Social control Theory

Social control Theory

Order Description

This is a 15 page essay as follow:
the 1st 5 pages are about the book ” Vold’s Theoretical criminology” sixth edition. Authors: Thomas J. Bernard, Jeffrey B. Snipes, and Alexander L. Gerould.
Please refer to the end of the chapter’s 5 questions in order to write the 1st 5 pages.

the 2nd 5 pages are from the book ” Of Crime and Criminality:The Use of Theory in Everyday Life.” Editor: Simpson, Sally.
Chapter 8: The Social Control of Corporate Criminals (Shame and Informal Sanction Threats).
From pages 141 to 157.
Again, please refer to the end of the chapter’s 5 questions to base the 5 pages.

3rd 5 pages: Is a Summary of the attached article, implication, results, how were the data gathered etc.

Department of Sociology and Institute for Studies of Religion
Baylor University
Department of Sociology
Baylor University
KEYWORDS: spirituality, religiousness, emerging adulthood, violent
crime, property crime
Although prior research has had a tendency to confirm a negative
association between religiousness and crime, criminologists have been
slow to incorporate new concepts and emergent issues from the scientific
study of religion into their own research. The self-identity phrase “spiritual
but not religious” is one of them, which has been increasingly used
by individuals who claim to be “spiritual” but disassociate themselves
from organized religion. This study first examines differences in crime
between “spiritual-but-not-religious” individuals and their “religiousand-
spiritual,” “religious-but-not-spiritual,” and “neither-religious-norspiritual”
peers in emerging adulthood. Specifically, we hypothesize that
the spiritual-but-not-religious young adults are more prone to crime
than their “religious” counterparts, while expecting them to be different
from the “neither” group without specifying whether they are more
or less crime prone. Second, the expected group differences in crime
are hypothesized to be explained by the microcriminological theories
* Additional supporting information can be found in the listing for this article in the
Wiley Online Library at
An earlier version of this article was presented at the 2012 annual meeting of
the American Society of Criminology, Chicago, IL. The authors are grateful
for insightful comments and suggestions of the three anonymous reviewers and
the Editor-in-Chief. Direct correspondence to Sung Joon Jang, Department of
Sociology, Baylor University, One Bear Place #97326, Waco, TX 76798 (e-mail:
Sung Joon [email protected]).
C 2013 American Society of Criminology doi: 10.1111/1745-9125.12013
CRIMINOLOGY Volume 51 Number 3 2013 595
of self-control, social bonding, and general strain. Latent-variable structural
equation models were estimated separately for violent and property
crimes using the third wave of the National Longitudinal Study of
Adolescent Health. The overall results tend to provide a partial support
for the hypotheses. Implications for criminology and future research are
“‘I’m spiritual but not religious’ is a cop-out,” says Miller (2012: para.
2), an author on the CNN Belief Blog. Being “especially prevalent in
the younger population in the United States,” (para. 2) he opined, “the
spiritual but not religious reflect the ‘me’ generation of self-obsessed,
truth-is-whatever-you-feel-it-to-be thinking, where big, historic, demanding
institutions that have expectations about behavior, attitudes and
observance and rules are jettisoned yet nothing positive is put in replacement”
(para. 17). Gallup (2003) echoed this portrayal of the increasingly
common religious/spiritual self-identity based on a national poll showing
“evidence of the focus on self in spirituality” (para. 4) among Americans,
which he attributes to “the influence of individualism [that] extends into
American spirituality” (para. 6). Although young adults are known to be
more religiously unaffiliated than their older counterparts (Pew Research
Center, 2010), thus, likely being more crime prone, this identity has never
been examined in terms of its implication for criminology.
Criminologists have rarely studied spirituality as a separate concept from
religiousness or religion, but this may have been partly because of data constraints
(Johnson and Jang, 2010). That is, a crime survey typically contains
only a small number of “standard” religion items (e.g., service attendance
and religious salience besides denomination) and infrequently, if ever, items
reflecting spirituality. In addition, criminologists have been slow in not only
collecting new data but also in creatively reinventing existing data to address
current issues in the scientific study of religion as they intersect with
the interdisciplinary study of crime. We intend to do the latter by focusing
on the emergent group of American adults who call themselves “spiritual
but not religious” to examine this religious/spiritual identity’s relevance for
To address the issue, we first examine differences in criminal offenses
between “spiritual-but-not-religious” individuals and their “religiousand-
spiritual,” “religious-but-not-spiritual,” and “neither-religious-norspiritual”
counterparts. Specifically, the “spiritual but not religious” are hypothesized
to be at higher risk of committing violent and property crimes
than their more religious (i.e., “religious and spiritual” and “religious but
not spiritual”) peers, whereas they are expected to be different in the risk
from the “neither-religious-nor-spiritual” ones. Furthermore, the expected
group differences in crime are hypothesized to be explained by general
theories of crime: self-control, social bonding, and general strain theory. To
test the hypotheses, we estimate separate latent-variable structural equation
models for violent crimes and property crimes, analyzing the third wave
of the National Longitudinal Study of Adolescent Health (Add Health)
data collected when respondents were young adults. Given the unfamiliarity
of the concept of “spirituality” as distinct from religiousness in criminology,
we begin with a discussion of the two concepts.
In 1918, Georg Simmel (1997: 20) wrote that in a modernizing culture,
people are drawn to mysticism, a highly individualized religious belief or
“way of life” because it allows “the fixed definition and delimitation of religious
forms to be suspended” when themore “objectively defined” religious
life is no longer satisfying. Some people become dissatisfied with institutionalized
religion and take on highly individualized religious beliefs. Although
the relative importance of the individualized and institutionalized dimensions
of religion has ebbed and flowed since Simmel’s day, the former dimension
has come to be increasingly referred to as “spiritual” as opposed
to as “religious,” conveying at least partly different meanings (Idler et al.,
2003; Miller and Thoresen, 2003; Underwood and Teresi, 2002).
Initially, some theorized that the religious and spiritual dimensions of
faith were largely independent of each other (Wuthnow, 1998), but research
has uncovered evidence of an overlap between the two. For example,
Zinnbauer et al. (1997) found that the term spirituality was most often described
in personal or experiential terms, whereas religiousness was used in
relation to institutional beliefs and practices, such as church membership
or attendance and commitment to the belief system of organized religion.
However, based partly on spirituality and religiousness being “modestly”
correlated (r = .21), they concluded that the terms are “not fully independent”
(p. 561) and thus should not be thought of as mutually exclusive
categories but as concepts tapping different dimensions of faith with a fair
amount of overlap between them. This conceptualization is further backed
by other scholars (Good, Willoughby, and Busseri, 2011; Hill et al., 2000;
Schlehofer, Omoto, and Adelman, 2008), and they note that most Americans
consider the two concepts as related but distinct from each other (Marler
and Hadaway, 2002). For this reason, Dougherty and Jang (2008) found
more than half of Americans (57 percent) to be “religious and spiritual”
and approximately one quarter (27 percent) to be either only “religious”
(17 percent) or only “spiritual” (10 percent) with the remainder (16
percent) being neither.
Thus, recognizing a conceptual distinction as well as an overlap between
the “religious” and “spiritual” dimensions,1 we treat them as two related
constructs (see also Hill et al., 2000; Marler and Hadaway, 2002; Schlehofer,
Omoto, and Adelman, 2008; Zinnbauer, Pargament, and Scott, 1999). The
religious dimension concerns church-centered beliefs and practices (Wink
et al., 2007; Wink and Dillon, 2008), faith that is oriented toward tradition
(Pepper, Jackson, and Uzzell, 2010; Saucier and Skrzypinska, 2006), institutional
and societal pressures (Zwissler, 2007), and a socially oriented and
directed faith (Piedmont, 1999; Schlehofer, Omoto, and Adelman, 2008;
Zinnbauer et al., 1997).
The spiritual dimension, however, connotes an individual’s search for a
connection with the sacred (Piedmont, 1999), a generalized spiritual seeking
(Wink and Dillon, 2008; Wink et al., 2007; Wuthnow, 1998), a subjective
religious orientation (Saucier and Skrzypinska, 2006), and an emphasis on
personal experiences and individuality (Roof, 1993; Zwissler, 2007). It also
could imply some nontheistic notion of a higher power (Schlehofer, Omoto,
and Adelman, 2008), a faith with an internal source (Fuller, 2001), an interest
in mysticism, experimentation with unorthodox beliefs and practices,
and negative feelings toward both clergy and churches (Zinnbauer et al.,
1997). That is, unlike the religious dimension that is relatively stable in its
relation to something institutional in orientation, the spiritual dimension
has two different sides to it. One side of being “spiritual” blends well with
institutional religious practices, as noted by Good, Willoughby, and Busseri,
(2011). The other side is antagonistic to organized religious faith (Garelli,
2007; Zinnbauer et al., 1997), is associated with a feeling of alienation from
religion (Roof, 1999), and may be characterized as “deconverting” from
something (Harrold, 2006).
How then are we to know whether the term “spiritual,” when used, refers
to the antagonistic or proinstitutional side? Because it could connote either
favorable or unfavorable attitudes toward religion, the meaning can be
determined by whether it is used in conjunction with the term “religious” or
not (e.g., “spiritual” as well as “religious” or “spiritual” without being “religious”).
