You are on page 1of 31

RELIGION AND CRIME REEXAMINED: THE

IMPACT OF RELIGION, SECULAR


CONTROLS, AND SOCIAL ECOLOGY ON
ADULT CRIMINALITY*

T. DAVID EVANS
University of North Carolina at Wilmington

FRANCIS T. CULLEN
University of Cincinnati

R. GREGORY DUNAWAY
Mississippi State University

VELMER S. BURTON, JR.


Washington State University

Since Hirschi and Stark's (1969) surprising failure to find religious


("hellfire") effects on delinquency, subsequent research has generally
revealed an inverse relationship between religiosity and various forms
of deviance, delinquency, and crime. The complexity of the relation-
ship and conditions under which it holds, however, continue to be
debated. Although a few researchers have found that religion's infiu-
ence is noncontingent, most have found support—especially among
youths—for effects that vary by denomination, type of offense, and
social and/or religious context. More recently the relationship has been
reported as spurious when relevant secular controls are included. Our
research attempts to resolve these issues by testing the religion-crime
relationship in models with a comprehensive crime measure and three
separate dimensions of religiosity. We also control for secular con-
straints, religious networks, and social ecology. We found that, among
our religiosity measures, participation in religious activities was a per-
sistent and noncontingent inhibiter of adult crime.

In their landmark study, "Hellfire and Delinquency," Hirschi and Stark


(1969:210) found that, contrary to expectations, frequent church attenders
and "students who believe in the Devil and in life after death are as likely

* This study was supported by a grant from the University Research Council and
by funding from the Departments of Criminal Justice and Sociology, University of
Cincinnati. We thank colleagues at the University of North Carolina, Wilmington and
several anonymous reviewers for their valuable comments.

CRIMINOLOGY VOLUME 33 NUMBER 2 1995 195


196 EVANS ET AL.

to commit delinquency as are students who do not believe in a supernatu-


ral world." Hirschi and Stark were not the first to question the link
between "hellfire" and crime (Barnes and Teeters, 1951; Bonger, 1916;
Falk, 1961; Fitzpatrick, 1967; Kvarceus, 1944; Lombroso, 1911; Reckless
and Smith, 1932; Tappan, 1949), but their empirical data were provocative
and prompted a line of research that continues today.
In contrast to Hirschi and Stark's findings, later research revealed a reli-
gious impact on various forms of deviance, including crime (for early
examples, see Albrecht et al., 1977; Higgins and Albrecht, 1977; Jensen
and Erickson, 1979; Peek et al., 1985; Rohrbaugh and Jessor, 1975). The
precise effects of religion on crime, however, are complicated and, some
scholars argue, present only under certain circumstances. As Grasmick et
al. (1991b:251) have noted, "sociologists have concluded that at least some
aspects of religion inhibit at least some kinds of illegal behavior at least
under some conditions."
Based on existing research, and employing a self-report crime survey of
adults, our study was designed to bring data to bear on various debates
and underresearched issues. Toward that end, we examine (1) the extent
to which religion, independently, or in combination with other factors,
inhibits adult crime; (2) the extent to which social and religious contexts
mediate the effects of personal religiosity; and (3) whether the effects of
religion are general across crime types or specific to crimes in violation of
religious ascetic standards.
Our data are especially suitable for addressing these issues. First, com-
pared with most previous research, we included comprehensive measures
of religion—at the individual and group levels—and a wide range of adult
criminality. Second, we also included variables used in previous research
to specify the religion-crime relationship. Accordingly, we were able to
construct models that estimate whether religiosity and religious contexts
inhibit criminal involvement when controls for several other relevant fac-
tors are introduced.

CONSENSUS AND CONTROVERSY


While research supports consensus on a weak-to-moderate inverse rela-
tionship between religiosity and criminality (especially ascetic juvenile
offenses), there are still several major points of controversy. Since Hirschi
and Stark's (1969) "hellfire" study, research has progressed by examining
the independent effects of religion on general crime or delinquency, varia-
tion in effects by offense type and by denominational affiliation, and the
conditioning of direct effects by broader religious and social contexts.
RELIGION AND CRIME REEXAMINED 197

VARIATION BY OFFENSE TYPE: ANTIASCETIC ,


SPECIFICATION

Failure to find consistent, strong religious effects on general crime moti-


vated early efforts to detect limited and/or conditional effects. In a repli-
cation of Hirschi and Stark (1969), Burkett and White (1974) found
especially strong religious effects for "victimless crimes" that also violated
religious "ascetic" traditions (cf. Middleton and Putney, 1962). Many later
studies also reported similar support for the antiascetic deviance thesis,
particularly in the area of teenage abuse of alcohol and drugs (e.g.,
Albrecht et al., 1977; Bock et al., 1987; Burkett, 1977; Elifson et al., 1983;
Hadaway et al., 1984; McLuckie et al., 1975; Nelson and Rooney, 1982;
Rohrbaugh and Jessor, 1975). Such findings have often been interpreted
as evidence for strong religious effects only on behavior that is uniquely
proscribed within a religious context.
The ascetic specification has been tested in a number of different ways.
For example, following Middleton and Putney (1962), Burkett and White
(1974) suggested that special religious effects on ascetic crime might be
discovered either by controlling for bonds arising from secular institutions
or by testing specifically for effects on the allegedly unique ascetic
offenses. Although they chose the later approach and found such effects
for ascetic crime, absent a wide range of crime—both ascetic and secu-
lar—any claim for such limited and unique effects lacks merit. Cochran
(1988) also criticized the literature for making a "semantic error": Follow-
ers of Middleton and Putney (1962), including Burkett and White (1974),
failed to clearly distinguish ascetic from victimless offenses. Even so,
neither Cochran (1988) nor subsequent researchers of the religion-ascetic
crime link have empirically grounded and validated the ascetic crime con-
cept in a manner that clearly establishes it as a distinct type of crime. In
our investigation, our respondents self-reported involvement in a wide
range of what have been called ascetic, as well as secular, crimes.

VARIATION BY DENOMINATION: PROSCRIPTIVE MESSAGES

Following research on differentiation of effects by offense type, investi-


gators focused on the consequences of membership in conservative reli-
gious groups. Such denominations presumably broadcast proscriptive
moral messages, inculcating their members with moral values that inhibit
crime. Findings for the moral messages model have been mixed. Some
researchers have found that denomination "makes a difference" (e.g..
Bock et al., 1987; Cochran et al., 1988; Hadaway et al., 1984; Nelson and
Rooney, 1982). Others have found little support (Cochran and Akers,
1989; Linden and Currie, 1977; Mclntosh et al., 1981; Tittle, 1980). Since
198 EVANS ET AL.

the data for our study included denominational affiliation, we were able to
examine such direct group-level religious effects on criminality.

