Abstract
While research conducted in Western nations suggests that religiosity and spirituality protect against delinquent behavior, few studies have examined these relationships in developing world contexts. Using a community sample of 290 high-risk and gang-involved adolescents (11-17 years) and young adults (18-25 years) in San Salvador, El Salvador, structural equation modeling and logistic regression are used to examine the relationships between religious coping, spirituality, social developmental factors, and delinquency. Results suggest that spirituality and, to a lesser extent, religious coping, protect Salvadoran youth at risk for involvement in delinquent behavior. The relationship between spirituality and delinquency was completely mediated by social developmental factors as no direct association was identified between spirituality and delinquency. Spirituality and religious coping are both relevant protective factors among Salvadoran youth; however, spirituality is more consistently protective against delinquency in terms of its relationship to social developmental mediating factors and its direct associations to particular delinquent behaviors.
Although substantial evidence has accrued that establishes a protective role for religiosity and spirituality against adolescent and young adult involvement in delinquent and criminal behavior (Baier & Wright, 2001; Koenig, King, & Carson, 2012), much less is known about the universality of these effects. Young people who are involved in public religious practices and who endorse elevated levels of private religiosity and spirituality tend to avoid the pitfalls of delinquency and related antisocial behaviors more often than their less religious and less spiritual counterparts (Salas-Wright, Vaughn, Hodge, & Perron, 2012). In addition, religious coping, which relates to the use of religious engagement as a means of managing stressful life events (Pargament, Koenig, & Perez, 2000) and shares substantial conceptual overlap with religiosity (Eisenberg et al., 2011), has been found to have important associations with self-regulatory and behavioral factors linked with antisocial and risky behavior (Ano & Vasconcelles, 2005; Brechting & Giancola, 2007). Notably, however, while a limited number of studies have looked cross-nationally at these concepts (Antonaccio, Tittle, Botchkovar, & Kranidiotis, 2010), the vast majority of research on this topic has been conducted among adolescents and young adults in the United States. As such, important questions remain in terms of establishing the general effect and nature of these relationships in cross-national contexts, particularly in developing nations that experience extremely high rates of violent and delinquent activities, such as El Salvador, Central America (United Nations Office on Drugs and Crime [UNODC], 2011). This is critical for criminological knowledge as cross-national tests are important additions to the convergent validity of theoretical constructs (Farrington, 2000).
The dearth of research on the relationships between religiosity, spirituality, and delinquency in less developed nations is marked. With respect to El Salvador, it is particularly noteworthy, as we know that religiosity and spirituality are salient factors in the lives of many young Salvadorans (Instituto Universitario de Opinión Pública [IUDOP], 2009). In a nationally representative study conducted by Santacruz Giralt and Carranza (2009), Salvadoran youth reported higher levels of participation in religious youth groups than in all other social groups except athletic teams. Santacruz Giralt and Carranza also found that more than three quarters of Salvadoran adolescents and young adults consider religion to be a “very important” aspect of their lives. These figures for young people are in keeping with overall levels of religious involvement among the Salvadoran population as a whole, as an estimated 90% of Salvadorans report belonging to a religious community (IUDOP, 2009) and nearly 7 in 10 Salvadorans report attending religious services at least once per month (World Values Survey, 1999).
Scholars have also noted that religion and spirituality are relevant not only to Salvadoran youth in general but also to high-risk youth engaged in delinquency and antisocial behavior. For instance, while few active Central American gang members are welcome in most religious communities, entrance into an evangelical Christian community is one of the few viable options for gang members interested in terminating gang membership and reentering mainstream society (Bellanger & Rocha, 2007). Indeed, interesting community-level relationships can be observed in terms of religious participation and the incidence of youth gangs in communities. A recent study by Equipo de Reflexión, Investigación, y Comunicación (ERIC; 2005) in Honduras found that higher levels of community participation in churches were associated with lower levels of gang participation in communities. Moreover, it can also be noted that many of the social service organizations that are actively involved in providing psychosocial services to high-risk and gang-involved Salvadoran and Central American youth are either faith-related or faith-based (Wolf, 2008). While perhaps controversial, the integration of religious or spiritual elements into social interventions may have its benefits in the Salvadoran context. Indeed, Brenneman (2012) suggested that for many former gang members in Central America, the entrance into evangelical churches can be a genuinely transformative process by which troubled youth are able to address feelings of chronic shame and other issues relating to male violence.