For this reason, a two-by-two typology is a better approach to operationalizing
the concept of “spiritual” rather than spirituality being measured
as a continuous variable independent of religiousness (e.g., Giordano
et al., 2008; Piedmont et al., 2009). That is, we can ask survey respondents
whether they are “spiritual” and whether they are “religious,” and then
1. Each dimension can be measured by using items designed to tap an individual’s
religiousness and spirituality (e.g., Hodge, Andereck, and Montoya, 2007) or by
asking an individual to self-rate how religious and how spiritual he or she is (e.g.,
Dougherty and Jang, 2008; Zinnbauer et al., 1997). This study is based on the latter
we can more clearly distinguish discrete categories reflecting “spiritual but
not religious,” “spiritual and religious,” “religious but not spiritual,” and
This typological approach allows for an overlap of the constructs for
those respondents who have both an institutional “religious” faith as well
as an individually invested “spiritual” approach to faith. More importantly,
it can distill out and separate those with a more hostile “spirituality” from
those whose “spirituality” is not antagonistic because of their attraction to
being “religious.” This typology has been used in previous studies (Chatters
et al., 2008; Hodge, Andereck, and Montoya, 2007; Marler and Hadaway,
2002; Roof, 1999; Storm, 2009; see Garelli, 2007, for a somewhat expanded
To the extent that the four-part typology is not random in such a way
that the four groups of individuals systematically differ in propensity for
crime, we are led to ask questions for which criminology has no ready answers.
How would we rank order the four groups in terms of their probability
of criminal offending? Is the emerging group of “spiritual-but-notreligious”
individuals more or less likely to commit crime than the others?
If that group is more or less likely to commit crime, then why? Criminological
theories and research have suggested that, all other things being equal,
religious individuals should be less likely to commit crime than their nonreligious
counterparts (Baier and Wright, 2001; Johnson and Jang, 2010).
However, criminologists have not studied a relationship between criminal
offending and being “spiritual,” providing little help with hypothesizing the
Although criminological research on religion has been conducted at both
the macrolevel and the microlevel, this study focuses on the latter because
we intend to examine and explain differences in crime among individuals of
differential religious self-identification. In fact, Johnson and Jang’s (2010:
121) systematic review revealed that “theoretical explanations of the influence
of religion on crime have been predominantly social psychological,”
which they attributed partly to an intellectual climate at the time of the
publication of Hirschi and Stark’s (1969) “hellfire” study, which generated
much research on religion and crime. As a result, microcriminological theories
have been applied to explain the effect of religion, particularly individual
religiousness, on crime. Johnson and Jang (p. 122, emphasis in original)
summarized as follows:
While none of [the] theories focus on religion as a key cause or
correlate of crime, they all offer explanations of the religion-crime
relationship by identifying processes, by which religiosity is expected
to decrease the probability of crime. Specifically, religious individuals
are less likely to commit crime than their less religious counterparts . . . ,
because they are more likely to: (1) fear supernatural sanctions as well
as this temporal criminal punishment and feel shame and embarrassment
associated with deviance; (2) be bonded to conventional society in
terms of attachment, commitment, involvement, and beliefs; (3) exercise
high self-control attributable to effective child-rearing by their parents
likely to be also religious; (4) have frequent and intimate associations
with peers who reinforce conventional definitions and behaviors and become
a model for imitation relative to those who do not; and (5) cope
with life’s strains or stressors and their resultant negative emotions in a
legitimate, non-deviant manner. Thus, individual religiosity is expected
to be negatively associated with crime.
Because of data constraints (see footnote 4), we incorporate three of the
preceding theories (Agnew, 2006; Gottfredson and Hirschi, 1990; Hirschi,
1969) into our explanations for expected differences in crime between the
“spiritual but not religious” and three other groups of individuals: “religious
and spiritual,” “religious but not spiritual,” and “neither religious nor
Given the conceptual duality of being “spiritual,” the spirituality–crime
relationship is to be explained by criminological theories in the same way
as its religiousness–crime counterpart to the extent that self-claimed spirituality
is associated positively with religiousness, connoting subjective religious
orientation and personal experience with organized religion. This
would be the case with those who claim to be religious as well as spiritual.
However, for those who claim to be spiritual but deny being religious, being
“spiritual” is likely to refer to experimenting with unorthodoxy and being
negative toward religion or even hostile to its rules and authorities. In such
a case, the theories would predict a positive association between spirituality
and crime. This antagonistic meaning of being “spiritual” may be what
those who claim only to be religious (i.e., “religious but not spiritual”) have
in mind when they disavow being “spiritual” because it is dissonant with
their proreligion attitudes.
Taken together, criminological theories would posit that those who are
“religious and spiritual” and “religious but not spiritual” should be less
likely to commit crime than their “spiritual-but-not-religious” peers. In
contrast, those who deny both being religious and spiritual are expected
to be less criminal than their “spiritual-but-not-religious” counterparts if
their claiming to be “not spiritual” meant a lack of interest in religion as
well as experimenting with unorthodox beliefs and practices. If, however,
it is intended to emphasize their negative, antagonistic attitudes toward
religion more than indicating no religious affiliation, then the “neitherreligious-
nor-spiritual” individuals might be somewhat more criminal than
the “spiritual-but-not-religious” individuals.
Although little criminological work has taken into account self-identified
spirituality and religiousness, one notable exception is Giordano et al.’s
(2008) life-course study on the effect of religiosity on criminal desistance,
where they distinguished between two “dimensions of religiosity”: “religious
participation” and “spirituality.” The former dimension is roughly
equivalent to the concept of being “religious” as we discussed previously,
but the latter differs from what we referred to earlier as being “spiritual.”
Their “spirituality” concept is neither discrete nor dualistic in meaning (i.e.,
antagonistic to institutional religion or not). Furthermore, “spirituality” was
“defined as perceived closeness to God” (p. 109) and “measured by feelings
of being close to God” (p. 125) that may tap the more orthodox side of being
“religious,” like one’s attachment to God (Pargament, 1997) rather than
being “spiritual.” This explicit focus on “God” has traditional and orthodox
religious undertones (Zinnbauer et al., 1997), thereby potentially excluding
the very population of interest here: those identifying as “spiritual but not
religious.” After all, for Giordano et al. (2008), spirituality is a dimension of
religiosity and, thus, is neither conceptually distinct from being “religious”
nor the same as being “spiritual” in the sense discussed previously. As a
result, the relevance of Giordano et al.’s study to ours is limited, but it is
still helpful for our purposes as they pointed out that the private nature
of spirituality may not be helpful in the desistance process because, unlike
communally connected coreligionists, others are not present to challenge
and support them.
Similarly, measuring spirituality “defined as an experiential relationship
with God (or Ultimate Transcendence)” with six items, Hodge, Cardenas,
and Montoya (2001: 154) found “spirituality” not to be associated with alcohol
use, whereas a measure of religiousness was positively associated
with the odds of never using alcohol. Another study, however, showed that
“spirituality” tended to be positively associated with alcohol use (Hodge,
Andereck, and Montoya, 2007). Specifically, based on a cluster analysis of
the same six “spirituality” items plus two religiousness items andANCOVA
tests, they found the spiritual-but-not-religious cluster to report higher levels
of alcohol use than its neither-spiritual-nor-religious counterpart. Because
the two clusters were not different in religiousness, the higher level
of alcohol use can be attributed to the part of being “spiritual.”
Although these findings about the relationship of spirituality with crime
and alcohol use seem to be contradictory to one another, the mixed findings
are, perhaps, actually congruent with the characterization of self-claimed
spirituality in relation to criminal propensity. As discussed, claiming to be
“spiritual” could, in some cases, have a negative association with criminal
behaviors as it reflects an individual’s positive self-esteem, conventional values,
and self-transcendent relationship with something sacred in life (e.g.,
Hodge, Cardenas, and Montoya, 2001; Miller and Thoresen, 2003; Wink
et al., 2007). In other cases, however, if it indicates narcissism, unorthodoxy,
and antireligious attitudes, then being “spiritual” is likely to be positively
associated with antisocial behaviors (e.g.,Marler and Hadaway, 2002;
Wink et al., 2007; Zinnbauer et al., 1997) and mental disorder (King et al.,
2013). As such, we continue our argument that greater specificity is necessary
when examining spiritual as well as religious dimensions of beliefs.
In sum, whereas directly linking spirituality and antisocial behavior is a
challenge because of the current dearth of extant research, we can extrapolate
from what we know of being “spiritual” and existing theories of crime.
If some “spiritual” individuals are opposed to organized religion, then they
would not reap the same benefit of protection from criminality that is observed
for “religious” individuals. They may be less constrained by religious
institutions and the connections that come with it. Thus, “spiritualbut-
not-religious” individuals do not have the same communal dimension
that reinforces prosocial connections with others, unlike their “religiousand-
spiritual” counterparts—who are institutionally embedded as well as
personally invested in religion—and to a lesser extent, their “religious-butnot-
spiritual” peers.
Criminologically speaking, “spiritual-but-not-religious” individuals are
likely to be exposed to lower levels of control (Gottfredson and Hirschi,
1990; Hirschi, 1969) and higher levels of strain (Agnew, 2006) compared
with their “religious” peers. In contrast, whether “spiritual-but-notreligious”
individuals are more or less likely to be criminal than their
“neither-religious-nor-spiritual” peers depends on what they mean when
they claim to be “spiritual” because they both are “not religious.” Thus, we
test between these two alternative hypotheses.
Emerging adulthood refers to “the period from the late teens through
the twenties, with a focus on ages 18–25” (Arnett, 2000: 469). According to
Arnett (2005: 239, italics added for emphasis), emerging adulthood is “the
age of identity exploration . . . instability . . . feeling in-between, neither adolescent
nor adult” and “the most self-focused age of life.” These features
imply potentially negative outcomes because it could become a time of narcissism,
unorthodoxy, and “innovation” (Merton, 1938), as well as of low
self- and social control relative to full-fledged adulthood.
Despite their declining pattern, crime rates are still relatively high during
emerging adulthood (Laub and Sampson, 2003), and criminologists have
found the variables of control and strain to explain criminal behaviors during
this period (Piquero et al., 2002). For example, Sampson and Laub
(1993, 1996) and Laub and Sampson (2003) found measures of adult social
bonds, such as marital attachment, job stability, and military service,
to have turning point effects in emerging adulthood (see also Uggen, 2000).
