Religion: The Forgotten Factor in Cutting Youth Crime and Saving At-Risk Urban Youth

By David B. Larson, M.D., M.S.P.H., President, National Institute for Healthcare Research and Adjunct Professor, Duke University Medical Center

Byron R. Johnson, Ph.D., Director, Center for Crime and Justice Policy and Senior Fellow, Vanderbilt Institute for Public Policy Studies, Vanderbilt University, Nashville, Tennessee

Introduction by
John J. DiIulio, Jr., Senior Fellow, Manhattan Institute, Director, The Jeremiah Project

Introduction

Religion Reduces Deviance

John J. DiIulio, Jr.

My friend and occasional co-author Dr. William J. Bennett has half-jokingly defined social science as “the elaborate demonstration of the obvious by methods that are obscure.”  To some readers, this report may justify that definition. Do we really need social science to “prove” that religion reduces many forms of deviance, including but not limited to crime and delinquency and other problems that are highly concentrated among inner-city youth?

Yes, we do, and for a host of reasons. First, in every major field of intellectual endeavor, including social science and criminology, the expert consensus has been that religion has either no effect on human behavior and social outcomes — indeed, most research on social problems omits any measure of religion as either an explanatory or a control variable – faith is indeed “the forgotten factor” – or a net negative effect on human behavior and social outcomes.

Essentially, the dominant theory has been that religiosity (almost universally assumed to be highly correlated with low intelligence and less than average formal education) probably conditions people to be more rather than less prone to poor physical and mental health outcomes, and no less likely to commit deviant, delinquent or criminal acts.

Second, even if one accepts that religion probably reduces deviance, it is important to know both the conditions under which this relationship holds and how robust the relationship is over time. For example, does “religion” as in mere church-going have any effect?  Does it have the same effect as church going in concert with daily prayer or other behavioral manifestations of religiosity? Other things held equal, does exposure to faith-based programs among unchurched children have the same effect on deviance as church going among otherwise comparable young persons?  Do prisoners exposed to Bible studies commit fewer crimes after release than otherwise comparable prisoners do one year out of prison?  Two years? More?

Third, for the last three decades, most major social programs, especially those supported with government monies, have been developed, institutionalized, and funded without much in the way of research indicating they might actually work or the conditions under which they might eventually succeed.  In areas from teen pregnancy to teen illiteracy, youth crime to welfare dependency, child abuse to substance abuse, Americans have paid a terrible price for this triumph of ideological advocacy over empirical analysis, policy-pushing dogma over policy-relevant data.  That a generation of social engineers “got away with it” is all the more reason for persons interested in faith-based approaches to avoid, not repeat, the mistake.  We should proceed enthusiastically but realistically, inductively, and incrementally, always with genuine openness to whatever serious social science, obscure methods and all, can tell us about the efficacy (or lack thereof) of churches and religious non-profits in relation to avoiding violence, achieving literacy, promoting employment, and achieving other desirable secular social goals among disadvantaged urban youth and young adults.

There is also a fourth reason to pursue faith-factor efficacy research, both statistical (as in the present paper) and field-based (as in other work of The Jeremiah Project), namely, the religiosity of most Americans, the majority’s belief in the efficacy of faith-based approaches in a day when “decline of morals, values” consistently tops the list of major public concerns.

From the famous nineteenth-century observations of Alexis de Tocqueville to the latest findings of survey researchers and social scientists, it is abundantly clear that Americans have been, and continue to be, a religious people. “The United States,” observes George Gallup, Jr., “is one of the most devout nations of the entire industrialized world, in terms of religious beliefs and practices.”1   Belief in God remains the norm in America, with levels of belief ranging between 94 percent and 99 percent over the past five decades. Claims of membership in a church, synagogue or similar place of worship have ranged from a high of 75 percent in 1947 to a low of 65 percent in 1988 and 1990.

Black Americans are in many ways the most religious people in America. Some 82 percent of blacks (versus 67 percent of whites) are church members; 92 percent of blacks (versus 55 percent of whites) say that religion is “very important in their life;” and 86 percent of blacks (versus 60 percent of whites) believe that religion “can answer all or most of today’s problems.”2 

All reports of the death of organized religion and religious sentiment in America have been greatly exaggerated.  Since the end of the Second World War, we have witnessed what Roger Finke and Rodney Clark have aptly described as the “churching of America,” resulting by the mid-1990s in a nation with an estimated half a million churches, temples, and mosques, 2,000 or more religious denominations, and an unknown number of independent churches.3   In 1995, Gallup’s Religion Index, an on-going measurement of eight key religious beliefs and practices of the American public, hit a ten-year high.4   That same year, Nobel economist Robert W. Fogel of the University of Chicago speculated that the United States was in the midst of “its Fourth Great Awakening,” a “new religious revival . . . fueled by a revulsion with the corruptions of contemporary society.”5 

Great Awakening or not, public laws have grown more faith-friendly.  For example, the federal government’s latest welfare reform overhaul measure, the Personal Responsibility and Work Opportunities Reconciliation Act of 1996, contains Section 104, the so-called Charitable Choice provision, which encourages states to utilize “faith-based organizations in serving the poor and needy,” requires that religious organizations be permitted to receive contracts, vouchers, and other government funding on the same basis as any other non-governmental provider, and “protects the religious integrity and character of faith-based organizations that are willing to accept government funds.”6   As enacted in 1996, Charitable Choice covers each of the major federal anti-poverty and social welfare programs (Temporary Assistance to Needy Families, Medicaid, Supplemental Security Income and Food Stamps). Congressional efforts begun in 1998 could eventually expand its scope to juvenile justice programs and other federal policy domains. Many states, most notably Texas, have moved aggressively to reorient their anti-poverty programs around Charitable Choice and kindred state laws favoring church-state cooperation.7 

Philanthropy, too, has been gradually tilting toward religion. Over the last few years, many foundations have either launched new grant-making initiatives focused on religion, increased their support for research on religion, or increased grants for technical or direct financial assistance to community-serving ministries.  For example, in 1996 three foundations with long-standing programs in religion, Lilly, Pew, and Irvine, made record religion grants of $60 million, $13 million, and $7.7 million, respectively.  In 1997, the Robert Wood Johnson Foundation made its 800th $25,000 grant to its “Faith in Action” program, which mobilizes interfaith networks of religious volunteers to serve some 200,000 elderly and disabled Americans.  The program’s director describes it as “the first mega program undertaken” by the foundation, and “the largest program” in the foundation’s history.8 

It is hardly surprising that networks of interfaith volunteers are the backbone of such a program. As Father Andrew Greely has aptly summarized the evidence, research has consistently shown that both “frequency of church attendance and membership in church organizations correlate strongly with voluntary service.  People who attend services once or more are approximately twice as likely to volunteer as those who attend rarely if ever.”9   As Greely also notes, the best available data suggest that religious organizations and “relationships related to their religion” are clearly the major forces in mobilizing volunteers in America; even a third of purely secular volunteers (persons who did not volunteer for specifically religious activities) also relate their service “to the influence of a relationship based on their religion.”10 

What, if any, broader social consequences for blacks and other Americans flow from religiosity and faith-based charitable, volunteer and community-service work?  In social science terms, what, if any, data are there to indicate that, other things being equal, religious belief, church-going, community-serving ministry, or some combination thereof, varies inversely with poverty, joblessness, crime, substance abuse, or other social ills? In plain English, is there any scientific evidence to show that religious do-gooding does any good, or to justify the faith of most black Americans that religion can “answer all or most of today’s problems”?

Over the last several years, journalists seem to have become more interested in this question.  For example, Newsweek magazine wrote a cover story on the inner-city ministry of Boston’s Rev. Eugene Rivers, “God v. Gangs,” following stories on the same ministry by a diverse set of writers, including Joe Klein of The New Yorker and columnists George Will of The Washington Post, Bob Herbert of The New York Times, and several others.  In each case, they credited the ministry with working cooperatively with police and probation officials, working one-on-one with the city’s most severely at-risk youngsters, and thereby helping to engineer a dramatic drop in youth crime and the virtual elimination of gun-related homicides. Two months before the Newsweek story, Time magazine featured “In the Line of Fire,” the tale of Brother Bill, a Catholic lay worker who “repeatedly walks into gunfire to stop the shooting – and love the unloved.”11   Two years earlier, the cover of U.S. News & World Report asked “Can Churches Cure America’s Social Ills?,” and the story answered largely in the affirmative.12 

While such “faith factor” journalism is out ahead of the empirical research on religion and social action, it is hardly pure hype. As UCLA’s James Q. Wilson has succinctly summarized the small but not insignificant body of credible evidence to date, “Religion, independent of social class, reduces deviance.”13 

Exhibit A is the present report, summarizing the latest research of David Larson, the medical research scientist who pioneered the development of the scientific “faith factor” research on public health outcomes (physical health, mental health, addictions) that led to new training programs at Harvard and three dozen other medical schools.14   With criminologist Byron Johnson, Larson has reviewed some four hundred juvenile delinquency studies published between 1980 and 1997. They have found that the better the study design and measurement methodology, the greater the likelihood the research will produce statistically significant results associated with “the faith factor.”  In other words, the more scientific the study, the more optimistic are its findings about the extent to which religion reduces deviance.

