Religion, Self-Regulation, and Self-Control: Associations ...

Psychological Bulletin
© 2009 American Psychological Association
2009, Vol. 135, No. 1, 69 –93
0033-2909/09/$12.00
DOI: 10.1037/a0014213
Religion, Self-Regulation, and Self-Control: Associations, Explanations,
and Implications
Michael E. McCullough and Brian L. B. Willoughby
University of Miami
Many of the links of religiousness with health, well-being, and social behavior may be due to religion’s
influences on self-control or self-regulation. Using Carver and Scheier’s (1998) theory of self-regulation
as a framework for organizing the empirical research, the authors review evidence relevant to 6
propositions: (a) that religion can promote self-control; (b) that religion influences how goals are
selected, pursued, and organized; (c) that religion facilitates self-monitoring; (d) that religion fosters the
development of self-regulatory strength; (e) that religion prescribes and fosters proficiency in a suite of
self-regulatory behaviors; and (f) that some of religion’s influences on health, well-being, and social
behavior may result from religion’s influences on self-control and self-regulation. The authors conclude
with suggestions for future research.
Keywords: religion, self-control, self-regulation, motivation, personality
Religion is a potent social force. History testifies to religion’s
(2003) likewise concluded that frequent religious service atten-
ability to focus and coordinate human effort, to create awe and
dance was associated with a 25% reduction in mortality.
terror, to foster war and peace, to unify social groups, and to
Studies published after McCullough et al.’s (2000) and Powell
galvanize them against each other. In addition to religion’s social
et al.’s (2003) reviews have yielded similar conclusions in random
power, however, religion is a psychological force that can influ-
samples of U.S. adults (Musick, House, & Williams, 2004); older
ence the outcomes of individual human lives. Indeed, the range of
Mexican Americans (T. D. Hill, Angel, Ellison, & Angel, 2005); and
health-related, behavioral, and social outcomes with which reli-
adults in Denmark (la Cour, Avlund, & Schultz-Larsen, 2006), Fin-
giousness is associated is both provocative and puzzling.
land (Teinonen, Vahlberg, Isoaho, & Kivela, 2005), and Taiwan
Consider these well-replicated findings. First, religiousness
(Yeager et al., 2006). These latter studies have also revealed 25% to
(measured variously as high levels of traditional religious belief;
30% reductions in mortality for religiously active people, even
frequent involvement in religious institutions such as churches,
after controlling for reasonable sets of potential confounds and
synagogues, mosques, and temples; and engagement in religious
explanatory variables.
practices such as reading scripture, worship, and prayer) has a
Relatedly, youths and adults (including Christians, Jews, and
small, positive association with longevity. In a meta-analytic re-
Muslims) who score higher on measures of religiousness are less
view of 42 independent effect sizes, McCullough, Hoyt, Larson,
likely to drink and smoke and are more likely to wear their
Koenig, and Thoresen (2000) discovered that people who were
seatbelts, see their dentists, and take their vitamins than are their
highly religious were, on average, 29% more likely to be alive at
less religious counterparts (T. D. Hill, Burdette, Ellison, &
any given follow-up point than were less religious people (95%
Musick, 2006; Islam & Johnson, 2003; Shmueli & Tamir, 2007;
confidence interval: 1.20 to 1.39). In a traditional narrative review
Wallace & Forman, 1998). These behavioral correlates of reli-
of studies considered by the authors to be of high quality (e.g.,
giousness may help to explain why religious people tend to live
studies with large, representative samples and adequate statistical
slightly longer lives, because behaviors like these are among the
control of potential confounds), Powell, Shahabi, and Thoresen
major behavioral risk and protective factors vis a` vis early mor-
tality (Bogg & Roberts, 2004).
Religiousness is also significantly (albeit weakly) associated
with psychological well-being. In a meta-analytic review of 147
Michael E. McCullough and Brian L. B. Willoughby, Department of
independent effect sizes that was published in this journal, Smith,
Psychology, University of Miami.
This work was supported by a grant from the John Templeton Founda-
McCullough, and Poll (2003) found that religiousness (measured
tion and support from the Center for the Study of Law and Religion at
in a wide variety of ways) was associated with lower rates of
Emory University to Michael E. McCullough and by a doctoral fellowship
depressive symptoms (mean effect size of r
.09, p
.000001).
from the Social Sciences and Humanities Research Council of Canada to
Moreover, this overall mean effect size obscures the fact that some
Brian L. B. Willoughby. We thank Roy Baumeister, Jessie Bering, Charles
measures of religiousness (e.g., positive religious coping, intrinsic
Carver, Terrence Hill, Patrick McNamara, Mark Regnerus, Brent Roberts,
religious motivation, and positive God concepts) had even stronger
Rich Sosis, and Kathleen Vohs for helpful comments on drafts of this
negative associations with depressive symptoms (i.e., mean effect
article.
sizes in the range of r
.20 to
.18), whereas measures of
Correspondence concerning this article should be addressed to Michael
E. McCullough, Department of Psychology, University of Miami, P.O.
negative religious coping and extrinsic religious motivation were
Box 248185, Coral Gables, FL 33124-0751. E-mail: mikem@miami.edu
positively related to depressive symptoms (i.e., mean effect sizes
69