VARIATION BY RELIGIOUS AND SOCIAL CONTEXT: MORAL


COMMUNITIES
The moral communities model suggests that it is neither the degree of
personal religiosity, nor the type of offense (ascetic or secular), nor even
religious affiliation that matters so much as community-level religiosity.
For example. Stark and his associates (Stark et al., 1980, 1982) claimed
that Hirschi and Stark's (1969) failure to find a hellfire factor in delin-
quency, in contrast with "commonsense" and with much earlier research,
was due to differences in the moral climates of the places from which sam-
ples were drawn (cf. Higgins and Albrecht, 1977). Stark et al. (1982:7)
asserted that "religion only binds people to the moral order if religious
infiuences permeate the culture and the social interactions of the individu-
als in question." Thus, strong personal religious effects would not be
expected in communities characterized by a low level of religiosity.
Tittle and Welch (1983), however, reported that for adults, at least, indi-
vidual religiosity is most salient in secular communities characterized by
disorganization and general normative ambiguity and minimally effective
in socially integrated, religious communities. In organized communities,
personal religious morality is redundant because there are other adequate
sources of moral authority and social control.
The finding of broad, albeit contingent effects, is in sharp contrast to the
limited ascetic specification discussed above. For example. Stark
(1987:111) expressed dismay that in his early work on religion and crime
and delinquency "about the only religious effects I could find were corre-
lations between orthodoxy and opposition to drinking, dancing, and gam-
bling [ascetic offenses] among American Protestants. Whenever I
searched for religious effects on behavior or attitudes more remote from
religiousness per se, I found little or nothing."

ADDITIONAL ISSUES
Aside from the issues that have been raised in the major specifications
reviewed above, three other matters are relatively unexplored and unset-
tled: (1) the possibility of different effects for adults than for juveniles,
(2) the relative efficacy of religion as an inhibitor of crime, and (3) the
operationalization of crime and religion.

ADULT VERSUS JUVENILE SAMPLES


The vast majority of research has been conducted with juvenile samples.
RELIGION AND CRIME REEXAMINED 199

yet there is reason to suspect that the relationship may not be the same for
adults. Welch et al. (1991:161) claim that since the
content, intensity, and expression of adult religiosity . . . are strongly
conditioned by the characteristic of the religious community in which
an individual functions . . . it seems likely that contextual variation in
the inhibiting effects of religion may appear in data collected from
adult samples, even though it may be absent from samples of youth.
Further, the religious involvement of youth may be reflective of parental
and peer infiuences, thus confounding the religion-delinquency association
(Burkett and Warren, 1987; Elifson et al., 1983; Tittle and Welch, 1983).
Church attendance may also be "less than voluntary" for youth and, there-
fore, an unreliable indicator of religious commitment (Bock et al., 1987).
Thus, "research on adults . . . potentially more exclusively tests religiosity's
effect" (Grasmick et al., 1991b:252). Unfortunately, there are few studies
of adult deviance and religion, and even fewer that assess the relationship
for adult crime.
In their review. Tittle and Welch (1983) reported only six studies of reli-
gion and adult deviance of all types, all of which found moderate-to-strong
inverse relationships. Of the studies of religion and adult criminality since
Tittle and Welch's 1983 review, four used ecological data sets (Bainbridge,
1989; Pettersson, 1991; Stack and Kanavy, 1983; and Stark et al., 1980) to
test the moral communities thesis. In general, they found that church
membership rates correlate negatively with crime: Crime rates are gener-
ally lower in regions of the country where the religious ecology is dense.
Aside from the ecological studies, and more germane to our research,
five other investigations used adult samples with individual-level and/or
contextual data to test the relationship between deviance and religiosity.
For example. Tittle (1980) found that self-reported probability of future
engagement in nine deviant acts varied inversely by church attendance.
As noted above. Tittle and Welch (1983) found that levels of secularity
and social disorganization mediated the association between religion and
crime.
In another specific test of the moral communities hypothesis, Welch et
al. (1991) found that parish-level religiosity was directly associated with
individual deviance for their all-Catholic sample. However, no interaction
was found between individual and aggregate parish-level religiosity. In
contrast, as reported above. Stark et al. (1982) found stronger crime-inhib-
iting effects for individual religiosity in "religious/moral communities"
than in secular ones.
Focusing on a single form of criminal behavior, self-reported "inclina-
tion to cheat on taxes," Grasmick et al. (1991b) traced the process
whereby religion may act as a sanctioning system. Rather than fears of
200 EVANS ET AL.

hellfire, these investigators posited that shame, based on internalization of


religious values, and embarrassment, imposed by others in a religious net-
work, are the religious sanctions that curb crime. In another study of reli-
gious effects on one form of adult deviance. Bock et al. (1987) established
that religiosity predicts the use, but not abuse, of alcohol, and that the
effects of religiosity on alcohol use are greater for affiliates of conservative
denominations.
In summary, while research with adult samples is just beginning, limited
support has been found for religiosity as a moderate inhibitor of adult
crime and deviance. Our study continues this trend toward the use of
adult samples.

THE RELATIVE EFFICACY OF RELIGION: SECULAR


CONSTRAINTS
While simply using an adult sample may reduce the possibility of some
spurious effects (due to, for example, parental influences and coerced
church attendance for juveniles), the power of religion as a personal con-
trol relative to other specific forms of secular controls—such as informal
peer and family social controls and formal legal deterrents—has not been
explored for adults. Even for youths, Albrecht et al. (1977:265) warned
that "the debate concerning the efficacy of religious attitudes . . . in pre-
dicting delinquent behavior becomes a moot point. . . researchers should
be developing more complex models of delinquency in which religious
attitudes are treated in combination with other [social control] factors."
The relatively few multivariate studies (e.g., Elifson et al., 1983) with
juvenile samples have found that personal religiosity loses influence when
nonreligious moral effects of such bonds as attachments to peers and par-
ents are controlled (see also, Cochran et al., 1994). Burkett and Warren
(1987) speculated that religiosity has only indirect influences—through
selection of peers, which in turn may dampen delinquency. To determine
the relative effects of religiosity for adults, our research incorporates for-
mal (legal deterrents) and informal social/secular constraints in mul-
tivariate models.

MEASUREMENT PROBLEMS: CRIME AND RELIGIOSITY


In addition to divergent theoretical approaches and model specifications
addressed above, inconsistencies in findings and controversies about the
strength and conditioning of religious effects may stem from methodologi-
cal problems (e.g.. Tittle and Welch, 1983, cited such flaws in the early
research). First, much of the religion and delinquency research has
examined the impact of personal religiosity on deviant behavior that can
hardly be termed "criminal." As noted above, most of the research has
RELIGION AND CRIME REEXAMINED 201

focused on ascetic and victimless offenses. In fact, the range of items used
to represent delinquency and crime has generally been quite limited as a
sampling of potential illegalities. For example, in recent research on
adults the effects of religion have been examined for alcohol use and
abuse, tax cheating, and littering and theft (Bock et al., 1987; Grasmick et
al., 1991a, 1991b). Agreeing with Stark (1987:117), we believe that the
"acid test of the moral integration thesis rests on demonstrating that reli-
gion prevents acts of criminal non-conformity."
The measurement of religiosity in most prior work is also potentially
problematic. As with delinquency and crime, indicators of religiosity have
usually been restricted. Multiple-item indicators are rare in the literature,
suggesting that either religiosity is unidimensional or that church attend-
ance, the most common measure, serves as an adequate proxy for religios-
ity. In fact, several researchers have acknowledged the deficiencies of
their religiosity measures—usually attendance or religious salience—in
tapping the full domain of the concept (Cochran, 1989; Higgins and
Albrecht, 1977; Rohrbaugh and Jessor, 1975; Tittle and Welch, 1983;
Welch et al., 1991). Still, truncated measures have been justified as good
single-item proxies for religious commitment.
Research in the sociology of religion indicates that since religiosity is
multifaceted it may only be well represented by use of multiple-item
indicators (Davidson and Knudsen, 1977; Glock, 1959; King and Hunt,
1972; Stark and Glock, 1968) and different dimensions. While reliability
cannot be assessed with single-item measures, validity is also in doubt.
Since both religion and crime appear to be more multidimensional than
previously represented, we included a wide range of both key variables.
This strategy allowed us to assess the effects of diverse religiosity indica-
tors across a variety of crimes.