This conclusion squares with the results of another major review of the relevant research literature as it pertains to adult criminals: “Our research confirms that the religiosity and crime relationship for adults is neither spurious nor contingent . . . [R]eligion, as indicated by religious activities, had direct personal effects on adult criminality as measured by a broad range of criminal acts. Further, the relationship held even with the introduction of secular controls.”15   In other words, “religion matters” in reducing adult crime, too.

But in relation to black inner-city poverty and related social ills, perhaps the single most illustrative line of “religion reduces deviance” research begins with a 1985 study by Harvard economist Richard Freeman, runs through Larson’s work, and continues through the community development, mentoring, and “faith-factor” research of analysts with Public/Private Ventures, a Philadelphia-based national nonprofit youth policy research organization, and the Manhattan Institute’s Jeremiah Project.

In 1985, Freeman reported that church-going, independent of other factors, made young black males from high-poverty neighborhoods substantially more likely to “escape” poverty, crime, and other social ills.16   In this analysis and extension of Freeman’s work, Larson and Johnson mine national longitudinal data on urban black youth and find that, using a more multi-dimensional measure of religious commitment than church-going, religion is indeed a powerful predictor of “escaping” poverty, crime, and other social ills, more powerful even than such variables as peer influences.

Like Freeman, Larson and Johnson conjecture that the potential of church-going and other religious influences to improve the life prospects of poor black urban youth is in part a function of how church-going and other “faith-factors” influence how young people spend their time, the extent of their engagement in positive structured activities, and the degree to which they are supported by responsible adults.

Larson and Johnson’s report on the “forgotten factor” is organized into two main sections.  Section one, “Religiosity and Youth Crime,” previews the findings of a research paper in which they examine the findings of virtually all extant scientific studies of the relationship between religiosity and delinquency in the United States.  Section two, “Religiosity and At-Risk Urban Youth,” extends the aforementioned work of Freeman and mines national data on religiosity and young black urban makes, black church youth and community outreach efforts, and a nationally representative sample of youth.

The bottom line of their research is that religion reduces deviance under a wide range of conditions and for diverse populations of youth and young adults. Their work is, as it were, hardly the final word about the social efficacy of religious belief, but it speaks powerfully to the need for researchers and others to start remembering the faith factor and to take religion as an ally in repairing lives, saving children, and resurrecting the civil society of inner-city America.  The Manhattan Institute’s Center for Civic Innovation and Jeremiah Project are grateful to the authors and pleased to have sponsored this paper and publicized their findings.

John J. DiIulio, Jr.
Senior Fellow, The Manhattan Institute
Director, The Jeremiah Project

Section I

Religiosity and Youth Crime17 

The relationship between religiosity and delinquency has been an area lacking research review, study, and explanatory consensus in the research literature (Evans et al., 1995; Johnson, 1987; Title and Welch, 1983). While some studies have found a strong negative or beneficial relationship between religion and delinquency (Benda, 1995; Brownfield and Sorenson 1991; Tittle and Welch, 1983), others have suggested that religion has only a weak or insignificant effect on delinquency (Cochran, Wood, and Arneklev, 1994). Another issue of equal importance but often overlooked in researching the relationship between religion and delinquency is how well religiosity or religious commitment is measured.  To provide an accurate and unbiased summary of the research on religion and delinquency, a review method is needed that is systematic yet sufficiently flexible to encompass and review a wide range of studies using diverse methodologies as well as using different measures of religion.

Systematic Reviews of Research
 
An innovative review strategy called a systematic review (SR) uses a method that permits a quantitative, or replicable, review of a specific research literature. In this way, the SR minimizes the opportunity for bias found in more traditional research review approaches. In this approach, key aspects of the review design are quantified, including inclusion and exclusion criteria regarding the published studies, or the study “subjects” to be sampled, the method for analyzing the methodologies of each study sampled, determining and specifying interrater reliabilities, and finally, summing the results across all reviewed studies. Results can be simply presented and understood as numeric items. Thus, the review and its results, like any good research protocol, are replicable. In areas of controversy such as the study of religion, replication of a literature review should be an available option.

The SR surveys a specified sample of representative research usually over a certain specified time period, reviewing and evaluating the field’s leading, or “tenure-granting,” journals (Bareta, Larson, Zorc, & Lyons, 1990; Beardsley et al., 1989; Larson, Lyons, Hohmann, et al., 1989; Lyons et al., 1990). These leading journals are often those most frequently cited and thus often define or at least provide the lead for clarifying the state of research in a certain field.  In addition, the SR can sample various types of studies with diverse samples and different research methodologies as long as they all have the study factor of interest (e.g. religiosity) in common.

Like another type of analytic review method, or meta-analyses, systematic reviews are objective, with reported interrater reliabilities generally above 0.90.  A clear advantage of SRs is that, like meta-analyses, the review method and thus the study (i.e. review) findings can be replicated.  Therefore, the systematic review represents an objective, accurate method for reviewing within a particular scientific literature how a specific and potentially controversial factor like religion is handled both in terms of frequency of study and quality of study.  One of the authors and numerous collegues have pioneered a series of studies using the SR methodology to gain insights on religion, beginning with another field that has shown some difficulty in handling and measuring religion - psychiatry (Larson 1993; Larson et al., 1986; Larson et al., 1989; Larson et al., 1992; Larson and Larson 1994). In this effort, our goal was to follow-up on previous systematic reviews by using this innovative strategy to review and critique the state of research on religiosity and juvenile delinquency.

Methodology

Article Selection

The study population of interest for the present SR is a specified group of study publications rather than a population of individuals.  Our population consists of journal articles that examined the effect of religion on juvenile delinquency published from January of 1980 to December 1997. By utilizing an online library database, we searched for published research, or quantified articles (i.e. not commentaries or reviews) with key terms: religion, spirituality, church, delinquency and matched these terms with deviant behavior, deviance, or delinquency. After identifying articles utilizing the database, we then checked the references of each article to determine if additional articles could be identified. In the current study, we reviewed peer-reviewed journals in relevant study fields (i.e. addiction, adolescence, criminology, psychology, sociology) because the leading journals in criminology published very few studies on religion and delinquency.

To be selected to our sample for systematic review, an article must have met the
following criteria:

1. Contained at least one quantified variable of any kind.  A quantified variable was defined as one about which data were collected for a group of subjects. The sample primarily consisted of juveniles under the age of 18, although some studies looked at young adults up to age 20.
2. Published in a peer-reviewed journal in the United States between January 1980 and December 1997.
3. Used a sample collected from the USA. Studies using international and cross-
national samples are excluded.
4. Analyzed both religiosity and juvenile delinquency measures.

All total, we located 402 articles that quantitatively studied an aspect of delinquency and were published between January, 1980 and December 1997.  Each of the 402 studies was read independently by two different reviewers to determine how many of the studies contained any measures of religion or religious variables. Forty articles, or approximately 10 percent out of the pool of 402 studies examined the relationship between religiosity and juvenile delinquency (see Appendix A). The focus of the current systematic review, as with previous SRs, was to systematically examine these 40 studies, not the remaining studies that did not include religious variables.  Our sample of forty articles provides us with the opportunity to assess how well religion was treated in relevant peer-reviewed journals in a very recent eighteen year time period.

Characteristics of the Study Sample

The average sample size of the forty studies reviewed was 2,324, with a maximum sample size of 34,129 and minimum sample size at 123.  Only five of the forty studies used samples smaller than 300 and ten studies had samples smaller than 500.  None of the studies used a small group sample—a sample smaller than 50.  The samples studied in these published articles varied in scope and type.  Eighteen studies, or 45 percent, used samples collected from a population within the boundaries of a state. Sixteen, or 40 percent, of the studies were based on regional samples drawn from the populations of two or more states.  Only six, or 15 percent, of the studies used nationally representative samples.

There were also notable differences in sampling methods and sample response rates. The majority of the studies—(25 of 40), or 63 percent—failed to adopt a random sampling procedure.  Fourteen, or 35 percent, of the studies did not report sample response rates.  Of those that specified sample response rates, 14, or 35 percent, of the studies had response rates higher than 70 percent, nine, or 25 percent, of the studies had response rates between 50 percent and 70 percent, with three studies having response rates smaller than 50 percent.

Quality of Research

Specific criteria were used to measure the quality of research methodology in the current sample of articles that were systematically reviewed.  These criteria were derived primarily from Cook and Campbell’s well known methodological text on quasi-experimental research (1979).  Eleven items or criteria are used to rate the methodology of the articles in the current SR: (1) no ambiguity about causal inference; (2) the use of prospective data; (3) specification of response rate; (4) specification of missing data; (5) specification of race of subjects; (6) specification of gender of subjects; (7) specification of reliability of measures; (8) no mono-operation bias (refers to the use of multiple measures to represent a particular possible cause or effect construct); (9) no mono-method bias (refers to the use of diverse methods to collect data for operational representation of a construct); (10) use of multivariate statistics; and (11) interpretation of statistical findings. These eleven items were chosen since they represent basic criteria from which researchers are able to draw acceptable causal inference and to achieve optimal reduction of measurement problems (Cook and Campbell, 1979).  For coding purposes, the eleven criteria are dichotomized thus facilitating the construction of an index that quantifies the quality of research methodology.  If a study includes or utilizes one of these eleven methodological procedures (or criteria), it is coded one (1), while the absence of a procedure resulted in the coding of zero (0).