---

70
MCCULLOUGH AND WILLOUGHBY
in the range of r
.14 to .16). It is important to note that the
relevant to health, well-being, achievement, and social flourishing.
religiousness– depression association was as strong among African
However, these associations present an interesting puzzle, because
Americans as it was among Americans of European descent and
an overarching explanatory mechanism that might explain them
Northern Europeans, suggesting that the association is not limited
has not been identified. Historically, theorists have focused on
to the White Protestants who are overrepresented in U.S. studies
explanations such as (a) religion’s ability to prescribe health-
(Smith et al., 2003).
promoting behaviors and proscribe health-compromising ones
Similarly, a 2-decades-old meta-analysis (Witter, Stock, Okun,
(T. D. Hill, Burdette, Ellison, & Musick, 2006; Strawbridge,
& Haring, 1985) concluded that religiousness was positively as-
Shema, Cohen, & Kaplan, 2001); (b) religion’s ability to confer
sociated with subjective well-being (mean effect size of r
.16).
social support (Joiner, Perez, & Walker, 2002); (c) religion’s
A more recent meta-analytic review of data from 49 studies (Ano
ability to socialize children to conform with society’s norms (Baier
& Vasconcelles, 2005) also concluded that “positive” forms of
& Wright, 2001); and (d) religion’s ability to promote effective
religious coping (e.g., benevolent religious reappraisals of stres-
coping with stress (Ano & Vasconcelles, 2005; Pargament, 1997).
sors, collaborative religious coping, and active religious surrender)
To be sure, all of these explanations are important, but the list is
were positively associated with scores on measures of positive
incomplete. George, Larson, Koenig, and McCullough (2000) sur-
psychological outcomes, such as satisfaction with life and happi-
mised that only 35% to 50% of the relationship between religious-
ness (mean r
.33), and negatively associated with measures of
ness and various measures of health and well-being could be
negative outcomes, such as anxiety and depression (mean r
explained on the basis of the explanatory variables such as social
.12). “Negative” forms of religious coping (e.g., appraising one’s
support, health behaviors, and coping. If George et al.’s 35% to
difficulties as due to demonic influences or God’s punishment), on
50% estimate comes close, then social scientists have only gone
the other hand, were not associated with positive outcomes (mean
one third to one half of the way in explaining scientifically how
r
.02), but they were positively associated with negative out-
religion affects health, well-being, and social behavior. Clearly,
comes (mean r
.22).
there is room for conceptual innovation in this research domain.
Consider also the negative association of religiousness with
crime, delinquency, and youth sexual behaviors, all of which are
Goals of the Present Article
risk factors for poor outcomes in adulthood. A meta-analysis of 60
effect sizes revealed that religiousness is associated with lower
In this article, we provide a systematic and comprehensive
rates of crime and delinquency (mean effect size of r
.12) and
review of empirical evidence surrounding an underappreciated
is even more strongly associated with lower rates of “victimless”
psychological process that may help explain why religious
crimes such as gambling and drug use (Baier & Wright, 2001).
people tend to live slightly longer lives; suffer less from de-
Moreover, studies of representative samples of U.S. youths show
pressive symptoms; avoid trouble with sex, drugs, and the
that White, African American, Hispanic, and Asian American
police; do better in school; enjoy more stable and more satis-
adolescents who are religious (measured as frequency of church
fying marriages; and more regularly visit their dentists. Specif-
attendance and self-rated importance of religion) wait longer be-
ically, we review evidence that is relevant to the ideas that (a)
fore first intercourse (Regnerus, 2007; Rotosky, Regnerus, &
some types of religious belief, behavior, and cognition foster
Wright, 2003), have lower rates of having ever given or received
self-regulation and, more specifically, self-control and (b) that
oral sex (Regnerus, 2007), and have lower rates of ever having
it is partly through its associations with self-regulation and
been pregnant (Nonnemaker, McNeely, & Blum, 2003). Recent
self-control that religion obtains its associations with health,
systematic and meta-analytic reviews (Lucero, Kusner, Speace, &
well-being, and social behavior. The idea that religion is effec-
O’Brien, 2008; Rotosky, Wilcox, Comer Wright, & Randall, 2004)
tive at reining in socially non-normative behavior and promot-
concur with these conclusions.
ing socially normative behavior has a long history in the sci-
Religious youths also tend to have higher grade point averages
entific study of religion (Durkheim, 1965/1912; Malinowski,
and standardized test scores than do their less religious counter-
1935). In contrast, the notion that religion fosters self-
parts (Regnerus, 2000; Regnerus & Elder, 2003). Indeed, a meta-
regulation and control of the self has attracted little theoretical
analysis of 15 studies on the association of religiousness and
attention (for interesting exceptions, see Ainslie, 2004; Geyer &
school achievement in Black and Hispanic American youths (Jey-
Baumeister, 2005). Nevertheless, as our review indicates, many
nes, 2002) found that religiousness was positively associated with
propositions arising from these basic ideas appear to enjoy
grade point average (mean r
.21) and achievement test scores
empirical support (though that support is preliminary in some
(mean r
.15).
instances).
The association of religion with measures of social adjustment
Self-regulation and self-control are crucial for success in many
also extends to the marital realm. A meta-analytic review of
life domains. College students with high self-control have better
dozens of studies showed that married religious adults are more
psychological adjustment, better interpersonal relationships, and
likely to stay married over time (mean correlation of religious
better performance on achievement-related tasks (Tangney,
service attendance with divorce, r
.13) and have higher levels
Baumeister, & Boone, 2004), and self-control appears to be a
of marital satisfaction (mean effect size of r
.15) and marital
better predictor of academic performance than even intelligence
commitment (mean effect size r
.19; Mahoney, Pargament,
(Duckworth & Seligman, 2006). People with high self-control also
Tarakeshwar, & Swank, 2001).
have lower alcohol and substance use, lower rates of crime and
As this brief survey of well-replicated findings shows, many
delinquency, better self-assessed health, and better health behav-
measures of religiousness are associated consistently (albeit, in
iors (for a review, see Baumeister & Vohs, 2004). Moreover,
most cases, weakly) with a wide variety of outcomes that are
Conscientiousness and its varied facets (including, notably, self-

