Developmental Patterns of Religiosity in Relation to Criminal Trajectories among Serious Offenders moving from Adolescence to Young Adult

Abstract

Using Group-Based Trajectory Models and Growth Curve Models, this study aimed to identify distinctive trajectories of religious attendance, religious importance and spirituality and how these trajectories related to changes in crime. The findings varied by measures of religiosity and types of crime. Generally, different measures of religiosity had little to do with initial offenses. A few trajectories of religious attendance and importance associated with both violent and income offenses, while changes in spirituality only related to violent offenses. Losses in religiosity may associate with elevated risks of recidivism. A small range of gains and losses in religiosity may increase the risk of recidivism, while maintaining high religiosity over time may result in a smaller growth change in recidivism.

Keywords

religiosity recidivism group-based trajectory models growth curve models

Introduction

Individual religiosity is often dynamic and fluid, such that religious beliefs, values, salience and practices are developed and reflexively practiced over a lifetime (Atchley, 1999). Religious behaviors and attitudes change, especially as adolescents make the transition to young adulthood (Desmond et al., 2010). During this transitional period, individuals begin to reconsider religious beliefs and values transmitted from their parents and then develop their own value and belief structures on the basis of experiences, backgrounds and interests they possess (Arnett, 2000; Koenig et al., 2008). Once they become young adults, they may alter their religious affiliations and decrease religious participation because of increasing autonomy and independence achieved during this period (Petts, 2007; Regnerus & Uecker, 2006). However, with the increased self-awareness, capacity for cognitive complexity and maturity of judgment in young adulthood, the salience of religious beliefs at this point in time may be more strongly internalized into decision-making and behaviors, regulating their impulsive or deviant behaviors (Yonker et al., 2012).

Given individuals’ proclivity to changes in religiosity, only investigating religiosity at one time point may overlook the relevance of these changes on individual behaviors, including delinquent and criminal behaviors. For instance, the exclusive focus on the consequences of current low religiosity ignores deviance-amplification effects of prior high religiosity (Charles et al., 1985). Stated differently, abrupt decreases in religiosity from initially high levels may contribute to increased criminal behavior that surpasses the level of crime committed by those adolescents whose religiosity remains steadily low throughout the time period. That is, these changes in religiosity could be of consequence, as opposed to just the absolute level of religiosity at any given point in time.

Within this context, longitudinal designs examining changes in religiosity over the life course in relation to changes in crime would be of particular interest. It is recognized, however, that few studies actually consider this longitudinal relationship accounting for both change in religiosity and change in crime over time, with even fewer exploring these changes over a longer time period (Desmond, Kikuchi, et al., 2010; Guo & Metcalfe, 2019; Jang, 2019; Petts, 2009; Pirutinsky, 2014). The individual heterogeneity of religious development, in particular, has received limited attention, especially during the important transitionary phase from adolescence to early adulthood (Koenig et al., 2008; Pearce & Denton, 2011; Regnerus & Uecker, 2006). In addition, existing evidence, although not expansive, reveals that the relationship between religiosity and crime may vary by measures of religiosity (Benda et al., 2006; Chu, 2007) and depend on the type of crime being assessed (Chu, 2007; Cochran, 1988; Jang, 2019; Jang & Johnson, 2001).

Expanding upon the existing research, the current study presents an exploratory analysis using the Pathways to Desistance Study (PDS), a 7-year investigation of serious offenders, to further examine the longitudinal relationship between religiosity and crime as these offenders transition from adolescence to early adulthood. In this way, individual pathways of religious development can be captured and an investigation of time-based changes of religiosity and crime during an important phase of the life course can be conducted. It is also determined whether the longitudinal relationship between religiosity and crime varies across three different measures of religiosity—religious attendance, religious importance, and spirituality—as well as type of self-reported offending—violent offending and income offending.

Theorizing Religiosity and Crime

Theories of social control, self-control, life course, social learning, and general strain, as well as various combinations of these perspectives, have been used to explain the religiosity-crime link by identifying various theoretical mechanisms whereby religiosity reduces the likelihood of offending. These theoretical perspectives suggest that religious individuals are less likely to be offenders than their less- or non-religious counterparts, since they are more prone to: (1) be strongly bonded to conventional society and adopt conventional beliefs and values reinforced by religious commitments (Cochran et al., 1994); (2) practice and develop high self-control to regulate impulsive behaviors in accord with moral directions emphasized by religiosity (Laird et al., 2011); (3) be strongly bonded to religious institutions or personal religious beliefs that work as turning points to keep them from future recidivism and shorten their criminal careers (Bakken et al., 2013); (4) closely associate with peers who share common conventional definitions and behaviors and develop a more favorable identity through positive reinforcement (Adamczyk & Palmer, 2008); and (5) use positive social and coping skills provided by religiosity that assist in overcoming individuals’ strain and negative emotions in a legitimate, non-delinquent manner (Jang & Johnson, 2003).

Religiosity likely entails complex processes of socialization and identity formation, and its effect on youth problem behaviors may or may not be reducible to the effects of social bonds, self-control, noticeable life events, peer influences, or coping strategies addressing strain. These relevant theories, though, point to important variables that may be related to both religiosity and crime, which should be included in research on the religiosity-crime link. Although the current study is designed to be inductive, consideration of these theoretical variables will help provide relevant explanations for differences in the risk of crime across distinctive trajectory groups of religiosity, as well as give salience to the continued exploration of the religiosity-crime link.

Prior Research Focusing on the Religiosity-Crime Relationship

According to numerous surveys and public opinion polls, religion plays an important role in the lives of adolescents in the United States (Denton et al., 2008). Religiosity is often considered as “those spiritual thoughts, feelings, and behaviors that are specifically related to a formally organized and identifiable religion” (Pargament & Saunders, 2007, p. 904). Given that “religious experience is inward, subjective, and highly individualized” (Chu, 2007, p. 4), it is difficult to define religiosity. When exploring the religiosity-crime link, studies focus more on operational rather than theoretical definitions of religiosity. There is still some debate regarding which of the measures and their accompanying operational definitions is best. Despite this, operational measures that have been widely used include organizational/objective/public religiosity representing public or organizational religious behaviors (e.g., attendance at religious services, participation in religious groups/activities) and intrinsic/subjective/private religiosity which indexes the perceived importance of religiosity, religious beliefs, and/or spirituality, the latter of which is characterized by how much of a person’s actions are influenced by a belief in a God (e.g., feeling close to God) (Moscati & Mezuk, 2014; Salas-Wright, Vaughn, Maynard, Clark, et al., 2014).

These studies have identified religiosity as a potentially important factor that protects against an extensive range of criminal behaviors, such as violence, petty or felony theft, and arrest (Baier & Wright, 2001; Ellison et al., 2007; Johnson et al., 2000). Involvement in religious activities or organizations may provide an external locus of control oriented toward individual behavior, limiting opportunities for deviant or criminal activities (Evans et al., 1996). The internal motivation to live life based on faith may initiate a transformative change in self-identity and promote an internal locus of control or moral constraints, which function to constrain individual behavior (Ellison & Levin, 1998). The degree to which adolescents internalize religious beliefs and values may lead to decreases in deviant behavior over time (Pearce et al., 2003).

However, the nature of the empirical relationship between religiosity and crime still remains unclear. Some researchers have found that religiosity is inversely related to the level of crime (Baier & Wright, 2001; Cochran et al., 1994; Rodell & Benda, 1999), as would be theoretically anticipated, while others have found that a positive or null relationship exists (Benda & Corwyn, 1997; Cochran et al., 1994; Hirschi & Stark, 1969). For those studies reporting a negative association between religiosity and crime, there are inconsistent results regarding whether religiosity directly affects crime or whether the relationship is indirect or spurious (Desmond et al., 2008, 2011; Jang et al., 2008; Mason & Windle, 2002). The discrepancy of these findings may be attributable to multiple ways in which religiosity and crime have been measured and confounding factors that were included. For instance, some studies demonstrated that only one measure of religiosity—sometimes religious attendance, sometimes religious importance—serves as a protective factor against particular expressions of crime (e.g., substance use or violence) (Laird et al., 2011; Salas-Wright, Vaughn, Maynard, Clark, & Snyder, 2014).

An overarching concern of existing empirical literature regarding the religiosity-crime link is that the vast majority of research has relied heavily on cross-sectional designs. Only a relatively small number of studies have used longitudinal data to examine this relationship (Bakken et al., 2013; Chu, 2007; Desmond et al., 2008, 2011; Giordano et al., 2008; Guo & Metcalfe, 2019; Jang, 2019; Jang et al., 2008). Some of these longitudinal studies have considered the long-term effect of adolescent religiosity on subsequent involvement in and/or dynamics of crime (e.g., abstinence vs. initiation and persistence vs. desistence). However, these longitudinal studies only treat religiosity as a time-invariant variable, in which constructs used to measure religiosity assess baseline or current religious involvement and/or beliefs. These measures thus provide little information about the development of religiosity, which may limit our understanding of whether changes in religiosity would have distinctive influences on criminal involvement over time.