Measures of Religiosity
 
In parallel with a previous review (Larson et al., 1986), the focus of the present review is on the measurement of religious variables and their evaluated relationship within the published delinquency research. Studies were reviewed to determine whether they contained at least one quantified variable of delinquency.  Delinquency was defined as referring to any criminal, delinquent, or status offense committed by a juvenile (and in several study instances by young adults). Studies were also reviewed to determine whether they reported at least one quantified religious variable. This approach permits comparison of occurrence, results and quality of the measure of religious variables with the delinquency variables.

Role of Religiosity Measures
    
The forty published articles in the sample were reviewed to determine the role assigned to the religious variable or variables within each study. All of the study articles assessed religion as an independent variable impacting other variables. As an independent variable, religion can be treated in one of three ways: (1) a central explanatory variable, (2) a peripheral explanatory variable, or (3) a covariate used for statistical control. We reviewed every article in our sample to determine how religious measures were treated by researchers over the recent eighteen years covered by the systematic review. 

Effect of Religiosity Measures

Each of the articles was also examined with a view toward identifying the relationship (if any) of the religiosity measure upon the dependent variable of delinquency.  Specifically, we were interested in identifying: (1) if the relationship between religiosity and delinquency was not specified; (2) if there was no relationship between religiosity and delinquency; (3) if there was an inverse relationship between religiosity and delinquency; (4) if there was a positive relationship; or (5) if there was a mixed or reciprocal relationship between religiosity and delinquency.

Religiosity Measures

Six categories of religious measures were examined in the current SR: 
(1) Attendance.  As a straightforward measure, church or synagogue attendance has been found to be one of the most commonly used single-item measures for religiosity; (2) Salience.  Researchers sometimes incorporate measures of religious salience  (e.g., importance of one’s religion or God in one’s life) into studies.  Such measures of religiosity can operate independently of other religiosity measures which might for example focus on church attendance (see above) or prayer or Bible study (see below);  (3) Denomination. This particular variable refers to the denominational affiliation of the study subjects (e.g. Catholic, Baptist, Jewish, Muslim, other, or no religious affiliation);  (4) Prayer. This variable typically refers to the degree to which one indicates that prayer is an active and/or meaningful part of one’s life; (5) Study of scripture.  This measure usually refers to the frequency of reading or studying sacred scriptures such as the Bible, Koran or Torah; And, finally, (6) Religious activities.  Generally refers to the recognition that an individual participates in various religious activities both inside and outside of typical church settings.

Number of Research Items used to Measure Religiosity

Often times religiosity has been measured with a single-item like church attendance or level of participation in various religious activities (e.g., small groups, study or mid-week meetings).  In fact, much of the previous research in the religiosity and delinquency area has used church attendance as the sole measure of religiosity (see Evans et al, 1995; Johnson, 1987; Title and Welch, 1983). 

A continuing concern among researchers deals specifically with the question of whether religiosity is best measured as a unidimensional or multidimensional concept.  And though we recognize that single-item measures like church attendance remain a frequently used measure within the literature, the frequent use does not make such a unidimensional assessment an acceptable research practice. It is important to acknowledge that treating religion as a multidimensional concept should be a more methodologically desirable goal (Gorsuch and McFarland 1972).  Therefore, the current systematic review examined the forty published studies to determine how many research items were used to measure religiosity.  For example, we were interested in determining if church attendance only was used (one factor) or if salience and prayer were both used (i.e. two factors). Or finally, if several indicators were used to develop a multidimensional measure of religiosity (i.e. three factors, or four or more factors).

Religiosity Measures and the Influence of Religiosity on Delinquency

One of the goals of the current study was to examine the differences in methodologies between published studies of religiosity and delinquency and to assess if these differences influence researcher findings in regard to the relationship between religiosity and delinquency. We first sought to identify how religiosity was operationalized in the study sample (e.g. attendance, salience, prayer) and then to determine the effect of religiosity measures on delinquency outcomes.  A preferred approach would be to analyze the effect size of each of these studies, whereby we are able to make comparisons between the statistical levels of significance in each of these studies. The issue of effect size is an important feature when examining any body of literature.  However, the forty studies reviewed in our sample do not all contain data necessary to compute statistical effect size that can be generalized across these studies.  As an alternative we computed crosstabulations to determine if the operationalization of religiosity affects research outcomes. While this approach has limitations such as our inability to assess levels of significance and the impact of sample size due to insufficient data, it does allow us to gain important insights into how research on religiosity and delinquency may potentially be impacted by research methodology.

Reliability of Religiosity Measures and the Effect of Religiosity on Delinquency

We also sought to determine whether studies that assessed reliability of religiosity measures would generate results systematically different from those produced by the studies that did not administer reliability tests. Data collection was performed by two trained raters, who reviewed all 40 articles independently. 

Rater reliability was calculated as the percentage of agreement between the two raters for decisions made independently about dimensions of religious measures, effect of religiosity, and quality of research methodology.  This reliability check allowed us to determine the degree to which these independent reviewers reached similar conclusions after reading and coding the 40 studies. Interrater reliability averaged 0.83 for all measures assessed. Interrater reliabilities for the separate variables were as follows: quality of research methodology 0.75, dimensions of religious measures 0.83, and effect of religiosity 0.91. In other words, the coding decisions of the independent reviews were very sufficiently similar to warrant confidence in the reliability or consistency of our  findings.

Results

Quality of Research

Examining a basic frequency distribution of the eleven items outlined on page seven reveals that there were extreme high and low scores, indicating that this group of articles performed well against some criteria but underperformed in some other areas. On the positive side, 85 percent of the studies had no ambiguity about the causal order that they were intended to test (see item 1) and 92.5 percent of the studies used multivariate statistics to test the causal relationship (see item 10). On the negative side, only 12.5 percent of the studies used prospective, longitudinal data, while the remaining 87.5 percent were all based on cross-sectional data (see item 2). Further, only five percent of the studies controlled for mono-method bias (see item 9), and in only half of the articles were there test for the reliability of the study’s measurements (see item 7).

A composite measure of the quality of research methodology is computed by taking the average of these individual items across all studies. The scale of this composite measure ranges from 0 to 1. The 40 articles as a whole had an average score of 0.59 on the quality index. 

Religiosity Measures

The Role of Religiosity Measures

Of the 40 studies reviewed, 27 articles (67 %) treated religion as the central explanatory variable.  To assess religion as a key or central explanatory variable is somewhat unusual given past SRs. Nine treated religion as a peripheral variable and four studies treated religion as a control variable.

The Effect of Religiosity Measures

The vast majority (75%) of studies reviewed revealed that religious measures consistently had a negative, or beneficial, effect on delinquency, whereby higher levels of religiosity were associated with inhibiting or reducing delinquency. Only one of the 40 studies found that religiosity had a positive impact on delinquency, whereby religiosity was associated with increases in delinquency.  This lone study, however, was one of the four studies that utilized religiosity as a control variable. The remaining studies found that the effect of religion was either not significant or inconclusive depending on its interaction with other variables.

Dimensions of Religiosity

Consistent with past systematic reviews of religious measures, salience and attendance were the two most frequently used variables to measure religion (85 percent and 65 percent respectively).  Prayer was used to measure religiosity in 35 percent of the studies.  Participation in religious activities was utilized in 27.5 percent of the studies to measure religiosity, while denomination and Bible study were both used in only 22.5 percent of the forty articles, respectively.

For the sample as a whole, only three studies took account of all six dimensions. Five out of the forty studies included five dimensions and only one study examined four different dimensions.  Thus, 15 of the 40 assessed four or more dimensions of religiosity. The majority of the studies (24 or 60%) measured only 1 or 2 dimensions, usually religious participation and/or religious salience. Twenty-one studies included measures of both religious participation and religious salience.

Number of Items used to Measure Religiosity

Although it has been suggested that it is preferable to use multiple questions or items to measure religion (Gorsuch and McFarland, 1972), most of the studies in our study sample failed to do so.  Less than half of the studies (19 of 40) used more than two items to assess or measure religiosity.  Slightly more than half of the articles (21 of 40) reviewed in the current study measured religiosity with one or two items. 

Dimensions of Religiosity and the Effects of Religiosity on Delinquency

Utilizing a crosstabulation analysis we examined the interaction between dimensions of religiosity and the effects of religiosity on delinquency.  Of the nine studies that measured four or more dimensions of religiosity, all nine found that religiosity had a negative or beneficial effect on delinquency.  Among the 31 studies that measured three or fewer dimensions of religiosity, 21 (67.74%) found that religiosity negatively affected delinquency, five (16.13%) found that religiosity had no association with delinquency (an additional study did not specify an effect), three (9.67%) found interactive or mixed effects, and only one (3.13%) found a positive or detrimental effect between religiosity and delinquency.  A summary of these findings reveal that studies utilizing more religious dimensions to measure religiosity were clearly more likely to demonstrate a negative, or beneficial, relationship finding between religiosity and delinquency. Though a majority of the studies using three or fewer dimensions to measure religiosity continued to indicate this negative relationship, the likelihood of religiosity to have mixed effects or even positive or harmful effects only occurred in those studies utilizing fewer dimensions of religiosity. Thus, the more religiosity is treated as a multidimensional variable, the more likely researchers will find a negative, or beneficial, relationship between religiosity and delinquency.  