More recently, Jang and Rhodes (2012) found that low self-control and variables
of general strain theory explain violent and property crimes among
emerging adults while controlling for social bonding and social learning
Emerging adulthood is also a unique period in life when religious beliefs
and practices are in flux. For example, Smith (2005, 2009) found a
highly individualistic belief “system,” likely related to “spirituality” independent
of a religious community, to be held in adolescence and to persist
into the emerging adult years. In a twin study, Koenig, McGue, and Iacono
(2008) showed that religiousness, especially church attendance, decreases
with age for emerging adults. Similarly, finding a reduction primarily in religious
practices, Uecker, Regnerus, and Valler (2007) reported that irreligious
teens tend to solidify their irreligion during emerging adulthood, who
are likely to then become “neither religious nor spiritual.”
This study tests the following hypotheses, focusing on violent and serious
property offenses because social group differences (e.g., sex, race, and class
differences) tend to be minimal for nonserious, trivial offenses.
Hypothesis 1: Spiritual-but-not-religious individuals are more likely
to commit crimes than their “religious” counterparts, whether “spiritual”
or not, whereas they are expected to be different in criminal
offending from the neither-religious-nor-spiritual individual.
Hypothesis 2: The expected differences in criminal offending are likely
to be explained by group differences in religious involvement as well
as by the variables of self-control, social bonding, and general strain
To test these hypotheses, latent-variable structural equation models
were constructed. The key exogenous variables are three religious/spiritual
dummy variables with the spiritual-but-not-religious group as the reference
category, and a latent construct of violent or property crimes is the
ultimate endogenous variable. First, controlling for sociodemographic variables,
we estimate a baseline model to establish relationships among the
three dummy variables and the latent variable of crime, separately for violent
and property offenses. Then, we add six variables, five latent and one
manifest, one at a time (models 1 to 6) and then all together (full model),
to determine whether the addition changes not only the size but also the
significance of the initially estimated relationship between the key exogenous
and the ultimate endogenous variables.
The current data come from the restricted-use sample of the National
Longitudinal Study of Adolescent Health (Add Health), a nationally representative
survey of more than 20,000 students, all of whom were in grades
7 through 12 when the first wave was collected during the 1994–1995 school
year. Add Health contains an exhaustive array of survey items measuring
the respondents’ social, economic, psychological, and physical well-being,
including an extensive battery of questions asking about delinquency and
drug use. Respondents were selected using a multistage, stratified, schoolbased
cluster sampling procedure. Eighty high schools, drawn from the
Quality Education Database, were stratified by region, urbanicity, sector,
racial composition, and size. Attempts were made to match each of these 80
schools with a feeder school, usually a middle school or junior high school.
In total, 52 feeder schools were added to the sample of schools, resulting in
a sample of 132 schools.
From September 1994 through April 1995, all available students in the
132 schools (n = 90,118) completed an In-School Survey. A representative
sample of these students (n = 20,745) then participated in the In-Home Interview
at wave 1 using laptop computers, headphones, and audio recordings
to encourage accurate and honest answers to many of the survey’s most
personal questions.Wave 2 data were collected from those who had not yet
graduated from high school (n = 14,738) from April through August 1996.
Wave 3 was collected in 2001–2002 when the respondents were emerging
adults, 18–28 years of age. In total, 1,507 partners of the original sample
were added, resulting in a sample size of 15,197. Finally, wave 4, collected
in 2007–2008, has a sample size of 15,701 (see Harris et al., 2009, for details
on sampling design).
We focus on the third wave: the only survey that included the two key
items of self-identified religiousness and spirituality needed to construct our
The religious/spiritual identity variable was created using two self-rated
items. One regards religiousness (“To what extent are you a religious person?”)
and the other spirituality (“To what extent are you a spiritual person?”).
Each question was answered based on a 4-point scale: not at all
(=0), slightly (=1), moderately (=2), and very religious/spiritual (=3).
These categories were dichotomized by combining the first two responses
into “not religious/spiritual” and the other two into “religious/spiritual” to
construct a 2 × 2 typology table that consists of four self-identity categories:
religious and spiritual, religious but not spiritual, spiritual but not religious,
and neither religious nor spiritual. A system of dummy variables was created,
and three categories were included in the subsequent analysis with the
spiritual-but-not-religious group, which is of our key interest in comparison
with the others, as the reference category.
The first explanatory variable, religious involvement, is a latent construct
that has four indicators—religious service attendance, participation in nonservice
religious activities, religious salience (i.e., perceived importance of
religious faith), and frequency of prayer (see table S.1 in the online supporting
information for details2), all of which have been found to be negatively
associated with crime (Johnson and Jang, 2010). An exploratory factor
analysis of the four indicators generated a one-factor solution with high
loadings, ranging from .55 to .80, with a high Cronbach’s alpha (.80).
A latent variable of low self-control (Gottfredson and Hirschi, 1990)
was measured by two items asking about the respondent’s tendency of
risk-taking and little thought for the future,3 whereas that of attachment
to parents, one of two social bonding variables (Hirschi, 1969), was operationalized
by three indicators: feeling close to parents, parents being warm
and loving, and enjoying doing things with parents. An exploratory factor
analysis revealed that the three have high loadings (.82, .87, and .84) and
that they were found to have an excellent interitem reliability (a = .88).
2. Additional supporting information can be found in the listing for this article
in the Wiley Online Library at
3. The content validity of our low self-control measure is limited in that it operationalizes
only two of the six “elements of self-control” (Gottfredson and Hirschi,
1990: 89, emphasis in original): “exciting, risky, or thrilling” and “little skill or planning.”
Because low self-control is to bemeasured by tapping all six dimensions, our
measurement is likely to have resulted in false positives (e.g., “risk takers” who
have high risk tolerance but not low self-control or respondents with attention
deficit disorder who are poor at planning but have high self-control). Positively
put, as a result, our test is deemed conservative with respect to the relationship
between low self-control and crime to the extent that false positives weaken the
The other social bonding variable, commitment to conventional activities,
is an index of four items asking whether the respondent was attending or
enrolled in school, had received any vocational education or job training,
was working for pay for at least 10 hours a week, and/or was serving in the
full-time active-duty military when the survey was conducted. Although
each item was measured as a dichotomous variable (0 = no, 1 = yes), all
of them are measures of adult social bonds that Sampson and Laub (1993)
suggested in their age-graded theory of informal social control.4
The last two explanatory variables come from Agnew’s (2006) general
strain theory. Among various types of strain, this study focused on criminal
victimization, one of the most criminogenic strains for adults, modeling it as
a latent construct with an indicator of vicarious victimization (“saw someone
shoot or stab another person”) and six of experienced victimization
(e.g., being pulled a knife or gun on, shot, stabbed, or beaten up) during
the 12 months prior to the survey. An exploratory factor analysis showed
that the seven indicators were loaded on a common factor with moderateto-
high loadings, ranging from .40 to .66, and a relatively good interitem
reliability (a = .70). Another latent variable measures negative emotions
with six items of state depression (i.e., depressive mood and malaise during
the last week prior to survey) and had generally high factor loadings and
Cronbach’s alpha (.80).5
Next, two latent endogenous variables of crimes were constructed using
three items of violent crimes and four items of property crimes committed
during the last 12 months prior to the survey. The indicators of violent
crime measure how often the respondent engaged in armed robbery, physical
fights, and assault, whereas those of property crime tap the frequency
of vandalism, theft (two items involving something worth more than $50
and less than $50), and burglary (0 = never, 1 = 1 or 2 times, 2 = 3 or 4
times, 3 = 5 or more times). Whereas the latter indicators have relatively
high loadings, ranging from .50 to .72, with an acceptable reliability (a =
.67), the former have loadings mixed in size (.37, .74, and .46) with a low
reliability (a = .44). This might reflect a lack of versatility of violent crimes
relative to property crimes.
4. We initially included in our model a variable of Akers’s (1998) social learning theory:
peer drinking, including binge drinking. However, in response to the Editorin-
Chief’s and the reviewers’ concerns about the measure being limited to drinking
and based on the respondent’s perception of peer deviance rather than peerreported
deviance, we removed the social learning variable, and we estimated a
model including the variable as well (see footnote 9).
5. Whereas anger tends to have greater construct validity than nonangry emotions
(Agnew, 2006), we could not include the emotion in the analysis because the third
wave of Add Health data do not contain items of state anger.
Sociodemographic variables were constructed for statistical control: race
(0 = non-White, 1 = White), age (at the time of survey), sex (0 = male, 1 =
female), education (whose response categories range from 6 = 6th grade
to 22 = 5 or more years of graduate school; see table S.1 in the online supporting
information), and marital status (0 = never married, divorced, separated,
widowed, or in marriage-like relationship; 1 = married). In addition,
a methodological control was constructed that measures whether anyone
besides the respondent and the interviewer was present (=1) or not (=0)
when the survey was being conducted to adjust for any reporting bias caused
by reactivity.
Finally, to control for the respondent’s religious affiliation, we created
a variable similar to Steensland et al.’s (2000) RELTRAD. Our modified
RELTRAD assigns respondents to one of nine religious affiliation groups:
evangelical Protestant, mainline Protestant, Black Protestant, Catholic,
Jewish, Mormon, other religion, no religious affiliation, or unknown. Those
who said they were “just Christian” or ”just Protestant” and those who did
not know their religious affiliation were sorted into the evangelical Protestant,
mainline Protestant, or Black Protestant categories according to racial
backgrounds and “born again” self-identification. If we still could not classify
the respondent, then they were coded as “unknown.”
To test our hypotheses, we used Mplus 7 (Muth´en and Muth´ en, 1998–
2012), a structural equation modeling (SEM) software that enables the
analysis of “complex survey” data like the Add Health. The latent-variable
SEM approach tends to generate more valid and reliable results than other
methods (e.g., path analysis) because it allows controlling for measurement
errors. Latent-variable modeling also is suitable to test our hypotheses,
where most concepts are abstract and thus are not directly observable.