Research Context: El Salvador
Over the last decade, El Salvador has received great attention with its international reputation as a hotbed for youth gang activity, and widespread crime and violence (Decker & Pyrooz, 2010). Indeed, the tiny Central American nation is considered one of the most violent countries in the world (United Nations Development Program [UNDP], 2009) with an astoundingly high homicide rate of 66 per 100,000 inhabitants—a rate nearly 3 times that of Latin America in general and almost 10 times greater than the rate worldwide (UNODC, 2011). Any discussion of crime and delinquency in El Salvador must also allude to the influence of the nation’s two primary youth gangs: La Mara Salvatrucha (MS-13) and El Barrio 18 (“the 18th Street Gang”). Conservative estimates suggest that Salvadoran gang members from these two groups total more than 29,000 and—despite accounting for less than one third of one percent of the nation’s total population—are believed to be responsible for as much as 25% of the nation’s homicides, violent crimes, and delinquency (Cruz, 2005; Demoscopía, 2007). Indeed, in keeping with trends among high-risk and gang-member youth in the United States, youth gang members in El Salvador have been found to be significantly more likely than their nongang counterparts to report involvement in delinquent and antisocial behavior (Olate, Salas-Wright, & Vaughn, 2011). Salvadoran youth are situated at the center of this social crisis as no age group is more likely to be victimized by or perpetrate violence than Salvadoran adolescents and young adults (Cruz, 2005).
In the face of this dire reality, scholars have begun to identify risk and protective factors that can be used in predicting the likelihood that Salvadoran youth will participate in delinquency and other problem behaviors (Cruz, 1999; Olate, Salas-Wright, & Vaughn, 2012). As an alternative to the enforcement-based approaches—which rely primarily on law enforcement strategies to address delinquency and gang activity—frequently implemented in Central America (Olate & Salas-Wright, 2010), the identification of such factors in target populations is an essential component of crafting effective prevention and intervention programs designed to address delinquency. Empirically identified risk factors for youth problem behaviors in El Salvador include callousness, school expulsion, low future orientation, delinquent peer affiliation, gang membership, violent victimization, and low social support (Cruz, 1999; Olate et al., 2012). Notably, few studies have systematically examined protective factors for delinquency in the Salvadoran context.
Research in the United States has identified religiosity and spirituality as important individual characteristics that can serve to buffer the adverse outcomes of psychosocial risk factors in youth. Such effects have been found to be particularly robust in sociocultural contexts in which the cultural importance of religious practice is paramount (Regnerus, 2003b), as is the case in El Salvador and throughout much of Latin America (IUDOP, 2009). Moreover, a growing body of literature also suggests that religiosity may function as an important factor in the lives of potential, current, and former gang members in the United States (Flores, 2009; Marsal, 2009; Morris, 2012). Although identifying such factors is an important first step in understanding the dynamics of juvenile delinquency and crime, it has been noted that this approach alone is insufficient as it is often done without a coherent theoretical foundation and typically fails to identify mediating mechanisms (Farrington, 2000).
Religious Coping, Spirituality, and Delinquency
As noted above, an increasingly robust body of literature points to the inverse relationship between religiosity, spirituality, and a variety of delinquent or criminal behaviors (Baier & Wright, 2001). As is the case for multiple problem behaviors, while scholars have begun to focus on the protective effect of religious coping (Brechting & Giancola, 2007; Eisenberg et al., 2011), the most commonly examined religiosity and spirituality variables are religious service attendance and religious or spiritual salience. Relatively few studies have systematically examined the relationship between religious coping and problem behavior among youth. Eisenberg and colleagues (2011) examined the relationship between trajectories of religious coping and externalizing/prosocial behavior among a cohort of Italian youth between late adolescence (age 16-17) and young adulthood (age 22-23). Overall, youth categorized as consistently having high levels of religious coping were found to exhibit more prosocial behavior and fewer externalizing behaviors than youth with less consistently high levels of religious coping. Along the same lines, Brechting and Giancola (2007) found, in a study of primarily non-Hispanic White adolescent boys in the United States, that religious coping during early adolescence (age 12-14) was predictive of lower rates of drug use and abuse during mid- to late adolescence (age 15-16). Although the evidence on the relationship between religious coping and delinquency is relatively thin, there nevertheless is strong conceptual evidence to suggest that religious coping may function to buffer against delinquent behavior inasmuch as religious coping is associated with self-regulatory, psychological, and social factors associated with delinquency and antisocial behavior (Ano & Vasconcelles, 2005). As noted above, religious coping is also a construct that shares substantial overlap with religious salience, which has been well-documented as a deterrent for delinquency and crime (Baier & Wright, 2001).
Theoretical Framework
The theoretical framework for this study draws upon the social development model (SDM; Catalano & Hawkins, 1996). The SDM draws from social learning theory, social control theory, and differential association theory in explaining the development of pro- and antisocial behaviors, including delinquency. The SDM hypothesizes that an individual learns behaviors, be they prosocial or antisocial, through a process of socializing and bonding with key socializing units such as family, school, peers, and other community groups. According to the theory, when youth perceive opportunities to take part in social activities or interactions and have sufficient skills to participate successfully over time, a process of social bonding ensues. Once such bonds are established, the bonds themselves have the capacity to subsequently influence beliefs and behaviors as youth seek to conform to the basic pro- or antisocial norms of the bonded group. The SDM stands out as a particularly useful theory in terms of the exploration of the research questions central to this study given that scholars have consistently emphasized the importance of conceptualizing the protective mechanisms associated with youth religiosity and spirituality within the framework of social relationships and social bonding (Regnerus, 2003c).