METHODS
SAMPLE
Data on adult criminality were gathered through a self-report survey of
the general population, aged 18 and older, residing in a midwestern, urban
area. Ouestionnaires were randomly sent to 1,500 individuals within our
sampling frame. Following Dillman's (1978) "Total Design Method," sam-
pled individuals were sent a reminder letter shortly after the anticipated
arrival date of the initial questionnaire mailing. Subsequently, two addi-
tional mailings of questionnaires were sent to nonrespondents. On the
fifth mailing, we included a pen as an incentive to complete the survey.
Finally, a private firm was hired to call each nonrespondent to solicit par-
ticipation. Another questionnaire was mailed to those individuals who
agreed to participate in the survey.
202 EVANS ET AL.

For various reasons (e.g., change of address, death), 303 questionnaires


could not be delivered. Of the 1,197 delivered surveys, 555 completed
surveys were returned, a response rate of 49%.i Our response rate for
white subjects, however, appears to exceed 60%, and thus our analysis is
confined to whites. We return to this issue below.
The sample generally represents the community from which it is drawn
on a number of key attributes. The median age for the population under
analysis (individuals 18 years of age and older) is 40.5, and the median
sample age is 41 (Bureau of Census, 1992). Moreover, males over age 17
constitute 45% of the population, whereas they are 42% of the sample.
Our sample's economic characteristics approximate the population: Per-
sonal median income is nearly $23,000, while the population's personal
median income is $21,006; median family income is $30,000 and the popu-
lation's median family income is $26,774.
The sample's main limitation, however, is that it underrepresents racial
minorities: Nonwhites constitute 35% of the community but only 14% of
the sample. Accordingly, we decided to restrict the analysis reported here
to whites. In this regard, it is possible to estimate the approximate
response rate for whites in our sample. As noted earlier, 1,197 people
were eligible to return surveys. According to the U.S. Census, the commu-
nity is 65% white and 35% nonwhite. Thus, of the 1,197 possible respon-
dents, we would expect 778 whites (65%) and 419 nonwhites (35%). Since
we received 477 surveys from whites, their estimated response rate is 61%
(477 of 778 possible respondents). In contrast, only 78 of the expected 419
nonwhite respondents returned a completed survey, a response rate of
19%. As a result, we confined the analysis to white respondents.2

CRIME MEASURE
Since self-report crime surveys have been criticized for restricting crime
to less serious offenses, we derived our crime scale from the diverse set of
items used by Elliott et al. (1983) in their comprehensive National Youth
Survey. While retaining the essence of their work, we eliminated those
items that referred to status offenses and delinquent acts that do not con-
stitute adult crimes. In addition, we adapted their school delinquency

1. Since those individuals failing to receive a survey could not participate, it is


acceptable to use such a net sample size of potential respondents in the calculation of
response rates (Babbie, 1980; Frey, 1989; Sosdian and Sharp, 1980).
2. The response rate for whites is, as noted, an estimate. If nonwhites were more
likely to have moved and thus been a disproportionate percentage of the 303 respon-
dents to which a survey was nondeliverable, then the base of eligible white respondents
would be larger than 778; if so, the response rate reported here would be a high esti-
mate. In any case, however, the response rate for whites appears to be in excess of
50%.
RELIGION AND CRIME REEXAMINED 203

items to measure adult workplace crime and added white-collar crimes to


form the crime scale. (See Appendix 1 for scale items and reliability.)
Crime scores were calculated by asking respondents how many times
during the past 12 months they had committed each of 43 diverse criminal
acts. The response sets were open-ended. To preserve the full range of
criminal involvement, while preventing relatively minor offenses from
overweighting and distorting the scale,3 a factor (weighted) score was
derived for each respondent on each of the crime items by multiplying the
standard score by the factor score coefficient. By allowing each item to
represent its relative contribution to the scale, this procedure helped
improve the distributions and increased the scale's reliability, compared
with a simple additive scale of raw frequencies.^ Finally, all 43 factor-
weighted items were combined into a composite crime index.

RELIGION MEASURES
Religion was measured as personal religiosity, denominational affilia-
tion, and interpersonal religious networks. (See Appendix 1 for religious
measures and reliabilities.)

RELIGIOSITY

Most prior research on religion and crime has used only one-or two-
item indicators of religiosity—usually church or Sunday school attend-
ance. Three multi-item dimensions of religion were included in this study
to form religiosity scales: religious activity, religious salience, and "hell-
fire" beliefs. These aspects of religiosity capture important involvement,
attachment, and belief elements identified in research on the measurement
of religiosity. We also developed a general religiosity scale composed of
all religious items.s All religiosity scales were constructed with factor-

3. Since minor offenses were both more likely to be reported and engaged in
more frequently, a simple additive approach to frequency of criminal involvement
yields a "general crime" scale in which minor, and especially victimless, acts dominate.
Further, and for the same reason, failure to use a weighting procedure might lead to the
warrantless claim that religion's power to inhibit general crime is greater than is actu-
ally the case.
4. Logarithmic transformations were also performed on the highly positively
skewed crime items prior to applying factor weights. This procedure resulted in sub-
stantial reductions in skewness and kurtosis. Examination of residuals and univariate
and multivariate plots indicated that the frequency distributions of the crime variables
were more normal with more equally distributed variances, both assumptions underly-
ing ordinary least squares regression. Other salutary effects included the improvement
in linearity of relationships and reduction in extreme outliers, either of which could also
deteriorate the flt of the model to the data and distort the parameter estimates.
5. Factor analysis revealed that all of the religiosity items loaded well (ranging
from .41 to .84 and with a mean loading of .67) on the general religiosity scale of all
204 EVANS ET AL.

weighted items, using the same procedure as described above for the crime
scale. This strategy enabled us to test for potential unique effects for each
separate aspect of religiosity as well as for the general religiosity scale.
Religiosity variables were recoded as necessary so that a high score indi-
cates a high degree of religiosity. The response sets and dimensions are
described below.

Religious Activity. Respondents were asked to indicate past year


attendance at religious services and social events, reading of religious
material, and listening to religious broadcasts. Possible scores on each
item ranged from 1 ("about every week") to 5 ("never"). The score repre-
sents a frequency category, not total frequency over the past year.

Religious Salience. This dimension, which indicates the practical infiu-


ence of religion in daily life, was measured by "agreement-disagreement"
(along a 6-category scale) to statements regarding the extent to which reli-
gious beliefs have impacts on daily behavior and the degree to which one
refers to a set of religious beliefs or to a religious community in daily life.

Hellfire. "Hellfire" was used by Hirschi and Stark (1969) to indicate


specific beliefs in and fear of supernatural sanctions. In this study, we
measured a similar dimension with respondents indicating 1 of 6 categories
of agreement or disagreement with the following beliefs: evil people will
suffer in hell, God punishes those who have sinned, God is omniscient,
personal fear of God's punishment for wrongdoing, there is life after
death, and AIDS sufferers are being punished for sins.