Furthermore, Muth´ en’s (1983) “general structural equation model” and its
full information maximum likelihood (FIML) estimation, incorporated into
Mplus, allow not only for continuous but also for dichotomous and ordered
polytomous variables to be indicators of latent variables, particularly when
the variables are skewed. Because our measures of offending are ordered
categorical variables, we employed the estimator of MLR: “maximum
likelihood parameter estimates with standard errors . . . that are robust
to non-normality and non-independence of observations” (Muth´en and
Muth´ en, 1998–2012: 484).We also used FIML to treat missing data (Baraldi
and Enders, 2010; Graham, 2009). Finally, for data-model fit assessment, we
focus on joint criteria using three types of fit index (Hu and Bentler, 1999)—
incremental (comparative fit index [CFI]), absolute (standardized rootmean-
squared residual [SRMR]), and parsimonious (root-mean-squared
error of approximation [RMSEA])—rather than relying on any one of
them. Specifically, a model is determined to have a good fit to data if one
of two joint criteria, (CFI = .96 and SRMR = .09) or (SRMR = .09 and
RMSEA = .06), is met.
Table 1 presents descriptive statistics of variables included in the analysis.
It shows that the current sample (n = 14,322) is 67.6 percent White, 49.2
percent female, and 16.4 percent married.6 The respondents were, on average,
21.8 years of age with the youngest and the oldest being 18 and 28, respectively,
whereas the mean of their highest education was approximately
1 year of college completed (13.072). Interestingly, although approximately
one quarter of the total sample (26.8 percent)—“none” (22.0 percent) or
“affiliation unknown” (4.8 percent)—reported no religious affiliation, 43.8
percent of the sample called themselves “neither religious nor spiritual.”
This finding indicates that reporting religious affiliation is one thing, but it
is another to self-identify as “religious” and/or “spiritual.” Also, as anticipated,
we found the percentage of the “neither religious nor spiritual” to be
higher than what Dougherty and Jang (2008) reported for a general population
of adults (44 percent vs. 16 percent), which is consistent with previous
findings: a general increase of religious “nones” among young adults
(Baker and Smith, 2009), and they become stable affiliates as they age (Lim,
MacGregor, and Putnam, 2010).7 However, the percentage of the spiritualbut-
not-religious category was about the same (11 percent vs. 10 percent),
whereas both “religious” groups were found to be smaller, whether “spiritual”
(38 percent vs. 57 percent) or not (7 percent vs. 17 percent).
6. The current sample consists of only those respondents enrolled in grades 7–12 during
1994–1995 (i.e., wave 1), for whom the Add Health provides sampling weight
(GSWGT3 2) to be used in cross-sectional analysis (Chantala, 2006). As a result,
875 of the 15,197 respondents who participated in the third survey could not be
included. Also, as anticipated, a relatively small percentage of the emerging-adult
respondents was married at the time of survey.
7. In addition, there are two potential methodological reasons for why comparisons
with the Baylor Religion Survey (BRS) data, whichDougherty and Jang (2008) analyzed,
differ from those of the Add Health. First, the BRS does not have nearly
the same number of respondents in the same age category as the Add Health for
good comparisons as there is reason to think a cohort effect exists (Marler and
Hadaway, 2002). Second, more importantly, this might have been, in part, an artifact.
Specifically, the Add Health asks the religious/spiritual questions immediately
after one another, more clearly juxtaposing the two and implying that they
are not the same concepts. The BRS, however, presents the self-rated religiousness
and spirituality questions at different points in the survey, decreasing the perception
that they are conceptually distinct. So we believe that the Add Health items
are more reliable for constructing our 2 × 2 typology than the BRS.
Table 1. Descriptive Statistics of Variables Analyzed and
Group Mean Comparisonsa
Variables Mean (SD) Minimum Maximum SBNR SBNR SBNR
vs. RAS vs. RBNS vs. NRNS
Race (White) .676 (.498) .000 1.000 .103* .104* –.003
Age 21.822 (1.764) 18.000 28.000 .123 .206 .188*
Sex (Female) .492 (.499) .000 1.000 -.022 -.029 .065*
Education 13.072 (1.965) 6.000 22.000 .057 .548* .553*
Married .164 (.375) .000 1.000 -.092* -.084* -.015
Third Person Present .233 (.420) .000 1.000 .003 -.017 -.022
Religious Affiliation
.223 (.396) .000 1.000 -.172* -.098* .017
Mainline Protestant .104 (.300) .000 1.000 -.020* -.026 -.028*
Black Protestant .101 (.343) .000 1.000 -.101* -.055* .009
Catholic .237 (.435) .000 1.000 -.004 -.128* -.027
Jewish .008 (.084) .000 1.000 .009 -.010 .005
Mormon .009 (.093) .000 1.000 -.009 .003 .002
Other religion .050 (.214) .000 1.000 .024* .041* .053*
None .220 (.402) .000 1.000 .241* .238* -.063*
.048 (.214) .000 1.000 .032* .035* .032*
Religious and
.379 (.489) .000 1.000 — — —
Religious but not
.068 (.248) .000 1.000 — — —
Spiritual but not
.115 (.315) .000 1.000 — — —
Neither religious
nor spiritual
.438 (.492) .000 1.000 — — —
Religious Involvement -.050 (.792) -1.166 2.223 -.824* -.460* .303*
Low Self-control 2.822 (.816) 1.000 5.000 .074* -.017 -.121*
Attachment to Parents .001 (.898) -4.722 .881 -.214* -.140* -.014
Commitment to
.198 (.202) .000 1.000 -.028* .032* .025*
Criminal Victimization .026 (.095) .000 1.000 .015 .015 .019
State Depression .486 (.495) .000 3.000 .046* .054* .083*
Violent Crimes .082 (.248) .000 3.000 .008* .000 -.004
Property Crimes .072 (.230) .000 3.000 .051* .044* .032*
NOTE: The number of observations for each variable analyzed range from 14,087 to 14,322,
and all analyses were weighted and corrected for survey design.
ABBREVIATIONS: NRNS = “neither religious nor spiritual”; RAS = “religious and spiritual”;
RBNS = “religious but not spiritual”; SBNR = “spiritual but not religious”; SD =
standard deviation.
aThe tests of difference in group means were based on the coefficients of three dummy variables
being RAS, RBNS, and NRNS (see the last three columns), regressed on each variable
with SBNR being the omitted category.
*p < .05 (two-tailed test).
Table 1 also reports results from comparing means between the “spiritual
but not religious” (SBNR) and each of the other three groups (see
the last three columns): “religious and spiritual” (RAS), “religious but not
spiritual” (RBNS), and “neither religious nor spiritual” (NRNS). We estimated
group-mean differences in terms of the structural relationships between
dummy variables representing the three groups and each variable
listed in the table, where a positive (negative) coefficient indicates the reference
group’s (i.e., SBNR’s) mean being larger (smaller) than a comparison
group’s mean.
Significant group differences tend to be consistent with what we discussed
previously: SBNR emerging adults are likely to be at higher risk for criminal
offending than their RAS and RBNS peers, whereas they are expected
to be at higher risk in some aspects but at lower risk in others than their
NRNS counterparts. At the time of survey, for example, the SBNR were
less likely than the “religious”—RAS (–.092) and RBNS (–.084)—to have
been married, which tends to decrease crime in emerging adulthood, particularly
when marriage is cohesive (Sampson and Laub, 1993). In contrast,
as expected, the SBNR–NRNS comparison results were mixed: The SBNR
were not different from the NRNS (–.015) in marital status, but differences
were found in social class (.553), being higher in terms of education. The
SBNR also reported higher grade or year of regular school completed than
the RBNS (.548), although not the RAS (.057).8 These demographic differences
between the SBNR and the others tend to be similar to previous
findings based on a national sample of American adults (Dougherty and
Jang, 2008).
Furthermore, interestingly, the SBNR respondents were more likely than
their RAS and RBNS peers to report affiliations other than Christian denominations
as their “present religion,” such as Islam, Eastern religion, paganism,
or New Age spirituality, whereas not surprisingly, they were less
likely to be either Protestant or Catholic. They also were more likely to
have “unknown” religion as well as being “none,” atheistic, or agnostic than
their “religious” counterparts, whereas there was no difference in terms of
affiliation with Jewish and Mormon religions. Similarly, the SBNR were
more likely to have “other” (.053) or “unknown” religion (.032) than the
NRNS, and they were less likely to be “none” (–.063).When taken together,
8. In addition, the SBNR respondents in the sample were more likely to be White
than their “religious” counterparts, whether they call themselves “spiritual” (.103)
or not (.104), whereas they were not significantly different from the NRNS (–.003).
However, the SBNR were older (.188) andmore likely to be female (.065) than the
the SBNR seem to be “not religious” with respect to the Protestant or the
Catholic religion but tend to be open or favorable to non-Christian, non-
Western, postmodern religion compared with the others (Fuller, 2001).
More importantly, the SBNR were found to be at higher risk for criminal
offending than the RAS and RBNS in terms of the explanatory variables,
while their comparisons with the NRNS generated mixed findings. First, the
SBNR were less likely to be religiously involved than the RAS and RBNS
(–.824 and –.460), although more likely than the NRNS (.303). This confirms
that the SBNR were “not religious” if it concerns the Protestant or
the Catholic denomination, but they were found to be more religious than
the NRNS as long as it is not the Christian religion, although they are still
less involved in religion than the RAS and RBNS.
Next, the SBNR respondents were at higher risk of criminal offending
in terms of low self-control and negative emotional state, and they were
less constrained by adult social bonds compared with their “religious”
counterparts. Specifically, they reported a significantly higher mean of low
self-control than the RAS (.074), although not the RBNS (–.017), and they
were found to have experienced depressive mood and malaise more often
during the last week prior to the survey than both “religious” groups (.046
and .054), which is consistent with a previous study (King et al., 2013).