The Present Study
The objective of this study is to examine the relationships between religious coping, spirituality, and delinquency among high-risk and gang-involved adolescents, and young adults in San Salvador, El Salvador. Although a robust body of empirical research suggests that religiosity and spirituality function as important protective factors, the relationships between variables related to these constructs have rarely been examined in developing world nations such as El Salvador. As such, drawing from the theoretical framework of the SDM, this study uses a community sample of high-risk adolescents (11-17) and young adults (18-25) to examine the relationships between religious coping, spirituality, and severe and minor delinquency with antisocial bonding and antisocial beliefs included as mediating factors. In addition, the study also examines the associations between religious coping, spirituality, and a variety of particular delinquent behaviors. To this end, three research hypotheses will be examined:
Hypothesis 1: Higher levels of religious coping and spirituality, as mediated by antisocial bonding and antisocial beliefs, are associated with lower levels of severe delinquency.
Hypothesis 2: Higher levels of religious coping and spirituality, as mediated by antisocial bonding and antisocial beliefs, are associated with lower levels of minor delinquency.
Hypothesis 3: When controlling for antisocial bonding and antisocial beliefs, higher levels of religious coping and spirituality are associated with the decreased likelihood of youth involvement in particular severe and minor delinquent behaviors.
Testing these hypotheses builds upon prior research on the role of religious coping and spirituality as protective factors for delinquency by examining the relationships in an underresearched population in a less developed society. In all, the goal of this study is to examine the influence of religious coping and spirituality on delinquency, both in terms of direct associations and with regard to the pathways of key social developmental factors to further understand the universality of these effects.
Method
Sampling Procedures
This study examines survey data from a purposive sample of 290 respondents (adolescents: n = 116; young adults: n = 174) in San Salvador, El Salvador, recruited between June and November 2011. The study was carried out in collaboration with a Salvadoran high-risk youth development organization that works closely with active gang members and high-risk youth across El Salvador. All survey respondents were conceptualized as “high-risk” due to the fact that they resided in communities experiencing high levels of various social risk factors for delinquency and crime (e.g., socioeconomic marginalization, high levels of community violence, etc.). Survey respondents were recruited by staff members of the youth organization by means of snowball or chain-referral sampling, which is an acceptable method of nonprobability sampling in the study of hard-to-reach populations (Volz & Heckathorn, 2008). Participants involved in the youth development organization were recruited to participate and were also encouraged to recruit high-risk and gang-involved neighbors, peers, friends, and family members living in the local community to participate in the study as well. Given the extreme conditions of marginalized communities in El Salvador and the difficulty in accessing gang-involved Salvadoran youth, more readily generalizable sampling strategies were not considered viable. Study procedures were approved by the Institutional Review Boards of Boston College and the Ohio State University.
Survey Design and Translation
The original survey instrument was designed for English-speaking North American youth only. To make the survey accessible for Salvadoran youth, a cross-cultural translation procedure was carried out. This multistep process included careful translation, back translation, cognitive interviews, and pilot testing with high-risk and gang-involved youth in El Salvador and in the United States. Importantly, survey items relating to religious coping and spirituality were translated and back translated, but no cognitive interviews or pilot tests were carried out for these constructs.
Data Collection Procedures
Data were collected in community-based youth development centers in a variety of marginalized communities in San Salvador, El Salvador. Given the dangers involved in Salvadoran gang members traveling beyond their designated communities, such on-site data collection was essential to gain access to gang-involved youth. Surveys were administered individually and in small groups by a team of three Salvadoran youth development staff members trained at the university level in the social sciences and survey methodology. For the majority of youth, the survey required between 30 and 45 min for completion. Youth who were identified as having difficulties in terms of concentration, reading, or writing were administered the survey individually by a member of the research team. All respondents were given US$5.00 for their participation. Consent for respondents under the age of 18 was granted by a parent or, in the absence of such a figure, by a supervising member of the youth development program staff. All respondents who agreed to the informed consent procedure participated fully in the survey administration.
Measures
Independent Variables
Two primary independent variables were examined in this study: religious coping and spirituality. These two variables share substantial overlap but nevertheless are unique: Religious coping relates to the use of religious engagement as a means of managing stressful life events (Pargament et al., 2000); spirituality, in contrast, refers to the relationship between an individual and God (or a perceived transcendence), and may or may not be linked with institutional religious communities or practices (Hodge, 2000).
