit and yielded acceptable fit indices (NFI of
0.89 and 0.88 for M2 and M3 ; NNFI of 0.90 for both M2 and M3 ). I note that modification indices
suggested error covariances only within latent variables reflecting Values (V2 –V3 ), Forgiveness
(F1 –F2 , F2 –F3 ), and within the Positive Coping latent variable (PC1 –PC2 ), thus suggesting further
specification of within-factor dependencies between indicators rather than suggesting dependen-
cies across latent variables.
Substantively, these tests indicate that, while a more complex model separating out “daily
spirituality” from “values/beliefs” may fit somewhat better than the more parsimonious combined
model, the overall fit of the two models is approximately equal (approaching 0.90 for each).
Further, the analyses suggest that factor loadings may vary across groups. Further analyses were
conducted to explore the implications of these CFA results.
First, while the CFA results suggest significant variation in factor loadings between whites
and African Americans, it is instructive to examine the observed factor loadings for the two
groups. Table 2 presents maximum likelihood estimates of the unconstrained loadings (unstan-
dardized with standard errors and standardized values) for each loading. Standardized loadings
are provided to allow clearer between-group comparisons in a common metric. Presented here
are loadings estimated for M3 , combining spiritual values/beliefs and daily spirituality, though
estimates computed under M2 show similar patterns.
What is most striking is the consistency of these loadings across racial groups. Loadings for
an individual item may be lower in one group and higher in another, but these variations do not
fall into consistent patterns and loadings generally fall in similar rankings across groups. One
way to examine consistency was to compare the rankings of loadings within latent variables for
FETZER MULTIDIMENSIONAL MEASURE 455

TABLE 2
MAXIMUM LIKELIHOOD FACTOR LOADINGS FOR MODIFIED FETZER MODEL
M3 (COMBINED DAILY SPIRITUALITY AND SPIRITUAL VALUES/BELIEFS) FOR
WHITES AND AFRICAN AMERICANS
Unstandardized Loadings Standardized
(SEs in Parentheses) Loadings

African African
Whites Americans Whites Americans
Combined Daily Spirituality and Values/Beliefs
I believe in a God who watches over me (V1 ) 0.56∗ (0.02) 0.29∗ (0.04) 0.70 0.46
I feel a deep sense of responsibility for reducing pain 0.31∗ (0.02) 0.29∗ (0.06) 0.38 0.35
and suffering in the world (V2 )
I try hard to carry my religious beliefs over into all my 0.60∗ (0.03) 0.47∗ (0.05) 0.69 0.60
other dealings in life (V3 )
I desire to be closer or in union with God (D1 ) 1.37∗ (0.04) 0.95∗ (0.08) 0.85 0.74
I feel God’s love, directly or through others (D2 ) 1.37∗ (0.04) 1.11∗ (0.08) 0.86 0.80
I am spiritually touched by the beauty of 0.98∗ (0.04) 0.85∗ (0.09) 0.65 0.59
creation (D3 )
I feel God’s presence (D4 ) 1.38∗ (0.04) 0.88∗ (0.08) 0.82 0.68
I find strength and comfort in religion (D5 ) 1.44∗ (0.04) 1.23∗ (0.08) 0.87 0.83
I feel deep inner peace or harmony (D6 ) 0.93∗ (0.04) 0.85∗ (0.09) 0.67 0.64
Forgiveness
I have forgiven myself for things that I have done 0.36∗ (0.03) 0.47∗ (0.08) 0.42 0.51
wrong (F1 )
I have forgiven those who hurt me (F2 ) 0.44∗ (0.03) 0.54∗ (0.07) 0.57 0.65
I know that God forgives me (F3 ) 0.68∗ (0.03) 0.28∗ (0.05) 0.86 0.56
Positive coping
I think about how my life is part of a larger spiritual 0.60∗ (0.03) 0.59∗ (0.07) 0.57 0.56
force (PC1 )
I work together with God as partners (PC2 ) 0.81∗ (0.03) 0.69∗ (0.06) 0.79 0.74
I look to God for strength, support, and 0.94∗ (0.03) 0.66∗ (0.05) 0.86 0.77
guidance (PC3 )
Private religiosity
How often do you pray privately in places other than at 2.11∗ (0.07) 1.53∗ (0.13) 0.84 0.75
church or synagogue (PR1 )
Within your religious or spiritual tradition, how often 1.42∗ (0.08) 1.59∗ (0.19) 0.53 0.57
do you meditate (PR2 )
Public religiosity
How often do you go to religious services (P1 ) 2.59∗ (0.06) 2.45∗ (0.14) 0.93 0.93
Self-rated religiosity
To what extent do you consider yourself a religious 0.81∗ (0.02) 0.77∗ (0.05) 0.86 0.85
person (SR1 )
To what extent do you consider yourself a spiritual 0.66∗ (0.03) 0.75∗ (0.06) 0.72 0.76
person (SR2 )
Error covariances
V2 –V3 0.11∗ (0.02) 0.12∗ (0.03) 0.16 0.19
F1 –F2 0.13∗ (0.02) 0.06 (0.05) 0.20 0.07
F2 –F3 −0.07∗ (0.02) −0.03 (0.03) −0.12 −0.08
PC1 –PC2 0.16∗ (0.02) 0.04 (0.04) 0.26 0.06
∗
Denotes loadings at least two times their standard error.
456 JOURNAL FOR THE SCIENTIFIC STUDY OF RELIGION