Also, the SBNR were less likely to be attached to their parents than the
RAS (–.214) and the RBNS (–.140), whereas they were less committed to
conventional activities than the RAS (–.028) but more so than the RBNS
(.032). In contrast, the SBNR were at a lower risk of criminal offending
than the NRNS in terms of low self-control (–.121) and commitment to
conventional activities (.025), but they were at a higher risk with respect to
state depression (.083). However, the SBNR were not different in criminal
victimization from the others.
Finally, consistent with the preceding group differences in the correlates
and causes of crime, the SBNR were found to have committed crimes more
often than the other groups during the last 12 months prior to the survey,
which was more pronounced in property than in violent crimes. Specifically,
the SBNR reported higher levels of violent criminal offending than the
RAS (.008), although not compared with the RBNS (.000) and the NRNS
(–.004). On the other hand, the SBNR committed property crimes more
often than the RAS (.051), the RBNS (.044), and the NRNS (.032).
Measurement and Structural Models
Before interpreting results from estimating our model, we need to assess
the model’s overall fit to the current data. First, both models of violent
and property crimes are found to have a good fit, meeting one of
Hu and Bentler’s (1999) joint criteria, with standardized root-mean-squared
residuals (SRMRs) being smaller than .090 (ranging from .008 to .023 in the
violent crime model and from .011 to .023 in the property crime model) and
root mean square error of approximations (RMSEAs) smaller than .060
(from .013 to .033 and from .017 to .031), despite comparative fit indices
(CFIs) (from .865 to .973 and from .885 to .965) mostly not reaching the
minimum cutoff (.960). Overall, the model fit is adequate, which supports
interpreting the results.
Beginning with the latent variables’measurement models, we found most
of their indicators to have moderate-to-high loadings in both the violent and
property crime models, ranging from .362 to .867 as shown in table 2 (see
the top panel), which also presents correlations among the explanatory variables
(see the bottom panel of table 2). Most correlations were significant
in the expected direction.
Table 3 shows the unstandardized coefficients of the structural models
of violent crimes, estimated first with no additional explanatory variables
(baseline model) and then with one additional explanatory variable at a
time (models 1 to 6) as well as all six of them simultaneously (full model).
Significant coefficients of sociodemographic variables are mostly in the expected
direction as race (White), age, sex (female), education, and being
married are all negatively associated with violent crimes. In contrast, the
methodological control (third person present) and the dummy variables of
religious affiliation (RELTRAD) were found to be not significant with one
exception: Mormon respondents were less likely to commit violent crime
than their evangelical Protestant peers in the baseline (-.013) and all but
one intermediate models (-.013 or -.017), though not in the full model
The baseline model provides partial support for the first hypothesis. The
RAS dummy variable was found to have a negative coefficient (–.009), as
hypothesized: That is, the SBNR respondents were more likely to have engaged
in violent crimes than their RAS peers but not the RBNS (–.002).
Also, there was no significant difference in violent criminal offending between
the SBNR and the NRNS (–.001). Although these findings tend to be
consistent with the results from estimating group differences without controlling
for the sociodemographic variables (see table 1), only one of the
three hypothesized group differences was found to be significant.
Next, six intermediate models (models 1 to 6) show whether the religious/
spiritual dummy variables’ coefficients change not only in the size
but also in significance as explanatory variables were added to the baseline
model one at a time. The first two intermediate models show that the
coefficient size of RAS reduced in absolute value, becoming nonsignificant,
when religious involvement (33 percent reduction from |–.009| to |–.006|)
and low self-control (22 percent reduction from |–.009| to |–.007|) were
added each to the baseline model. These indicate the two variables’
Table 2. Estimated Measurement Models (Top Panel) and Correlations among Explanatory
Variables (Bottom Panel) (N = 14,322)
Variables Commitment to
Religious Low Attachment to Conventional Criminal State Violent Property
Involvement Self-control Parents Activities Victimization Depression Crime Crime
b (SE) b (SE) b (SE) b (SE) b (SE) b (SE) b (SE) b (SE)
Indicator (1) .724* (.008) .362* (.035) .815* (.006) — (—) .533* (.031) .605* (.014) .386* (.043) .523* (.031)
Indicator (2) .509* (.010) .563* (.048) .867* (.006) — (—) .627* (.025) .746* (.011) .684* (.037) .681* (.038)
Indicator (3) .828* (.007) — (—) .844* (.008) — (—) .677* (.024) .522* (.012) .542* (.033) .571* (.043)
Indicator (4) .752* (.008) — (—) — (—) — (—) .437* (.061) .824* (.009) — (—) .721* (.037)
Indicator (5) — (—) — (—) — (—) — (—) .504* (.053) .392* (.015) — (—) — (—)
Indicator (6) — (—) — (—) — (—) — (—) .420* (.035) .765* (.008) — (—) — (—)
Indicator (7) — (—) — (—) — (—) — (—) .412* (.051) — (—) — (—) — (—)
1.000 (—) –.298* (.022) .167* (.015) .201* (.017) –.027 (.014) .006 (.014)
Low Self-control -.297* (.022) 1.000 (—) –.105* (.021) -.187* (.032) .195* (.033) .144* (.027)
Attachment to
.167* (.015) -.105* (.021) 1.000 (—) .086* (.012) -.089* (.018) -.195* (.016)
Commitment to
.202* (.017) -.189* (.031) .085* (.012) 1.000 (—) -.014 (.015) -.054* (.011)
-.031* (.014) .194* (.034) -.087* (.018) -.016 (.016) 1.000 (—) .091* (.020)
State Depression .006 (.014) .144* (.026) -.195* (.016) -.054* (.011) .093* (.019) 1.000 (—)
NOTE: Standardized coefficients and their standard errors (in parentheses) from estimating the full model of violent crime (except those of the
indicators of property crime) are presented in the top panel, whereas correlations from the models of violent (below diagonal) and property crime
(above diagonal) are in the bottom panel; see table S.1 in the online supporting information for the description of indicators of each latent variable.
ABBREVIATION: SE = standard error.
*p < .05 (two-tailed test).
Table 3. Estimated Structural Models of Religious/Spiritual Self-identity and Violent Crime
(N = 14,322)
Variables Baseline Model Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Full Model
b (SE) b (SE) b (SE) b (SE) b (SE) b (SE) b (SE) b (SE) ß
Race (White) -.010* (.004) -.011* (.004) -.009* (.004) -.010* (.004) -.010* (.004) -.007* (.004) -.009* (.004) -.006 (.004) -.039
Age -.003* (.001) -.003* (.001) -.003* (.001) -.003* (.001) -.003* (.001) -.002* (.001) -.003* (.001) -.002* (.001) -.053*
Sex (female) -.035* (.006) -.034* (.006) -.024* (.005) -.035* (.006) -.035* (.006) -.023* (.004) -.036* (.006) -.017* (.004) -.113*
Education -.003* (.001) -.003* (.001) -.001 (.001) -.003* (.001) -.003* (.001) -.002* (.001) -.003* (.001) -.000 (.001) -.008
Married -.009* (.003) -.008* (.003) -.004 (.003) -.009* (.003) -.009* (.003) -.006* (.003) -.008* (.003) -.003 (.003) -.014
Third Person Present .001 (.003) .000 (.003) -.001 (.003) .000 (.003) .001 (.003) -.001 (.003) .000 (.003) -.002 (.003) -.011
Mainline Protestant .000 (.004) -.001 (.004) -.002 (.004) .000 (.004) .000 (.004) .002 (.004) .000 (.004) .001 (.004) .005
Black Protestant .003 (.006) .003 (.006) .005 (.006) .003 (.006) .004 (.006) .001 (.005) .004 (.006) .002 (.005) .007
Catholic .002 (.003) -.000 (.004) -.002 (.004) .002 (.003) .002 (.003) .002 (.003) .002 (.003) -.000 (.004) -.001
Jewish -.007 (.005) -.010 (.005) -.013 (.007) -.006 (.005) -.007 (.005) -.002 (.005) -.008 (.005) -.006 (.006) -.007
Mormons -.013* (.005) -.013* (.005) -.010 (.006) -.013* (.005) -.013* (.005) -.017* (.007) -.013* (.005) -.015 (.008) -.019
Other Religion .002 (.006) .001 (.006) .000 (.006) .002 (.006) .002 (.006) .003 (.006) .001 (.006) .002 (.006) .005
None -.002 (.003) -.006 (.004) -.005 (.004) -.002 (.003) -.002 (.003) -.002 (.003) -.002 (.003) -.003 (.004) -.016
Affiliation Unknown .001 (.006) -.000 (.006) -.003 (.006) .001 (.006) .001 (.006) -.001 (.006) .002 (.006) -.003 (.005) -.009
Religious and Spiritual -.009* (.004) -.006 (.004) -.007 (.004) -.008* (.004) -.009* (.004) -.008* (.004) -.008* (.004) -.008 (.005) -.049
Religious but Not Spiritual -.002 (.005) -.001 (.005) -.003 (.005) -.002 (.005) -.002 (.005) .001 (.005) -.001 (.005) .000 (.005) .002
Neither Religious Nor
-.001 (.003) -.002 (.004) -.003 (.004) -.000 (.003) -.000 (.003) .002 (.004) .001 (.003) .001 (.004) .007
Religious Involvement — -.005 (.003) — — — — — .001 (.004) .007
Low Self-control — — .039* (.008) — — — — .029* (.008) .147*
Attachment to Parents — — — -.002 (.002) — — — .002 (.002) .022
Commitment to
Conventional Activities
— — — — .004 (.006) — — .004 (.007) .010
Criminal Victimization — — — — — .332* (.074) — .324* (.073) .513*
State Depression — — — — — — .012* (.003) .001 (.003) .005
R2 .091 .092 .132 .092 .091 .370 .096 .376
?2 112.308 1,633.447 444.821 401.078 117.412 793.913 946.579 4,066.412
d.f. 34 98 55 76 36 170 145 602
p value .000 .000 .000 .000 .000 .000 .000 .000
RMSEA .013 .033 .022 .017 .013 .016 .020 .020
CFI .950 .903 .865 .973 .949 .875 .944 .914
SRMR .008 .019 .014 .010 .008 .020 .014 .023
ABBREVIATION: SE = standard error.