---

RELIGION, SELF-REGULATION, AND SELF-CONTROL
71
control) are negatively associated with many health-risk behaviors,
Table 1
including physical inactivity, attempted suicide, risky sex, un-
Key Propositions Regarding Religion, Self-Regulation,
healthy eating, substance use, risky driving, and violence (Bogg &
and Self-Control
Roberts, 2004).
Research has also shown that young children who do well at
Proposition
delaying gratification (i.e., forgoing a small reward in the present
1. Religiousness can promote self-control
so that they might obtain a larger reward after time has passed)
2. Religion influences self-regulation by influencing people’s goals
perform better years later on measures of academic achievement
a. Religion influences goal selection
and social adjustment (Mischel, Shoda, & Rodriguez, 1989). Some
b. Religion increases the importance of some goals by sanctifying
them
social scientists consider delay of gratification to be an important
c. Religion reduces conflict among goals
dynamic underlying the behavioral choices of people who believe
d. Religion influences how goals are internalized
in an afterlife in which their behavior during this life will be
3. Religion influences self-regulation by promoting self-monitoring
judged. For people with strong beliefs in such an afterlife, it would
a. Perceived interaction with (and monitoring by) supernatural
indeed be rational to deny short-term gains that might come from
entities fosters self-monitoring
b. Religious communities, as moralistic audiences, foster self-
engaging in behavior that is proscribed by one’s religion because
monitoring
the long-term (eternal) gains of not engaging in the behavior might
c. Many religious rituals deliberately activate self-monitoring
outweigh the short-term gains associated with engaging in the
4. Religion influences self-regulation by building self-regulatory
behavior (Azzi & Ehrenberg, 1975; Iannaccone, 1998). In keeping
strength
a. Involvement in religious communities fosters the development of
with this idea, Turkish (Muslim) undergraduates who are highly
self-regulatory strength
religious tend to report considering the future in their present
b. Many religious rituals foster the development of self-regulatory
decision making to a greater extent than do less religious students
strength
(Oner-Ozkan, 2007). If religion is robustly related to self-control
5. Religions influence self-regulation by prescribing and promoting
and self-regulation over the life course, then these connections
mastery with specifically religious outputs for self-change
6. Religion affects health, well-being, and social behavior through self-
might go far in explaining the associations of religiousness with
regulation and self-control
many important life outcomes.
Method and Organization of the Review
article concludes with a series of recommendations for future
research.
In this article, we use Carver and Scheier’s (1998) model of
self-regulation to organize a review of research on the associations
Definitions of Religion, Self-Regulation, and Self-Control
of religion with self-regulation and self-control. The Carver–
Scheier framework enabled us to integrate several literatures (e.g.,
Before proceeding, it is useful to define the terms religion,
the literatures on the personality correlates of religion, on religion
self-regulation, and self-control. Following James (1958) and J. B.
and goals, on religion and self-monitoring, and on the affective,
Pratt (1934), we define religion as cognition, affect, and behavior
cognitive, and behavioral effects of various religious rituals) that
that arise from awareness of, or perceived interaction with, super-
scholars have not considered as relevant to a unitary subject.
natural entities that are presumed to play an important role in
We organize our review of the literature around six major
human affairs. The psychological components of religion can be
propositions, and some subsidiary propositions, that emerge from
measured in a variety of ways (P. C. Hill & Hood, 1999), including
the basic idea that religion is related to high self-control and better
(a) beliefs about the existence of gods or spirits and their involve-
self-regulation. These propositions appear in Table 1. In address-
ment in human life, (b) level or quality of engagement in activities
ing some of the propositions, we found ourselves unable to locate
that are traditionally motivated by the awareness of supernatural
adequate published research, but we had the good fortune to locate
forces (e.g., frequency of prayer or religious service attendance),
data sets that enabled us to evaluate those propositions directly.
and (c) strength of commitment to a particular religious belief
We report the analyses of those data as appropriate.
system.
The studies we review herein resulted from an exhaustive search
From a psychometric point of view, these aspects of religion are
of PsycINFO through July 2008 and the additional published and
often so strongly intercorrelated as to be interchangeable
unpublished works that we subsequently located by examining the
(D’Onofrio et al., 1999; Gorsuch, 1984; McCullough, Enders,
reference sections of those articles. We also contacted authors who
Brion, & Jain, 2005), especially in samples that are homogeneous
were prominent researchers in this area and requested access to
with respect to religious tradition (e.g., Christian, Jewish, Bud-
relevant work through e-mail discussion groups for personality,
dhist). Nevertheless, it is widely acknowledged that two people
social, and developmental psychologists. We explicitly tried to
can adhere to the same religious belief system for very different
locate research studies whose results contradicted the propositions
reasons (Allport, 1950; Ryan, Rigby, & King, 1993) and that these
that form the backbone of this review. In other words, it was our
different ways of being religious can lead to different motivational
intent to incorporate all of the empirical evidence that was relevant
and behavioral outcomes. Moreover, some aspects of religious
to the propositions in a thorough, unbiased, and systematic way. In
belief, behavior, and experience might foster self-control and self-
addition, we worked assiduously to locate studies of non-White,
regulation, whereas others might hinder them. These distinctions
non-Christian, and non–North Americans samples so that we could
become important later in this article.
evaluate (informally, at least) the extent to which our propositions
We define self-regulation as the process by which a system uses
held up across a variety of ethnicities, religions, and cultures. The
information about its present state to change that state (McCullough &