In this study, both measures were dichotomized so as to facilitate a clear distinction between the presence and absence of each of these hypothesized protective factors. To the extent possible, scores for these protective factors within the appropriate upper quartile were identified as “protective” and coded as “1” whereas the other three quartiles of the sample were considered to be beyond the designated spectrum of protection and thus coded as “0.” In both these cases, a recoding at the exact 75th percentile was not possible, and thus the closest cut-point available was selected. While some variation exists in terms of the precise operationalized point of risk exposure (Thornberry, Krohn, Lizotte, Smith, & Tobin, 2003), this method of dichotomization was selected as it is likely to identify only those youth who are genuinely “protected” inasmuch as it is designed to be a cautious measure of exposure.
Religious coping
The measure of religious coping comprises four positive religious coping items derived from the Measure of Religious Coping (RCOPE; Pargament et al., 2000). For the purposes of this study, neither the RCOPE nor the brief RCOPE was used in their entirety. However, three of the four RCOPE items used had been previously used together as part of a six-item positive/negative religious coping scale included in the 1998 General Social Survey (GSS). The fourth item—designed to measure comprehensive positive religious coping—was slightly altered as the term God was substituted for “religion” to measure de-institutionalized theistic coping as it was deemed necessary to account for the fact that many gang-involved youth in El Salvador, though perhaps intrinsically religious, tend to not be involved in formal religious communities. Three of the items measured the degree to which respondents were in agreement with religious statements such as, “I look to God for strength, support, and guidance in crises” and “I work together with God as partners to get through hard times.” These items had the response format of 1 = not at all, 2 = somewhat, 3 = quite a bit, and 4 = a great deal. The fourth item asked respondents, “To what extent is God helping you to understand or deal with stressful situations in your life?” This item had the response format of 1 = God does not help me, 2 = God helps me a little, 3 = God somewhat helps me, and 4 = God helps me a lot. The four-item RCOPE was found to have acceptable reliability and validity with an average Cronbach’s alpha value of .800 (α = .811 for adolescents; α = .746 for young adults) and all items loading cleanly on a single factor (.705-.863).
Spirituality
The Intrinsic Spirituality Scale (Hodge, 2003), which is designed to measure spirituality as a motivational construct for theistic and nontheistic populations alike, is the measure of spirituality. While this six-item scale has not been validated in the Spanish language, research into the psychometric properties of the scale conducted in English with a predominantly non-Hispanic White sample of university students (ages 17-25) has demonstrated high reliability (α = .96) and acceptable validity in terms of goodness of fit and concurrent validity performance (Hodge, 2003). In the English version of the scale, three items were reverse coded; however, in light of previous research as to the dubious psychometric properties of reverse-coded items with Salvadoran high-risk youth as well as marginalized and educationally disadvantaged youth in general (Salas-Wright, Olate, & Vaughn, 2013), these three items were adjusted so that all items were positively phrased. The response options for the Intrinsic Spirituality Scale range from 0 to 10, with 0 indicating low levels of spirituality and 10 indicating the highest levels of spirituality. The scale includes items such as, “My spiritual beliefs affect,” with response categories ranging from 0 = no aspect of my life to 10 = absolutely every aspect of my life. An additional sample item prompts, “Spirituality is,” with response categories ranging from 0 = not part of my life to 10 = the most important part of my life. Other items tap constructs such as spiritual growth, the impact of spirituality on decision making, and the use of spirituality as an interpretive lens. A reduced five-item scale was found to have acceptable reliability and validity with an average Cronbach’s alpha value of .912 (α = .908 for adolescents; α = .928 for young adults) and all items loading cleanly on one factor (.828-.888).
Family life
Two variables measured well-being in the domain of family: positive family life and single-mother household status. The positive family life variable is a binary measure of the experience of the family life of respondents derived from the combination of two binary variables: living with family and significant problems at home. Youth who reported that they (a) lived with their family of origin and (b) have no significant problems at home were considered to have a “positive experience living with family” (47.79%) and were coded as “1.” Youth who responded “no” to either variables or to both were considered to not meet the criteria of a positive experience living with family (52.21%) and were coded as “0.” In terms of single-mother household status, youth who reported that they lived in a single-parent family headed by their mother (62.71%) were coded as “1,” and all other youth (37.29%) were coded as “0.”
Antisocial peer bond
The measure of antisocial peer bonds was derived from the Eurogang Program Gang Affiliation Index (Klein, Kerner, Maxson, & Weitekamp, 2001). This two-item adaptation measures the degree to which a youth’s peers tacitly endorse and actively carry out illegal or antisocial activities. Peer tacit endorsement of antisocial activity was measured by asking, “Is doing illegal things accepted by or okay for your group?” Peer active engagement in illegal or antisocial activities was measured by asking, “Do people in your group actually do illegal things together?” This measure was found to have a Cronbach’s alpha value of .783 indicating acceptable internal consistency.
Antisocial beliefs
The measure of antisocial beliefs was derived from two binary items included in the Grasmick, Tittle, Bursik, and Arneklev’s (1993) Self-Control Scale. Within the SDM framework, antisocial beliefs refer to the beliefs of youth related to cheating, lying, and carrying out behaviors believed to be “wrong” to achieve a self-serving end (Catalano & Hawkins, 1996). An example of the antisocial beliefs measure is, “I sometimes find it exciting to do things that might get me into trouble.” This two-item measure was found to have a Cronbach’s alpha value of .713, indicating acceptable internal consistency.