A few existing longitudinal studies have attempted to examine whether and how changes in religiosity relate to subsequent involvement in crime (Desmond et al., 2010; Jang, 2019; Petts, 2009; Pirutinsky, 2014). These studies have generally indicated that significant decreases in religiosity coincide with increases in criminal behavior, and vice versa. Jang (2019) also indicated that juvenile offenders whose religiosity increased were more likely to increase their crime slowly, if they were on the rise, or decrease quickly, if declining, as compared to those with relatively stable religiosity. However, the analytic approach in these studies focuses more on a dramatic and overall trend of religious changes (e.g., a significant overall decrease or increase in religiosity), which can overlook a great deal of variability in individual-level changes of religiosity. These changes have been strongly supported by studies of religious development (Koenig et al., 2008; Pearce & Denton, 2011; Regnerus & Uecker, 2006).

Additionally, some studies model changes in individual religiosity by subtracting religiosity at Time 1 from that at Time 2 (Charles et al., 1985; Moscati & Mezuk, 2014; Ulmer et al., 2012), with results indicating that adolescents who decrease their level of religiosity have higher levels of later delinquency than those who have consistently been low in religiosity. Although these studies have recognized the inherent heterogeneity of religious development within individuals, examination of change between only two time points may not fully capture the real change of religiosity over one’s life course. For instance, it may be premature to classify individuals who are high in religiosity at Time 1 and remain stable at Time 2 into a stable high group. Ultimately, identifying more nuanced changes of religiosity through trajectory models would bring greater clarity to the relationship between changes in religiosity over the life course and changes in crime. Trajectory models are beneficial in that they allow for an examination of both small and large changes in religiosity and provide an illustration of particular pathways of religious development that individuals may experience from early adolescence though young adulthood.

Existing longitudinal studies also vary considerably in their measurement of religiosity. Some studies assess only one measure of religiosity (Moscati & Mezuk, 2014 [religious salience]; Petts, 2009 [religious attendance]; Pirutinsky, 2014 [spirituality]), which does not adequately capture its complexity. Religiosity is a multidimensional phenomenon including aspects such as frequency of prayer, Bible study participation, or commitment to religious organizations. Others assess overall religiosity by creating composite scores of subjective or objective religiosity as proxies (Desmond et al., 2010; Jang et al., 2008; Jang & Johnson, 2001; Mason & Spoth, 2011; Ulmer et al., 2012). Again, this approach precludes inferences about individuals’ development in each measure of religiosity, as well as which measures are more closely tied to crime changes.

Although measures of religiosity tend to be related, different measures of religiosity may follow distinctive trajectories, in which some aspects of religiosity may increase, while others stay the same or continue decreasing over time. For instance, research suggests drops in the frequency of participation in religious activities/services, but stability or increases in personal religious commitment and experiences, such as belief salience and close personal relationships to God, in the transition from adolescence into young adulthood (Stoppa & Lefkowitz, 2010). These studies provide insights into the importance of separately examining individual measures of religiosity, such as religious attendance, religious salience, and spirituality (e.g., emotional connections to God). With respect to the religiosity-crime link, previous studies have shown that the results are inconsistent in terms of individual measures of religiosity (Benda & Corwyn, 1997; Benda et al., 2006). In some instances, only certain measures are associated with criminal behaviors, thus making it relevant to consider each measure. For instance, Laird and colleagues (2011) found that religious salience, but not religious attendance, was associated with lower levels of antisocial behavior among adolescents. Thus, it seems relevant to examine different measures of religiosity separately to capture the nuances of change in religiosity and its connection to criminal behavior.

The relationship between religiosity and crime is also dependent on the forms of criminal behaviors that adolescents actually engage in, or stated differently, the relationship exists only for certain types of crime. A number of studies indicate that religiosity has a stronger relationship to victimless or minor crimes (e.g., substance use or status offenses) than to more serious forms of crime against people or property, such as assault offenses (Burkett & White, 1974; Cochran, 1988; Elifson et al., 1983; Rodell & Benda, 1999). Despite this, existing literature examining the role of changes in religiosity on dynamics of criminal behavior has looked at relatively few outcomes. Almost all of them focus on these minor forms of offending, including substance abuse, delinquency or misdemeanors, instead of serious crime (Desmond et al., 2010; Mason & Spoth, 2011; Moscati & Mezuk, 2014; Petts, 2009; Ulmer et al., 2012). A focus on the effect of religiosity on more serious forms of crime is needed.

Furthermore, few studies have attempted to look at the relationship among serious juvenile offenders (Jang, 2019; Pirutinsky, 2014). Given their more extensive involvement in crime, serious juvenile offenders who are placed in unique contextual and social milieus may be more likely to have lower levels of religiosity than other conventional adolescents. Investigating a sample of serious juvenile offenders provides an opportunity to test whether differential patterns exist in the religiosity-crime relationship, especially whether religiosity operates differently to help offenders stay away from crime over the life course.

Finally, there is still much debate about whether the association between religiosity and adolescent crime is spurious. It has been found that the relationship between religiosity and crime decreases or becomes insignificant after accounting for important confounding variables, such as peer, family, and school influences (Burkett & Warren, 1987; Cochran et al., 1994; Desmond et al., 2008; Elifson et al., 1983; Mason & Windle, 2002). Including relevant control variables in research is essential to resolve this debate. However, much of the research—both cross-sectional and longitudinal—does not account for a variety of variables that may confound this relationship. Without controlling for relevant confounding factors, it is difficult for researchers to make a convincing conclusion that the religiosity-crime relationship is not spurious.

The Current Study

In order to fill the gaps in prior studies of the religiosity-crime link, the current study examined the relation of religiosity with crime in a sample of adjudicated adolescents aged 14 to 18 through the use of the Pathways to Desistance Study, a seven-year longitudinal dataset, paying close attention to individual-level changes of religiosity and their relationships with changes in crime over time. The following exploratory research questions were used to guide the research:

Are distinct trajectories of religiosity related to changes in crime among serious offenders transitioning from adolescence into early adulthood? If so, how?

Do these relationships vary across measures of religiosity (i.e., religious attendance, religious importance, and spirituality) and depend on the types of crime being studied (i.e., violent offending and income offending)?

Do these relationships differ between two groups of adolescents who start from different developmental time periods—mid-adolescence and late-adolescence?

Do these relationships vary after accounting for important confounding variables?

The current study contributed to the existing body of literature on the religiosity-crime link via a number of important ways: (1) investigating whether and how changes in religiosity relate to changes in crime by identifying subgroups of individuals who follow distinct trajectories of religiosity; (2) modeling changes in various measures of religiosity; (3) looking into the changes in different types of criminal activities; (4) focusing on the changes that occur in the transition from adolescence into early adulthood, where changes in religiosity and crime are most likely to occur; (5) considering the relevance of religiosity among a serious group of juvenile offenders, and (6) testing whether the longitudinal religiosity-crime link remains even after accounting for theoretically relevant confounding factors.

Methods

Data and Sample

The PDS is a longitudinal dataset that followed 1,354 serious adolescent offenders over 7 years—from mid-adolescence through early adulthood (Mulvey & Schubert, 2012; Schubert et al., 2004). The enrolled adolescents were recruited from the juvenile and adult court systems in Maricopa County (Phoenix), AZ (n = 654) and Philadelphia County, PA (n = 700). Adolescents were selected if they were between the ages of 14 and 18 at the time of their involvement in crime and have been adjudicated or found guilty of committing a serious crime. Almost all included offenses were felony crimes, with the exception of less serious property offenses, sexual assaults, and weapons offenses. Demographic information about the sample is displayed in Table 1.

Table 1.

Descriptive Statistics for the Sample.

	All sample	Younger cohort group (aged 14–16)	Older cohort group (aged 17–19)
Dependent variables
Violent offending	0.48	0.49	0.45
Income-related offending	0.37	0.37	0.37
Time-variant variables
Religious attendance	1.93 (1.19)	1.94 (1.19)	1.91 (1.19)
Religious importance	3.11 (1.32)	3.10 (1.32)	3.12 (1.33)
Spirituality	3.09 (1.23)	3.05 (1.21)	3.14 (1.26)
Age	19.42 (2.56)	18.72 (2.43)	20.59 (2.34)
Educational achievement	5.62 (2.35)	5.26 (2.45)	6.22 (2.03)
Enrolment status	0.48	0.53	0.38
Weeks employed	12.83 (17.42)	12.09 (16.97)	14.06 (18.07)
Romantic relationship status	0.45	0.42	0.50
Parenthood status	0.37	0.30	0.47
Moral disengagement	1.47 (0.35)	1.47 (0.35)	1.46 (0.34)
Peer delinquency	−0.001 (0.93)	−.001 (0.93)	−0.002 (0.92)
Low self-control	−1.60e-10 (0.89)	2.38e-10 (0.89)	−8.14e-10 (0.89)
Incarceration length	69.25 (120.94)	61.42 (115.94)	82.20 (127.75)
Time supervised	99.22 (128.24)	98.13 (127.04)	101.00 (130.20)
Baseline variables
Male	0.86	0.85	0.88
Race
White	0.20	0.22	0.17
Black	0.41	0.40	0.43
Hispanic	0.34	0.34	0.34
Others (reference category)	0.05	0.04	0.06
Family structure
Biological-parent family	0.15	0.16	0.13
Step-parent family	0.20	0.21	0.20
Single-parent family	0.47	0.47	0.46
Others (reference category)	0.18	0.16	0.21
Site
Philadelphia	0.52	0.50	0.55
Socioeconomic status	51.41 (12.29)	51.73 (12.31)	50.89 (12.26)
Community involvement	0.26 (0.58)	0.29 (0.61)	0.21 (0.51)
School involvement	0.81 (1.11)	0.82 (1.10)	0.79 (1.11)
School attachment	3.19e-09 (0.85)	2.63e-09 (0.85)	4.09e-09 (0.85)
Mother-child relationships	−3.42e-09 (0.84)	−4.85e-09 (0.85)	−1.10e-09 (0.81)
Parental monitoring	2.80 (0.86)	2.96 (.81)	2.52 (.88)
Offending history	2.92 (2.14)	2.64 (1.97)	3.36 (2.33)
Early onset of behavior problems	1.52 (1.19)	1.53 (1.18)	1.51 (1.21)
N (individuals)	1,354	829	525
N (person waves)	10,832	6,632	4,200