---

72
MCCULLOUGH AND WILLOUGHBY
Boker, 2007). Baumeister and Vohs (2004) defined it as “how a
Personality Research on Religiousness and Self-Control
person exerts control over his or her own responses so as to pursue
goals and live up to standards” (p. 500). Barkley (1997) defined
In the first known study of the link between religion and
self-regulation as “any response, or chain of responses, by the
self-control, Hartshorne, May, and Maller (1929) found small (and
individual that serves to alter the probability of the individual’s
not uniformly statistically significant) positive associations be-
subsequent response to an event and, in doing so, functions to alter
tween children’s length of Sunday school attendance (controlling
the probability of a later consequence related to that event” (p. 68).
for children’s ages) and a performance-based measure of persis-
What these definitions have in common is that when people
tence (but not a performance-based measure of inhibition). Since
self-regulate, they are guiding or adjusting their behavior in pursuit
then, many other studies have examined the association of reli-
of some desired end state or goal (see Carver & Scheier, 1998).
giousness and self-control, albeit with rating-based measures of
Self-regulation need not be a deliberative, effortful process: Much
self-control rather than with performance-based ones.
of it occurs in a relatively effortless and automatic fashion
In addition to Hartshorne et al. (1929), we found 12 studies that
(Fitzsimmons & Bargh, 2004). This point is relevant to the present
examined the association of measures of religiousness with mea-
article because religious constructs can be activated outside of
sures of general self-control (Aziz & Rehman, 1996; Bergin,
awareness (Shariff & Norenzayan, 2007; Wenger, 2003, 2004) and
Masters, & Richards, 1987; Bouchard, McGue, Lykken, & Telle-
thereby influence cognition and behavior—perhaps through the
gen, 1999; Desmond, Ulmer, & Bader, 2008; French, Eisenberg,
self-regulatory process described presently.
Vaughan, Purwono, & Suryanti, 2008; Longshore, Chang, Hsieh,
We reserve the term self-control for situations in which people
& Messina, 2004; McClain, 1978; Pfefferbaum & Wood, 1994;
engage in behaviors designed to counteract or override a prepotent
Richards, 1985; C. Walker, Ainette, Wills, & Mendoza, 2007;
response (e.g., a behavioral tendency, an emotion, or a motiva-
Welch, Tittle, & Grasmick, 2006; Wills, Gibbons, Gerrard, Murry,
tion), such as assaulting someone who has angered them, resting
& Brody, 2003). Most of these studies measured self-control with
after a hard day at work instead of painting the kitchen, or playing
previously published self-report measures (e.g., Gough, 1975;
hooky instead of going to school. In other words, when people
Rosenbaum, 1980) or self-report measures that authors created for
exert self-control, they modify their response tendencies in a
these specific studies, although in one study (French et al., 2008)
fashion that involves suppressing one goal so as to pursue another
self-control ratings were obtained from teachers who rated their
one that is judged to have greater long-term utility. Self-control is
students using a previously published instrument (Capaldi & Roth-
thus a more specific concept than is self-regulation. Not all psy-
bart, 1992). Of these 12 studies, 11 found positive associations
chological states that are self-regulated involve self-control in the
between religiousness and self-control, with associations (either
sense for which we reserve the term here; however, self-control
correlation coefficients or standardized regression coefficients)
may rely on mechanisms that are also involved in self-regulation
ranging from .21 to .38.
per se.
For instance, Bergin et al. (1987) found that intrinsic religious-
Self-control is not simply a process: It can also be conceptual-
ness was positively related to Rosenbaum’s (1980) Self-Control
ized as a property of systems that possess effective self-control
Schedule (r
.38) and the Self-Control scale of the California
capabilities. In other words, most human beings have self-control
Psychological Inventory (r
.32) in a sample of Mormon college
in the sense that they can pursue one goal that conflicts with
students. More recently, using data from the Add Health study (a
another, prepotent one, but inasmuch as people differ in the effi-
nationally representative study of students from 132 U.S. middle
ciency with which the mechanisms governing self-control operate,
schools and high schools), Desmond et al. (2008) found that
they also differ in self-control. When we use the term self-control
self-reported religiousness (measured in terms of self-rated impor-
in this latter sense, we refer to “the internal resources available to
tance of religion, frequency of prayer, and frequency of church
inhibit, override, or alter responses that may arise as a result of
attendance) was positively and significantly associated with a
physiological processes, habit, learning, or the press of the situa-
multi-item measure of self-control, even after controlling for sex,
tion” (Schmeichel & Baumeister, 2004, p. 86).
age, race, socioeconomic status, parental education, family struc-
ture, attachment to parents, and religious denomination.
Also, French et al. (2008) found that a latent variable measuring
Proposition 1: Religiousness Can Promote Self-Control
religiousness based on self-reports and parent reports was posi-
tively associated (standardized coefficient
.36) with a latent
The first proposition (see Table 1) is that some forms of reli-
variable representing self-control (as based on teacher reports of
gious belief, behavior, and institutional involvement can promote
students’ inhibitory control and attentional control) in a sample of
self-control. Evidence relevant to this proposition comes from four
Muslim 8th and 9th graders in Indonesia. Likewise, Aziz and
quarters. First, personality research shows that people with higher
Rehman (1996) found that religiousness (measured with a 27-item
scores on measures of self-control and personality dimensions that
self-report measure of Muslim religiousness) among postgraduate
subsume self-control also tend to be more religious. Second,
Pakistani Muslims was associated with higher self-reported self-
family research shows that religious parents and families have
control (r
.35). It is worth pointing out that the relationships of
children with high self-control and low impulsiveness. Third,
religiousness and self-control generally appeared to be as strong in
several longitudinal studies shed light on the causal relations
samples of adolescents, university students, community-dwelling
between religiousness and personality variables that subsume self-
adults, and convicted drug offenders in North America as they
control. Fourth, a single published experiment suggests that reli-
were, for example, among Muslim adolescents and postgraduates
gious cognition is automatically activated as a form of self-control
from Indonesia and Pakistan (Aziz & Rehman, 1996; French et al.,
in the face of temptation.
2008). Contradicting this trend for positive associations between