DENOMINATIONAL CONSERVATISM

As a measure of proscriptive moral messages, respondents were asked


to identify their religious denomination. Protestant denominations were
ranked along a conservative-moderate-liberal continuum based on catego-
ries devised by Smith (1990) for the General Social Survey. In accord with
this same scheme (Smith, 1990), Catholics were ranked as moderates and
Jews were classified as liberals. (See Grasmick et al., 1991a, and Elifson et
al., 1983, for similar classifications.) Members of conservative denomina-
tions were given the highest rank score of 3, while liberals were coded 1.

items. In addition, a three-factor solution generally confirmed the three separate postu-
lated dimensions. One item in the hellfire scale, "There is life after death," loaded
moderately on hellfire but more heavily on salience. Nonetheless, because we surmised
that a belief in an afterlife is necessary for fear of hellfire to be salient, the item was
retained as part of the hellfire scale. At any rate, the results did not change substan-
tively when this item was deleted from hellfire and added to salience.
RELIGION AND CRIME REEXAMINED 205

INTERPERSONAL RELIGIOUS NETWORKS

Two indices measure the extent to which respondents are embedded in


religious networks. We first asked respondents to estimate how many
among their five closest friends and family members attend church regu-
larly (every week or every other week). As a second, more distant mea-
sure of religious context, we asked respondents to estimate the number of
their neighbors who attend church regularly (with the response set ranging
from 1 to 4, or nearly none to nearly all) and the overall religiosity of their
neighbors (a 6-category, agreement-disagreement response set to the
statement that "Most of the people in my neighborhood are religious").

SECULAR CONSTRAINTS

Religious infiuences do not operate in a vacuum, especially in modern,


secular societies. Some studies have found that secular sources of morality
attenuate the religion-delinquency relationship (Albrecht et al., 1977;
Cochran et al., 1994; Elifson et al., 1983). Thus, other social and political
forces—informal social constraints and more formal legal deterrents—were
controlled. (Refer to Appendix 1 for a complete list of scale items and
reliabilities.) Drawing on previous research on control theory (Eve, 1978;
LaGrange and White, 1985; Rosenbaum, 1987; Simons et al., 1980), infor-
mal social constraints were indicated by a factor-weighted, summed scale
of responses to a 6-category, agree-disagree set of statements expressing
respect for father and mother, quality of relationships with parents, and
the probability that friends would intervene to keep one from breaking the

Another form of secular constraint was represented by items gauging


the impact of formal legal deterrents (that is, fear of detection, apprehen-
sion, and sanctioning by the law). Respondents were assigned a legal
deterrence score based on their summed responses to each statement in
the 6-category, agree-disagree scale. As with other scales, we used a fac-
tor-weighting procedure.

6. We recognize that for adults anticipated social disapproval of friends and


strong attachments to parents, held to foster indirect control (Rankin and Wells, 1990),
does not necessarily mean that such intimates will actually sanction wayward behavior.
Still, these items cohered well as a scale (reliability = .82) and had significant effects in
all of the multivariate models. We acknowledge that the analysis would have been
strengthened by indicators of "adult social bonds," such as attachments to spouse and
work (see Sampson and Laub, 1993), but our data did not have good measures of these
bonds. We suggest that future research on religion and crime seek to incorporate more
appropriate measures of adult bonds.
206 EVANS ET AL.

SOCIAL ECOLOGY MEASURES


While religious context may have direct or interactive effects with per-
sonal religiosity and alternative sources of morality may render religion
redundant, it has also been reported that religiosity's effects on crime are
conditioned by social context. Social ecologies in which religiosity may be
more or less salient were measured by (1) respondents' subjective percep-
tions of social integration in their neighborhoods and (2) objective census-
tract measures of community-level disorganization. (See Appendix 1.)
Specifically, perceived social integration was indicated by agreement or
disagreement with statements about the organization, order, and civility of
the neighborhood in which respondents resided (Skogan, 1990). The
responses were factor-weighted and then summed to form the social inte-
gration scale.
In addition to neighborhood ecology as indicated by perceptions of
social integration, social ecology was measured by two U.S. Census struc-
tural measures: percent female-headed households and average percent
renter-occupied housing units. These figures were matched with the
respondents residing in the corresponding census tracts. We acknowledge
that census-tract-level measures may not constitute ideal ecological units
or "reference groups" comparable to neighborhoods. Such census-tracts
also may overlap, but seldom duplicate, "real living neighborhoods."
Nonetheless, census tracts are more proximate community-level reference
groups than most other groups used in the limited research on ecological
effects. (For a notable exception, see Welch et al., 1991.)
Our rationale for using such objective measures of social disorganiza-
tion is derived in part from Sampson (1985, 1986), who used percent
female-headed households as a key indicator of family instability and
neighborhood disorganization. Sampson (1986:278) claimed that "high
levels of family disruption in a community may interfere with individual
and collective efforts of families to link youth to the wider society [that is,
to achieve social integration] through institutional means such as schools,
religion and sports" (emphasis added). Informal controls are also weak-
ened by family dissolution because there are fewer adult caretakers in the
community to maintain surveillance over potentially wayward members of
the community.
Our other census-derived measure of social integration, the proportion
of rental housing units in respondents' census tracts, represents socioeco-
nomic status of the neighborhoods. An increase in rental housing relative
to homeowner-occupied units is a partial indicator of social disorganiza-
tion (Shaw and McKay, 1931). High levels of rental housing also indicate
population turnover, lower neighborhood stability, and transience. Such
population turnover may ultimately affect the ability of the community to
RELIGION AND CRIME REEXAMINED 207

organize institutions of formal or informal social control, including


churches.
Of relevance to both of our objective census measures of social ecology,
Taylor and Covington (1988:561) proposed that
declines in stability [indicated by population turnover and home own-
ership] or familism will be most strongly linked to increasing violence.
Decreased stability or familism, as indicated by lower proportions of
married-couple households and home ownership, increases anonymity,
thereby impeding the development of local standards or informal
social control (emphasis added).
We add that such instability may also mitigate against the formation of
churches and the maintenance of cohesive, long-term religious groups in
the community.
DEMOGRAPHIC CONTROL VARIABLES
Sex, age, and family income were used as control variables in the subse-
quent analyses because they may be related to both variables of direct
interest—crime and religion. Sex was coded 1 for male, and family
income is the combination of respondent's income and spouse's income.

RESULTS
Table 1 reports the results for three separate regression models of gen-
eral religiosity and crime.^ In the first model, general religiosity—our
composite measure of all religiosity items—has a significant negative
effect on crime. As the model is more fully specified, with controls for
secular constraints, personal religiosity is rendered statistically nonsignifi-
cant. In the third, fully specified model, no religious variable has any sig-
nificant impact on crime.s Variation explained in the models of general
religiosity and general crime ranged from 21% to 24%.
In order to discover potential differential effects by the three religiosity
dimensions, fully specified models were separately estimated for each
dimension (see Table 2).9 Two of the three religiosity subscale measures—
hellfire and salience—had no significant effects on general crime when all
7. Diagnostic tests for multicollinearity, requested as part of the SPSS regression
output for all models in the analyses, indicated that intercorrelations among predictor
variables did not appear to unduly influence the results.
8. For reasons discussed earlier, we restricted our analysis to whites only. As
noted, we believe that the estimates for nonwhites in our sample may be unreliable due
to small numbers of respondents, especially relative to the number of predictors in the
full models. In addition, the low response rate of this group makes us skeptical of
including it in the analysis or of making any inferences to the nonwhite population.
9. As will be seen later in the results, the "best" religious predictor—in fact the
only significant one among the religiosity measures—was religious activities. Due to
208 EVANS ET AL.