*p < .05 (two-tailed test).
significant explanation of the observed difference in violent crime between
the SBNR and the RAS as hypothesized (hypothesis 2), whereas the other
explanatory variables failed to account for the SBNR–RAS difference (see
models 3 to 6). That is, the SBNR were more likely to commit violent
crime than the RAS partly because the former tend to have lower levels of
religious involvement and self-control. The full model that includes all six
variables shows the RAS coefficient reduced by 11 percent, from –.009 to
–.008, and became nonsignificant. On the other hand, differences between
the SBNR and the other two groups, found to be nonsignificant in the
baseline model (–.002 and –.001), remained so across the intermediate and
full models.
In sum, the overall results provide partial support for our hypotheses.
First, as expected, SBNR was more likely to engage in violent crime
than RAS, but it was not significantly different from RBNS and NRNS.
Second, we found that only one third of our hypothesized variables (religious
involvement and low self-control) significantly explained the observed
SBNR-RAS difference. Although it failed to explain the difference,
the strain of criminal victimization had a stronger relationship with violent
criminal offending (ß = .513; see the full model) and explained it more than
any other explanatory variable, including religious/spiritual self-identity, in
the model (R2 of .279 = .370 to .091 compared with R2 of .285 = .376 to
.091 for all six variables combined).
Turning to property crimes, table 4 presents structural models estimated
in the same way as the models of violent crime. The baseline model shows
empirical support for hypothesis 1 not only for the hypothesized differences
involving the “religious” groups (–.048 and –.040) but also for the difference
involving the “neither” group (–.036). That is, SBNR respondents were
more likely to report property crimes than their NRNS peers as well as
the “religious” ones. The next six models, however, show no support for
the second hypothesis. Specifically, the three coefficients of group difference
remained significant across the models as we entered the explanatory
variables, whether one at a time (models 1 to 6) or simultaneously (full
model), whereas the size of coefficients somewhat reduced: RAS (e.g., 17
percent, from |–.048| to |–.040| in model 1), RBNS (e.g., 10 percent, from
|–.040| to |–.036| in model 1 and the full model), and NRNS (e.g., 11 percent,
from |–.036| to |–.032| in models 5 and 6).
In sum, we found full support for the first hypothesis with respect to
property crimes. As expected, the SBNR respondents were more likely to
commit the crimes than the RAS and RBNS counterparts, and they were
found to be different from the NRNS in the same direction as the “religious”
groups. In contrast, we found no support for the second hypothesis.
That is, the variables of self-control, social bonding, and general strain
theory failed to explain the observed group differences. In addition, the six
Table 4. Estimated Structural Models of Religious/Spiritual Self-identity and Property Crime
(N = 14,322)
Variables Baseline Model Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Full Model
b (SE) b (SE) b (SE) b (SE) b (SE) b (SE) b (SE) b (SE) ß
Race (White) -.008 (.007) -.010 (.007) -.005 (.007) -.006 (.007) -.008 (.007) -.003 (.007) -.006 (.007) .001 (.008) .001
Age -.012* (.002) -.012* (.002) -.012* (.002) -.012* (.002) -.012* (.002) -.011* (.002) -.012* (.002) -.011* (.002) -.106*
Sex (female) -.060* (.009) -.058* (.009) -.031* (.008) -.061* (.009) -.060* (.009) -.047* (.009) -.065* (.010) -.032* (.009) -.081*
Education .004* (.002) .005* (.002) .011* (.002) .005* (.002) .004* (.002) .006* (.002) .006* (.002) .011* (.002) .112*
Married -.026* (.005) -.025* (.005) -.014* (.006) -.026* (.005) -.026* (.005) -.023* (.005) -.025* (.005) -.013* (.006) -.024*
Third Person Present -.011 (.007) -.012 (.007) -.016* (.007) -.012 (.007) -.011 (.007) -.013 (.007) -.011 (.007) -.017* (.007) -.035*
Mainline Protestant -.001 (.009) -.005 (.009) -.006 (.010) .000 (.009) -.001 (.009) .001 (.009) -.001 (.009) -.000 (.009) -.001
Black Protestant .003 (.013) .002 (.013) .007 (.013) .003 (.013) .003 (.013) .000 (.011) .004 (.013) .004 (.012) .006
Catholic .009 (.009) .004 (.009) -.001 (.010) .010 (.009) .009 (.009) .010 (.009) .009 (.009) .005 (.010) .010
Jewish -.005 (.023) -.013 (.024) -.022 (.022) .001 (.022) -.005 (.023) -.001 (.023) -.009 (.023) -.009 (.022) -.004
Mormons -.016 (.013) -.016 (.013) -.008 (.014) -.014 (.013) -.016 (.013) -.017 (.012) -.018 (.013) -.012 (.014) -.006
Other Religion .018 (.016) .015 (.015) .013 (.016) .017 (.016) .018 (.016) .018 (.016) .016 (.016) .013 (.015) .014
None .011 (.008) -.001 (.011) .002 (.009) .008 (.008) .011 (.008) .011 (.008) .010 (.008) .005 (.011) .009
Affiliation Unknown .000 (.014) -.004 (.013) -.010 (.014) -.002 (.014) .000 (.013) -.002 (.014) .001 (.013) -.009 (.014) -.010
Religious and Spiritual -.048* (.011) -.040* (.012) -.044* (.011) -.044* (.011) -.049* (.011) -.047* (.011) -.047* (.011) -.042* (.012) -.101*
Religious but Not Spiritual -.040* (.014) -.036* (.013) -.041* (.014) -.038* (.013) -.040* (.013) -.037* (.014) -.038* (.013) -.036* (.014) -.044*
Neither Religious Nor
-.036* (.011) -.039* (.012) -.041* (.011) -.035* (.011) -.036* (.011) -.032* (.011) -.032* (.011) -.035* (.011) -.086*
Religious Involvement — -.015 (.009) — — — — — .002 (.010) .006
Low Self-control — — .102* (.020) — — — — .086* (.020) .167*
Attachment to Parents — — — -.021* (.004) — — — -.015* (.004) -.060*
Commitment to
Conventional Activities
— — — — .009 (.017) — — .023 (.018) .023
Criminal Victimization — — — — — .316* (.077) — .274* (.077) .160*
State Depression — — — — — — .041* (.009) .017 (.009) .035
R2 .050 .050 .084 .057 .050 .082 .057 .110
?2 269.059 1,753.679 589.637 549.925 286.038 930.133 1166.797 4,340.881
d.f. 53 121 76 98 56 196 170 644
p value .000 .000 .000 .000 .000 .000 .000 .000
RMSEA .017 .031 .022 .018 .017 .016 .020 .020
CFI .937 .901 .896 .965 .935 .885 .937 .912
SRMR .011 .019 .016 .012 .011 .020 .016 .023
ABBREVIATION: SE = standard error.
*p < .05 (two-tailed test).
explanatory variables jointly explained property criminal offending (R2 of
.060 = .110 to .050) less than a quarter (21 percent) of what they did for violent
offending (R2 of .285 = .376 to .091). This limited explanation might
be partly because of the data constraints that did not allow us to include
other theoretical variables relevant to property crimes, especially those of
social learning (Akers, 1998), like association with deviant peers, which was
expected to have substantial influence on behaviors during emerging adulthood
(Warr, 1998).9
Although researchers of religion, health, and well-being have increasingly
studied and have found utility in the “spirituality” concept as distinct
from religion or religiousness (e.g., Hill and Pargament, 2003; Idler et al.,
2003), criminologists have yet to add this dimension to the explanation
of crime. In this context, the current study intended to examine whether
individuals who say that they are “spiritual” while denying that they are
“religious” differ from others in criminal propensity, focusing on emerging
adulthood. Our findings tend to indicate that those respondents did not randomly
choose the term “spiritual” over “religious” for self-identification
and were more prone to commit violent crime than their “religiousand-
spiritual” peers and property crime than their “neither” as well as
“religious” counterparts. Furthermore, the difference in violent criminal
offending was found to be partly a result of the spiritual-but-not-religious
individuals being low on religious involvement and self-control, whereas
the difference in property offending failed to be explained by the variables
of religious involvement, self-control, social bonding, and general strain
The overall results show that claiming to be “spiritual” explained one’s
criminal propensity independently of being “religious.” Specifically, our
finding that the “spiritual-but-not-religious” individuals were more likely
to commit property crimes than their “neither” as well as “religious-butnot-
spiritual” peers implies that claiming to be “spiritual” is a potential
risk factor of criminal offending for emerging adults. With this in mind,
should we expect the “religious-and-spiritual” individuals to be more crime
9. When we added the variable of peer drinking (see footnote 4) to the model, the
results remained practically the same (see table S.2 in the online supporting information).
Although the effect of peer drinking on violent crime was found to be
significant (.005), the addition increased explained variance only by .010, from .376
to .386. In contrast, the effect on property crime was not even significant (–.002).
This weak finding for the alleged measure of social learning theory seems in line
with the editor’s and reviewers’ concern that the peer drinking variable would not
do justice to the theory.
prone than their “religious-but-not-spiritual” counterparts? No. In fact, we
found the former to bemore conventional (religious involvement and social
bonds) relative to the latter.10 This is because, as we suggested, self-claimed
spirituality is dualistic in meaning. Claiming to be “spiritual” while being
“religious” is denied as a risk factor for deviance if the claim connotes unorthodox
or “deviant” `a la Merton (1938) and antireligious beliefs and attitudes
(Marler and Hadaway, 2002; Zinnbauer et al., 1997). It is, however, a
protective factor when it is accompanied by claiming to be “religious,” indicating
an interest in and striving for the private dimension of faith—such
as self-transcendence and “ultimate concerns,” including life’s meaning and
purpose (Emmons, 2005).