Note. Values in parentheses represent standard deviations from the mean. Negative values are a result of standardizing the indices.

All enrolled adolescents were required to complete follow up computer-assisted interviews every 6 months for the first 3 years and annually thereafter. Thus, there were 11 total waves of data collected over a period of 7 years. At each wave, the average retention rate was about 90%. All waves of data were employed for the analyses reported in this study. Data from the waves covering 6-month time periods (the first 6 waves) were combined into 1-year periods, so that the intervals between time periods were equal across the full length of the study. In addition, the sample was divided into two separate groups—younger cohort (baseline age: 14–16 years old) and older cohort (baseline age: 17–19 years old)—to address the heterogeneity within the sample across quite a large age range, as well as to make comparisons between age groups from different developmental time periods—mid-adolescence and late-adolescence.

Measurements

Criminal behavior

Criminal behavior was based on the Self-Reported Offending (SRO; Elliott, 1990) items used to assess the respondents’ involvement in 24 different types of offenses over the 7-year study period. These items were broken into two measures of offending—violent offending and income offending (the full list of offenses is available in Appendix A). Violent Offending was measured by a binary outcome assessing whether the respondent engaged in any of 11 violent offenses during each recall period. These items had acceptable internal consistency (alpha) ranging from 0.68 to 0.76 across studied waves. Income Offending was a binary outcome asking the respondent to indicate whether s/he engaged in any of 10 income offenses during each recall period. These items showed good internal consistency (alpha), which ranged from 0.69 to 0.79 across studied waves. As Table 1 indicates, 48% of all observations reported Violent Offending and 37% reported Income Offending.

Religiosity

Religiosity was operationalized by three measures: Religious Attendance, Religious Importance, and Spirituality during each interview period. Religious Attendance was assessed by the question, “During the past year, how often did you attend church, synagogue, or other religious service?” The item was rated on a 5-point response scale ranging from “1 = never” to “5 = several times per week”. Religious Importance was evaluated by the question “How important has religion been in your life?” with a 5-point response scale ranging from “1=not at all important” to “5 = very important”. The measure of Spirituality was based on a highly reliable scale created by Maton (1989), which includes three items asking respondents to indicate how much their everyday experiences are influenced by religion and beliefs in God. The items included: (a) “I experience God’s love and caring on a regular basis”; (b) “I experience a close personal relationship to God”; and (c) “Religion helps me to deal with my problems”. These items were rated on a 5-point scale ranging from “1 = not at all true” to “5 = completely true,” and their reliability coefficients ranged from 0.88 to 0.95 across the waves. A mean index was created with higher scores indicating a greater degree of spirituality.

Baseline covariates

Gender was a binary variable indicating whether the respondent was male (1) or female (0). Race was classified as Black, White, Hispanic, and Other (reference group). Site was a dummy variable indicating Philadelphia (1) or Phoenix (0). Four categories of Family Structure were created to reflect respondents with varying family dynamics: biological-parent family, step-parent family, single-parent family, and others (reference group; including circumstances, such as two adoptive parents, other adult relatives, etc.). Socioeconomic Status relied on a pre-constructed parental Index of Social Position ranging from 11 to 77, which was computed based on both education and occupation obtained by the respondent’s parents. Official criminal history data was used to construct a measure of Offending History, which equated to the number of arrests prior to the arrest that lead to the adolescent’s entry into the Pathways study. Early Onset of Behavior Problems was a count variable indicating the number of problem behaviors the respondent engaged in before age 11, consisting of five items, such as getting into trouble for cheating, disturbing class, being drunk/stoned, stealing, and fighting. This scale ranged from 0 to 5, with higher scores indicating greater early onset behavior problems.

Community Involvement was based on the Community Involvement scale (Elliott, 1990) assessing the extent of the adolescent’s involvement in four different community organizations (e.g., sports teams, scouts, and volunteer work). This scale ranged from 0 to 4, with higher scores indicating more community involvement. School Involvement was a pre-constructed variable, in which respondents were asked to report the total number of extra-curricular school activities. School Attachment was assessed by two pre-constructed variables (Cernkovich & Giordano, 1992) used to evaluate the adolescent’s educational experience consisting of Bonding to Teachers and School Orientation. An Exploratory Factor Analysis (EFA) indicated that the two pre-constructed variables loaded on one factor with factor loading scores all above 0.67 and were thus grouped into a standardized index (alpha = .62).

The quality of the Mother-Child Relationship was constructed by two pre-constructed variables (Conger et al., 1994), Maternal Warmth and Maternal Hostility, in which mean scores of the maternal warmth scale and the maternal hostility scale were calculated separately. This measure displayed adequate internal consistency (alpha = .63) and a single-factor EFA model was acceptable with factor loading scores all above 0.67. The two pre-constructed variables were thus grouped into a standardized index. Parental Monitoring was measured by a pre-constructed index, in which a mean score of the 9-item Parental Monitoring Inventory (Steinberg et al., 1992) was calculated for each respondent. Responses were based on a 4-point scale ranging from “1 = doesn’t know at all/never” to “4 = knows everything/always”. Higher scores reflected more parental monitoring.

Time-varying covariates

Age was operationalized as the respondent’s age at the time of the interview and represented the interview date minus the subject’s date of birth truncated to a whole number. Educational Achievement reflected the highest grade the respondent achieved during each recall period ranging from 6th grade or less (1) to a college degree (10). Responses were left on a continuous scale, with higher values representing greater educational attainment. Enrollment Status was constructed by an item asking respondents to report whether they were enrolled in school during each recall period (1 = yes, 0 = no). Weeks Employed was operationalized as the number of weeks where a respondent worked in any legal job, which was used to evaluate the respondent’s employment status during each recall period. Incarceration Length, measured by the total number of days that the respondent spent in jail or prison, was used to indicate the length of imprisonment during each recall period. Time Supervised, or the time supervised in all institutional settings, was designated by the number of days that the respondent was supervised institutionally (i.e., removed from the community) during each recall period.

The measure of Romantic Relationship Status included both marital and non-marital relationships. Respondents were asked to report if they were currently married in the recall period. If not currently married, respondents were asked if they were currently involved in a serious romantic relationship. Then, the measure was coded 0 = not married or not in a romantic relationship, and 1 = married or in a romantic relationship. The number of children reported by respondents acted as a proxy for Parenthood Status. A dichotomous variable was created for whether the respondent reports having, at least, one child (1) or none (0). Moral Disengagement was a pre-constructed index assessing adolescents’ moral beliefs, in which the mean of 32 items was computed with higher scores reflecting a greater moral disengagement (Bandura et al., 1996). These items showed good internal consistency, in which reliability coefficients ranged from 0.88 (baseline) to 0.92 (24 month).

Peer Delinquency was assessed by two pre-constructed variables: Peer Antisocial Behavior and Peer Antisocial Influence (Thornberry et al., 1994). These two constructed variables were captured by separate calculated mean scores on the Peer Antisocial Behavior scale and the Peer Antisocial Influence scale. EFA results suggested a single-factor model with factor loading scores all above 0.84. A standardized index was thus created, in which reliability coefficients ranged from 0.82 to 0.87 over the interviewed time period. Low Self-control was designated by two pre-constructed variables of the Weinberger Adjustment Inventory (WAI; Weinberger & Schwartz, 1990): Impulse Control and Suppression of Aggression. EFA results indicated that the two variables loaded on one factor with factor loading scores all above 0.76, and were thus computed as a standardized index. The reliability coefficients ranged from 0.69 to 0.75 over the interviewed time period.