---

RELIGION, SELF-REGULATION, AND SELF-CONTROL
73
religiousness and self-control was a study yielding a near-zero
variety of potential confounds, people who were highly agreeable
correlation (r
.04) between a single-item measure of the per-
in early adulthood went on to have high levels of religiousness
sonal importance of religion and the Self-Control scale of the
(based on expert ratings) through adulthood. Conversely, people
California Psychological Inventory in a sample of 296 U.S. un-
with low Agreeableness in early adulthood tended to have low
dergraduates (Pfefferbaum & Wood, 1994).
levels of religiousness through adulthood. Moreover, survey re-
Two other exceptional findings merit mention. Bergin et al.
searchers rate religious respondents as more “cooperative,”
(1987) found that extrinsic religiousness, as opposed to intrinsic
“open,” and “enjoyable to interview” and less “suspicious” than
religiousness, was negatively associated (though not significantly
their less religious counterparts (Brennan & London, 2001; Elli-
so) with two measures of self-control (rs
.19 and
.13).
son, 1992; Morgan, 1983), all of which are indicative of Agree-
Bouchard et al. (1999) also found extrinsic religiousness to be
ableness.
virtually uncorrelated with the California Psychological Inventory
High Agreeableness and high Conscientiousness also seem to
measure of self-control (r
.01). Thus, whereas general religiosity
form the core of religious people’s (at least Christians’) personality
and intrinsic religious motivation seem to be associated cross-
ideals. For instance, people rate their prototypes of highly religious
sectionally with higher self-control, these latter findings suggest
individuals as being particularly high in Agreeableness and Con-
that extrinsic motivation for being religious is either negatively
scientiousness (L. J. Walker, 1999). In addition, in a study of
correlated or uncorrelated with self-control.
Christian university students and a study of (mostly Christian)
adults from Illinois (Nielsen, 2000; Nielsen & Stevens, 2001),
Research on Higher Level Personality Traits That
participants completed measures of the Big Five to indicate how
Subsume Aspects of Self-Control
they felt about themselves during “important religious experi-
ences.” In both studies, participants indicated that during their
Higher level personality traits that subsume aspects of self-
important religious experiences, they perceived themselves as high
control also tend to be positively correlated with religiousness, as
in Agreeableness and high in Conscientiousness (as well as low in
well as with people’s religious ideals and their self-perceptions
Neuroticism). In other words, these studies suggest that religious
during important discrete religious experiences.
people are not only more conscientious and agreeable than are
Religiousness and the Big Five personality taxonomy.
Saro-
their less religious counterparts, but also that Agreeableness and
glou (2002) and Lodi-Smith and Roberts (2007) conducted inde-
Conscientiousness are held up as exemplary religious traits and are
pendent meta-analytic reviews of research on the Big Five corre-
experienced as temporary personality changes associated with
lates of religious involvement, incorporating data from 13 and 38
memorable religious experiences.
independent studies, respectively (mostly, though by no means
Religiousness and Eysenck’s personality taxonomy.
Research-
exclusively, conducted with students from the United States and
ers have also evaluated the links between measures of religious-
Canada). Both meta-analytic efforts led to the conclusion that
ness and the personality variables in Eysenck’s (1991) P-E-N
among the Big Five dimensions of personality (i.e., Openness,
(Psychoticism, Extraversion, Neuroticism) model of personality.
Conscientiousness, Extraversion, Agreeableness, and Neuroti-
In the Eysenck model, it is Psychoticism (which Costa & McCrae,
cism), it is Agreeableness and Conscientiousness that are the most
1995, argued is largely a conflation of Conscientiousness and
reliable correlates of religiousness (mean rs
.20 and .21 for
Agreeableness) that is most strongly (and negatively) related to
Agreeableness and mean rs
.17 and .12 for Conscientiousness in
religiousness (e.g., Francis, 1997; Hills, Francis, Argyle, & Jack-
the Saroglou, 2002, and Lodi-Smith and Roberts, 2007, meta-
analyses, respectively). Agreeableness and Conscientiousness sub-
son, 2004). Indeed, Lodi-Smith and Roberts’s (2007) meta-
sume aspects of self-control because they involve the ability to
analysis of 19 correlations revealed that Psychoticism had a mean
adapt one’s behavior to the wishes and feelings of others and to
correlation of r
.20 with various measures of religiousness.
task demands, respectively. Indeed, self-control may be the tem-
The association of religiousness with low Psychoticism has ob-
peramental basis for the development of Agreeableness and Con-
tained not only in samples of Christian youths and adults but also
scientiousness over the life course (Cumberland-Li, Eisenberg, &
in Muslim and Israeli samples (Francis & Katz, 1992; Wilde &
Reiser, 2004; Jensen-Campbell et al., 2002).
Joseph, 1997).
Although many studies relating religiousness to measures of
Religiousness and Cattell’s personality taxonomy.
We also
personality traits such as Agreeableness and Conscientiousness
located four studies that evaluated the links of religiousness with
have relied on self-reports (e.g., Ashton, Kibeom, & Goldberg,
the measures in Cattell’s personality taxonomy. In a study involv-
2004), even studies in which personality was measured with in-
ing U.S. college students and a replication with high school stu-
formant reports (e.g., parents and teachers) or expert ratings have
dents from New Zealand (Barton & Vaughan, 1976; McClain,
shown that Conscientiousness and Agreeableness are positively
1970), active churchgoers (versus nonchurchgoers) had higher
associated with religiousness (McCullough, Tsang, & Brion, 2003;
scores on scale “G” (variously called “Superego,” “Conformity”
Saroglou & Fiasse, 2003; Wink, Ciciolla, Dillon, & Tracy, 2007).
and “Expedient vs. Conscientious”) of Cattell’s 16PF question-
McCullough et al. (2003) found that parent-rated and teacher-rated
naire, which also measures self-control (Hofer, Horn, & Eber,
Conscientiousness predicted religiousness over a 19-year longitu-
1997). Francis and Bourke (2003) also found that among English
dinal follow-up from childhood to early adulthood in an archival
secondary school pupils, Scale G was the strongest 16PF correlate
sample of gifted children from California who were born early in
of attitudes toward Christianity (r
.19). Finally, in a sample of
the 20th century (The Terman Life Cycle Study of Children With
British students, the “G” scale was correlated r
.60 with a
High Ability). In another longitudinal study using the Terman data,
measure of traditional Christian religious belief and behavior
McCullough et al. (2005) found that even when controlling for a
(Rasmussen & Charman, 1995).