To test this conceptual dualism of self-claimed spirituality in relation to
criminal behavior, we conducted two supplemental analyses. First, we replaced
the “religious” and “spiritual” dummy variables with the original
items of self-claimed religiousness and spirituality as continuous measures.
The results showed that neither the continuous measures nor their interaction
term was significant, although the replacement, in general, did not
affect the relationships between the explanatory variables and criminal offending
in terms of significance (see tables S.3 and S.4 in the online supporting
information). This indicates that our dichotomous measure of selfclaimed
spirituality, consistent with the concept’s dualism, was more construct
valid than its continuous counterpart.
Second, we constructed an alternative 2 × 2 typology of religious/spiritual
identity by reclassifying “slightly religious” and “slightly spiritual” as “religious”
and “spiritual” rather than not, coding only the “not at all” category
as “not religious” and “not spiritual.” Then, we reestimated models
using alternative dummy variables based on the reconstructed typology to
determine whether this operationalization would generate the identity concept’s
measure that had higher construct validity than what we had employed.
The results from estimating the alternative models of both violent
and property crimes showed no significant difference between the spiritualbut-
not-religious group and its comparison groups (see tables S.5 and S.6 in
the online supporting information). That is, the alternative typology failed
to reflect a more construct-valid measure of religious/spiritual identity than
the original.
10. We reestimated group differences in the explanatory variables as well as violent
and property crimes, using the religious-and-spiritual group as the reference
category to compare directly with the religious-but-not-spiritual group. The latter
was found to be lower than the former on religious involvement (–.326) and commitment
to conventional activities (–.159) but higher on low self-control (.061).
They were, however, not significantly different in attachment to parents, criminal
victimization, negative emotions, and violent and property crimes.
A closer observation revealed that the reclassification resulted in more
changes in the religious-and-spiritual and neither-religious-nor-spiritual
categories than the other two. Specifically, the proportion of the religiousand-
spiritual respondents in the sample increased by .151, from .379 to .530
(40 percent increase), whereas that of the neither-religious-nor-spiritual
decreased by .137, from .438 to .301 (31 percent decrease). In contrast,
the religious-but-not-spiritual and the spiritual-but-not-religious groups increased
(9 percent, from .068 to .074) and decreased (9 percent, from .115
to .105), to a lesser extent. In sum, the alternative typology had classified
more respondents as being “religious,” whether “spiritual” as well or not,
and fewer as being “spiritual but not religious” or “neither religious nor
spiritual.” This more lenient classification of respondents as being “religious”
and more strict classification of those as being “spiritual” eliminated
the significant group differences in violent and property offending observed
previously, and thus, the original typology seems to be better than the alternative
in terms of its resultant measure’s construct validity. Furthermore, a
contingency table analysis showed no significant (positive) relationship between
the two typologies (chi-square = 16.742, degrees of freedom [d.f.] =
9, p = .369), indicating that they were not even remotely similar.
Although not related to our hypothesis testing, an interesting observation
was that only one third of the “neither” group had no religious affiliation
(36.13 percent), whereas most of them reported they were either Protestant
(30.81 percent)—evangelical (14.28 percent), mainline (11.08 percent),
Black Protestant (5.45 percent)—or Catholic (24.19 percent), with still others
reporting their affiliation as Jewish (.79 percent), Mormon (.49 percent),
other (3.20 percent), or unknown (4.41 percent). Thus, for emerging adults,
being “neither religious nor spiritual” is unlikely to be coterminous with
whether they have a religious affiliation.
The current study is the first ever conducted to examine the relationship
between self-claimed spirituality as well as religiousness and criminal behaviors,
so future research should replicate our study not only for emerging
adulthood but also for other stages of life. When replicated, panel data
with short intervals would be best to establish causal ordering between the
self-identity and criminal offending, which we could not show with crosssectional
data. Also, although we included in our model variables reflecting
self-control, social bonding, and general strain theory as explanations of the
differences between the “spiritual but not religious” and other groups, some
key predictors of crime, especially social learning variables, could not be included
in our analysis. Future research should examine whether our limited
explanation of the group differences is partly a result of the omission of key
criminological variables.
Despite these limitations, the current study has a potential contribution
to criminology not only because it reports empirical evidence of a concept
prior research has largely ignored (i.e., spirituality as distinct from religiousness)
but also because it points to “a new criminological perspective
that focuses on human personhood (i.e., moral, believing animal), motivation,
and action, which often has been neglected by modern criminology
under the influence of the naturalistic, utilitarian, and noncultural tradition
of Western social theory (Smith, 2003)” (Johnson and Jang, 2010: 129).
Unlike criminologists and sociologists, other social scientists, especially
psychologists, have examined a broadly redefined concept of religion,
often called “spirituality,” finding the concept to be significantly related
to human behaviors, independent of religiousness (Emmons, 2005; Hill
and Pargament, 2003; Miller and Thoresen, 2003; Wong, 2012). The new
concept is unlikely to trump the key, more proximate causes of crime (e.g.,
control and strain variables) in the relative strength of “direct” explanation
as implied by the current findings, but it is likely to enhance our understanding
of crime causation as it not only adds more explanation, although
perhaps mostly indirect via the proximate causes, but also conditions relationships
between the proximate causes and crime. Thus, future research
on the new “spirituality” concept has a potential to contribute to the development
of “a systematic criminology of religion,” which Cullen (2010: 158)
recently called for, suggesting the “study of religion should be an integral
part of the criminological enterprise and a vibrant subfield within our
In conclusion, is being “spiritual” enough to reduce criminal propensity
without also being religious? Our study suggests the answer is no—at least
during emerging adulthood. That is, being “spiritual” without being religious
was found to be positively, rather than negatively, associated with the
probability of engaging in violent and, to a greater extent, property crimes.
In contrast, consistent with prior research on religion and crime (Johnson
and Jang, 2010), being religious was inversely related to criminal offending
whether or not in conjunction with being “spiritual.” Thus, according to the
current study, it would be a mistake to treat the term “spiritual” as criminologically
synonymous with “religious” given that they are likely to connote
different beliefs and attitudes.
Agnew, Robert. 2006. Pressured into Crime: An Overview of General Strain
Theory. Los Angeles, CA: Roxbury.
Akers, Ronald L. 1998. Social Learning and Social Structure: A General
Theory of Crime and Deviance. Boston, MA: Northeastern University
Arnett, Jeffrey J. 2000. Emerging adulthood:Atheory of development from
the late teens through the twenties. American Psychologist 55:469–80.
Arnett, Jeffrey J. 2005. The developmental context of substance use in
emerging adulthood. Journal of Drug Issues 35:235–53.
Baier, Colin J., and Bradley R. E. Wright. 2001. “If you love me, keep my
commandments”: A meta-analysis of the effect of religion on crime.
Journal of Research in Crime and Delinquency 38:3–21.
Baker, Joseph O., and Buster G. Smith. 2009. The nones: Social characteristics
of the religiously unaffiliated. Social Forces 87:1251–63.
Baraldi, Amanda N. and Kraig K. Enders. 2010. An introduction to modern
missing data analyses. Journal of School Psychology 48:5–37.
Chantala, Kim. 2006. Guidelines for Analyzing Add Health Data. Chapel
Hill: Carolina Population Center, University of North Carolina at
Chapel Hill.
Chatters, Linda M., Robert Joseph Taylor, Kai M. Bullard, and James
S. Jackson. 2008. Spirituality and subjective religiosity among African
Americans, Caribbean Blacks, and non-HispanicWhites. Journal for the
Scientific Study of Religion 47:725–37.
Cullen, Francis T. 2010. Toward a criminology of religion: Comment on
Johnson and Jang. In Contemporary Issues in Criminological Theory
and Research: The Role of Social Institutions, eds. Richard Rosenfeld,
Kenna Quinet, and Crystal Garcia. Belmont, CA: Wadsworth.
Dougherty, Kevin D., and Sung Joon Jang. 2008. Spirituality: Religion and
spirituality are not mutually exclusive. In What Americans Really Believe,
ed. Rodney Stark. Waco, TX: Baylor University Press.
Emmons, Robert A. 2005. Striving for the sacred: Personal goals, life meaning,
and religion. Journal of Social Issues 61:731–45.
Fuller, Robert C. 2001. Spiritual, but Not Religious: Understanding
Unchurched America. New York: Oxford University Press.
Gallup, George H., Jr. 2003. Americans’ Spiritual Searches Turn Inward.
Garelli, Franco. 2007. Research note between religion and spirituality: New
perspectives in the Italian religious landscape. Review of Religious Research
Giordano, Peggy C.,Monica A. Longmore, Ryan D. Schroeder, and Patrick
M. Seffrin. 2008. A life-course perspective on spirituality and desistance
from crime. Criminology 46:99–132.
Good, Marie, Teena Willoughby, and Michael A. Busseri. 2011. Stability
and change in adolescent spirituality/religiosity: A person-centered approach.
Developmental Psychology 47:538–50.
Gottfredson, Michael R., and Travis Hirschi. 1990. A General Theory of
Crime. Stanford, CA: Stanford University Press.
Graham, John W. 2009. Missing data analysis: Making it work in the real
world. Annual Review of Psychology 60:549–76.
Harris, Kathleen M., Carolyn T. Halpern, Eric Whitsel, Jon M. Hussey,
Joyce Tabor, Pamela Entzel, and J. R. Udry. 2009. The National Longitudinal
Study of Adolescent Health: Research Design. Chapel Hill: Carolina
Population Center, University of North Carolina at Chapel Hill.
Harrold, Philip. 2006. Deconversion in the emerging church. International
Journal for the Study of the Christian Church 6:79–90.
Hill, Peter C., and Kenneth I. Pargament. 2003. Advances in the conceptualization
and measurement of religion and spirituality—implications for
physical and mental health research. American Psychologist 58:64–74.