Analytic techniques

The data were analyzed in three linked steps. First, Group-based Trajectory Models (GBTM) were conducted to identify distinct developmental trajectories of religiosity in terms of religious attendance, importance and spirituality for the younger and older cohort groups, respectively, via the use of a SAS procedure (PROC TRAJ) (Jones et al., 2001). GBTMs account for missing data through maximum likelihood techniques. Second, unconditional linear and quadratic Growth Curve Models (GCM) were fitted to the data to examine which models best represented observed patterns of growth in violent and income offending. Unconditional GCM can be used to represent differences over time, taking into account the initial status (i.e., intercept) of crime and the shape and rates of change in crime over time (i.e., slope).

Finally, multiple-group GCMs were used to examine whether inter-individual differences in average crime for the first wave (“intercepts”) and inter-individual changes in crime across all observed waves (“slopes”) can be explained by the trajectory groups of religiosity. The multi-group GCM was estimated in two steps. First, religiosity trajectory groups found in the GBTMs were coded as dummy variables. These trajectory groups were then entered into the model without covariates as grouping variables to test their differences in both initial levels (intercepts) and change rates (slopes) of crime over time. Second, the model with both time-varying and time-invariant covariates was estimated to examine whether any of the religiosity trajectory group differences in the intercepts and slopes of crime can be attributed to these control variables. Particularly, a series of multi-group GCMs were estimated for violent offending and income offending based on distinct trajectories of religious attendance, religious importance, and spirituality, respectively. All these analyses were conducted in both younger and older cohort groups. Missing data patterns were checked, and data were identified as missing at random. As such, the technique of multiple imputation was used to deal with missing data, in which average parameter estimates were computed within 10 imputed datasets (Allison, 2001).

Results

The BIC and Log Bayes Factor and the size of each trajectory group were used as a guide to determine the best-fitting model of each measure of religiosity (available in Appendix B). Overall, a seven-group model emerged as the best-fitting model for each measure of religiosity across both the younger and older cohort groups (The specific trajectories of each measure of religiosity are available in Supplemental Appendix D: Figures 1–6). A seven-group model showed the largest BIC and Log Bayes factors were larger than 5 (Jones et al., 2001). The sizes of each identified trajectory group were all above 5% (Nagin, 2005). The model adequacy of the final seven-group model with specific shapes indicated that individuals were well assigned to their groups, in which the Average Posterior Probabilities were all around or above 0.70 (Nagin, 2005).

In order to determine the individual and comparative fit of the GCMs of crime, BIC values, chi-square difference tests and latent growth curve factors were used (available in Appendix C). The quadratic models appeared to fit the data better than the linear models with a significant improvement in chi-square and smaller BIC in both cohort groups. Significant negative linear slopes and positive quadratic slopes suggested both violent and income offending followed a similar trajectory of an initial decrease and a subsequent increase (acceleration) in both cohort groups. In addition, there were significant variances in the initial level, linear slope and quadratic slope for both types of offending (except for the initial level variance on the quadratic model of income offending), which indicated that the shape and rate of linear and nonlinear change significantly varied across individuals. This information provided the foundation for the next step in the analysis, which was to examine whether religiosity trajectories were associated with individual differences in growth in crime over time.

Multiple-group Growth Curve Model

The multiple-group GCM of each measure of religiosity and crime (i.e., violent and income offending) with a quadratic function was estimated in both cohort groups, respectively. Table 2 provides summaries of the model estimates for each multiple-group GCM with control variables included. The reference group is non-attenders for religious attendance and stable low for religious importance and spirituality. It appeared that only a few trajectory groups for each measure of religiosity (i.e., religious attendance, religious importance and spirituality) were significantly related to the growth of violent and income offending. The results were substantively similar to those found in models without controls for the confounding variables (available upon request).

Table 2.

Religiosity Trajectory Group Differences in Growth Trajectory Estimates.

Religious attendance	Violent offending	Income offending	Religious importance	Violent offending	Income offending	Spirituality	Violent offending	Income offending
Younger cohort group (aged 14–16)
Intercept
Gradual increasing attenders	−0.904 (0.569)	−0.342 (0.613)	Medium increasing	−0.007 (0.639)	−0.177 (0.418)	Low-medium increasing	0.002 (0.770)	0.135 (0.579)
Late increasing attenders	0.633 (0.616)	0.245 (0.306)	Gradual increasing	−1.115 (0.927)	−0.513 (0.583)	High-medium declining	−0.529 (0.583)	−0.249 (0.478)
Parabolic attenders	0.145 (0.648)	0.629 (0.573)	Gradual declining	−0.890 (0.788)	−0.822^† (0.471)	Medium-high increasing	0.472 (0.757)	−0.096 (0.644)
Early declining attenders	−0.339 (0.471)	−0.477 (0.394)	Late declining	−1.356 (0.874)	−0.484 (0.552)	Stable medium	−0.353 (0.576)	−0.133 (0.449)
Frequent attenders	−0.268 (0.610)	0.707 (0.542)	Declining/increasing	−0.264 (0.649)	−0.323 (0.474)	Medium-low declining	−0.407 (0.603)	0.136 (0.445)
Gradual declining attenders	0.126 (0.555)	−0.422 (0.411)	Stable high	0.109 (0.746)	−0.514 (0.452)	Stable high	−0.177 (0.626)	−0.227 (0.515)
Linear slope
Gradual increasing attenders	0.332 (0.335)	−0.014 (0.334)	Medium increasing	−0.007 (0.363)	0.022 (0.259)	Low-medium increasing	−0.061 (0.43)	−0.235 (0.348)
Late increasing attenders	−0.178 (0.325)	−0.289 (0.320)	Gradual increasing	0.525 (0.494)	0.025 (0.358)	High-medium declining	0.363 (0.322)	0.266 (0.304)
Parabolic attenders	−0.109 (0.357)	−0.242 (0.359)	Gradual declining	0.644 (0.416)	0.170 (0.307)	Medium-high increasing	−0.022 (0.432)	0.029 (0.402)
Early declining attenders	0.085 (0.281)	−0.141 (0.259)	Late declining	0.820^† (0.482)	0.537 (0.353)	Stable medium	0.243 (0.320)	0.141 (0.280)
Frequent attenders	0.033 (0.369)	−0.445 (0.350)	Declining/increasing	0.344 (0.363)	0.261 (0.302)	Medium-low declining	0.185 (0.334)	−0.272 (0.285)
Gradual declining attenders	0.184 (0.287)	0.172 (0.269)	Stable high	0.190 (0.395)	0.141 (0.291)	Stable high	0.151 (0.350)	0.013 (0.317)
Quadratic slope
Gradual increasing attenders	−0.019 (0.042)	0.004 (0.046)	Medium increasing	−0.005 (0.044)	−0.003 (0.034)	Low-medium increasing	0.002 (0.053)	0.021 (0.045)
Late increasing attenders	0.035 (0.040)	0.030 (0.041)	Gradual increasing	−0.051 (0.059)	−0.004 (0.049)	High-medium declining	−0.042 (0.040)	−0.060 (0.041)
Parabolic attenders	0.036 (0.043)	0.025 (0.046)	Gradual declining	−0.069 (0.05)	−0.001 (0.041)	Medium-high increasing	−0.005 (0.054)	−0.033 (0.054)
Early declining attenders	−0.001 (0.034)	0.041 (0.034)	Late declining	−0.107^† (0.061)	−0.098* (0.048)	Stable medium	−0.034 (0.039)	−0.027 (0.037)
Frequent attenders	0.004 (0.048)	0.042 (0.047)	Declining/increasing	−0.037 (0.045)	−0.036 (0.040)	Medium-low declining	−0.022 (0.040)	0.024 (0.038)
Gradual declining attenders	−0.019 (0.034)	−0.022 (0.037)	Stable high	−0.037 (0.048)	−0.011 (0.039)	Stable high	−0.022 (0.043)	−0.011 (0.043)
Older cohort group (aged 17–19)
Intercept
Declining-increasing-declining attenders	0.621 (0.795)	0.581 (1.154)	Gradual increasing	0.255 (0.715)	0.646 (0.886)	Low-medium increasing	1.132 (0.972)	−0.571 (1.166)
Parabolic attenders	0.667 (1.023)	−4.489* (2.210)	Medium declining	−0.835 (0.608)	−0.119 (0.685)	High-medium declining	−0.003 (0.747)	−0.722 (1.041)
Late increasing attenders	−0.334 (1.107)	−0.376 (1.137)	Stable medium	−0.676 (0.600)	−0.364 (0.659)	Medium-low declining	1.015 (0.786)	−0.726 (1.074)
Early declining attenders	−0.559 (0.478)	−1.486 (1.106)	Early increasing	0.208 (0.755)	−1.159 (.883)	Medium declining	0.266 (0.766)	−1.509 (1.036)
Increasing-declining-increasing attenders	−0.475 (1.191)	−0.296 (1.404)	Late declining	−0.524 (1.197)	−2.597^† (1.425)	High-medium declining	0.655 (1.072)	−2.946* (1.445)
Frequent attenders	0.156 (0.697)	1.259 (1.342)	Stable high	−0.096 (0.660)	−0.309 (0.742)	Stable high	0.677 (0.785)	−2.005^† (1.092)
Linear slope
Declining-increasing-declining attenders	−0.004 (0.418)	−0.430 (0.761)	Gradual increasing	−0.270 (0.462)	−0.773 (0.545)	Low-medium increasing	−10.026* (0.518)	−0.288 (0.693)
Parabolic attenders	−0.393 (0.609)	2.868* (1.360)	Medium declining	0.020 (0.358)	−0.204 (0.405)	Medium-high increasing	−0.664 (0.422)	−0.216 (0.568)
Late increasing attenders	0.582^† (0.315)	−0.817 (0.669)	Stable medium	0.053 (0.346)	0.024 (0.401)	Medium-low declining	−1.206** (0.449)	−0.086 (0.568)
Early declining attenders	0.527^† (0.288)	0.224 (0.620)	Early increasing	−0.299 (0.425)	0.551 (0.568)	Medium declining	−0.914* (0.428)	0.298 (0.568)
Increasing-declining-increasing attenders	0.638 (0.616)	0.015 (1.101)	Late declining	−0.247 (0.760)	0.965 (0.810)	High-medium declining	−1.227* (0.585)	0.685 (0.823)
Frequent attenders	0.266 (0.436)	−2.078* (0.854)	Stable high	0.157 (0.410)	−0.336 (0.447)	Stable high	−0.730 (0.464)	0.307 (0.614)
Quadratic slope
Declining-increasing-declining attenders	0.003 (0.054)	0.080 (0.112)	Gradual increasing	0.050 (0.064)	0.100 (0.074)	Low-medium increasing	0.162* (0.066)	0.062 (0.090)
Parabolic attenders	0.030 (0.078)	−0.474* (0.217)	Medium declining	0.032 (0.046)	0.029 (0.053)	Medium-high increasing	0.114* (0.056)	0.044 (0.069)
Late increasing attenders	−0.065 (0.042)	0.146 (0.095)	Stable medium	0.008 (0.046)	−0.013 (0.055)	Medium-low declining	0.187*** (0.059)	0.028 (0.070)
Early declining attenders	−0.066^† (0.039)	0.011 (0.081)	Early increasing	0.056 (0.054)	−0.069 (0.075)	Medium declining	0.155** (0.055)	−0.009 (0.070)
Increasing-declining-increasing attenders	−0.092 (0.078)	−0.048 (0.152)	Late declining	0.043 (0.095)	−0.124 (0.108)	High-medium declining	0.190** (0.076)	−0.053 (0.101)
Frequent attenders	−0.045 (0.055)	0.286** (0.109)	Stable high	−0.024 (0.054)	0.058 (0.059)	Stable high	0.023 (0.085)	−0.005 (0.077)