---

74
MCCULLOUGH AND WILLOUGHBY
In summary, then, across the Five Factor, Eysenck, and Cattell
second religion scale was created by summing six items addressing
personality taxonomies, personality dimensions associated with
importance of religion, religious involvement of one’s friends,
the ability to regulate one’s behavior in a manner consistent with
reading of sacred scriptures and religious literature, frequency of
one’s goals or out of concern for the wishes and feelings of others
prayer, relationship with God, and experiences with God (
(e.g., high Agreeableness, high Conscientiousness, and low Psy-
.76). The RCI-10 and the six-item religion scale were highly
choticism) are associated with religiousness. Although most of the
correlated (r
.76), so we standardized them and used the mean
research on this topic has been conducted on samples from North
of the two standardized scores as a “religion composite.”
America and Western Europe (which implies that most people in
Participants also completed the Self-Transcendence scale of the
those studies were affiliated with some form of Christianity), the
Temperament and Character Inventory (TCI; Cloninger, Svrakic,
basic associations have been replicated in Israeli and Muslim
& Przybeck, 1993), which measures spirituality as distinct from
samples as well (Francis & Katz, 1992; Wilde & Joseph, 1997) and
conventional religion. The Self-Transcendence scale includes
with a variety of measures of religiousness and traits relevant to
items such as “Sometimes I have felt my life was being directed by
self-control.
a spiritual force greater than any human being” and “I sometimes
Religiousness versus spirituality and self-control.
It is impor-
feel a spiritual connection to other people that I cannot explain in
tant to know whether the associations of religiousness with self-
words.” The Self-Transcendence scale’s correlation with the com-
control, and related constructs such as Agreeableness and Consci-
posite measure of religion was r (N
257)
.46, p
.05.
entiousness, extend to measures of spirituality as distinct from
The correlation coefficients in the first two columns of Table 2
more conventional manifestations of religiosity: Although reli-
show significant positive relationships of the religion composite
giousness and spirituality have some overlap, they also appear to
with Conscientiousness, Agreeableness, and Self-Control. Self-
have different personality correlates and different social conse-
transcendence showed significant positive relationships with
quences (Saucier & Skrzypinska, 2006). Therefore, we analyzed
Openness and Agreeableness, but it was not related to self-control
data from 257 undergraduate psychology students enrolled in a
or Conscientiousness. The partial correlation coefficients in the
medium-sized southeastern university to address this question.
last two columns of Table 2 show that when self-transcendence is
Participants’ ages ranged from 17 to 46 years (M
19.04, SD
partialed out of the religion composite, the correlations of the
2.98), and they were diverse in gender (60% female, 40% male)
religion composite with Conscientiousness and self-control stay
and ethnicity (63% White non-Hispanic, 15% Hispanic, 12%
essentially the same (and, in fact, increase slightly) relative to
African American, 10% other ethnicities). Although we do not
those correlations prior to partialing. In addition, the correlation of
have the breakdown of religious affiliations for the participants in
the religion composite with Openness and Extraversion become
this particular sample, the students were drawn from a religiously
significant and negative. Conversely, when the religion composite
diverse student body (46% Catholic, 25% Protestant, 17% Jewish,
is partialed out of Self-Transcendence, the correlations of Self-
2% Muslim, 2% Hindu, and 8% “other”).
Transcendence with Conscientiousness and self-control become
As part of a larger study, participants completed a packet of
significant and negative, the association of Self-Transcendence
questionnaires on a single occasion. The Big Five Inventory (John,
with Agreeableness becomes nonsignificant, and the association of
Donahue, & Kentle, 1991) measured the Big Five personality
Self-Transcendence with Openness is unaffected. These results
factors. The Self-Control Scale (Tangney et al., 2004) measured
suggest that the personality core of religiousness, as distinct from
trait self-control. The Religious Commitment Inventory–10 (RCI-
spirituality, is high Conscientiousness, high Agreeableness, and
10; Worthington et al., 2003) measured participants’ strength of
high self-control (and, to a lesser extent, low Openness and Ex-
commitment to their religious values (
.95). In addition, a
traversion). In contrast, the personality core of spirituality, as
Table 2
Zero-Order and Partial Correlations of the Big Five Personality Factors and Self-Control With
a Religion Composite and Self-Transcendence
Zero-order correlations
Partial correlations
Religion composite,
Self-transcendence,
Religion
Self-
controlling for self-
controlling for the
Personality factor
composite
transcendence
transcendence
religion composite
Openness
.02
.36
.15
.38
Conscientiousness
.17
.02
.22
.14
Extraversion
.10
.06
.15
.11
Agreeableness
.22
.18
.15
.08
Neuroticism
.07
.11
.03
.10
Self-Control
.20
.04
.26
.16
Note.
The religion composite was computed by averaging standardized scores on the Religious Commitment
Inventory (RCI-10) and the standardized means of six single items addressing importance of religion, religious
involvement of friends, reading of sacred scriptures and religious literature, frequency of prayer, relationship
with God, and experiences with God.
p
.05.
p
.01.
p
.001.