Reliability of Religiosity Measures and the Effect of Religiosity on Delinquency

Each of the studies in the sample was reviewed in order to determine if reliability tests had been performed on the study’s religiosity measures.  Of particular interest was determining if those studies that assessed the reliability of the religiosity measures generated results different from those produced by the studies that did not administer reliability tests.

The 13 studies that assessed the reliability of their religious measures all found that religion had a negative or beneficial effect on juvenile delinquency.  In contrast, the 27 studies that did not administer reliability tests yielded somewhat mixed results. Among those studies, 17 (62.96%) still found that religion had a negative effect on delinquency, five  (18.52%) found no effect, three (11.11%) found mixed effect, one article (3.70%) found a positive or harmful effect, and in one last study, no effect was specified. The SR results reveal that the studies using demonstrated reliable measures of religiosity are more likely to find a negative or beneficial relationship between religiosity and juvenile delinquency than those studies that failed to access the reliability of their measurement.  In sum, studies which included reliability measures reached a unanimous consensus that religiosity is inversely related to delinquency, while those which failed to include reliability measures were not able to make such a pervasive claim. Interestingly, the majority of those studies that did not administer reliability measures still found the negative or beneficial relationship between religiosity and delinquency.

Conclusion

The current systematic review of the state of the art of religious measures in crime and delinquency research documented that the role of religiosity in explaining and understanding juvenile delinquency has been an overlooked factor in many studies.  In fact, the present study found that only 10 percent of the quantitative delinquency studies in an eighteen year sample frame included religious variables as either central, peripheral, or covariate measures. Since public opinion polls have consistently shown for decades that a majority of Americans, particularly adolescents, indicate that religion is an important part of their lives (Gallup and Bezilla, 1992), it is indeed intriguing that only 10 percent of the 402 published delinquency studies located from 1980 to 1997 included religious measures.  Of the 40 articles we systematically reviewed, only four included religiosity as a control variable.  By so frequently excluding, and, in particular, not controlling for religious measures, researchers may have misspecified their theoretical models, especially if religiosity serves as a common cofounder to a target relationship.

While there exist some good studies, research on religiosity and delinquency has often been plagued with many methodological problems.  The same remains in the recent studies, representing the present, state of the art of the research.  Many studies in our sample did not use random sampling, did not use multiple indicators to control measurement errors, and did not test for the reliability of their religious measures.  Almost all of the studies had mono-method bias. In addition, very few studies were based on longitudinal data.

The current review reminds us that research methodology can have an important effect on research findings. Studies that adopted multiple indicators to measure religion consistently found that religiosity was negatively related to delinquency.  Likewise, studies that selected religious measures by means of reliability tests also found that religion consistently had a negative effect on delinquency. In contrast, studies that generated mixed findings regarding the impact of religiosity on delinquency did not use multiple indicators and did not administer reliability tests.  The results of the current review suggest that the previously assumed “inconsistent findings” inconsistent findings regarding the role of religion in explaining delinquency are due at least in part to different research strategies employed in the state of the art sociological and criminological research.  With improvement in measurement and analytic methods, we should expect more consistent empirical results.

Finally, most of the studies we reviewed showed that religion had an inverse, or beneficial, impact on delinquency. Notably, this was particularly true with the studies that demonstrated higher quality of research methodology as discussed above. Survey research has long indicated that a majority of American youth are exposed to religion early in their lives. As this systematic review found, given such population prevalence, commitment to religious values and beliefs can have both an immediate and a long-term impact on deviant or delinquent behavior.  Religion is a large part of many adolescents' lives, but it remains is a small part of the criminological research.  Unless this disparity is reconciled, researchers will limit unnecessarily their ability to understand this phenomena in delinquency research.

References

Bareta, J.C., D.B. Larson, J.J. Zorc, J.S. Lyons. 1990. “A Comparison of the MEDLARS and Systematic Review of the Consultation-Liaison Psychiatry Literature.” American Journal of Psychiatry 147:1040-1042.

Beardsley, R.S., D.B. Larson, J.S. Lyons, G. Gottlieb, P.V. Rabins, and B. Rovner. 1989. “Health Services Research in Nursing Homes: A Systematic Review.” Journal of Gerontology: Medical Sciences 41:30-35.

Benda, B. 1995. “The Effect of Religion on Adolescent Delinquency Revisited.”
Journal of Research in Crime and Delinquency 32:446-466.

Brownfield, D. and A. M. Sorenson. 1991. “Religion and Drug Use among
Adolescents: A Social Support Conceptualization and Interpretation.” Deviant Behavior 12:259-276.

Cochran, J., P. Wood, and B. Arneklev. 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.

Cook, T. and D. Campbell. 1979.  Quasi-Experimentation: Design & Analysis Issues for Field Settings. Boston: Houghton Mifflin.

Evans, T. D., F. Cullen, R.G. Dunaway, and V. S. Burton, Jr. 1995.  “Religion and Crime              Reexamined:   The Impact of Religion, Secular Controls, and Social Ecology on Adult Criminality.” Criminology 33:195-217.

Gallup, G. and R. Bezilla. The Religious Life of Young Americans. Princeton, NJ: The George Gallup International Institute

Gorsuch, R. L. and S. G. McFarland. 1972. “Single vs. Multiple-Item Scales for Measuring Religious Values.” Journal for the Scientific Study of Religion 11:53-63.

Johnson, B.R. 1987. “Religiosity and Institutional Deviance: The Impact of Religious Variables Upon Inmate Adjustment,” Criminal Justice Review 12 (1):21-30, 1987.

Larson, D.B. 1993.  The Faith Factor, Volume II: An Annotated Bibliography of Systematic Reviews and Clinical Research on Spiritual Subjects.  Rockville, MD: National Institute for Healthcare Research.

Larson, D.B. and S.S. Larson. 1994 The Forgotten Factor in Physical and Mental Health: What Does the Research Show.  Rockville, MD: The National Institute for Healthcare Research.

Larson, D.B., J.S. Lyons, A.H. Hohmann, R.S. Beardsley, W.M. Huckeba, P.V. Rabins, and B.D Lebowitz. 1989. “A Systematic Review of Nursing Home Research in Three Psychiatric Journals.” International Journal of Geriatric Psychiatry 4:129-134.

Larson, D.B., E.M. Pattison, D.G. Blazer, A.R. Omran, and B.H. Kaplan. 1986. “Systematic Analysis of Research on Religious Variables in Four Major psychiatric Journals, 1978-1982.” American Journal of Psychiatry 143:329-334.

Larson, D.B., K.A. Sherrill, J.S. Lyons, F.C. Craigie, S.B. Thielman, M.A. Greenwold, and S.S. Larson. 1992. “Associations between Dimensions of Religious Commitment and Mental Health Reported in the American Journal of Psychiatry and Archives of General Psychiatry: 1878-1989.” American Journal of Psychiatry 149:129-134.

Lyons, J.S., D.B. Larson, J.C. Bareta, A.H. Hohmann, I.Y. Liu, and C.H. Sparks. 1990. “A Systematic Analysis of the Quality of AIDS Publications and the Quality of Research Methods in Three General Medical Journals.” Evaluation and Program Planning 13:73-77.

Tittle, C. and M. Welch. 1983. “Religiosity and Deviance: Toward a Contingency Theory of Constraining Effects.” Social Forces 61: 653-682

Appendix A

Articles Comprising the Systematic Review:
January, 1980 – December, 1997

1. Barnes, G. M., Farrell, M. P., and Banerjee, S. (1994). Family influences on alcohol abuse and other problem behaviors among black and white adolescents in a general population sample. Special issue: preventing alcohol abuse among adolescents: preintervention and intervention research.  Journal of Research on Adolescence 4: 183-201.

2. Bahr, S. J., Hawks, R. D., and Wang, G. (1993). Family and religious influences on adolescent substance abuse. Youth and Society 24: 443-465.

3. Benda B. B. (1997). An examination of a reciprocal relationship between religiosity and different forms of delinquency with a theoretical model. Journal of Research in Crime and Delinquency 34: 163-186.

4. Benda, B. B. (1994). Testing competing theoretical concepts: Adolescent alcohol consumption. Deviant Behavior 15: 375-396.

5. Benda, B. B. (1995). The effect of religion on adolescent delinquency revisited. Journal of Research in Crime and Delinquency 32: 446-466.

6. Benda, B. B., and Whiteside, L. (1995). Testing an integrated model of delinquency using LISREL. Journal of Social Service Research 21: 1-32.

7. Bliss, S. K., and Crown, C. L. (1994). Concern for appropriateness, religiosity, and gender as predictors of alcohol and marijuana use. Social Behavior & Personality 22: 227-237.

8. Brownfield, D., and Sorenson, A. M. (1991). Religion and drug use among adolescents: A social support conceptualization and interpretation. Deviant Behavior 12: 259-276.