Note. Unstandardized coefficients are reported. Values in parentheses represent the standard errors. All models reported include controls for the time-variant and time-invariant variables noted in Table 1. Reference group is non-attenders (for religious attendance) and stable low (for religious importance and spirituality).

p ≤ .05. **p ≤ .01. ***p ≤ .001. ^†p < .1 (two-tailed).

Regarding religious attendance, no relationship was found with the growth of either type of self-reported offending in the younger cohort group. Nevertheless, there were a few trajectory groups in the older cohort group associated with the initial level and the rate of change in either type of self-reported offending. Parabolic attenders attended religious services at a somewhat low rate at baseline, increased their participation until approximately the age of 20 to 22, and then declined throughout the remainder of the observed life course (see Supplemental Appendix D: Figure 2). Although parabolic attenders had a pretty low frequency of religious attendance at baseline, they still participated in religious service at a higher rate relative to non-attenders. Not surprisingly, parabolic attenders who started with an initial higher frequency of religious attendance had 98.9% lower odds of engaging in income offenses than non-attenders (b = −4.489, p < .05). In addition, the odds of engaging in income offending deceased more initially (b = 2.868, p < .05) and accelerated quicker subsequently (b = −0.474, p < .05) among parabolic attenders who followed an initial increase and a subsequent decrease in the frequency of religious attendance over time than those who maintained a low frequency of religious attendance. Furthermore, frequent attenders experienced a smaller growth rate for both the initial decrease and the subsequent increase of income offending (linear: b = −2.078, p < .05; quadratic: b = 0.286, p < .01), since they had quite stable and higher levels of religious attendance relative to non-attenders.

Although late increasing attenders had a relatively low frequency of religious participation at the beginning of the observed life course, they experienced a steady increase during the ages of 21 to 23. Thus, late increasing attenders with an increasing frequency of religious attendance over time decreased the odds of engaging in violent offenses more so (b = 0.582, p < .1) than those who maintained a low frequency of religious attendance. With respect to early declining attenders, they participated in religious services at a relatively high frequency at baseline but experienced a rapid decline in the frequency of religious attendance one year later. Accordingly, the odds of violent offending initially deceased more (b = 0.527, p < .1) and then accelerated quicker (b = −0.066, p < .1) among individuals who possessed an initial high and overall decrease in the frequency of religious attendance.

Moving to the younger cohort group and focusing on religious importance, involvement in the gradual declining group of religious importance was marginally associated with the initial level of income offenses. With higher initial levels of religious importance relative to the group of stable low, individuals in the gradual declining group held 56.1% slightly lower odds of income offending initially (b = −0.822, p < .1). In addition, late declining of religious importance was associated with the linear and quadratic slope of violent offending, as well as the quadratic slope of income offending. Individuals within the late declining trajectory maintained a relatively high level of religious importance in the first four years and then experienced a gradual decline (see Supplemental Appendix D: Figure 3). Not surprisingly, the odds of engaging in violent offending initially decreased more (b = 0.820, p < .1) and then accelerated quicker (b = −0.107, p < .1) among these individuals. In addition, individuals following the trajectory of late declining accelerated the odds of engaging in income offending quicker over time (b = −0.098, p < .05) than those who maintained low levels of religious importance throughout the waves. In the older cohort group, the late declining trajectory of religious importance was marginally associated with the initial likelihood of income offenses. This finding indicated that the late declining group, which had higher initial religious importance, had 92.6% lower likelihood of engaging in income offenses in the initial level compared to the trajectory of stable low (b = −2.597, p < .1).

With respect to spirituality, no relationship was found with the growth of either type of self-reported offending in the younger cohort group. In the older cohort group, involvement in the high-medium declining and stable high groups was associated with the initial level of income offenses. Both trajectory groups tended to have higher initial levels of spirituality relative to the stable low group. Thus, the odds of engaging in income offending would be 94.7% lower for the high-medium declining group (b = −2.946, p < .05) and 86.5% lower for the stable high group (b = −2.005, p < .1).

Furthermore, the trajectory groups of spirituality were more related to linear and quadratic changes in violent offending as opposed to income offending. Regarding the declining trajectory groups of spirituality, such as medium-low declining (linear: b = −1.206, p < .01; quadratic: b = 0.187, p < .001), medium declining (linear: b = −0.914, p < .05; quadratic: b = 0.155, p < .01), and high-medium declining (linear: b = −1.227, p < .05; quadratic: b = 0.190, p < .01), each of these groups was associated with the linear slope negatively and the quadratic slope positively for violent offending. Individuals in these trajectory groups exhibited an overall decrease in spirituality, and their spirituality was relatively higher over time than those who maintained low levels of spirituality (see Supplemental Appendix D: Figure 6). Unsurprisingly, these adolescents then experienced a smaller rate of initial decrease and a slower rate of subsequent acceleration in the likelihood of engaging in violent offenses compared to the spirituality group of stable low.

Interestingly, like the declining trajectories mentioned above, two increasing trajectory groups of spirituality, including the low-medium increasing and medium-high increasing groups, had a similar relationship with the growth of violent offending. The odds of violent offending decreased less initially (b = −1.026, p < .05) and accelerated slower subsequently (b = 0.162, p < .05) among individuals who held a low-medium increasing spirituality than those who maintained low levels of spirituality over time. Accordingly, individuals who followed a medium-high increasing spirituality experienced a slower acceleration in the odds of violent offending (b = 0.114, p < .05).

Discussion

Despite the recent emphasis on longitudinal research, the relationship between changes in both religiosity and crime over time remains unclear. This study aimed to fill the gaps in prior studies by examining the longitudinal religiosity-crime relationship in a sample of adjudicated adolescents via the use of the Pathways to Desistance study. Using GBTMs and GCMs, this study identified different developmental trajectories of religious attendance, religious importance and spirituality and assessed their relationships with changes in violent and income offending. To reduce the possibility of spurious findings, an extensive number of control variables were considered that could confound the relationship of interest.

Given the initial levels of criminal behavior, religiosity had little relationship with the likelihood of crime in the younger cohort group. Only offenders whose perceptions of religious importance gradually declined were slightly less prone to engage in income offending at the beginning of the observed time period. Similarly, a few trajectory groups of religiosity were only associated with income offenses in the older cohort group. For instance, offenders whose religious importance or spirituality was higher before later declining were less likely to recidivate in income-offenses at the beginning of the observed time period. These significant relationships, to some extent, suggested that offenders with higher religiosity initially had a lower likelihood of income offenses than those who were less religious or nonreligious. Overall, though, it seems that different measures of religiosity had little or no relationship with violent and income offenses in the initial level. This might be explained by the nature of the sample, which consisted of adolescents who had previously committed serious crime. There were no significant variations on the baseline level of self-reported crime in each religiosity trajectory group, in which the majority of adolescents reported having violent offenses (98%) or income offenses (90%) at the beginning of the interview period. The initial levels of crime were less likely, then, to be due to preexisting factors, like religiosity.