Hill, Peter C., Kenneth I. Pargament, Ralph W. Hood, Michael E.
McCullough, James P. Swyers, David B. Larson, and Brian J.
Zinnbauer. 2000. Conceptualizing religion and spirituality: Points of
commonality, points of departure. Journal for the Theory of Social Behaviour
Hirschi, Travis. 1969. Causes of Delinquency. Berkeley: University of
California Press.
Hirschi, Travis, and Rodney Stark. 1969. Hellfire and delinquency. Social
Problems 17:202–13.
Hodge, David R., Kathleen Andereck, and Harry Montoya. 2007. The protective
influence of spiritual-religious lifestyle profiles on tobacco use,
alcohol use, and gambling. Social Work Research 31:211–9.
Hodge, David R., Paul Cardenas, and Harry Montoya. 2001. Substance use:
Spirituality and religious participation as protective factors among rural
youths. Social Work Research 25:153–61.
Hu, Li-tze and Peter M. Bentler. 1999. Cutoff criteria for fit indexes in covariance
structure analysis: Conventional criteria versus new alternatives.
Structural Equation Modeling 6:1–55.
Idler, Ellen L., Marc A. Musick, Christopher G. Ellison, Linda K. George,
Neal Krause, Marcia G. Ory, Kenneth I. Pargament, Lynda H. Powell,
Lynn G. Underwood, and David R.Williams. 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.
Jang, Sung Joon, and Jeremy R. Rhodes. 2012. General strain and nonstrain
theories: A study of crime in emerging adulthood. Journal of
Criminal Justice 40:176–86.
Johnson, Byron R., and Sung Joon Jang. 2010. Crime and religion: Assessing
the role of the faith factor. In Contemporary Issues in Criminological
Theory and Research: The Role of Social Institutions, eds.
Richard Rosenfeld, Kenna Quinet, and Crystal Garcia. Belmont, CA:
King, Michael, Louise Marston, Sally McManus, Terry Brugha, Howard
Meltzer, and Paul Bebbington. 2013. Religion, spirituality and mental
health: Results from a national study of English households. British
Journal of Psychiatry 202:68–73.
Koenig, Laura B., Matt McGue, and William G. Iacono. 2008. Stability and
change in religiousness during emerging adulthood. Developmental Psychology
Laub, John H., and Robert J. Sampson. 2003. Shared Beginnings, Divergent
Lives: Delinquent Boys to Age 70. Cambridge, MA: Harvard University
Lim, Chaeyoon, CarolA.MacGregor, and RobertD. Putnam. 2010. Secular
and liminal: Discovering heterogeneity among religious nones. Journal
for the Scientific Study of Religion 49:596–618.
Marler, Penny Long, and C. Kirk Hadaway. 2002. “Being religious” or “being
spiritual” in America: A zero-sum proposition? Journal for the Scientific
Study of Religion 41:289–300.
Merton, Robert K. 1938. Social structure and anomie. American Sociological
Review 3:672–82.
Miller, Alan. 2012. “My take: ‘I’m spiritual but not religious’ is a cop-out.”
Miller, William R., and Carl E. Thoresen. 2003. Spirituality, religion, and
health—an emerging research field. American Psychologist 58:24–35.
Muth´ en, Bengt O. 1983. Latent variable structural equation modeling with
categorical data. Journal of Econometrics 22:43–65.
Muth´ en, Linda K. and Bengt O. Muth´ en. 1998–2012. Mplus User’s Guide,
7th ed. Los Angeles, CA: Muth´en & Muth´ en.
Pargament, Kenneth I. 1997. The Psychology of Religion and Coping: Theory,
Research, Practice. New York: Guilford Press.
Pepper, Miriam, Tim Jackson, and David Uzzell. 2010. A study of multidimensional
religion constructs and values in the United Kingdom. Journal
for the Scientific Study of Religion 49:127–46.
Pew Research Center. 2010. Religion Among the Millennials: Less Religiously
Active Than Older Americans, But Fairly Traditional In Other
Piedmont, Ralph L. 1999. Does spirituality represent the sixth factor of personality?
Spiritual transcendence and the five-factor model. Journal of
Personality 67:985–1013.
Piedmont, Ralph L., Anna T. Ciarrochi, Gabriel S. Dy-Liacco, and Joseph
E. G. Williams. 2009. The empirical and conceptual value of the spiritual
transcendence and religious involvement scales for personality research.
Psychology of Religion and Spirituality 1:162–79.
Piquero, Alex R., Robert Brame, Paul Mazerolle, and Rudy Haapanen.
2002. Crime in emerging adulthood. Criminology 40:137–70.
Roof,Wade Clark. 1993. A Generation of Seekers: The Spiritual Journeys of
the Baby Boom Generation. San Francisco, CA: HarperSanFrancisco.
Roof, Wade Clark. 1999. Spiritual Marketplace: Baby Boomers and the
Remaking of American Religion. Princeton, NJ: Princeton University
Sampson, Robert J., and John H. Laub. 1993. Crime in the Making: Pathways
and Turning Points through Life. Cambridge, MA: Harvard University
Sampson, Robert J., and John H. Laub. 1996. Socioeconomic achievement
in the life course of disadvantaged men: Military service as a turning
point, circa 1940–1965. American Sociological Review 61:347–67.
Saucier, Gerard, and Katarzyna Skrzypinska. 2006. Spiritual but not religious?
Evidence for two independent dispositions. Journal of Personality
Schlehofer, Michele M., Allen M. Omoto, and Janice R. Adelman. 2008.
How do “religion” and “spirituality” differ? Lay definitions among
older adults. Journal for the Scientific Study of Religion 47:411–25.
Simmel, Georg. 1997. Essays on Religion. New Haven, CT: Yale University
Smith, Christian. 2003. Moral, Believing Animals: Human Personhood and
Culture. New York: Oxford University Press.
Smith, Christian. 2005. Soul Searching: The Religious and Spiritual Lives of
American Teenagers. New York: Oxford University Press.
Smith, Christian. 2009. Souls in Transition: The Religious and Spiritual Lives
of Emerging Adults. New York: Oxford University Press.
Steensland, Brian, Jerry Park, Mark D. Regnerus, Lynn Robinson, W.
B. Wilcox, and Robert Woodberry. 2000. The measure of American
religion: Toward improving the state of the art. Social Forces 79:
Storm, Ingrid. 2009. Halfway to heaven: Four types of fuzzy fidelity in
Europe. Journal for the Scientific Study of Religion 48:702–18.
Uecker, Jeremy E., Mark D. Regnerus, and Margaret L. Vaaler. 2007. Losing
my religion: The social sources of religious decline in early adulthood.
Social Forces 85:1667–92.
Uggen, Christopher. 2000. Work as a turning point in the life course of criminals:
A duration model of age, employment, and recidivism. American
Sociological Review 65:529–46.
Underwood, Lynn G., and Jeanne A. 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.
Warr, Mark. 1998. Life-course transitions and desistance from crime. Criminology
Wink, Paul, Lucia Ciciolla, Michele Dillon, and Allison Tracy. 2007. Religiousness,
spiritual seeking, and personality: Findings from a longitudinal
study. Journal of Personality 75:1051–70.
Wink, Paul, and Michele Dillon. 2003. Religiousness, spirituality, and psychosocial
functioning in late adulthood: Findings from a longitudinal
study. Psychology of Religion and Spirituality 18:916–24.
Wong, Paul T. P. 2012. The Human Quest for Meaning: Theories, Research,
and Applications. New York: Routledge.
Wuthnow, Robert. 1998. After Heaven: Spirituality in America since the
1950s. Berkeley: University of California Press.
Zinnbauer, Brian J., Kenneth I. Pargament, Brenda Cole,Mark S. Rye, Eric
M. Butter, Timothy G. Belavich, Kathleen M. Hipp, Allie B. Scott, and
Jill L. Kadar. 1997. Religion and spirituality: Unfuzzying the fuzzy. Journal
for the Scientific Study of Religion 36:549–64.
Zinnbauer, Brian J., Kenneth I. Pargament, and Allie B. Scott. 1999.
The emerging meanings of religiousness and spirituality: Problems and
prospects. Journal of Personality 67:889–919.
Zwissler, Laurel. 2007. Spiritual, but religious: “Spirituality” among religiously
motivated feminist activists. Culture and Religion 8:51–69.
Sung Joon Jang is an associate professor of sociology and a distinguished
senior fellow in the program on prosocial behavior in the Institute
for Studies of Religion at Baylor University. His research focuses on
the effects of family, school, peers, religion, and spirituality on crime and
deviance. Recent publications have appeared in Journal of Criminal Justice,
Deviant Behavior, Youth & Society, and Sociological Focus. He is coprincipal
investigator on a longitudinal study examining the long-term effectiveness
of seminary programs in maximum security prisons.
Aaron B. Franzen is a PhD candidate in the Department of Sociology
at Baylor University. His research focuses on links between sociology of
religion and personal identity, criminology, and more recently, the interaction
between physicians and patients. His recent study on the impact that
the relationship among identity, context, and Bible reading has on moral
and political views is forthcoming in Review of Religious Research.
Additional Supporting Information may be found in the online version
of this article at the publisher’s web site:
Table S.1. Items Used for Analysis
Table S.2. Estimated Structural Models of Religious/Spiritual Self-identity
and Violent/Property Crime: Peer Drinking Included (n = 14,322)
Table S.3. Estimated Structural Models of Religious/Spiritual Self-identity
(as continuous) and Violent Crime (n = 14,322)
Table S.4. Estimated Structural Models of Religious/Spiritual Self-identity
(as continuous) and Property Crime (n = 14,322)
Table S.5. Estimated Structural Models of Religious/Spiritual Self-identity
(Alternative Measure) and Violent Crime (n = 14,322)
Table S.6. Estimated Structural Models of Religious/Spiritual Self-identity
(Alternative Measure) and Property Crime (n = 14,322)