9. Carlucci, K., Genova, J., and Rubackin, F. (1993). Effects of sex, religion, and amount of alcohol consumption on self-reported drinking-related problem behaviors. Psychological Reports 72: 983-987.

10.  Chadwick, B. A., and Top, B. L. (1993). Religiosity and delinquency among LDS adolescents. Journal for the Scientific Study of Religion 32: 51-67.

11.  Cochran, J. K. (1989). Another look at delinquency and religiosity. Sociological Spectrum 9: 147-162.

12.  Cochran, J. K. (1988). The effect of religiosity on secular and ascetic deviance. Sociological Focus 21: 293-306.

13.  Cochran, J. K. (1993). The variable effects of religiosity and denomination on adolescent self-reported alcohol use by beverage type. Journal of Drug Issues 23: 479-491.

14.  Cochran, J. K., and Akers, R. L.  (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.

15.  Cochran, J. K., Wood, P. B., and Arneklev, B. J. (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.

16.  Donahue, M. J. (1995). Religion and the well-being of adolescents. Journal of Social Issues 51: 145-160

17.  Dudley, R. L., Mutch, P. B., and Cruise, R. J. (1987). Religious factors and drug usage among seventh-day adventist youth in north America. Journal for the Scientific Study of Religion 26: 218-233.

18.  Ellis, L., Thompson, R. (1989). Relating religion, crime, arousal and boredom. Sociology and Social Research 73: 132-139.

19.  Engs, R. C., Diebold, B. A., and Hanson, D. J. (1996). The drinking patterns and problems of a national sample of college students, 1994. Journal of Alcohol & Drug Education 41: 13-33.

20.  Evans, T. D., Cullen, F. T., Burton, V. S., Dunaway, R. G. et al (1996). Religion, social bonds, and delinquency. Deviant Behavior 17: 43-70.

21.  Fernquist, R. M. (1995). A research note on the association between religion and delinquency. Deviant Behavior 16: 169-175.

22.  Free, M. D. (1991). Clarifying the relationship between the broken home and juvenile delinquency: a critique of the current literature. Deviant Behavior 12: 109-167.

23.  Free, M. D. (1994). Religiosity, religious conservatism, bonds to school, and juvenile delinquency among three categories of drug users. Deviant Behavior 15: 151-170.

24.  Free, M. D. (1992). Religious affiliation, religiosity, and impulsive and intentional deviance. Sociological Focus 25: 77-91.

25.  Free, M. D. (1993). Stages of drug use: A social control perspective. Youth and Society 25: 251-271.

26.  Hardert, R. A., and Dowd, T. J. (1994). Alcohol and marijuana use among high school and college students in Phoenix, Arizona: A test of Kandel’s socialization theory.  International Journal of the Addictions 29: 887-912.

27.  Hawks, R. D., Bahr, S. J., and Wang, G. (1994). Adolescent substance use and codependence. Journal of Studies on Alcohol 55: 261-268.

28.  Humphrey, J. A., Leslie, P., and Brittain, J. (1989). Religious participation, southern university women, and abstinence. Deviant Behavior 10: 145-155.

29.  Ingram, A. L. (1993). Type of place, urbanism, and delinquency: further testing the determinist theory.  Journal of Research in Crime and Delinquency 30: 192-212.

30.  Johnson, R. E., Marcos, A. C., and Bahr, S. J. (1987). The role of peers in the complex etiology of adolescent drug use. Criminology 25: 323-340.

31.  Mitchell, J., Dodder, R. A., and Norris, T. D. (1990). Neutralization and delinquency: A comparison by sex and ethnicity.  Adolescence 98: 487-497.

32.  Nagin, D. S., and Paternoster R. (1991). On the relationship of past to future participation in delinquency. Criminology 29: 163-189.

33.  Peek, C. W., Curry, E. W., and Chalfant, H. P. (1985). Religiosity and delinquency over time: Deviance deterrence and deviance amplification. Social Science Quarterly 66: 120-131.

34.  Perkins, H. W. (1987). Parental religion and alcohol use problems as intergenerational predictors of problem drinking among college youth. Journal for the Scientific Study of Religion 26: 340-357.

35.  Ross, L. E. (1994). Religion and deviance: Exploring the impact of social control elements. Sociological Spectrum 14: 65-86.

36.  Sloane, D. M., and Potvin, R. H. (1986). Religion and delinquency: Cutting through the maze.  Social Forces 65: 87-105.

37.  Stark, R. (1996). Religion as context: hellfire and delinquency one more time. Sociology of Religion 57: 163-73.

38.  Walsh, A. (1995). Parental attachment, drug use, and facultative sexual strategies. Social Biology 42: 95-107.

Yarnold, B. M., Patterson, V. (1995). Factors correlated with adolescents’ use of crack in public schools. Psychological Reports 76: 467-474.

Zimmerman, M. A., and Maton, K.I. (1992). Life-style and substance use among male African-American urban adolescents: A cluster analytic approach.  American Journal of Community Psychology 20: 121-137.

Section II

Religiosity and At-Risk Urban Youth

In Section I we systematically reviewed the recent religiosity and delinquency research and summarized a number of methodological shortcomings present in many of these studies.  In Section II of this paper we present in summary fashion the research results from three forthcoming studies which offer important theoretical and methodological improvements to the research publications systematically reviewed in Section I.

Religiosity and Young Black Males

In the first study we examine the efficacy of churchgoing to help young black males residing in poverty tracts in Boston, Chicago, and Philadelphia, to escape criminal and delinquent activity so often associated with at-risk youth living in urban communities. Drawing upon relevant criminological theories and utilizing hierarchical regression analysis, the findings confirm that churchgoing is significantly associated with lower levels of crime and delinquency.  This finding holds after controlling for various demographic, family, and other social control variables, suggesting that churchgoing has its own unique effect on crime prevention.

Black Churches and Youth Outreach

The second study builds upon the beneficial finding of churchgoing among young black males in poverty tracts by examining the role of the African-American church to reduce delinquency among black youth residing in urban communities in a national sample.  Improving on earlier research by incorporating a multidimensional measure of religiosity and utilizing a longitudinal study design, the hierarchical regression models demonstrate that religiosity is significantly linked to reduced levels of minor and general delinquency over time. The historical legacy of the African-American church as an agency of social control is well documented in the sociology literature, but the current longitudinal finding is the first to demonstrate its significance in reducing delinquency over time.  Indeed, the beneficial legacy of the African-American church is one that continues to the present.

Religiosity and Youth

The third study examines the indirect and directs effects of religiosity upon delinquency in a nationally representative sample of youth in a longitudinal panel study. Religiosity, a multidimensional variable, was found to have a positive direct effect on adolescent belief and a negative indirect, or beneficial, effect on delinquency in three separate time periods.  The effects of religiosity among both white and black youth remain significant even after controlling for peer association, belief, and other background variables.  The results suggests that religiosity can influence current behaviors and attitudes of youth, and thus can affect their involvement in delinquency.

Escaping from the Crime of Inner-Cities:
Religiosity and Young Black Males18 

Sociologists and criminologists have often studied the propensity for crime or deviance among at-risk adolescent groups in inner-cities, but the role of religious institutions has often been overlooked or ignored (Larson 1993; Larson and Larson 1994). Freeman’s (1986) finding of the importance of churchgoing in helping inner-city black male youth escape from the world of poverty, drug use, and crime is a rare, much needed research effort with substantial implications for criminological theory.  For example, the influence of religious belief in social learning as well as the role of the faith community in enhancing social control, particularly in urban areas often typified by social disorganization and decay is worthy of much more extensive study.

The present study is intended to extend and modify Freeman’s economic model into one that fits into a criminological framework, especially within the context of research on the effects of religiosity on youth deviance. The core hypothesis of the present study concerns whether a youth’s church attendance has any independent effect on delinquent behavior, especially among inner-city black youth.  Although church attendance might be found to have little or no significant direct effect on deviance, we are still interested in identifying potential intervening variables between church attendance and deviance.

Data from the National Bureau of Economic Research
 
The present study will be primarily focussed on reanalyzing data collected on inner-city black youth by the National Bureau of Economic Research (NBER), a coalition of research economists interested in studying black youth joblessness in America. NBER researchers developed a survey (Survey of Inner-City Black Youth) that made it possible to study the problems facing black youth including delinquency, drug use, and school dropout. NBER researchers developed this survey as a response to the perceived inadequacy of existing governmental data,19  to study and understand the problem of employment among American black youth in the inner-city.20 

Measures of Delinquency

In the NBER data set, delinquency is measured on a dichotomous, or “yes – no,” scale. The respondents were asked whether or not they committed any of nine different illegal activities (played street numbers or gambling; sold or fenced stolen goods; sold marijuana or other drugs; committed burglary, larceny, or auto theft; shoplifted or stole from cars or trucks; committed muggings or purse snatchings; committed robberies or stickups; cashed or forged stolen checks; run or been part of any con games, swindles, or frauds) over the past 12 months. All nine of these activities can be determined as serious offenses.  Since we do not know the frequency of offenses and all the known offenses are serious ones, we use a simple dichotomous variable to measure delinquency involvement. Respondents who reported involvement in any of the offenses during the past 12 months are classified as delinquents.  Those who did not report any involvement are treated as non-delinquents.