When it came to changes in religiosity, a few religiosity trajectory groups were associated with the growth of criminal behavior. In cases of statistical significance, the relationships between changes in both religiosity and crime varied by measures of religiosity and the types of crime. There were more limited findings for the younger cohort group than the older cohort group. In the younger cohort group, only one of the trajectory groups of religious importance—late declining—was related to the growth of crime, while changes in religious attendance and spirituality were not. However, changes in both religiosity and crime showed different relationship patterns in the older cohort group. Specifically, there were a few trajectories of religious attendance associated with both violent offenses (marginally) and income offenses. Trajectories of religious importance were not associated with changes in either type of crime. Increasing and decreasing trajectories of spirituality were only related to the growth of violent offenses. In addition, these relationships were maintained despite the confounding variables controlled.

As for the relationships between changes in both religiosity and crime, the results generally suggested that an increase in religiosity was associated with a greater decrease or a smaller increase in recidivism over time (and vice versa). Notably, the relationships with growth patterns of violent offending were impacted by the degree of change in spirituality, which showed a subtle increase and/or decrease over time. That is, in addition to the overall trajectory of decline in spirituality, a small gain or loss in spirituality over time, even within low and medium levels, to some extent might increase the likelihood of violent offending. Relative to those who had stable low levels of religious attendance, offenders who maintained relatively high levels of religious attendance over time reported a smaller growth change in income offending. There might be deterrent effects among offenders with unchanging high religiosity for whom religious attendance continued to impede crime.

Despite these general patterns discussed, it should be recognized that there was not overwhelming support for the relationship between changes in both religiosity and crime throughout adolescence and early adulthood among serious juvenile offenders. It is still notable, though, that some of these relationships emerged despite the extent of cofounding factors controlled. Overall, the results contributed to the existing knowledge of the religiosity-crime link, suggesting that the longitudinal relationship between religiosity and crime among serious juvenile offenders varied by measures of religiosity and types of crime. The various measures of religiosity had little relationship with offenses at the beginning of the observed time period. Trajectories capturing changes in religious attendance and importance were associated with both violent and non-violent offenses, while changes in spirituality were only related to violent offenses. A decrease or only a minor increase in religiosity may account for increased risk of recidivism, while maintaining high religiosity or experiencing a more substantial increase in religiosity over time could result in a smaller growth change in recidivism.

Explanations of the Findings

The nature of religious attendance and the degree of religious commitment and beliefs at different periods of the life course might explain why religiosity was more relevant to explaining changes in the older cohort group. Younger offenders may attend religious services more because of coercion from family members or significant others and less because of their willingness or commitment to religious beliefs (Rhodes & Reiss, 1970). Nevertheless, spending time with family and friends involved in religious activities may provide an external locus of control oriented toward their behavior, limiting opportunities for criminal activities (Evans et al., 1996). As individuals age, older offenders’ religious involvement may be less dependent on others’ expectations (Koenig et al., 2008). They may be more prone to internalize the salience of their beliefs into their decision-making and behaviors (Yonker et al., 2012), since that they may come to have a stronger sense of identity and self-awareness and capacity for cognitive complexity (Arnett, 2007). Likewise, the self-control related to religiosity may matter more for older offenders in regulating criminal behaviors as their developed brain allows for better maturity of judgment than for younger offenders (Yonker et al., 2012). Overall, religiosity may carry more impact in the older cohort group. That is, the decreased risk of crime may be not only because of an increased attendance at religious services that provide social control over offenders’ behavior, but also due to an increased level of religious commitment and beliefs that are explicitly proscriptive for crime.

Such external and internal loci of control derived from religiosity can further explain the fact that the longitudinal religiosity-crime link varied by measures of religiosity and types of crime. Religiosity, when measured by religious attendance, appeared to be more relevant to non-violent offenses, such as income offenses, than violent offenses, which was consistent with previous studies (Baier & Wright, 2001). Subjective religiosity, such as spirituality, reflects a greater internal locus of control illustrated by individuals’ internal motivation to live their lives based on their faith. Subjective religiosity was found to having a greater relationship with the growth of violent offenses than objective religiosity, like religious attendance. Although religious attendance reflects an external locus of control that helps constrain delinquency/crime, these external regulators may or may not be readily accessible to adolescents when they are making a critical decision to engage in criminal behavior. Pearce et al. (2013) suggest that internal mechanisms of religiosity (i.e., indicators of private practices) may have more profound effects on problematic behavior than do external mechanisms (e.g., church attendance). Not surprisingly, internalized moral beliefs and values might result in greater social and moral constraints leading to less violent types of crime or no criminal behavior at all.

The effects of declining religiosity may seem straightforward. Religiosity is protective, so it is reasonable that a decreased level of religiosity is associated with an increase in the risk of criminal behavior (Pirutinsky, 2014). Specifically, the reduction of religious participation to some extent may reflect the loss of relevant positive social support/control and coping strategies that may continue to keep offenders away from criminal activities as they age. Involvement in religious activities may keep offenders from later crime because it occupies otherwise free time to engage in crime activities, imposes standards and guidelines of moral and righteous behavior, enhances the relationships with conventionally oriented peers and mentors, and provides positive social and coping skills that help overcome stress and strain in the life (Desmond et al., 2010). Subjective religiosity, particularly, spirituality may promote a healthy self-concept/control (McCullough & Willoughby, 2009), enhance a sense of self-forgiveness, and facilitate the development of new prosocial identities (Maruna, 2001). Conversely, the loss of religious commitment tied to feelings of purpose and meaning in life may contribute to a less positive self-concept/control. Offenders may therefore be less likely to forgive themselves for the things they have done wrong, inhibiting a transformation in an offender’s identity to a prosocial identity and motivation for good. The loss of spirituality may become the potential risk for crime.

There may be alternative explanations regarding the effects of declining religiosity on crime. In the typical age-crime curve, the prevalence of offending tends to increase from late childhood, peaks in the teenage years (around ages 15 to 19), and then declines from the early 20 s (Nagin & Tremblay, 2005). To the degree that religiosity restrains crime, high religiosity during early adolescence should delay entry of adolescents into this sequence. As their religiosity later declines, commencement of this sequence is more likely, but they would start it later. Because their peak years of crime would occur after those with an initial low religiosity, their current level of crime would be amplified later. Efforts to compensate for a former relative lack of crime may also explain later high levels of crime among initially religious adolescents. As such, “a decrease from prior high religiosity not only removes a previous deterrent but also may provoke an effort to “make up for lost time.” (Charles et al., 1985, p. 121).

The reduction of criminal behavior among offenders whose religiosity increases may indicate the continuity of religious deterrence. Emerging adults are caught in rapidly changing contexts, including but not limited to the decreased social control and support from parents, dramatic life-events, and evolving identity, that may increase the risk of crime (Arnett, 2000). Being active in a religious community may still provide positive social support and control to emerging adults, increasing the probability of following a trajectory of low-level crime throughout adolescence to young adulthood (Petts, 2009). The continued increase of spirituality to some extent reflects consistent beliefs and less alterations in beliefs. It may serve as an additional protective mechanism or coping strategy when people respond to the loss of certain direct and indirect protective factors due to changing life and social experiences. Not surprisingly, the gradual increase of religiosity may continue to attenuate the increased risk of being involved in criminal behavior during emerging adulthood.

Given the potential protective effect of religiosity, the increased likelihood of engagement in violent offending among those who possess a moderate gain in spirituality (i.e., increase from low spirituality to medium spirituality) may seem counterintuitive. Religiosity developed in adulthood is less dependent on others’ expectations, but instead depends on the commitment to religion and internalization of religious beliefs on their own volition (Koenig et al., 2008). However, as Diener et al. (2011) pointed out, religiosity is sometimes accompanied by difficult life circumstances. With respect to those who experience a gain in religiosity over time, the increased religiosity may signify the possibility to cope with stressful life events and stimuli that often result in criminal behavior as well (Jang & Johnson, 2003). The mechanism by which an increase in religiosity contributes to an increased risk of crime may align with reasons that individuals seek religion in adulthood to counteract the risk factors for crime. Additionally, this small increase in spirituality, to some extent, shows that offenders in this trajectory do not actually have strong commitment to religion. The increased spirituality may just be a form of seeking help from religion to reduce the stress from life circumstances.

Limitations and Directions for Future Studies

Although the findings extended previous research in many ways, there were several limitations inherent in the current study. First, a measure of religious affiliation is not available in this dataset. As a result, it is impossible to investigate the role of religious affiliation in the explanation of crime among this sample of serious adolescent offenders. Considering that some fundamentalist groups (e.g., Christians and Mormons) are more inclined to be involved in crime than other denominations (e.g., Catholics) (Jensen & Erickson, 1979), the relationship between religiosity and crime may not be uniform across different religious denominations.