whites and African Americans. The Daily Spirituality/Values-Beliefs rankings were identical
for six of nine loadings (rank-order correlation of 0.88). There was even greater consistency of
rankings for Positive Coping (3/3), Private Religiosity (2/2), and Self-Rated Religiosity (2/2).
The greatest discrepancies were found for Forgiveness. Here, while “forgiveness of self” had
the lowest loading in both groups, rankings of “forgiveness of others” and “forgiveness by God”
were reversed for whites and African Americans. It is unclear whether such differences are sub-
stantively meaningful—what is more striking are similarities across groups. While the overall
tests of invariance were significant, those omnibus tests reflect the sum total of differences across
groups—differences that may be negligible when considered separately but that become statis-
tically significant when considered in total. The absence of consistent patterns in the loadings
argues against systematic racial variation.
A second question, involving the nature and magnitude of correlations between latent factors,
pertains both to racial variation as well as to the possibility of a higher-order global religiosity-
spirituality factor. Specifically, correlations between factors obtained in model M2 , with Daily
Spirituality and Values/Beliefs separately, were compared for whites and African Americans
to assess the consistency of correlations. Factor correlations in both groups were consistently
high (averages = 0.75 and 0.74 for whites and African Americans, respectively (SDs = 0.13
and 0.12); medians = 0.75 and 0.77, respectively; ranges = 0.45–0.93 and 0.50–0.99). Specific
data on correlations are available from the author upon request, though a few variations are
suggestive. For example, Public and Private Religious Participation correlate more highly among
African Americans (0.81) than whites (0.65). Similarly, Self-Rated Religiosity correlates more
highly with both Public and Private Religious Participation among African Americans (0.76 and
0.90) than among whites (0.69 and 0.82). These findings would be consistent with research noting
higher levels of religiosity among African Americans than whites. In terms of CFA comparisons of
Models M2 and M3 , I note that Daily Spirituality and Values/Beliefs correlate highly in both groups
(0.89 for whites; 0.77 for African Americans), suggesting the appropriateness of combining these
possibly overlapping dimensions. All in all, comparisons of factor correlations suggest substantial
similarities between whites and African Americans.

DISCUSSION AND CONCLUSIONS

The focus of this research note has been empirical rather than theoretical. As the Fetzer
multidimensional measure has been developed to be useful in studying relationships between
religiosity/spirituality and health and well-being dimensions, the goal here has been to capitalize
upon the availability of national data on the measure provided by the 1998 GSS to go beyond
previous psychometric or exploratory examinations of the measure (Fetzer Institute 1999; Under-
wood and Teresi 2002; Idler et al. 2003). Specifically, I applied CFA techniques to examine the
dimensionality of the Fetzer items in order to help assess the measure’s potential utility. I must
begin the discussion with the caveat that the present study is limited in its focus by examining
“religiosity” and “spirituality” as measured by the available Fetzer items. As noted, the Fetzer
measure was not designed to be comprehensive and, further, only brief versions of proposed Fetzer
dimensions were included in the GSS. Given these facts, are confirmatory analyses of the mea-
sures really appropriate? While the present analyses are limited to the available data and available
items, my premise is that, as Fetzer (1999) and Idler et al. (2003) both present psychometric data
to support the utility of the brief measures used in the GSS, CFA is particularly important to assess
whether these brief ad hoc measures fit the data. Previous considerations of the internal consis-
tency reliability of individual scales do not address questions of uniqueness or overlap between
Fetzer scales; CFA techniques are particularly appropriate to address these issues. However, CFA
techniques cannot magically inject a theoretical framework if one was not present to begin with
(Idler et al. 2003). The present study simply addresses the question of whether the Fetzer items
used in the GSS data appear to fit the proposed dimensionality.
FETZER MULTIDIMENSIONAL MEASURE 457