Dependent Variables
Severe delinquency
Severe delinquency is measured by five yes/no items from the Self-Reported Delinquency Scale (SRD) originally developed for the National Youth Survey (Elliott, Huizinga, & Ageton, 1985). The SRD has been found to have acceptable psychometric properties (Huizinga & Elliott, 1986) and has been used in multiple large-scale studies of youth in the United States (Loeber, Stouthamer-Loeber, Kammen, & Farrington, 1991; Thornberry et al., 2003). That said, concerns have been raised as to the SRD’s reliability and validity across age, gender, and racial/ethnic differences (Piquero, Macintosh, & Hickman, 2002). The adapted measure used in this study comprises a variety of severe delinquent behaviors, including property destruction, weapon carrying, theft, and marijuana selling. Sample questions include, “In the last three months, have you carried a hidden weapon?” and “In the last three months, have you stolen money or things worth between $5.00 and $50.00?” In this study, the five-item measure was found to have acceptable psychometric properties with a Cronbach’s alpha value of .818 and all items loading cleanly on a single factor (.685-.832). In addition, four severe delinquent behaviors from the SRD were examined individually as dichotomous outcome variables: weapon carrying, theft, marijuana sales, and property destruction.
Minor delinquency
The six-item measure of minor delinquency was also derived from the SRD. This adapted measure comprises a variety of minor delinquent behaviors such as public disturbance, the purchase or sale of stolen items, asking for money from strangers, and the avoidance of payment for activities. Sample questions include, “In the last three months, have you been loud or rowdy in a public place where someone complained and you got into trouble?” and “In the last three months, have you avoided paying for things like a movie, taking bus rides, using a computer, or anything else?” In this study, the measure of minor delinquency was found to have acceptable internal consistency (Cronbach’s α = .817), and all items loaded cleanly on a single factor (.690-.793). Three individual minor delinquency items from the SRD were also examined individually as dichotomous outcome variables. These items include public disturbance, asking for money from strangers, and the sale or purchase of stolen goods. Notably, within the Salvadoran context, “asking for money from strangers” is more typically understood as a delinquent behavior than as a benign response to the nation’s harsh economic climate as, in practice, such requests frequently carry with them an implicit threat of violence or harm.
Statistical Analysis Plan
Structural Equation Modeling
Two sets of structural equation models were conducted that examined the relationship between religious coping, spirituality, and severe and minor delinquency. The strength of structural equation modeling is that it is a statistical technique that allows researchers to specify and test the goodness of fit between sample data and theoretical models designed to depict the causal relationships between observed and latent variables (Kline, 2011). While the data used in this study is cross-sectional and therefore cannot be used to infer causal relationships, mediation analysis is used here in exploratory correlational fashion.
Consistent with the theoretical framework of the SDM (Catalano & Hawkins, 1996), it is hypothesized that social developmental factors completely mediate the relationship between the exogenous variables and the problem behavioral outcome variables. However, conceptually plausible direct associations between the exogenous variables and dependent problem behavior variables were examined if justified empirically. Six exogenous variables were included in all of the hypothesized structural models: religious coping, spirituality, positive family life, single-mother household status, gender, and age. Given the documented associations between family makeup and youth religiosity/spirituality (Regnerus, 2003a), positive family life and single-mother household status were included as risk and protective covariates in the home/family domain. Gender and age were included as sociodemographic covariates as well.
In examining the goodness of fit of hypothesized and modified structural equation models, multiple indicators are typically examined. First, the chi-square statistic, although strongly influenced by sample size and other factors, should have a value close to the number of degrees of freedom and a probability value greater than .05 (Kline, 2011). However, given the instability of the chi-square statistic, other measures—such as the root mean square error of approximation (RMSEA), comparative fit index (CFI), and Tucker–Lewis index (TLI)—should also be considered in determining goodness of fit (Tran, 2009). It is recommended that the RMSEA value be below 0.05 and no greater than 0.08 (Tran, 2009), and that the CFI and TLI both be above 0.95 (Kline, 2011; West, Taylor, & Wu, 2012).
Logistic Regression
In addition to structural equation modeling, logistic regression was used to examine the relationships between religious coping, spirituality, and particular delinquent behaviors coded as binary variables. This method of statistical analysis allows for a more nuanced examination of the particular delinquent behaviors included in the data set as individual items and not as indexes.