Table 1. The Effects of Religion on General Crime (Betas


Reported)
Religion and Religion, Secular
Secular Constraints, and
Religion Constraints Ecology

Religion
General Rebgiosity -.11* -.07 -.02
Denominational Conservatism -.08 -.08 -.06
Status Characteristics
Age -.36* -.33* -.33*
Sex (1 = Male) .21* .18* .17*
Family Income -.17* -.17* -.18*
Secular Constraints
Legal Deterrents -.11* -.10
Social Constraints -.12* -.12*
Religious Networks
Friends and Family -.11
Neighbors -.08
Social Ecology
Social Integration .03
Percent Female-
Headed Households -.03
Percent Renter .07
R^ .21 .24 .24

* p< .05.
controls were included. Our third measure of religiosity, religious activi-
ties, had negative effects that persisted with all controls in place. (See
Table 2, column 1.) Variation in general crime explained in all full models
with religiosity subscale measures (activities, hellfire, and salience) ranged
from 25% to 28%.
Tables 1 and 2 report several other significant relationships in the religi-
osity and crime models. Across all models except the general religiosity
model (Table 1), both legal deterrents and informal social constraints
inhibited general criminal involvement. With general religiosity as the
religiosity indicator, legal deterrents were not significant predictors. Also,
and consistent with previous self-report research, crime was inversely
related to age and was more probable among males than females. Crimi-
nality also appearQd to diminish at higher levels of family income.

high correlations among the religiosity measures and resulting potential that multicol-
linearity would bias the estimates, it was not possible to place all religiosity measures
into the same equation to determine which was the best predictor.
RELIGION AND CRIME REEXAMINED 209

Table 2. The Effects of Religion on General Crime (Betas


Reported)
Measures of Religiosity
Religious Hellfire Religious
Activities Beliefs Salience
Religiosity -.16* .06 -.03
Denominational Conservatism -.07 -.06 -.06
Status Characteristics
Age -.32* -.34* -.35*
Sex (1 = Male) .17* .17* .16*
Family Income -.20* -.20* -.20*
Secular Constraints
Legal Deterrents -.12* -.13* -.12*
Social Constraints -.11* -.13* -.12*
Religious Networks
Friends and Family -.03 -.14* -.10
Neighbors -.09 -.09 -.09
Social Ecology
Social Integration .02 .05 .04
Percent Female-
Headed Households -.02 -.01 -.01
Percent Renter .06 .07 .06
R^ .28 .25 .26
* p< .05.

The claim that religiosity's effects on crime are mediated by moral com-
munities was tested by forming product-interaction terms for combina-
tions of personal religiosity dimensions, social ecology measures, and
religious networks. Then, crime was regressed on these terms, along with
all other variables in the full additive models."* Among the resulting
potential interactions, we detected no significant patterns of joint effects
or, in general, any additional significant contribution to explained variance
beyond the combined individual additive terms of the equations.

DISCUSSION
RECONSIDERING HELLFIRE
Three of our four measures of religiosity—general religiosity, religious
beliefs (hellfire), and religious values (salience)—failed to inhibit adult

10. Variables in the interaction models were "centered" (Aiken and West, 1991) to
minimize multicollinearity and corresponding computational problems.
210 EVANS ET AL.

criminality directly in full additive models." And, if conservative denomi-


nations are sending strong proscriptive messages to their members, such
admonitions are going unheeded when other sources of control are
accounted for. The religiosity factors representing beliefs and attitudes
are apparently overwhelmed or duplicated in their influence on crime by
other secular sources of morality.
Yet, the finding that religious activities has effects in our model with
secular controls confirms the efficacy of behavioral indicators of religiosity
(usually attendance) so prevalent in prior research. Since we did not find
significant effects for the other measures of religiosity, it appears that the
discovery of individual religious effects for adults at least may depend not
so much on where—that is, in what contexts—religiosity is measured, but
rather which variables are used to indicate religiosity.
In the models specified here, then, religious activities appears to be the
best predictive measure of religious control. Such acts "make a differ-
ence" for our respondents possibly because they are regularly involved in
church or church-related activities. Compared with the less religiously
active, they are more often and more intensely exposed to "proscriptive
moral messages" (Bock et al., 1987) that help curb criminal activity. Fre-
quent interaction with others—even "airwave" believers—who are them-
selves religious may enable close monitoring and sanctioning of
waywardness. And, continual reinforcement of religious moral values and
policing of behavior are more likely when one is embedded in such a com-
munity of fellow believers. As White (1968:27-28) noted, "the fact that an
individual believes strongly or even prays often, is not as effective in
directing his behavior as are the sanctions [religious and otherwise] he
receives from other people."
Those engaged in religious activities are also more likely to interact with
others who are able to apply these sanctions. Grasmick et al. (1991b) have
further suggested that it is through such primary social networks, as are
likely to develop for regular church-goers, that sanctioning threats are
made real. Thus, even if an individual professes hellfire beliefs or claims
an internalized religious orientation in daily life, such religiosity is a less
effective inhibitor for them than for those who regularly and directly
account to a religious community.12

11. Of course, our data do not permit the assessment of nonlinear relationships,
reciprocal effects, or effects over time. In other words, it is possible that involvement in
crime influences religiosity. Future research should examine this possibility.
12. A major specification of the religion-crime relationship as reviewed above is
that "believers" are exposed to proscriptive messages that motivate them to refrain
from antiascetic acts. Although we acknowledged the "ascetic tradition" in the litera-
ture—and corresponding conceptual and empirical (validity) problems in deflnition—
our findings for general crime indicate that religion's impact is quite broad. In our own
RELIGION AND CRIME REEXAMINED 211

RELIGIOSITY AND MORAL CONTEXTS


Aside from direct personal and denominational religious effects, we
investigated the possibility that the religion and crime relationship is con-
tingent on contextual factors—moral communities. We found no support
for such joint effects. As Stark et al. (1980, 1982) and Tittle and Welch
(1983) have suggested, religiosity may be more salient for those embedded
in a religious community, but it is also likely that those who are not merely
"believers," but also religiously active in church-sponsored events, are
subjected to religious-group controls. Accordingly, neither religious (hell-
fire) beliefs nor religious attitudes (salience) are as important in inhibiting
criminal involvement as actual behavior (religious activities) that requires
immersion in church networks—in a "community of believers."
Our research confirms that the religiosity and crime relationship for
adults at least is neither spurious nor contingent, and it is not limited to
the least serious or ascetic offenses. We also found that it is more likely
revealed with behavioral indicators of religiosity. Only with a clear behav-
ioral measure of religiosity—religious activities—were effects observed
for crime. We hasten to acknowledge that even with our comprehensive
crime measures, self-reports with limited sample sizes reveal relatively less
serious crime. We would argue, however, that most of the "reality" of
crime, though perhaps not indicative of the type citizens appear to fear the
most, is represented in our crime scale.
A somewhat anomalous finding in the hellfire model requires further
explanation. In this case, religiosity of friends and family had a significant
negative impact on crime, while hellfire had a nonsignificant powf/vc effect
on crime. Religiosity of friends and family members may attain signifi-
cance in this case due to the relative weakness of hellfire as an indicator of
personal religiosity. In other words, such networks may be serving as a
proxy for personal religious activity, which is not being controlled in this
model.
CONCLUSION
In this study we found that religion, as indicated by religious activities,
had direct personal effects on adult criminality as measured by a broad
data the items previously mentioned as ascetic by other researchers—alcohol and drug
abuse primarily—did not cohere as a scale in a factor analysis of all items. Nonetheless,
we estimated separate models for ascetic and secular crime, using a 7-item ascetic scale
comparable to prior work and relegating the remaining 36 items to a "secular" scale.
Similar to the findings reported here for general crime, religious activities inhibited
both secular and ascetic acts. This tends to confirm that religious inhibition of general
crime as found here, is not due merely to the inclusion of such items in the general
crime scale—when they were segregated from the general crime scale, we still observed
statistically significant effects.
212 EVANS ET AL.