---

RELIGION, SELF-REGULATION, AND SELF-CONTROL
75
distinct from religiousness, is high Openness, low Conscientious-
Longitudinal and Experimental Studies Addressing the
ness, and low self-control.
Causal Status of the Association Between Religiousness
These results lend credence to the idea that something about
and Self-Control
religious beliefs, behaviors, institutions, and rituals themselves
(irrespective of the feelings of spiritual connectedness that religion
It is important to consider not only the possibility that religious-
often fosters) may be responsible for the links between religion
ness precedes changes in self-control (or traits that are relevant to
and self-control. However, whether that “something” is (a) the fact
self-control) but also the possibility that self-control (or traits
that religion prescribes sets of rules that are legitimated on the
related to self-control) precedes changes in religiousness. That is,
basis of the preferences of an omniscient deity; (b) the conviction
if religiousness is predicated partly on the ability to exercise
self-control, then people with low self-control might have diffi-
that one’s behavior is being monitored by that omniscient deity,
culty becoming fully involved in (and committed to) religious
who can administer rewards and punishments; (c) the self-
beliefs, practices, and institutions. In light of this possibility,
discipline that is needed to maintain regular involvement in private
Hathaway, Douglas, and Grabowski (2003) conducted a study of
and public religious rituals; or (d) something else remains an open
249 children (ages 5–12 years), and their parents, whom they
question.
recruited from religious congregations in Eastern Virginia. The
researchers found that parents’ ratings of whether their children’s
Religion and Self-Control in Research on Families and
behavior was problematic in religious situations such as “preparing
to go to a religious service,” “quiet times during the service,” and
Child Development
“family devotions” distinguished between children who had been
We found four empirical articles addressing the associations
diagnosed with attention deficit/hyperactivity disorder (ADHD)
and those who had not. ADHD is commonly conceptualized as a
between measures of parents’ religiousness and their children’s
disorder of self-control or self-regulation (Barkley, 1997). We
self-control (Bartkowski, Xu, & Levin, 2008; Brody & Flor, 1998;
therefore suspect that people with low self-control (including those
Brody, Stoneman, & Flor, 1996; Lindner-Gunnoe, Hetherington,
with disorders of self-control such as ADHD) experience self-
& Reiss, 1999). Two of these studies involved African American
control difficulties in religious settings that reduce their religious
children and youths from the rural southeastern United States
interest and commitment. Longitudinal and experimental research
(Brody & Flor, 1998; Brody et al., 1996), one study predominantly
on the religion/self-control relationship, therefore, is especially
involved youth and adolescents from White, middle-class families
valuable for testing causal relations among these variables.
around the United States (Lindner-Gunnoe et al., 1999), and a
Longitudinal research suggesting that religiousness precedes
fourth study consisted of a nationally (U.S.) representative sample
changes in self-control–relevant traits.
We located only one
of first graders (Bartkowski, Xu, & Levin, 2008). In all four
study that investigated whether individual differences in religious-
studies, investigators discovered positive (though not always sta-
ness precede changes in self-control–relevant traits. Using an
tistically significant) relationships between parents’ religiousness
archival sample of (mostly White) California youths from upper
and their children’s self-control.
middle class families, Wink et al. (2007) discovered that religious-
In the most definitive study to date, Bartkowski et al. (2008)
ness in adolescence was linked to increases over the life course in
examined the links of parents’ religiousness and their children’s
Agreeableness for women (
.33, p
.001), although not for
self-control among 17,000 children (57% White, 14% Black, 17%
men (beta
.05, ns). On the other hand, the authors did not find
Hispanic, 6% Asian, and 5% other) from the Early Childhood
that religiousness in adolescence was associated with increases
Longitudinal Study, a representative survey of American 1st grad-
over the life course in Conscientiousness for either women or men.
ers (95% of the students in the sample were 1st graders). The
It is worth noting that the lag between the baseline measurement of
researchers found that parents who frequently attended church and
religiousness and the follow-up measures of personality in this
who frequently discussed religion in the home rated their children
study was more than 50 years.
as having higher self-control and lower impulsiveness. The chil-
Longitudinal research suggesting that self-control–relevant
traits precede changes in religiousness.
We located five longi-
dren of religious parents were also rated by their teachers as higher
tudinal studies examining whether individual differences in self-
in self-control and lower in impulsiveness than were children
control–related traits preceded changes in religiousness. First, as
whose parents were less religious. These associations obtained
noted above, McCullough et al. (2003) found in a 19-year longi-
even when controlling for children’s gender, ethnicity, and grade
tudinal study with data from the Terman study that children high
in school; parents’ gender, employment status, age, and educa-
in Conscientiousness (a trait that subsumes self-control) went on to
tional level; family income levels; a variety of family structure
become more religious as adults (
.14), even when background
variables; and several other potential confounds.
variables influencing religiousness (e.g., being raised in a religious
Similar results are reported in the three other studies with
home, gender) were statistically controlled. Second, McCullough
different measures of religiousness (e.g., multi-item measures of
et al. (2005) found that young adults with low Agreeableness
engagement in religious behaviors or self-rated importance of
(which is a personality trait reflecting the ability to control oneself
religiousness) and different measures of self-control (Brody &
out of concern for the feelings and desires of others) manifested a
Flor, 1998, r
.16; Brody et al., 1996, rs
.10 and .21;
pattern of low religiousness through adulthood.
Lindner-Gunnoe et al., 1999,
s
.40 and .49). Taken together,
Third, Wink et al. (2007) found that Conscientiousness (
these studies suggest that religious families in the United States
.17, p
.05) and Agreeableness (
.20 for women, p
.05;
tend to raise children and adolescents with higher self-control.
.05 for men, ns) in adolescence were associated with in-

---

76
MCCULLOUGH AND WILLOUGHBY
creased religiousness through late adulthood. Fourth, Regnerus
higher levels of self-control. These basic associations have been
and Smith (2005) found that among American youths, religious
replicated in samples of children, adolescents, college students,
students with high self-control (i.e., those who eschewed risk
and adults from a variety of religious backgrounds and nation-
taking and who reported making decisions in a deliberative, ef-
alities.
fortful fashion) remained more religious (as measured by self-rated
The one longitudinal study that evaluated whether religiousness
importance of religion and self-reported frequency of religious
might lead to increases over time in traits that are relevant to
service attendance) 1 year later than did those religious students
self-control yielded some evidence consistent with this notion
who initially had low self-control. Fifth, Heaven and Ciarrochi
(Wink et al., 2007), at least for women, but five studies also
(2007) found that Australian high school boys whose Psychoticism
suggest that traits relevant to self-control may influence later levels
scores declined over two time points reported higher scores on a
of religiousness. Thus, it seems plausible that religiousness and
multi-item measure of religiousness at a third time point (
self-control–relevant traits influence each other longitudinally. The
.17, p
.01) and that high school girls whose Conscientiousness
one experiment that has examined the link between religiousness
increased over two time points reported higher religiousness at the
and self-control indicates that religious mental content is automat-
third time point (
.19, p
.01). Taken together, these five
ically recruited when mental content related to sin and temptation
studies lend credence to the idea that self-control–relevant traits
has been previously activated (Fishbach et al., 2003). This exper-
precede changes in religiousness.
iment also showed that activating religious mental content makes
Experimental research on religion and self-control.
We found
sin and temptation stimuli less accessible, which may be one route
only one experimental study addressing the possibility that religion
by which religious content facilitates (automatic) self-control.
promotes self-control (Fishbach, Friedman, & Kruglanski, 2003).
Taken together, these results provide tentative support for the
In this experiment, which was part of a five-experiment investi-
proposition that the links of religiousness and self-control are
gation into the automatic nature of self-control in response to the
causal, although better longitudinal studies and experiments are
automatic activation of temptation-related stimuli, undergraduates
needed to advance our understanding.
at the University of Maryland were subliminally primed for 50 ms
with a temptation/sin-related concept (e.g., drugs, temptation, pre-
Religion and the Real-Time Process of Self-Regulation
marital sex), a religion-related concept (e.g., prayer, bible, religion,
and God), or a neutral word. After each prime, participants were
Thus far, we have proposed that religiousness is associated with
asked to identify religion-related words or temptation/sin-related
higher levels of self-control. However, some of religion’s putative
words as either words or nonwords as quickly as possible.
effects on behavior and well-being are not about “overriding
Fishbach et al. (2003) found that the subliminal presentation of
prepotent responses” (which defines self-control), but rather are
temptation/sin-related primes led to faster subsequent recognition
about steering one’s behavior according to goals more generally
of religion-relevant words than did the subliminal presentation of
(which defines the broader concept of self-regulation). In Carver
neutral primes. Conversely, subliminal presentation of religion-
and Scheier’s (1998) model of self-regulation, which was informed
relevant primes led to slower subsequent recognition of
by cybernetic theory (e.g., Wiener, 1948), self-control is concep-
temptation/sin-relevant words than did the subliminal presentation
tualized as a dynamical process by which people bring their
of the neutral primes. In the context of the four other experiments,
behavior into conformity with a standard through the operation of
the authors interpreted these results as evidence that people auto-
feedback loops consisting of several integrated functions. The first
matically recruit religious concepts to help them exercise self-
function is an input function that detects the system’s state. In
control in the face of temptation and, conversely, that the activa-
human terms, this is equivalent to one’s perceptions of the self and
tion of religious mental content reduces the accessibility of
the environment. The second function is a comparator function
temptation/sin-relevant mental content. Fishbach et al.’s study is
that compares the system’s state to a reference value. A reference
perhaps the best direct evidence available to date that religious
value can be conceptualized as a goal, a standard, or an ideal.
mental content is capable of increasing self-control.
When the comparator indicates that the state of a system matches
the reference value, the system changes nothing, and the existing
Summary of Studies on the Links Between
state is maintained. When the comparator registers a discrepancy
Religiousness and Self-Control
between the system’s actual state and its reference value, a third
function, referred to as an output function, is activated to reduce
Among 14 cross-sectional studies (13 of which were previously
that discrepancy. A self-regulating system continuously self-
published and the 14th of which is reported herein), 13 indicated
monitors for discrepancies and attempts to minimize those discrep-
that individual differences in general religiousness and intrinsic
ancies through its outputs.
religious motivation are positively associated with individual dif-
Effective human self-regulation, as conceptualized by Carver
ferences in self-control. In general, these associations are small to
and Scheier (1998), requires several abilities. First, it requires clear
medium in magnitude (i.e., standardized coefficients generally
standards or goals to pursue or preserve. These goals must be
range from .2 to .4). Two studies suggest that extrinsic religious
organized so as to permit effective management of conflict among
motivation is either negatively related or unrelated to self-control.
them (Fitzsimmons & Bargh, 2004). Second, it requires sufficient
In addition, the balance of more than 40 studies shows that
self-monitoring so that one can detect discrepancies between one’s
religiousness is associated cross-sectionally with self-control–
behavior and one’s goals. Third, it requires sufficient motivation or
relevant personality traits, such as high Agreeableness, high Con-
strength to change one’s behavior. Finally, it requires effective
scientiousness, and low Psychoticism. In addition, it appears that
mechanisms for behavioral change (Schmeichel & Baumeister,
religious parents and families tend to have children who have
2004). Insofar as religion influences self-regulation, it does so by