Missing Data
Missing data were addressed via two statistical techniques. Missing data were dealt with in the structural equation modeling analyses by means of the “method(mlmv)” command in Stata 12.1. This command stands for maximum likelihood with missing values and allows for the estimation of fitted parameters in the presence of missing values. Further information regarding this technique has been provided in detail elsewhere (StataCorp, 2011). Missing data for the logistic regression analyses were imputed to maximize the analytical sample size. Imputation was conducted using multiple imputation with chained equations (MICE) via the “ice” command in Stata 12.1. Multiple imputation is an increasingly common procedure by which missing values are estimated and averaged among multiple data sets such that sound parameter estimates and standard errors can be generated (McKnight, McKnight, Sidani, & Figueredo, 2007). A comparison between the coefficients from the imputed and nonimputed data sets showed similar magnitude and directionality, indicating that the imputation procedure did not appear to substantively alter the study results.
Results
Sample Demographics
Table 1 presents a summary of the descriptive statistics for the total population of respondents in the sample. The sample consisted of 290 adolescents (11-17 years) and young adults (18-25 years) between the ages of 11 and 25 with an average age of 18.54 years (SD = 3.3). The vast majority of respondents were male (81.6%), and more than two out of every five respondents (42.5%) reported having at least one child. In terms of educational and vocational involvement, roughly two in every five respondents were either employed (37.6%) or school enrolled (41.4%), with an average education level of 6.67 years (SD = 4.2). More than two fifths of respondents reported having dropped out of school (43.0%), and sizable proportions of respondents reported difficulty with reading (32.0%) and difficulty in school in general (17.5%). In the realm of family life, nearly two thirds of all respondents reported living in a single-parent female-headed household (63.4%), and just over 50% reported that they were living with their families of origin without significant intrafamily problems (54.6%).
Sociodemographic Characteristics and Risk/Protective Factors of Study Participants (N = 290).
Involvement in Problem Behavior
Figure 1 illustrates the comparative prevalence of delinquent behavior involvement among adolescent and young adult respondents. With the exception of marijuana and “hard drug sales,” the prevalence of delinquent behavior involvement for adolescent and young adult respondents was between approximately 20% and 40%. As for inter-age group differences, across the board, young adult respondents reported higher levels of delinquent behavior compared with their adolescent counterparts. Although most differences were relatively small, sizable inter-age group prevalence differences were identified in terms of weapon carrying (39.88% vs. 22.02%, p < .001), theft (32.32% vs. 21.70%, p < .001), and marijuana sales (26.99% vs. 17.76%, p < .001).
Percentage of Severe and Minor Delinquency Involvement Among Adolescent and Young Adult Respondents.
Structural Equation Modeling
Tests and Modification of Proposed Models
Table 2 presents the goodness of fit statistics for the proposed and modified severe and minor delinquency models. Initial analyses of the proposed models for severe and minor delinquency indicated moderate model fit. However, several modifications were suggested to improve the model fit for both the severe delinquency and minor delinquency models.
Goodness-of-Fit Results From Proposed and Modified Structural Equation Models for Spirituality, Religious Coping, and Severe/Minor Delinquency.
Note. RMSEA = root mean square error of approximation; CFI = comparative fit index; TLI = Tucker–Lewis index.
Coefficients in bold are statistically significant at p < .05 or lower.
p < .10. **p < .05. ***p < .01. ****p < .001.
First, modification indices suggested that the relationship between age and minor delinquency was not completely mediated by antisocial bonding and antisocial beliefs. This is noteworthy as the theoretical framework of the SDM hypothesizes that social developmental factors completely mediate the relationship between all exogenous factors and the dependent variable. Nevertheless, the direct relationship between age and minor delinquency is both intuitively plausible and theoretically coherent. As such, a path was added from age to minor delinquency.
Second, modification indices in both models also suggested that a path be added between antisocial bond and delinquency. Although the SDM hypothesizes that antisocial beliefs completely mediate the relationship between antisocial bond and all outcome variables, it is nevertheless intuitively plausible that antisocial bonding could directly influence delinquent behavior. Consequently, in both models, a path was added from antisocial bond to delinquency.
Finally, modifications were suggested in terms of the correlation of error terms. With respect to severe delinquency, error terms were found to be correlated between (a) theft of items of low value and theft of items of higher value, (b) the carrying of a hidden weapon and theft, and (c) the carrying of a hidden weapon and marijuana dealing. Given that the two measures of theft have a common origin and that the use of a hidden weapon is plausibly linked with both theft and marijuana dealing, it seems to be theoretically justified to allow these error terms to correlate. With respect to minor delinquency, error terms were found to be correlated between asking for money from strangers and cheating others to obtain money. Given that both of these behaviors involve the extortion of funds from other people, it seems to be plausible that these error terms are correlated. In all, in light of the plausibility of the aforementioned error term correlations, the severe and minor delinquency error terms were allowed to correlate in their respective modified models.
Modified Structural Model for Severe Delinquency
As seen in Table 2, multiple goodness-of-fit indicators suggest that the fit between the modified severe delinquency model and the sample data is acceptable. The chi-square statistic for this model was not significant, and all goodness of fit statistics fell within an acceptable range: RMSEA (90% confidence interval [CI]) = 0.000 [0.00, 0.03]; CFI = 1.00; TLI = 1.00. No identification problems or further plausible modifications were identified.