range of criminal acts. Further, the relationship held even with the intro-
duction of secular controls, and it did not depend on social or religious
contexts. Thus, measured as an individual behavioral trait, religion's
effects persist over a wide range of crime. Of course, as noted above, it is
likely that personal religious behavior (church attendance and related
activities) is entangled with and reinforced by association with other
believers, religious friends, family members, and fellow parishioners.
While the data provide confidence that religion "matters" for adult crime,
our knowledge of the religious factor in crime for adults and juveniles
would be enhanced by further explorations of reciprocal connections
between individual and group religious effects.
We also suggest further multivariate research with adequate measures,
adult samples, and nonreligious social controls at different levels of infiu-
ence. The moral communities hypothesis, for example, could be better
evaluated with better measures of religious and social networks and con-
texts than are available at this point. If moral communities are too proxi-
mate to the individual, their infiuence may be confounded with individual
religiosity; if they are too far removed, they lose sanctioning power. The
collection of panel data that would permit the further assessment of causal
linkages through time ordering (cf. Burkett and Warren, 1987; Free, 1994)
of direct, indirect, and reciprocal effects would also be helpful in under-
standing the relationship. It is likely that personal religiosity infiuences
the selection of friends and that friends infiuence individual religiosity, for
example. In addition, religious involvement may motivate respect for sec-
ular morality and authority, and strong social bonds may reinforce reli-
gious bonds.
Researchers are beginning to understand the processes that may link
religion and crime (see, e.g.. Bock et al., 1987; Clarke et al., 1990; Cochran
et al., 1988; Grasmick et al., 1991a), but much more work remains. Con-
sideration of religion within a context of competing theoretical perspec-
tives (see, e.g., Cochran et al., 1994; Free, 1994) should yield additional
and important insights into the relative efficacy of religion as an insulator
against crime and delinquency and the process by which it may serve as a
moral arbiter to dampen crime and delinquency.

REFERENCES
Aiken, Leona S. and Stephen G. West
1991 Multiple Regression: Testing and Interpreting Interactions. Newbury
Park, Caiif.: Sage.
Albrecht, Stan L., Bruce A. Chadwick, and David S. Alcom
1977 Religiosity and deviance: Application of an attitude-behavior contingent
consistency model. Journal for the Scientific Study of Religion
16:263-274.
RELIGION AND CRIME REEXAMINED 213