Second, some researchers argue that the configuration of the measures of religiosity at the individual level may be extremely complex, such that they cannot be captured in examining measures of religiosity in isolation or in combination (Pearce et al., 2013). In order to capture religiosity, inductive statistical methods such as cluster or latent class analyses have been encouraged to identify distinct religious profiles—unique combinations of individual measures of religiosity—that are meaningful to individuals in their life, yet shared by many people (Salas-Wright et al., 2014). Employing this approach to model multifaceted religiosity, individuals may be classified into not only straightforward religious profiles, such as irreligious and the highly organizationally or intrinsically involved, but also more nuanced profiles of religiosity, such as that of individuals with high subjective religiosity (e.g., religious importance or beliefs) but little objective religiosity (e.g., religious attendance), or vice versa. Thus, in addition to the examination of changes in individual components of religiosity, respectively, future studies need to assess the ways in which multifaceted religious profiles evolve over time when investigating the religiosity-crime link over time.

Third, although this study was conducted within a longitudinal design, it emphasized the contemporaneous effects of religiosity on crime. This study is limited in its ability to make a causal inference regarding the effect of religiosity on crime. It is difficult to identify whether religiosity or crime comes first and then rule out the possibility of reverse causality (i.e., the impact of crime on religiosity). Jang (2018) suggested that the relationship between religiosity and crime could be either bidirectional or unidirectional, depending on the developmental period being studied. Building upon this work, future research is needed to further explore the direction of the religiosity-crime link and clarify possible reciprocal effects.

Finally, this study included a variety of factors based on existing theoretical perspectives, arguing that religiosity can impact crime through the effects of social bonds, social learning, self-control, coping strategies and turning points. This study drew attention to the significant direct effects of religiosity on crime by controlling for possible spurious effects and indirect effects. It did not explicitly investigate the mechanisms that account for the effects of changes in religiosity on changes in crime. Developmental patterns of religiosity may have an impact on life changes, such as getting married and establishing a career, which in turn could contribute to desistance of crime (Bakken et al., 2013; Giordano et al., 2008). Given that the indirect effect of religiosity is also religious influence, these theoretical mechanisms or mediating effects need to be parsed out further in future research when investigating the long-term effects of religiosity on trajectories of crime.

Conclusions and Implications

Despite these limitations, this study to some extent indicated changes of religiosity over time may have the potential to stimulate long-term behavioral changes away from crime. Although this study did not specifically lead to a prevention strategy, it may make an important contribution by illustrating how changes in religiosity may be related to the trajectories of crime between adolescence and young adulthood. This knowledge may be useful in developing strategies to encourage at-risk adolescents to avoid criminal behavior throughout adolescence into young adulthood. The study indicated that certain measures of religiosity may not only inhibit the initial levels of criminal behavior but also deter their continued involvement. Programs designed to introduce religiosity into serious offenders’ lives, especially prison ministry programs, may take note of certain religiosity elements, like spirituality and religious attendance.

Aspects of religiosity, like religious attendance and spirituality, may be incorporated with preventive and rehabilitative initiatives of criminal behavior. For instance, religious institutions in the community may be encouraged to develop various youth programs and deliver services to prevent at-risk adolescents from the onset of crime, as well as reach out to individuals who have been involved in status offenses or income crime (e.g., sold marijuana or other illegal drugs, shoplifted or so on) by increasing participation in religious services. Given that released offenders face multiple challenges or difficulties when they return to their families and communities, it is challenging for many offenders to desist from crime. A strong sense of spirituality may serve as a guide for coping with the tumultuous life situations and circumstances that released offenders may encounter, “such as dealing with issues relating to substance use, unemployment, reconnecting with family and peers, and finding adequate housing” (Bakken et al., 2013, p. 14). Intrinsically motivated religiosity can further help juvenile offenders stay away from violent and more serious crimes.

Supplemental Material

Appendix – Supplemental material for Developmental Patterns of Religiosity in Relation to Criminal Trajectories among Serious Offenders moving from Adolescence to Young Adult

Supplemental material, Appendix for Developmental Patterns of Religiosity in Relation to Criminal Trajectories among Serious Offenders moving from Adolescence to Young Adult by Siying Guo in Crime & Delinquency

Footnotes

Acknowledgements

The author would like to thank Christi Metcalfe for her helpful comments on an earlier draft of this article.

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Siying Guo

Supplemental Material

Supplemental material for this article is available online.

Author Biography

Siying Guo is an assistant professor of Criminal Justice at the Department of History, Philosophy, and Social Sciences, Pittsburg State University. Her research interests include bullying, juvenile delinquency and justice, youth development, religion and crime, life-course criminology, and quantitative methodology. Email: sguo@pittstate.edu

References

Adamczyk

Palmer

(2008). Religion and initiation into marijuana use: The deterring role of religious friends. Journal of Drug Issues, 38(3), 717–742.

Allison

P. D.

(2001). Missing data (Vol. 136). SAGE.

Arnett

J. J.

(2000). Emerging adulthood: A theory of development from the late teens through the twenties. American Psychologist, 55(5), 469–480.

Arnett

J. J.

(2007). Emerging adulthood: What is it, and what is it good for? Child Development Perspectives, 1(2), 68–73.

Atchley

R. C.

(1999). Continuity and adaptation in aging: Creating positive experiences. Baltimore, MD: Johns Hopkins University Press.

Baier

C. J.

Wright

B. R.

(2001). “If you love me, keep my commandments”: A meta-analysis of the effect of religion on crime. Journal of Research in Crime and Delinquency, 38(1), 3–21.

Bakken

N. W.

Gunter

W. D.

Visher

C. A.

(2013). Spirituality and desistance from substance use among reentering offenders. International Journal of Offender Therapy and Comparative Criminology, 58(11), 1321–1339. https://doi.org/10.1177/0306624X13494076

Bandura

Barbaranelli

Caprara

G. V.

Pastorelli

(1996). Mechanisms of moral disengagement in the exercise of moral agency. Journal of Personality and Social Psychology, 71(2), 364–374.

Benda

B. B.

Corwyn

R. F.

(1997). Religion and delinquency: The relationship after considering family and peer influences. Journal for the Scientific Study of Religion, 36(1), 81–92.

10.

Benda

B. B.

Pope

S. K.

Kelleher

K. J.

(2006). Church attendance or religiousness: Their relationship to adolescents’ use of alcohol, other drugs, and delinquency. Alcoholism Treatment Quarterly, 24(1–2), 75–87.

11.

Burkett

S. R.

Warren

B. O.

(1987). Religiosity, peer associations, and adolescent marijuana use: A panel study of underlying causal structures. Criminology, 25(1), 109–132.

12.

Burkett

S. R.

White

(1974). Hellfire and delinquency: Another look. Journal for the Scientific Study of Religion, 13(4), 455–462.

13.

Cernkovich

S. A.

Giordano

P. C.

(1992). School bonding, race, and delinquency. Criminology, 30(2), 261–291.

14.

Charles

W. P.

Curry

E. W.

Chalfant

H. P.

(1985). Religiosity and delinquency over time: Deviance deterrence and deviance amplification. Social Science Quarterly, 66(1), 120–131.

15.

Chu

(2007). Religiosity and desistance from drug use. Criminal Justice and Behavior, 34(5), 661–679.

16.

Cochran

J. K.

(1988). The effect of religiosity on secular and ascetic deviance. Sociological Focus, 21(4), 293–306.

17.

Cochran

J. K.

Wood

P. B.

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(1), 92–123.

18.

Conger

R. D.

Elder

G. H.

Jr. Lorenz

F. O.

Simons

R. L.

(1994). Economic stress, coercive family process, and developmental problems of adolescents. Child Development, 65(2), 541–561.

19.

Denton

M. L.

Pearce

L. D.

Smith

(2008). Religion and spirituality on the path through adolescence. National Study of Youth and Religion.

20.

Desmond

S. A.

Kikuchi

Budd

(2010). Does change in teenage religiosity predict change in Marijuana use over time? Interdisciplinary Journal of Research on Religion, 6, Article 5.

21.

Desmond

S. A.

Morgan

K. H.

Kikuchi

(2010). Religious development: How (and why) does religiosity change from adolescence to young adulthood? Sociological Perspectives, 53(2), 247–270.

22.

Desmond

S. A.

Soper

S. E.

Kraus

(2011). Religiosity, peers, and delinquency: Does religiosity reduce the effect of peers on delinquency? Sociological Spectrum, 31(6), 665–694.

23.

Desmond

S. A.

Soper

S. E.

Purpura

D. J.

Smith

(2008). Religiosity, moral beliefs, and delinquency: Does the effect of religiosity on delinquency depend on moral beliefs? Sociological Spectrum, 29(1), 51–71.

24.

Diener

Tay

Myers

D. G.

(2011). The religion paradox: If religion makes people happy, why are so many dropping out? Journal of Personality and Social Psychology, 101(6), 1278–1290.

25.

Elifson

K. W.

Petersen

D. M.

Hadaway

(1983). Religiosity and delinquency. Criminology, 21(4), 505–528.