Drug use in the NBER data is also measured on a dichotomous scale. Respondents were asked whether they used marijuana or other drugs such as cocaine, heroin, barbiturates, amphetamines or LSD at the time of the interview.  Those who reported use of any drug are given the value of 1. Those who did not report drug use are assigned the value of 0. Ideally, regular and habitual users should not be distinguished from experimental users – those who use drugs four times or less (see Elliott, Huizinga, and Menard, 1989).

Two measures of alcohol use are also introduced in the analysis. In the descriptive analysis, a simple dichotomous measure is used to distinguish frequent users from nonusers and infrequent users. Respondents who used alcoholic beverages every day or almost every day are given a value of 1.  Otherwise,  they are given the value of 0.  In the hierarchical regression analysis to follow, an index of alcohol consumption was used to measure alcohol use in the NBER data.  The index consists of three items: frequency of drinking beer, frequency of drinking wine, and frequency of drinking hard liquor. The reliability coefficient (a) for the three-item index is 0.731.

Hierarchical Models for the NBER Sample

We conducted a series of hierarchical regression analyses to test the relationship between religiosity and delinquency. A hierarchical analysis allows us to adjust for different sources of study impact or influence so that the net effect of religious commitment or religiosity on delinquency can be determined.  For each delinquency measure, we first examined the effects of religious measures on the outcome. We then entered a set of  background variables into the equation to test whether the impact of religiosity on delinquency is spurious. Age, family structure, family size, welfare status, and illegal opportunities have been shown in past research to correlate strongly with delinquency. Previous studies have also suggested that urban residence has an impact on the commission of illegal behavior. Two dichotomous variables measuring urban residence (i.e., Boston and Chicago) were also entered into the equation to control for this geographic effect.

The results indicate that church attendance: (1) has a direct inverse, or beneficial, impact on delinquency, drug use, and alcohol use; and (2) has a positive influence on a juvenile’s behavior net of the impact of age, family structure, family size, urban residence, and illegal opportunities.  Regular attendance at religious services seems to produce many socially desirable outcomes, such as strong attachment to education and stable job records.  The results indicate that in addition to its direct effect on delinquency, church attendance also indirectly reduces delinquency involvement by fostering stronger social bonds, good peer relationships, and high involvement in productive social activities.  Though the inner-city may present the same challenges for all the kids who live in it, those who attend religious services seem to have a much better chance to build strong attachment and commitment to conventional values and activities such as the church and school, which in turn decreases their involvement in delinquency.

In sum, these findings suggest that when it comes to church attendance, the less deviant behavioral patterns among churchgoing inner-city black youth cannot be accounted for by either background factors or social process variables (i.e., social bonding and social learning variables).  These findings provide evidence that church attendance itself has its own unique effect on deviance among inner-city black youth.  Further, the findings lend support to the notion that religiosity, as measured by church attendance, appears to be a strong protective factor in insulating at-risk youth from the crime and delinquency of inner-city poverty tracts. 

Conclusions

We find that regular church attendance has a consistent direct impact on all three types of deviance (i.e., illegal activities, drug use, and alcohol use) for those high-risk youth. This pattern remains even after controlling for background and non-religious or secular bonding and learning variables. As expected, the effect of church attendance on deviance somewhat decreases when those variables are held constant, indicating that the effect is indirect as well as direct.  However, the lower levels of deviance reported by inner-city black male youth who are religiously committed could not be fully explained by their stronger bond to family and school, more conventional peer networks, or higher involvement in productive activities. 

Thus, a policy implication from this finding would be to bolster weakened social control mechanisms, such as the family, and to encourage utilization of other urban institutions, like churches, synagogues, or mosques, whose ability to influence youth through informal means of social control remain largely intact.  Criminologists have historically focused on the school and family, paying little attention to religion when they address the etiology of deviance among at-risk urban youth whose chance to escape from the substance abuse and crime of inner-cities is relatively low. Therefore, from a social disorganization perspective for at-risk youth in poor urban environments, the church can be viewed as a remnant of social organization amidst the otherwise disorganized and troubled areas so often found in inner-cities.

One might also argue that participating in a church community is much of what leads to the protective or beneficial influence of religion. From a life course perspective, it is the quality of relationships formed within the participation in that faith community (i.e., whether church, synagogue, mosque, or temple) that is important.  In this way, youth commitment to the church can be seen as analogous to other life changing events such as employment or marriage, that research shows can help shape in a pro-social or positive fashion, an individual’s behavioral trajectory over the life course.

Further, churchgoing may play a key a role as a protective factor that insulates inner-city black male youth from various forms of deviance.  Church teachings in general, tend to run counter to various forms and expressions of anti-social behavior such as that espoused by gangs.  Therefore, social support networks and the influence of mentors provide what criminologists call “buffers” that help protect otherwise vulnerable youth to the criminal and delinquent elements adversity of poverty tracts in the inner-city.  Indeed, if churchgoing matters and is a source of resiliency among youth, then we have a basis for arguing that we need to support public and private policies that strengthen inner-city churches and bring youth into these churches.

Finally, the authors believe these findings have both theoretical as well as policy implications.  First, this study adds to the small, but mounting, body of evidence already suggesting that religiosity is a neglected, but relevant, variable in delinquency research.  Second, the findings suggest that at a minimum, religiosity is a variable with considerable relevance for social disorganization, life course, and resilience perspectives.  Additional research is needed which examines additional ways religiosity may influence theories of social control as well as its unique contribution to pathways and turning points in the life course.  We need to know more about how religious commitment, religious practices, religious institutions are linked to protective factors and the resiliency of youth in general, and specifically, youth from socially disorganized communities.

Black Churches and Youth Outreach21

The historical significance of the African-American church as an agency of social control and organization is well documented in the sociology literature (see for example: DuBois, 1898, 1903; Frazier 1963; Lincoln, 1974; Lincoln and Mamiya, 1992; Mukenge, 1983; Nelsen, Yokley, and Nelsen, 1971; Paris, 1985; and J. Washington, 1964). However, one will search the criminological literature with difficulty to locate empirical studies examining the role of the black church as agency of social control with regard to crime and delinquency. 

Therefore, the current study examines the significance of the African-American church in impacting delinquency among black youth living in urban communities. Bursik and Grasmick (1993) point out that local religious institutions such as churches or mosques can play an important role as the agent of community socialization and informal social control. That is, the African-American church can provide relational networks of community social organization for urban black youth living in environments of structural disadvantages that have been shown to be predictors of adolescent deviance.

Urban communities are often characterized by various structural disadvantages, such as family disruption and unemployment (Wilson 1987), which put youth who live in those communities at increased risk of delinquency, especially criminal violence (Sampson 1987).  However, it is also true and should be underscored that many urban youth do not turn to delinquency as a result of living in disadvantaged communities. That is, there exists a significant proportion of children who develop through adolescence without serious problems, such as juvenile delinquency, in the face of great adversity during the early and formative years of life (e.g., Williams and Kornblum 1985).  Often referred to as “resilient youth,” developmental criminologists have recently begun to focus on factors that not only tend to protect at-risk youth from the risk factors associated with structural disadvantage, but help provide an explanation of the significant within-group variation in delinquent involvement (Rutter 1985, 1988; Rutter and Giller 1983; Smith et al. 1995; Werner 1989; Werner and Smith 1992).

Identifying protective factors which may serve to buffer or shield at-risk children and adolescents like urban black youth from negative behavioral outcomes (e.g. delinquency, drug use, school dropout) has strong implications for delinquency prevention and intervention policy as well as important theoretical implications.  Though previous research has mostly focused on risk factors, resilience research is far more promising in that it takes into consideration the contribution of often overlooked protective factors. For example, social disorganization theory can be expanded by incorporating community-level protective factors like local religious institutions given that social organization and social disorganization are conceptualized as opposite ends of the same continuum of systemic networks of community social control (Bursik and Grasmick 1993; Elliott et al. 1996; Sampson and Wilson 1995).

Resilience researchers have begun to examine various sources of protective factors such as family, school and peers (Smith et al. 1995), but like many other research fields (Larson 1993; Larson and Larson 1994; Larson et al., 1997) have largely ignored religion though it has been shown to protect children and adolescents from negative behavioral outcomes (Anthony and Cohler 1987; Gartner, Larson, and Allen, 1991; Stark 1996).  Criminological research on the effects of individual religiosity on adolescent deviance also provides evidence that religious adolescents (e.g., those who regularly attend church) are less likely than their non-religious peers (e.g., those who never or irregularly attend church) to engage in deviance, especially what has been referred to as “ascetic” deviance like alcohol and drug use (Benda and Corwyn 1997; Brownfield and Sorenson 1991; Burkett and Warren 1987; Cochran and Akers 1989; Elifson et al. 1983; Evans et al. 1995; Higgins and Albrecht 1977; Johnson et al. 1997a, 1997b; Tittle and Welch 1983).  What is relatively lacking, however, is research that investigates whether individual religiosity protects at-risk adolescents such as urban youth from engaging in delinquency.