Babbie, Earl
1990 Survey Research Methods. 2d ed. Beimont, Calif.: Wadsworth.
Bainbridge, William Sims
1989 The religious ecology of deviance. American Sociological Review
54:288-295.
Barnes, Harry E. and Negley K. Teeters
1951 New Horizons in Criminology. Englewood Cliffs, N.J.: Prentice-Hall.
Bock, E. Wilbur, John K. Cochran, and Leonard Beeghley
1987 Moral messages: The relative influence of denomination on the religios-
ity-alcohol relationship. The Sociological Quarterly 28:89-103.
Bonger, William A.
1916 Criminality and Economic Conditions. Bloomington: Indiana University
Press.
Bureau of the Census
1992 Census of Population and Housing 1990: Summary Tape File III (Ohio).
Machine-readable data files. Washington, D.C: Government Printing
Office.
Burkett, Steven R.
1977 Religion, parental influence, and adolescent alcohol and marijuana use.
Journal of Drug Issues 7:263-273.
Burkett, Steven R. and Bruce O. Warren
1987 Religiosity, peer associations, and marijuana use: A panel study of
underlying causal structures. Criminology 25:109-131.
Burkett, Steven R. and Mervin White
1974 Hellfire and delinquency: Another look. Journal for the Scientific Study
of Religion 13:455-462.
Clarke, Leslie, Leonard Beeghley, and John K. Cochran
1990 Religiosity, social class, and alcohol use: An application of reference
group theory. Sociological Perspectives 33:201-218.
Cochran, John K.
1988 The effect of religiosity on secular and ascetic deviance. Sociological
Focus 21:293-306.
1989 Another look at delinquency and religiosity. Sociological Spectrum
9:147-162.
Cochran, John K. and Ronald L. Akers
1989 Beyond hellfire: An exploration of the variable effects of religiosity on
adolescent marijuana and alcohol use. Journal of Research in Crime and
Delinquency 26:198-225.
Cochran, John K., Leonard Beeghley, and E. Wilbur Bock
1988 Religiosity and alcohol behavior: An exploration of reference group
theory. Sociological Forum 3:256-276.
Cochran, John K., Peter B. Wood, and Bruce J. Ameklev
1994 Is the religiosity-delinquency relationship spurious? A test of arousal and
social control theories. Journal of Research in Crime and Delinquency
31:92-123.
214 EVANS ET AL.
Davidson, James D. and Dean D. Knudsen
1977 A new approach to religious commitment. Sociological Focus 10:151-173.
Dillman, Don A.
1978 Mail and Telephone Surveys: The Total Design Method. New York:
Wiley-Interscience.
Elifson, Kirk W., David M. Petersen, and C. Kirk Hadaway
1983 Religiosity and delinquency: A contextual analysis. Criminology
21:505-527.
Elliott, Delbert S., Suzanne S. Ageton, David Huizinga, Brian A. Knowles, and
Rachelle J. Canter
1983 The Prevalence and Incidence of Delinquent Behaviors: 1976-1980.
Boulder, Colo.: Behavioral Research Institute.
Eve, Raymond A.
1978 A Study of the efficacy of interactions of several theories of explaining
rebelliousness among high school students. Joumal of Criminal Law and
Criminology 49:115-125.
Falk, Gerhard J.
1961 Religion, personal integration and criminality. Joumal of Education and
Sociology 35:159-161.
Fitzpatrick, Joseph P.
1967 The role of religion in programs for the prevention and correction of
crime and delinquency. The President's Commission on Law Enforce-
ment and Juvenile Justice, Task Force Report: Juvenile Delinquency and
Youth Crime. Washington, D.C: Govemment Printing Office.
Free, Marvin D., Jr.
1994 Religiosity, religious conservatism, bonds to school, and juvenile delin-
quency among three categories of drug users. Deviant Behavior
15:151-170.
Frey, James H.
1989 Survey Research by Telephone. 2d ed. Newbury Park, Calif.: Sage.
Glock, Charles Y.
1959 The religious revival in America? In Jane Zahn (ed.). Religion and the
Face of America. Berkeley: University of Califomia Press.
Grasmick, Harold G., Karyl Kinsey, and John K. Cochran
1991a Denomination, religiosity, and compliance with the law: A study of
adults. Journal for the Scientific Study of Religion 30:99-107.
Grasmick, Harold G., Robert J. Bursik, Jr., and John K. Cochran
1991b "Render unto Caeser what is Caesar's": Religiosity and taxpayer's
inclinations to cheat. The Sociological Quarterly 32:251-266.
Hadaway, C. Kirk, Kirk W. Elifson, and David M. Petersen
1984 Religious involvement and drug use among urban adolescents. Joumal
for the Scientific Study of Religion 23:109-128.
Higgins, Paul C. and Gary L. Albrecht
1977 Hellfire and delinquency revisited. Social Forces 55:952-958.
Hirschi, Travis and Rodney Stark
1969 Hellfire and delinquency. Social Problems 17:202-213.
RELIGION AND CRIME REEXAMINED 215
Jensen, Gary F. and Maynard L. Erickson
1979 The religious factor and delinquency: Another look at the hellfire
hypothesis. In Robert Wuthnow (ed.), The Religious Dimension: New
Directions in Quantitative Research. New York: Academic Press.
Kim, Jae-On and Charles W. Mueller
1978 Factor Analysis: Statistical Methods and Practical Issues. Newbury Park,
Calif.: Sage.
King, Morton B. and Richard A. Hunt
1972 Measuring Religious Dimensions: Studies of Congregational Involvement.
Dallas: Southern Methodist University.
Kvarceus, William C.
1944 Delinquent behavior and church attendance. Sociology and Social
Research 28:418-434.
LaGrange, Randy L. and Helen Raskin White
1985 Age differences in delinquency: A test of theory. Criminology 23:19-45.
Linden, Rick and Raymond Currie
1977 Religiosity and drug use: A test of social control theory. Canadian
Joumal of Criminology 5:346-355.
Lombroso, Cesare
1911 Crime, Its Causes and Remedies. Boston: Little, Brown.
Mclntosh, William Alex, Staria Fitch, J. Branton Wilson, and Kenneth L. Nyberg
1981 The effect of mainstream religious social controls on adolescent drug use
in rural areas. Review of Religious Research 23:54-75.
McLuckie, Benjamin, Margaret Zahn, and Robert A. Wilson
1975 Religious correlates of teenage drug use. Joumal of Drug Issues
5:129-139.
Middleton, Russell and Snell Putney
1962 Religion, normative standards, and behavior. Sociometry 25:141-152.
Nelsen, Hart M. and James F. Rooney
1982 Fire and brimstone, lager and pot: Religious involvement and substance
use. Sociological Analysis 43:247-256.
Peek, Charles W., Evans W. Curry, and H. Paul Chalfant
1985 Religiosity and delinquency over time: Deviance deterrence and deviance
amplification. Social Science Quarterly 66:120-131.
Pettersson, Thorleif
1991 Religion and criminality: Structural relationships between church involve-
ment and crime rates in contemporary Sweden. Journal for the Scientific
Study of Religion 30:279-291.
Rankin, Joseph H. and Edward L. Wells
1990 The effect of parental attachments and direct controls on delinquency.
Joumal of Research in Crime and Delinquency 27:140-165.
Reckless, Walter C. and Mapheus Smith
1932 Juvenile Delinquency. New York: McGraw-Hill.
Rohrbaugh, John and Richard Jessor
1975 Religiosity in youth: A personal control agent against delinquent
behavior. Joumal of Personality 43:135-155.
216 EVANS ET AL.
Rosenbaum, Jill Leslie
1987 Social control, gender, and delinquency: An analysis of drug, property,
and violent offenders. Justice Quarterly 4:117-132.
Sampson, Robert J,
1985 Neighborhood and crime: The structural determinants of personal
victimization. Journal of Research in Crime and Delinquency 22:7-40.
1986 Neighborhood family structure and the risk of personal victimization. In
James M. Byrne and Robert J. Sampson (eds,). The Social Ecology of
Crime, New York: Springer-Verlag,
Sampson, Robert J. and John H. Laub
1993 Crime in the Making: Pathways and Tbming Points Through Life,
Cambridge, Mass,: Harvard University Press.
Shaw, Clifford R, and Henry D, McKay
1931 Social factors in juvenile delinquency. Report on the Causes of Crime,
National Commission of Law Observance and Enforcement. Vol. 2,
Washington, D,C.: Govemment Printing Office,
Simons, Ronald L,, Martin G, Miller, and Stephen M, Aigner
1980 Contemporary theories of deviance and female delinquency: An empiri-
cal test, Joumal of Research in Crime and Delinquency 17:42-53,
Skogan, Wesley G,
1990 Disorder and Decline: Crime and the Spiral of Decay in American
Neighborhoods, Berkeley: University of Califomia Press,
Smith, Tom
1990 Classifying Protestant denominations. Review of Religious Research
31:225-245,
Sosdian, Carol P, and Laure M, Sharp
1980 Nonresponse to mail surveys: Access failure or respondent resistance.
Public Opinion Quarteriy 44:396-402,
Stack, Steven and Mary Jane Kanavy
1983 The effect of religion oh forcible rape: A structural analysis, Joumal for
the Scientific Study of Religion 22:67-74,
Stark, Rodney
1987 Religion and deviance: A new look. In James A, Day and William S,
Laufer (eds,). Crime, Values, and Religion, Norwood, N.J,: Ablex,
Stark, Rodney and Charles Y, Glock
1968 American Piety: The Nature of Religious Commitment, Berkeley:
University of Califomia Press,
Stark, Rodney, Daniel P, Doyle, and Lori Kent
1980 Rediscovering moral communities: Church membership and crime. In
Travis Hirschi and Michael Gottfredson (eds,). Understanding Crime:
Current Theory and Research, Beverly Hills, Calif,: Sage,
Stark, Rodney, Lori Kent, and Daniel P, Doyle
1982 Religion and delinquency: The ecology of a "lost" relationship, Joumal
of Research in Crime and Delinquency 19:4-24,
Tappan, Paul H,
1949 Juvenile Delinquency, New York: McGraw-Hill,
RELIGION AND CRIME REEXAMINED 217

Taylor, Ralph B, and Jeanette Covington


1988 Neighborhood changes in ecology and violence. Criminology 26:553-589,
Tittle, Charles R,
1980 Sanctions and Social Deviance: The Question of Deterrence, New York:
Praeger,
Tittle, Charles R, and Michael R, Welch
1983 Religiosity and deviance: Toward a contingency theory of constraining
effects. Social Forces 61:653-682,
Welch, Michael R,, Charies R, Tittle, and Thomas Petee
1991 Religion and deviance among adult Catholics: A test of the "moral
communities" hypothesis, Joumal for the Scientific Study of Religion
30:159-172,
White, Richard H,
1968 Toward a theory of religious infiuence. Pacific Sociological Review
11:23-28,

T, David Evans is Assistant Professor of Sociology at the University of North Caro-


lina at Wilmington, His research interests include testing theories of crime, religion and
crime, white-collar crime, and discretion in criminal charging, plea bargaining, and
sentencing,
Francis T, Cullen is Distinguished Research Professor in the Department of Criminal
Justice at the University of Cincinnati, where he also holds a joint appointment in soci-
ology. He is author of Rethinking Crime and Deviance Theory, and coauthor of Reaf-
firming Rehabilitation, Corporate Crime Under Attack, Criminological Theory, and
Criminology. He is past president of the Academy of Criminal Justice Sciences and
previously was editor of Justice Quarterly and the Joumal of Crime and Justice.
R, Gregory Dunaway is Assistant Professor of Sociology and Director of the Correc-
tions Program at Mississippi State University, His research areas include parental
transmission of crime ideology, social inequality and crime, and rural criminology,
Velmer S, Burton, Jr, is Assistant Professor of Political Science at Washington State
University, He has published in the areas of theories of crime, corrections, and mental
health, and is co-editor of Contemporary Criminological Theory.
218 EVANS ET AL.