---

RELIGION, SELF-REGULATION, AND SELF-CONTROL
77
influencing these functions, which leads to Propositions 2–5, pre-
Through a hierarchical process by which system concepts lead
sented in Table 1.
to principles, principles lead to programs, and programs lead to
sequences, action can emerge that then functions as feedback for
evaluating how well one is doing at minimizing discrepancies at
Proposition 2: Religion Influences Self-Regulation by
the appropriate levels of specificity. For example, proprioceptive
Influencing People’s Goals
feedback and the sound of the radio constitute feedback indicating
that (a) a certain sequence (i.e., turning on the radio) is completed,
Austin and Vancouver (1996) defined goals as
(b) progress has been made toward one’s goal of distracting
internal representations of desired states, where states are broadly
oneself from negative thoughts, (c) progress has been made toward
construed as outcomes, events, or processes. Internally represented
one’s goal of being a forgiving person, and (d) progress has been
desired states range from biological set points for internal processes
made toward becoming one’s ideal self.
(e.g., body temperature) to complex cognitive depictions of desired
Religion, principles, and programs: Uniqueness among reli-
outcomes (e.g., career success). Likewise, goals span from the mo-
gious systems.
Religion’s most pervasive influences on goal
ment to a life span and from the neurological to the interpersonal.
selection probably occur at the principle and program levels.
(p. 338)
Specific religions dictate specific principles and programs by
which adherents are supposed to organize their behavior. The
Religion may influence goals in four ways. First, religious belief
unique goals prescribed by each religion probably arise from the
systems may influence the goals that people adopt and assist
unique emphases in the written scriptures of each religion, the
people in prioritizing those goals. Second, religion may increase
social and physical ecology in which the religion emerged (e.g.,
people’s motivation to obtain their goals by sanctifying them, or
the surrounding cultural alternatives from which supporters of
imbuing them with a sacred significance (Mahoney, Pargament, et
various religions wish to distinguish themselves, along with the
al., 2005). Third, religion may foster goal integration, thereby
resources available), and the changing physical and social circum-
reducing the extent to which people experience conflict among
stances to which specific religions must continually adapt (Wilson,
their goals. Integrated goals are more easily obtained (Emmons,
2002).
1999). Fourth, religion may affect the manner in which goals
One example of unique religious influences on goals comes
are internalized. Below, we elaborate on these four subsidiary
from work on religious differences in valued emotional states. A
propositions.
study of self-identified Christians and Buddhists from various
North American universities revealed that Christian students val-
Proposition 2a: Religion Influences Goal Selection
ued high-arousal positive emotional states (e.g., “excited,” “enthu-
siastic,” “elated,” “euphoric”) to a greater extent than did Buddhist
Religions dictate the domains in which people should attempt to
students (Cohen’s d
.44, p
.001) and that Buddhist students
self-regulate. Religions also prescribe the goals to which people
value low-arousal positive emotional states (e.g., “calm,” “peace-
should aspire in those domains. To discuss the effects of religion
ful,” “serene,” and “relaxed”) to a greater extent than did Christian
on goal selection, we first introduce some basic ideas about how
students (Cohen’s d
.29, p
.05). That Buddhism and Chris-
goals are organized. Following Powers (1973), Carver and Scheier
tianity place differing emphases on the desirability of different
(1998) proposed that the goals that interest social scientists (as
emotional states was confirmed in two additional studies showing
opposed to scientists exclusively interested in motor control) can
that Christian religious texts place a higher value on the high-
be expressed with four-level goal hierarchies. At the highest level
arousal positive affective states and a lower value on the low-
of these hierarchies are broad goals that Carver and Scheier re-
arousal positive affective states than do Buddhi