The present study defines individual religiosity as the extent to which an individual is committed to religion he or she professes and its teachings so that the individual’s attitudes and behaviors reflect such commitment. Applying this definition to the present study, urban black youth who are religiously committed are expected to (1) more frequently attend religious services, (2) more often participate in religious activities besides regular services, and (3) attribute more significance to religion and religious activities in his or her life as compared to those who are less committed.

Measurement

This study incorporates four sets of variables, including: (1) measures of delinquency (divided into three categories: minor delinquency, general delinquency, and serious delinquency); (2) an index of religiosity (measured by four variables: frequency of attending religious services; time spent on community-based religious activities during weekends; salience; and importance of involvement in religious activities); (3) measures of social control and delinquent peer association; and (4) a set of variables measuring demographic characteristics and family background.

Data come from Wave 3, Wave 4, and Wave 5 of the National Youth Survey (NYS), a multi-wave panel survey based on a national probability sample. The survey started in 1977 with a sample of 1,725 youth ranging in age from 11 to 17.  The survey items used in the NYS have been validated in numerous refereed publications in criminology and sociology over the last decade. 

Bivariate Analysis

Bivariate scatterplots between religiosity and delinquency demonstrated that most minor, general, and serious delinquent acts were committed by juveniles who had low levels of religious commitment.  Those juveniles whose religiosity levels were in the middle to high levels committed very few delinquent acts.  However, to determine the actual impact of religiosity on the delinquency measures we ran a series of multivariate models that allowed us to control the influence of any number of variables.

Hierarchical Regression Analysis

We ran three separate models to test the effect of religiosity on minor, general, and serious delinquency when controlling for other explanatory variables in the model. In all three models, the effects of religiosity, social bond, and peer association were assumed to have similar effects on this group of individuals.  This is a reasonable approach because the subjects of this study belonged to a relatively homogenous group (i.e. urban black youth), compared to the general population. 

Both hierarchical regression models with minor and general delinquency as the response variables indicated that religiosity had a negative effect on delinquency, controlling for social bond, association with delinquent peer, demographic characteristics and family background. In these models, religiosity was one of only four variables (the others being age, gender, and belief in conventional values) that were significantly related to minor and general delinquency.

The effect of religiosity on serious delinquency was small and non-significant.  Our inability to detect a significant relationship between religiosity and serious delinquency might be due at least in part to the fact that very few respondents in the small sample we examined committed serious delinquency. Because serious delinquency is less prevalent than minor or general delinquency, a larger sample is needed to accurately test the relationship between religiosity and serious delinquency.

Conclusions

Prior delinquency research has not examined the role of the African-American church as an agency of local social control that might help black youth to be resilient to and thus resistant to delinquency.  The present study examined a national sample of urban black youth to determine the relationship of religious commitment to delinquency in urban areas.

Urban black youth tend to come from more disadvantaged backgrounds and report more involvement in deviant as well as non-productive activities than other black or white youth. We have found that higher levels of religiosity have a beneficial impact on lower levels of delinquency over time.  This pattern remains even after controlling for background and non-religious or secular bonding and learning variables.

The present findings have theoretical, methodological, and policy implications.  First, the current findings provide evidence that religiosity can be seen as a relevant protective factor for urban youth. As resilience researchers continue to pursue factors that may provide buffers from deviant activities for at-risk youth, religiosity measures should clearly not be overlooked in such future studies.  Similarly, more research is needed which examines more closely what contribution religious variables may make to our understanding of delinquency over the life-course.  

Second, much of the prior research on religiosity has relied upon single-item measures of religiosity rather than multidimensional measures. In the present study we have created an index of religiosity that consists of four observed variables that capture levels of religious participation as well as the intensity of religious commitment.  We believe that inadequate measurement and operationalization of religiosity in previous research has contributed at least in part, to the mixed results so often referred to in the literature. Further, utilizing longitudinal panel data allowed us to examine in a multilevel modeling approach, the effects of religiosity over time, a feature that has been missing from the religiosity-delinquency literature.

Third, the current study suggests that future delinquency prevention and intervention policies that overlook the role of the African-American church in reducing delinquency among black youth in urban areas will, at a minimum, be unnecessarily short-sighted. In the spirit of multidisciplinary and multifaceted approaches to various social problems, it would seem that common sense, coupled with empirical evidence, warrant the inclusion of the faith community in various partnership strategies to fight urban delinquency. This is quite important given the prevalence of religious belief among American adolescents. Though more research is needed in this area, the current study provides important nationally-relevant evidence that the African-American church can play a key role as an agency of local social control in communities too often typified by disorganization and disadvantage.

Religion and Youth22

In this article we make the theoretical argument that religious commitment can be seen as influencing the key processes of social learning: differential reinforcement, definitions, differential association, and imitation.  Drawing mainly upon Akers’ social learning perspective, we outline several hypotheses that lead us to expect direct as well as indirect effects of adolescent religiosity on delinquency. The behavior of adolescents who are religiously committed is expected to be strengthened through: (1) the learning from religious beliefs that are opposed to deviant behavior; (2) being exposed to religious as well as behavioral patterns that are pro-social (i.e. conventional or law-abiding); and (3) modeling the behavior and attitudes of pro-social spiritual role models that youth like and/or respect.

As outlined in Section I of this paper, prior research in the area of religiosity and delinquency has a number of methodological shortcomings. One of these flaws is the failure to pay adequate attention to both indirect and direct effects of religious commitment variables in regard to various measures of delinquency. For example, if religiosity is not found to be significantly related to delinquency in a particular study, one cannot state that there is no relationship between these two variables.  If, for example, increasing religiosity is directly related to a variable called belief, and belief is directly related to delinquency, then the indirect relationship between religiosity and delinquency is indeed measurable and important.  Further, the sum total of all the indirect effects can be substantial in explaining variance in the dependent variable even though there may not be a direct link between the two.

Data and Sample

This study examines the relationship between religiosity and delinquency using data from the National Youth Survey.  We selected data from Wave 3, Wave 4, and Wave 5 of the National Youth Survey (NYS) since the data in each of these waves contains an identical list of variables that can be used to assess both religiosity as well as delinquency.  Because the variables we selected represent the same attributes measured at different points of time, we can use them to test whether the same relationships hold true across time. The NYS is a multi-wave panel survey based on a national probability sample. The study started in 1977 with 1,725 youth aged 11 to 17.  The sample originally selected was representative of the total 11-year old through 17-year old youth population in the United States as established by the U.S. Census Bureau (Elliot & Ageton, 1980).

Variables and Measurement

The same set of variables is used in all three waves.  The variables are either composite measures or factors represented by several indicators. We deliberately incorporate latent variables into our analysis to control for measurement errors, which are often unexamined in previous studies of religion and crime (Tittle & Welch 1983; Burkett & Warren 1987).  Failure to control measurement errors can have serious consequences.  One of these is inconsistency in parameter estimation, which can result in either overestimation or underestimation of the population parameters (Bollen 1989:151-178). To minimize the effects of measurement errors, most of the variables in our model are measured by multiple indicators.

Social bond elements proposed by Hirschi (1969), including attachment, commitment, involvement, and belief, are often cited as important correlates of religion and crime (Brenda 1997; Burkett & Warren 1987). Researchers argue that a strong social bond attenuates the effect of religion because the restraining effect of religiosity may not be necessary when personal bond to family, school, and conventional values is sufficiently strong (Cochran et al. 1994).  Thus in addition to personal belief, we controlled several additional measures of social bond, including attachment to parents and school and commitment to conventional social activities, in a separate analysis. These social bond variables did not have significant effects on the outcome variables. This finding is consistent with Agnew’s (1991) finding that social bond does not contribute significantly to the explanation of delinquency in the NYS data.  Therefore, based on our preliminary analysis and previous literature, we eliminated the measures of attachment and commitment from the model.

Two sets of causal effects are specified in the study. The first set of effects consists of contemporaneous effects among the variables in the same wave. Consistent with our hypothesis, religiosity is proposed to have direct effects on delinquent attitudes, delinquent peer associations, and delinquency. Delinquent attitudes are expected to have direct effects on delinquent peer associations and delinquency. Finally, delinquent peer associations are expected to have a direct effect on delinquency.  The second set of relationships comprises stability effects among the same attitudinal and behavioral variables measured at different waves.

Direct Effects

Based on the development of the social learning processes, we expect the effect of religiosity on delinquency to remain significant after controlling for other religious as well as secular variables that we hypothesize mediate some portion of the behavioral impact of religiosity. In this study we focused on two key social learning concepts as intervening variables between adolescent religiosity and delinquency: (1) delinquent attitudes and (2) associations or relationships with delinquent peers. We hypothesize that delinquent attitudes have significant direct effects on delinquent relationships as an adolescent’s belief system plays a significant role in determining the normative or law-abiding behavior of his/her peer group (Hirschi 1969).

Indirect Effects

The centrality of relationships with delinquent peers (henceforth, delinquent relationships) in the social learning perspective is well documented in the literature (Akers 1985, 1997; Akers et al. 1979; Elliott, Huizinga & Ageton 1985; Thornberry et al. 1994; Warr 1993). The attitudes of adolescents and those of their friends, has a strong, independent effect on adolescents’ behavior (Thornberry et al. 1994).  Therefore, we hypothesize that the effect of religiosity on delinquency is partly mediated by delinquent relationships to the extent to which an adolescent’s religious commitment enhances the selection of conventional or law-abiding friends and the avoidance of delinquent friends.