Figure 2 presents the path diagram for the modified severe delinquency model and lends partial support to Hypothesis 1. Antisocial bond—but not antisocial beliefs—completely mediated the association between religious coping, spirituality, and severe delinquency. More precisely, religious coping was found to be marginally associated with antisocial bond (β = −0.12, p < .10), and spirituality was significantly associated with antisocial bond (β = −0.19, p < .01). Antisocial bond was directly associated with severe delinquency (β = 0.76, p < .001) and was associated with antisocial beliefs (β = 0.55, p < .001); however, the path from antisocial beliefs to severe delinquency was not statistically significant.
Modified Severe Delinquency Model: Structural Equation Model for Relationship Between Religious Coping, Spirituality, and Severe Delinquency.
Modified Structural Model for Minor Delinquency
Table 2 presents the goodness of fit indicators for the fit between the modified minor delinquency model and the sample data. Multiple goodness-of-fit indicators suggest that the model fit is acceptable. The chi-square statistic for this model was not significant, and all goodness of fit statistics fell within an acceptable range: RMSEA [90% CI] = 0.021 [0.00, 0.04]; CFI = 0.987; TLI = 0.983. No identification problems or further plausible modifications were identified.
Figure 3 presents the path diagram for the modified minor delinquency model and lends partial support to Hypothesis 2. While no significant direct or indirect effects were identified between religious coping and minor delinquency, the relationship between spirituality and minor delinquency was mediated entirely by social developmental factors. Spirituality was inversely associated with antisocial bonding (β = −0.20, p < .05), which, consistent with the SDM, was associated with antisocial beliefs (β = 0.56, p < .001), which, in turn, was marginally associated with minor delinquency (β = 0.17, p < .10). The indirect relationship between spirituality and minor delinquency was also mediated by the direct relationship between antisocial bonding and minor delinquency (β = 0.57, p < .001). Also, age was directly associated with minor delinquency (β = 0.13, p < .05), indicating that increases in age were associated with increases in minor delinquent behavior.
Modified Minor Delinquency Model: Structural Equation Model for Relationship Between Religious Coping, Spirituality, and Minor Delinquency.
Logistic Regression
Severe Delinquency
Lending partial support to Hypothesis 3, Table 3 presents the results of four logistic regression analyses examining the associations between spirituality, religious coping, and key manifestations of severe delinquent behavior. Spirituality was found to be significantly associated with the decreased likelihood of weapon carrying (odds ratio [OR] = 0.29, CI = [0.12, 0.73]) and property destruction (OR = 0.37, CI = [0.14, 0.94]). However, spirituality was not found to be significantly associated with the decreased likelihood of participation in theft or marijuana sales. Religious coping was found to be associated with decreases in theft (OR = 0.29, CI = [0.10, 0.87]) but was not found be significantly associated with the other severe delinquent behaviors examined.
Associations Between Religious Coping and Spirituality, Additional Risk and Protective Factors, and Severe Delinquent Behaviors Among Respondents 11-25 Years of Age.
Note. OR = odds ratio; CI = confidence interval.
ORs in bold are statistically significant at p < .05 or lower.
p < .10. **p <.05. ***p < .01. ****p < .001.
Minor Delinquency
Table 4 presents the results of three additional logistic regression analyses examining the associations between spirituality, religious coping, and three forms of minor delinquent behavior, and also provides partial support for Hypothesis 3. Spirituality was found to be significantly associated with the decreased likelihood of youth creating a public disturbance (OR = 0.37, CI = [0.17, 0.83]) but not with asking for money from strangers or with the sale or purchase of stolen goods. These findings indicate that the odds of highly spiritual youth creating a public disturbance are 63% lower than the odds of this behavior for their less spiritual counterparts. Religious coping was not significantly associated with any of the minor delinquent behaviors examined in this analysis.
Associations Between Religious Coping and Spirituality, Additional Risk and Protective Factors, and Minor Delinquent Behaviors Among Respondents 11-25 Years of Age.
Note. OR = odds ratio; CI = confidence interval.
ORs in bold are statistically significant at p < .05 or lower.
p < .10. **p < .05. ***p < .01. ****p < .001.
Discussion
Few studies have systematically examined the role of religiosity and spirituality in the lives of gang-involved youth (Flores, 2009; Marsal, 2009; Morris, 2012), and few, if any, studies have systematically examined the relationships between religious coping, spirituality, and delinquency among gang-involved youth in a developing society. The results of this analysis suggest that spirituality and, to a far lesser extent, religious coping function as protective factors in the lives of Salvadoran youth at risk for involvement in delinquent behavior. Why religious coping might operate differently than spirituality is open to speculation, but prosocial behavior seems embedded to a greater degree in persons seeking spirituality than in those who draw upon religious coping strategies. For instance, it does not seem out of the question for highly antisocial youth to look to God or religion in moments of terror or uncertainty, but it seems slightly more unlikely that highly antisocial youth will say that they are working hard to develop their spiritual life. Simply, the cultivation of spirituality seems to have strong normative and behavioral implications while the impact of religious coping on delinquency and crime is more ambiguous.