Appendix 1. Survey Scales and Items

CRIME (RELIABILITY = .77)*


1. Been drunk in public
2. Avoided paying for such things as movies, bus rides, food
3. Filed an insurance claim that you knew was false
4. Parked your car illegally in a "no parking" zone
5. Bought or provided liquor for a minor
6. Failed to return extra change that a cashier gave you by mistake
7. Been loud, rowdy or unruly in a public place
8. Hitchhiked where it was illegal to do so
9. Drove your car more than 15 miles above the speed limit
10. Begged for money or things from strangers
11. Damaged another car but did not try to notify the owner
12. Claimed a deduction on your income tax return that you did not fully
deserve
13. Did not report all of your income on your tax returns
14. Gambled illegally such as betting on sporting events or playing cards
15. Drove a car while drunk
16. Made obscene telephone calls, such as calling someone and saying
dirty things
17. Urinated in a public place (like behind a bush)
18. Been involved in gang fights
19. Sold marijuana or hashish
20. Sold hard drugs, such as heroin, cocaine, or LSD
21. Knowingly bought, sold, or held something that was stolen (or tried
to do any of these things)
22. Taken a vehicle other than your own for a ride (drive) without the
owner's permission
23. Thrown objects at cars or people
24. Stole or tried to steal things worth $5 or less at work
25. Stole or tried to steal things worth between $5 and $50 at work
26. Stole or tried to steal things worth more than $50 at work
27. Stole or tried to steal things worth $5 or less at places other than
work
28. Stole or tried to steal things worth between $5 and $50 at places other
than work
29. Carried a hidden weapon other than a plain pocket knife
30. Purposely damaged or destroyed property belonging to your family
members
31. Purposely damaged or destroyed property belonging to your
employer
RELIGION AND CRIME REEXAMINED 219

32. Purposely damaged or destroyed property that did not belong to you
(not counting your family or employer's property)
33. Broken into a building or vehicle (or tried to break in) to steal some-
thing or just look around
34. Used force (strong-arm methods) to get money or things from family
members
35. Used force (strong-arm methods) to get money or things from people
other than family members or people you work with
36. Hit or threatened to hit a family member
37. Hit or threatened to hit someone other than a coworker or family
member
38. Attacked someone with the idea of seriously hurting or killing him/
her
39. Had marijuana or hashish
40. Had hallucinogens
41. Had amphetamines
42. Had barbiturates
43. Had cocaine

RELIGIOSITY
RELIGIOUS ACTIVITIES (RELIABILITY = .79)

1. In the last twelve months, how often did you attend religious
services?
2. In the last twelve months, how often did you attend social events at
church?
3. In the last twelve months, how often did you read reUgious material?
4. In the last twelve months, how often did you listen to religious pro-
grams on radio or television?
RELIGIOUS SALIENCE (RELIABILITY = .85)

1. Religion is a very important part of my life.


2. Following God's commandments is important to me.
3. In times of personal trouble, I turn to religion for guidance.

HELLFIRB (RELIABILITY = .88)

1. After I do something wrong, I fear God's punishment.


2. People who are evil in this world will eventually suffer in Hell.
3. Following God's commandments is important to me.
4. God knows everything a person does wrong.
5. In the end, God punishes all those who have sinned.
6. There is life after death.
220 EVANS ET AL.

7. Many people with diseases like AIDS are being punished by God for
their sinfulness.

RELIGIOUS NETWORKS
FAMILY AND FRIENDS (RELIABILITY = .62)
1. Of your five closest friends, how many of them would you say attend
church regularly (every week or every other week)?
2. Of your five closest adult family members, how many of them would
you say attend church regularly (every week or every other week)?
NEIGHBORS (RELIABILITY = .78)
1. In your neighborhood, how many of your neighbors would you esti-
mate attend church on a regular basis (every week or every other
week)?
2. Most of the people in my neighborhood are religious.

SECULAR CONSTRAINTS
FEAR OF LEGAL SANCTIONS (RELIABILITY = .74)
1. If I broke the law, there is a good chance I'd be caught by the police.
2. If I broke the law and got caught, there is a good chance I'd go to
prison.

INFORMAL SOCIAL CONSTRAINTS (RELIABIL^Y - .82)


1. Throughout my life, I have had a lot of respect for my mother and
father.
2. If I were thinking of breaking the law, my friends would tell me not
to do it.
3. Throughout my life, I have gotten along well with my parents.

SOCIAL ECOLOGY
SOCIAL INTEGRATION (RELIABILITY = .80)
1. In my neighborhood it's easy to know who belongs and who is a
stranger.
2. My neighborhood is getting worse and worse all the time.
3. Most of my neighbors spend a lot of time keeping their houses in
good shape.
4. In my neighborhood, it's not safe to walk outside after dark.
5. My neighborhood is noisy and the streets always seem to have litter
in them.
6. In my neighborhood, people will call the police right away if they
think a crime is being committed.
RELIGION AND CRIME REEXAMINED 221

7. Most of the people living in my neighborhood are good, upstanding


citizens.

OBJECTIVE STRUCTURAL DISORGANIZATION (CENSUS)


1. Percent female-headed households
2. Percent renter-occupied housing units
* All reported scale reliabilities were calculated by the split-half method, using factor-
weighted items.
222 EVANS ET AL.

I CO

gll

o
o

a
00

« * * « «
a
a ^ ' ' ' ' r
* »« » *
§ §§^§§ 2 ss;
f r-; i' r r r r r i

o o yj ^o t"^ S c^ p*J ^
o

o
^5 ^5 ^5
^ * I
oJ co ^^ oJ f^ c**
I r
•§ cQ ^ ^ (O QO t C? C ; C;

1 r
O r~* O 1-^ O O C^ O C
r-; • • • r r r r r i' r

•3
U
E o 'a a w a> c c

•S •! S ^ 3 ,, o 3 a - 8 S
all i l l
u o o fli 4*
a. i-J C/3 C/3 O H CX<
V
tt.
RELIGION AND CRIME REEXAMINED 223

Appendix 3. Correlations Among Independent Variables


and Crime
Crime
1. Sex (1 = Male) .176*
2. Age -.420*
3. Family Income -.131*
4. Denominational Conservatism -.013
5. General Religiosity -.198*
6. Religious Acts -.282*
7. Hellfire -.087
8. Salience -.227*
9. Religious Networks (Family and Friends) -.248*
10. Religious Networks (Neighbors) .072
11. Legal Deterrents -.250*
12. Social Constraints -.270*
13. Social Integration -.182*
14. Percent Female-Headed Households .082
15. Percent Renter .109*
* p< .05.
224 EVANS ETAL.

Appendix 4. Descriptives for Study Variables


Mean Standard Deviation
Crime 36.27 103.11
Demographics
Age 46.31 19.18
Sex (1 = Male) .44 .50
Family Income 42,788.55 58,036.73
Religion
Denominational Conservatism 1.92 .54
General Religiosity 44.23 12.94
Religious Activities 10.13 4.45
Hellfire Beliefs 20.88 6.89
Salience 13.37 3.95
Religious Networks (Family and
Friends) 5.54 2.76
Religious Networks (Neighbors) 6.10 1.77
Secular Constraints
Legal Deterrents 8.88 2.17
Social Constraints 15.18 2.53
Social Ecology
Social Integration 30.90 5.41
Percent Female-
Headed Households 15.71 10.19
Percent Renter-Occupied
Housing Units 53.74 17.42

You might also like