Findings

Direct Effects of Religiosity

In general, the effects of religiosity are consistent with our hypotheses. Religiosity has a positive direct effect on delinquent attitudes and a negative direct effect on delinquency in all three waves.  In addition, religiosity also has a negative direct effect on association with delinquent peers in Wave 3.  All the direct effects, except the one from religiosity to delinquent attitudes in Wave 3, are moderate in strength.  The highest standardized coefficient in absolute term is 0.17 and the lowest is 0.07.


Indirect Effects of Religiosity

In addition to the direct effects, religiosity also has significant, indirect effects on delinquency in all three waves.  The indirect effects of religiosity on delinquency in Wave 3 are especially strong.  Through its direct impact on delinquent attitudes and peer association, religiosity in Wave 3 exerts a significant, negative effect on delinquency in Wave 3. It also has negative, indirect effects on the delinquency measures in Wave 4 and Wave 5 through other intermediate variables. The standardized coefficients of the three indirect effects are -0.23, -0.30, and -0.35, for Wave 3, 4, and 5 respectively indicating that the indirect effects of religiosity in Wave 3 on delinquency becomes progressively stronger from Wave 3 to Wave 5.

Total Effects of Religiosity

Overall, religiosity in Wave 3 has a strong, negative total effect on delinquency in all three waves. The effects of religiosity in Wave 4 and Wave 5 follow the same pattern. They all have a negative total effect on delinquency in the same wave, and, when applicable, on delinquency in the subsequent wave.  The indirect impact of religiosity in Wave 3 on delinquency is especially strong. More than two-thirds of its total impact on delinquency in any of the three waves consists of indirect effects through intermediate variables, including delinquent attitude, peer association, and religiosity in later waves.

In comparison with the other variables in the model, religiosity, especially religiosity in Wave 3, contributes significantly to the explanation of delinquency in the adolescent’s life.  The two variables most strongly related to delinquency are association with delinquent peers and religiosity.  These two variables directly affect delinquency across three waves.  In addition, the total effects of these two factors on the delinquency measures are all significant.  Peer association has a stronger effect on delinquency than religiosity, but the effects of religiosity, however, remain quite robust after controlling for peer association, delinquent attitudes, and the background variables.

The major findings regarding religiosity can be summarized as follows: (1) All three measures of religiosity have direct, negative effects on delinquency; (2) Religiosity in Wave 3 has the strongest effect on delinquency, suggesting that religious commitment and involvement in early age has a strong inhibiting effect on delinquency; (3) All three measures of religiosity have significant indirect effects on delinquency. The indirect effect of religiosity in Wave 3 is much stronger than its direct effect.  This result suggests that one can seriously underestimate the total effect of religiosity if one focuses on either only direct or else on only indirect effects; and (4) Religiosity remains an important explanatory variable of delinquency after the effects of peer association, delinquent attitudes, and the exogenous variables are controlled. Though association with delinquent peers has a stronger total effect on delinquency than religiosity, the total effect of religiosity remains substantially robust.

These results indicate that a youth’s current behavior and attitude, such as church attendance and conventional belief, can affect their involvement in delinquency one year or even two years later by increasing their future participation in the same type of behavior and by strengthening their belief in the same value system.  In sum, the variables in the model show both strong contemporaneous effects and stability over time.  The results suggest an adolescent’s behavior may be more affected by current rather than past events and beliefs.

Discussion and Conclusion

We proposed that religiosity would have a direct effect on delinquency independent of the effects of other social and economic variables. We find that religiosity has a consistent direct effect on delinquency, independent of the effects of all the other variables controlled in the model. We also find that the indirect effects of religiosity through delinquent attitudes and peer associations to be significant and substantial.

In addition to theoretical limitations, previous research concerning the effects of religiosity on deviance and crime has also been plagued by methodological limitations. Rarely have such studies used nationally representative or longitudinal data, and thus, tend to lack generalizability. Further, studies in this area tend to use data from relatively small samples, often having problems with response rates and missing data.  And, finally, the concept of religiosity has often been measured with only single items.

In the current study we have eliminated many of the methodological flaws that have too often plagued studies examining the relationship between religiosity and delinquency. The NYS is a longitudinal and nationally representative data set, making our findings more generalizable than previous studies.  Further, we have operationalized religiosity as a multidimensional variable, thereby measuring the influence of religious attendance and activities as well as indicators of salience.  Finally, given that religiosity is an abstract and, thus, not directly observable construct, we have modeled the concept as a latent variable and applied a structural equation model to examine the indirect and direct effects of religiosity.

In light of these findings we are intrigued that among the host of published delinquency studies using data from the NYS, we have yet to find one article which acknowledges the contribution of religiosity in the etiology of delinquency. Thus, it would seem prudent for delinquency researchers to reconsider any number of theoretical arguments by considering religious measures in future delinquency research.

Endnotes

1  George Gallup, Jr., Emerging Trends, Princeton Religious Center, Volume 18, Number 3, March 1996, p.5. Also see Richard Morin, “Keeping the Faith: A Survey Shows the United States Has the Most Churchgoing People in the Developed World,” The Washington Post Weekly Edition, January 12, 1998, p.37.
2  George Gallup, Jr., “Religion in America: Will the Vitality of Churches Be the Surprise of the Next Century?,” The Public Perspective, October/November 1995, p.4.
3  Roger Finke and Rodney Stark, The Churching of America (New Brunswick, NJ; Rutgers University Press, 1992).
4  Gallup, Emerging Trends, op. cit., p. 1.
5  Robert W. Fogel, “The Fourth Great Awakening and the Political Realignment of the 1990s,” paper prepared for presentation at the Seventh Annual Bradley Lecture Series of the American Enterprise Institute, September 11, 1995, p. 2.
6  Center for Public Justice, A Guide to Charitable Choice (Washington, D.C.: Center for Public Justice, 1997).  Also see Carl H. Esbeck, “A Constitutional Case for Governmental Cooperation with Faith-Based Social Service Providers,” Emory Law Journal, Volume 46, Number 1, Winter, 1997, pp. 1-83, and Stanley W. Carlson-Thies and James W. Skillen, eds, Welfare in America: Christian Perspectives on a Policy in Crisis (Grand Rapids, MI and Cambridge, UK:  William B. Eerdmans Publishing Company, 1996).
7  For example, see Governor’s Advisory Task Force on Faith-Based Community Service Groups, Faith in Action: A New Vision for Church-State Cooperation in Texas (State of Texas, December 1996).
8  Mike Jackson, “Faith in Action Supporting Volunteering with 800 Grantees,” Advances:  The Robert Wood Johnson Foundation Quarterly Newsletter, issue 1, 1998, p.10.
9  Andrew Greely, “The Other Civic America: Religion and Social Capital,” The American Prospect, Number 32, May-June 1997, p.70.
10 Ibid., p.72.
11 Ron Stodghill II, “In the Line of Fire,” Time, April 20, 1998, pp. 34-37.
12 Joseph P. Shapiro, “Can Churches Save America,” September 9, 1996, pp. 46-53.
13 James Q. Wilson, “Two Nations,” paper delivered as Francis Boyer Lecture, American Enterprise Institute, December 4, 1997, p. 10.
14 For example, see David B. Larson, et al., Scientific Research on Spirituality and Health (Radnor, PA: The John M. Templeton Foundation, October 1, 1997).
15 T. David Evans, et al., “Religion and Crime Reexamined: The Impact of Religion, Secular Controls, and Social Ecology on Adult Criminality,” Criminology, Volume 33, Number 2, 1995, pp. 211-212.
16 Richard B. Freeman, “Who Escapes? The Relation of Church-Going and Other Background Factors to the Socio-Economic Performance of Black Male Youths From Inner-City Poverty Tracts,” Working Paper, Number 1656, National Bureau of Economic Research, Cambridge, MA, 1985.
17 Based on a forthcoming article entitled, “A Systematic Review of the Religiosity and Delinquency Literature: A Research Note,” (Johnson, Li, Larson, and McCullough, 1999).
18 Based on a forthcoming article entitled, “Escaping from the Crime of Inner-Cities: Churchgoing Among At-Risk Youth,” (Johnson, Larson, Li, and Jang 1999).
Specifically, the Current Population Survey (CPS), produced the by Census Bureau.
20 The NBER survey was administered in 1979 and 1980, by Mathematica Policy Research, Inc., to black males, aged 16 to 24, residing in Boston, Chicago, and Philadelphia.  The sample encompassed persons living on city blocks identified by the 1970 Census as having at least 70 percent black residents and 30 percent of families living below the poverty line.  Well over 2,800 survey interviews were attempted in the worst poverty tracts of these three cities, and more than 2,300 interviews were completed.
21 Based on a forthcoming article entitled, “The ‘Invisible Institution’ and Urban Delinquency: The African-American Church as an Agency of Local Social Control,” (Johnson, Li, Jang, and Larson, 1999).
22 Based on a forthcoming article entitled, “Does Adolescent Religious Commitment Really Matter?: A Reexamination of the Effects of Religiosity on Delinquency,” (Johnson, Larson, Li, and Jang 1999).