26.

Elliott

(1990). Educational research in crisis: Performance indicators and the decline in excellence. British Educational Research Journal, 16(1), 3–18.

27.

Ellison

C. G.

Levin

J. S.

(1998). The religion-health connection: Evidence, theory, and future directions. Health Education & Behavior, 25(6), 700–720.

28.

Ellison

C. G.

Trinitapoli

J. A.

Anderson

K. L.

Johnson

B. R.

(2007). Race/ethnicity, religious involvement, and domestic violence. Violence Against Women, 13(11), 1094–1112.

29.

Evans

T. D.

Cullen

F. T.

Burton

V. S.

Jr. Dunaway

R. G.

Payne

G. L.

Kethineni

S. R.

(1996). Religion, social bonds, and delinquency. Deviant Behavior, 17(1), 43–70.

30.

Giordano

P. C.

Longmore

M. A.

Schroeder

R. D.

Seffrin

P. M.

(2008). A life-course perspective on spirituality and desistance from crime. Criminology, 46(1), 99–132.

31.

Guo

Metcalfe

(2019). Religion as a social control: A longitudinal study of religious involvement and substance use. Crime & Delinquency, 65(8), 1149–1181.

32.

Hirschi

Stark

(1969). Hellfire and delinquency. Social Problems, 17(2), 202–213.

33.

Jang

S. J.

(2018). Religiosity, crime, and drug use among juvenile offenders: A test of reciprocal relationships over time. International Journal of Offender Therapy and Comparative Criminology, 62(14), 4445–4464.

34.

Jang

S. J.

(2019). Religiosity, crime, and drug use among juvenile offenders: A latent growth modeling approach. Journal of Quantitative Criminology, 35(1), 27–60.

35.

Jang

S. J.

Bader

C. D.

Johnson

B. R.

(2008). The cumulative advantage of religiosity in preventing drug use. Journal of Drug Issues, 38(3), 771–798.

36.

Jang

S. J.

Johnson

B. R.

(2001). Neighborhood disorder, individual religiosity, and adolescent use of illicit drugs: A test of multilevel hypotheses. Criminology, 39(1), 109–144.

37.

Jang

S. J.

Johnson

B. R.

(2003). Strain, negative emotions, and deviant coping among African Americans: A test of general strain theory. Journal of Quantitative Criminology, 19(1), 79–105.

38.

Jensen

G. F.

Erickson

M. L.

(1979). The religious factor and delinquency: Another look at the hellfire hypothesis. In Wuthnow

(Ed.), The religious dimension; New directions in quantitative research (pp. 157–177). New York: Academic Press.

39.

Johnson

B. R.

Larson

D. B.

McCullough

M. E.

(2000). Religion and delinquency: A systematic review of the literature. Journal of Contemporary Criminal Justice, 16(1), 32–52.

40.

Jones

B. L.

Nagin

D. S.

Roeder

(2001). A SAS procedure based on mixture models for estimating developmental trajectories. Sociological Methods & Research, 29(3), 374–393.

41.

Koenig

L. B.

McGue

Iacono

W. G.

(2008). Stability and change in religiousness during emerging adulthood. Developmental Psychology, 44(2), 532–543.

42.

Laird

R. D.

Marks

L. D.

Marrero

M. D.

(2011). Religiosity, self-control, and antisocial behavior: Religiosity as a promotive and protective factor. Journal of Applied Developmental Psychology, 32(2), 78–85.

43.

Maruna

(2001). Making good: How ex-convicts reform and rebuild their lives. American Psychological Association.

44.

Mason

W. A.

Spoth

R. L.

(2011). Thrill seeking and religiosity in relation to adolescent substance use: Tests of joint, interactive, and indirect influences. Psychology of Addictive Behaviors, 25(4), 683–696.

45.

Mason

W. A.

Windle

(2002). A longitudinal study of the effects of religiosity on adolescent alcohol use and alcohol-related problems. Journal of Adolescent Research, 17(4), 346–363.

46.

Maton

K. I.

(1989). The stress-buffering role of spiritual support: Cross-sectional and prospective investigations. Journal for the Scientific Study of Religion, 28(3), 310–323.

47.

McCullough

M. E.

Willoughby

B. L.

(2009). Religion, self-regulation, and self-control: Associations, explanations, and implications. Psychological Bulletin, 135(1), 69–93.

48.

Moscati

Mezuk

(2014). Losing faith and finding religion: Religiosity over the life course and substance use and abuse. Drug and Alcohol Dependence, 136, 127–134.

49.

Mulvey

E. P.

Schubert

C. A.

(2012). Some initial findings and policy implications of the Pathways to Desistance study. Victims & Offenders, 7(4), 407–427.

50.

Nagin

D. S.

(2005). Group-based modeling of development. Harvard University Press.

51.

Nagin

D. S.

Tremblay

R. E.

(2005). Developmental trajectory groups: Fact or a useful statistical fiction? Criminology, 43(4), 873–904.

52.

Pargament

K. I.

Saunders

S. M.

(2007). Introduction to the special issue on spirituality and psychotherapy. Journal of Clinical Psychology, 63(10), 903–907.

53.

Pearce

M. J.

Jones

S. M.

Schwab-Stone

M. E.

Ruchkin

(2003). The protective effects of religiousness and parent involvement on the development of conduct problems among youth exposed to violence. Child Development, 74(6), 1682–1696.

54.

Pearce

L. D.

Denton

M. L.

(2011). A faith of their own: Stability and change in the religiosity of America’s adolescents. Oxford University Press.

55.

Pearce

L. D.

Foster

E. M.

Hardie

J. H.

(2013). A person-centered examination of adolescent religiosity using latent class analysis. Journal for the Scientific Study of Religion, 52(1), 57–79.

56.

Petts

R. J.

(2007). Religious participation, religious affiliation, and engagement with children among fathers experiencing the birth of a new child. Journal of Family Issues, 28(9), 1139–1161.

57.

Petts

R. J.

(2009). Family and religious characteristics’ influence on delinquency trajectories from adolescence to young adulthood. American Sociological Review, 74(3), 465–483.

58.

Pirutinsky

(2014). Does religiousness increase self-control and reduce criminal behavior? A longitudinal analysis of adolescent offenders. Criminal Justice and Behavior, 41(11), 1290–1307.

59.

Regnerus

M. D.

Uecker

J. E.

(2006). Finding faith, losing faith: The prevalence and context of religious transformations during adolescence. Review of Religious Research, 47(3), 217–237.

60.

Rhodes

A. L.

Reiss

A. J.

Jr. (1970). The “religious factor” and delinquent behavior. Journal of Research in Crime and Delinquency, 7(1), 83–98.

61.

Rodell

D. E.

Benda

B. B.

(1999). Alcohol and crime among religious youth. Alcoholism Treatment Quarterly, 17(4), 53–66.

62.

Salas-Wright

C. P.

Vaughn

M. G.

Maynard

B. R.

(2014). Profiles of religiosity and their association with risk behavior among emerging adults in the United States. Emerging Adulthood, 3(2), 67–84. https://doi.org/10.1177/2167696814539327

63.

Salas-Wright

C. P.

Vaughn

M. G.

Maynard

B. R.

Clark

T. T.

Snyder

(2014). Public or private religiosity: Which is protective for adolescent substance use and by what pathways? Youth & Society, 49(2), 228–253. https://doi.org/10.1177/0044118X14531603

64.

Schubert

C. A.

Mulvey

E. P.

Steinberg

Cauffman

Losoya

S. H.

Hecker

. . .Knight

G. P.

(2004). Operational lessons from the pathways to desistance project. Youth Violence and Juvenile Justice, 2(3), 237–255.

65.

Steinberg

Lamborn

S. D.

Dornbusch

S. M.

Darling

(1992). Impact of parenting practices on adolescent achievement: Authoritative parenting, school involvement, and encouragement to succeed. Child Development, 63(5), 1266–1281.

66.

Stoppa

T. M.

Lefkowitz

E. S.

(2010). Longitudinal changes in religiosity among emerging adult college students. Journal of Research on Adolescence, 20(1), 23–38.

67.

Thornberry

T. P.

Lizotte

A. J.

Krohn

M. D.

Farnworth

Jang

S. J.

(1994). Delinquent peers, beliefs, and delinquent behavior: A longitudinal test of interactional theory. Criminology, 32(1), 47–83.

68.

Ulmer

J. T.

Desmond

S. A.

Jang

S. J.

Johnson

B. R.

(2012). Religious involvement and dynamics of marijuana use: Initiation, persistence, and desistence. Deviant Behavior, 33(6), 448–468.

69.

Weinberger

D. A.

Schwartz

G. E.

(1990). Distress and restraint as superordinate dimensions of self-reported adjustment: A typological perspective. Journal of Personality, 58(2), 381–417.

70.

Yonker

J. E.

Schnabelrauch

C. A.

DeHaan

L. G.

(2012). The relationship between spirituality and religiosity on psychological outcomes in adolescents and emerging adults: A meta-analytic review. Journal of Adolescence, 35(2), 299–314.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.76 MB