Beyond the dimensionality question, I have considered whether the factor structure of the
Fetzer items differs between whites and African Americans. Given the availability of the 1998
GSS data, it has been possible to assess empirically racial (at least non-Hispanic white and
African-American) differences in the structure of the Fetzer measure.
In terms of dimensionality, while exploratory analyses suggest the possibility of a global
spirituality dimension, CFA analyses indicated that the single-factor model did not fit adequately.
Both Fetzer models (M2 and M3 in the analyses) had fit indices approaching or reaching 0.90,
suggesting that the multidimensional Fetzer measure may work rather well and that combining the
Daily Spirituality and Values/Beliefs dimensions provides a simpler model overall (compared to
the full Fetzer model) while providing comparable fit. High correlations observed between Daily
Spirituality and Value/Belief latent variables further support the combination of these dimensions
in model M3 . Psychometric data not presented here indicated marginal internal consistency relia-
bility (0.66) for the Values/Beliefs items by themselves, though the combined Values/Beliefs and
Daily Spirituality items yielded acceptable levels of reliability (Cronbach’s alpha > 0.90).
Omnibus tests for differences in factor loadings suggested significant racial variation; how-
ever, close inspection of exploratory and confirmatory analysis results between non-Hispanic
whites and African Americans yielded generally similar patterns. Exploratory analyses suggested
a global principal component with a similar secondary Forgiveness component in both groups.
CFA analyses for both whites and African Americans showed similar rankings of loadings within
dimensions. There was some suggestion of variation in loadings on the Forgiveness dimension,
though it is difficult to assess whether such findings are of substantive importance. It is, however,
intriguing that for whites factor loadings on “forgiveness of others” were lower than loadings for
“forgiveness by God,” while the reverse was observed among African Americans. This might be
consistent with suggestions regarding the importance of relationships with others among African
Americans (Fletcher 1998; Mattis 2000; Armstrong 1996), though the present findings are sugges-
tive at best. Further comparisons suggest higher correlations between Public and Private Religious
Participation dimensions among African Americans than for whites. While this may suggest higher
levels of religious participation and self-identification among African Americans, similarities in
the rankings of factor loadings themselves may suggest that observed statistically significant om-
nibus tests for racial differences in loadings may reflect an accumulation of smaller differences
rather than systematic racial differences in the dimensionality of “spirituality.” Overall, Fetzer
models appear potentially useful in research involving non-Hispanic white and African-American
subgroups.
Several specific issues merit discussion here. One notable finding in the analyses (both
exploratory and CFA) was the difficulty in obtaining convergent solutions when the three negative
coping items were included in analyses. Given Pargament’s (1997) extensive research on religious
coping, it is surprising that the negative coping subdimension did not “fit” in the analyses with
the other subdimensions. One possible explanation is that negative coping may have simply fallen
out of the analyses given its negative emphasis in contrast to the other more positive dimensions.
Alternatively, as negative coping items had a Cronbach’s alpha internal consistency estimate of
0.31, the negative coping items may not represent a single dimension. These items tapped “Feel
God is punishing me for sin,” “Wonder whether God has abandoned me,” and “Try to make
sense without relying on God.” While the first two items appear to tap negative dimensions, the
final item may tap more of a “self-directing” coping strategy. The addition of more items in each
negative coping subdimension may have allowed further exploration of the concept.
A second issue concerns the choice of reference model used in the present CFA analyses.
While a single-factor model has received considerable discussion in the analysis, it serves largely
as a reference model and is of little inherent conceptual interest—indeed, it represents a place
to start. Any number of intermediate models could have been examined for comparison. Hill
and Hood (1999) present over 100 models and measures of religiosity/spirituality and related
dimensions that may have offered useful comparisons. On the other hand, given the focus of this
458 JOURNAL FOR THE SCIENTIFIC STUDY OF RELIGION

study upon examining the Fetzer dimensionality, it would have been equally reasonable to skip
reference models completely and conduct a test of exact fit of the multidimensional Fetzer model.
Comparisons with the single-factor model are provided because exploratory analyses suggested
the existence of a global factor. The present analyses suggest that: (1) something more complex
than a single-factor model is needed to yield adequate fit; and (2) the Fetzer multidimensional
model and its variant offer suitable fit. While other multidimensional models might fit as well, for
purposes of brevity, only the Fetzer has been considered here. Again, there are new theoretically
derived measures in the literature (e.g., Rowatt and Kirkpatrick 2002) that offer potentially useful
conceptual and measurement directions for future research.
Further, I note that while the present CFA analyses generally support the use of the Fetzer
among non-Hispanic whites and African Americans, the Fetzer’s utility for other ethnic groups re-
mains to be seen. Unfortunately, there were too few Hispanics and others in the GSS data to include
in the analyses. All in all, however, the present analyses are consistent with previous examinations
(e.g., Idler et al. 2003); they provide support for the multidimensional nature of the Fetzer measure
and suggest its potential utility—at least among non-Hispanic white and African-American com-
munity populations. Admittedly, further analyses could have examined other variables—such as
religious group representations—in relation to the Fetzer. Such considerations, while potentially
interesting, are beyond the scope of this article.
Finally, the present analyses address the utility of the Fetzer measure, not its ultimate valid-
ity. While the measure goes beyond simple religiosity-spirituality distinctions and differentiates
among some dimensions of spirituality, the present analyses cannot establish that these are the
“right” dimensions. The present analyses have solely sought to take advantage of available data
to examine the utility of the Fetzer—not to argue that the Fetzer necessarily represents a gold
standard in the conceptualization and measurement of religiosity and spirituality. Further research
should address other theoretically derived multidimensional measures in large, diverse samples.

REFERENCES

Armstrong, T. 1996. Exploring spirituality: The development of the Armstrong measure of spirituality. In Handbook of
tests and measurements for black populations, edited by R. L. Jones, Vol. 2, pp. 105–15. Hampton, VA: Cobb and
Henry.
Bentler, P. and D. Bonett 1980. Significance tests and goodness of fit in the analysis of covariance structures. Psychological
Bulletin 88:588–606.
Byrne, B. M. 1989. A primer on LISREL: Basic applications and programming for confirmatory factor analysis. New
York: Springer-Verlag.
Chatters, L. M., J. S. Levin, and R. J. Taylor 1992. Antecedents and dimensions of religious involvement among older
Black adults. Journal of Gerontology 47(6):S269–78.
Ellison, C. 1983. Spiritual well-being: Conceptualization and measurement. Journal of Psychology and Theology 11:330–
40.
———. 1993. Religion, the life stress paradigm, and the study of depression. In Religion in aging and health: Theoretical
foundations and methodological frontiers. edited by J. Levin, pp. 78–121. Thousand Oaks: Sage.
Ellison, C. and D. E. Sherkat 1995. The semi-involuntary institution revisited: Regional variations in church participation
among black Americans. Social Forces 73(4):1415–37.
Fetzer Institute 1999. Multidimensional measurement of religiousness/spirituality for use in health research. A report of
the Fetzer Institute/National Institute on Aging Working Group. Fetzer Institute.
Fletcher, B. 1998. Spirituality, grieving, and mental well-being. In African American mental health. edited by R. Jones,
pp. 135–49. Hampton, VA: Cobb & Henry.
Gartner, J., D. Larson, and G. Allen 1991. Religious commitment and mental health: A review of the empirical literature.
Journal of Psychology and Theology 19:6–25.
Hill, P. and R. Hood, eds. 1999. Measures of religiosity. Birmingham: Religious Education Press.
Idler, E., C. Ellison, L. George, N. Krause, M. Ory, K. Pargament, L. Powell, D. Williams, and L. Underwood 2003.
Measuring multiple dimensions of religion and spirituality for health research: Conceptual background and findings
from the 1998 General Social Survey. Research on Aging 25:327–65.
Jöreskog, K. and D. Sörbom 1993. LISREL 8 user’s reference guide. Chicago: Scientific Software.
———. 2000. Interactive LISREL 8.5. Chicago: Scientific Software International.
FETZER MULTIDIMENSIONAL MEASURE 459

Levin, J., R. J. Taylor, and L. M. Chatters 1995. A multidimensional measure of religious involvement for African
Americans. Sociological Quarter 36(1):157–73.
Lincoln, C. E. and L. H. Mamiya 1990. The black church in the African American experience. Durham, NC: Duke
University Press.
Mattis, J. S. 2000. African American women definitions of spirituality and religiosity. Journal of Black Psychology
26(1):101–22.
Pargament, K. 1997. The psychology of religion and coping: Theory, research, practice. New York: Guilford Publications.
Rowatt, W. and L. Kirkpatrick, 2002. Two dimensions of attachment to God and their relation to affect, religiosity, and
personality constructs. Journal for the Scientific Study of Religion 41:637–51.
Underwood, L. and J. Teresi 2002. The daily spiritual experience scale: Development, theoretical description, reliability,
exploratory factor analysis, and preliminary construct validity using health-related data. Annals of Behavioral
Medicine 24:22–33.
Williams, D. R. 1993. The measurement of religion in epidemiologic studies. In Religion in aging and health: Theoretical
foundations of methodological frontiers. edited by Lewis. Thousand Oaks: Sage.
Witter, R., W. Stock, M. Okun, and M. Haring. 1985. Religion and subjective well-being in adulthood: A quantitative
synthesis. Review of Religious Research 26:332–42.