In examining the relationship between spirituality and delinquency as mediated by social developmental factors, spirituality was found to be indirectly related to both severe and minor delinquency. Notably, this relationship was entirely mediated by social developmental factors as no direct association was identified between spirituality and severe or minor delinquency. As noted above, a complete mediation of this relationship by social developmental factors is in keeping with the basic tenets of the SDM inasmuch as it is hypothesized that the relationship between exogenous factors and endogenous dependent variables is completely mediated by social developmental factors such as pro-/antisocial bonding and pro-/antisocial beliefs (Catalano & Hawkins, 1996). Worthy of note, however, is that the hypothesized path from antisocial bond to severe delinquency via antisocial beliefs was not found to be significant in terms of severe or minor delinquency. While the path from antisocial beliefs to minor delinquency was marginally significant (p < .10), strong direct paths from antisocial bonding to both manifestations of delinquency suggest that the relationship between spirituality and delinquency may be primarily accounted for via antisocial bonding rather than via antisocial beliefs. Finally, in contrast to spirituality, religious coping was not found to be directly or indirectly associated with minor delinquency and was only found to be marginally associated with antisocial bonds in the severe delinquency model.
While the structural equation modeling analyses provide important information in terms of key mediated relationships, results from the logistic regression analyses are also relevant as they provide a slightly more nuanced depiction of the direct, correlational relationships between religious coping, spirituality, and the spectrum of particular delinquent behaviors. For instance, although spirituality was indirectly associated with decreases in both severe and minor delinquency, results of the logistic regression analyses indicate that when controlling for antisocial bonding and antisocial beliefs, spirituality was not uniformly protective against all manifestations of delinquent behavior. In terms of severe delinquency, spirituality was found to be protective against weapon carrying and property destruction, but was not associated with other, more person-oriented forms of severe delinquent behavior such as theft and marijuana selling. This pattern may suggest that spirituality has a greater protective influence on delinquent behaviors that are either fundamentally passive in nature or are not directly related to other individuals. As for minor delinquency, spirituality was associated with the decreased likelihood of enacting public disturbances, but no such relationship was identified in terms of asking for money from strangers or the purchase or sale of stolen goods. Given that enacting a public disturbance, that is, being “rowdy” in a public setting, is arguably a more quotidian offense than the other minor delinquent behaviors examined, these findings may suggest that spirituality functions as a more protective factor in terms of relatively insignificant, rather than relatively consequential, minor delinquent behaviors. In contrast to spirituality, religious coping was not found to be protective against any minor delinquent behaviors but was strongly protective against theft. This suggests that, while no significant relationships were identified in terms of the composite measures of minor and severe delinquency, religious coping may be of relevance to certain delinquent behaviors. In sum, whereas spirituality tended to protect against both minor and severe delinquent behaviors, such as public disturbances, property destruction, and weapon carrying, religious coping protected only against one, arguably severe, manifestation of delinquency.
The identification of spirituality and, to a lesser extent, religious coping as protective factors against youth involvement in delinquent behavior is certainly in keeping with the bulk of the empirical research conducted in the United States (Baier & Wright, 2001). Indeed, previous studies utilizing structural equation modeling have identified related socio-cognitive factors as mediators of the relationship between religiosity, spirituality, and delinquency. For instance, Johnson, Jang, Larson, and De Li (2001), using a national sample of American adolescents from the National Youth Survey, found that adolescent religiosity was associated with lower levels of adolescent delinquent affiliation and higher levels of adolescent prosocial beliefs, both of which were, in turn, significantly associated with adolescent delinquent behavior. Similarly, Harris (2011), drawing from a national sample from the Monitoring the Future study, identified religiosity as associated with lower rates of antisocial peer bonding, which, in turn, was associated with delinquent behavior. Notably, however, no studies were identified that explicitly examined the relationship between spirituality and delinquency as mediated by socio-cognitive or social developmental factors.
While these findings in terms of the aforementioned mediated relationships are theoretically and empirically well-supported, the differential findings in terms of the relationships between religious coping, spirituality, and severe versus minor delinquent behaviors are less unambiguous. Religiosity and spirituality have been found to be associated with both minor (Benda & Corwyn, 1997, 2001; Smith & Faris, 2002) and more severe forms of delinquency (Smith & Faris, 2002). Few studies, however, have examined the differential impact of religiosity and spirituality on delinquent behavior, and regretfully, in such studies, the religious/spiritual constructs used are typically distinct from those used in this study as they tend to represent a comparison of church attendance versus intrinsic spirituality (Good & Willoughby, 2006). It is worth noting that some scholars have suggested that both religiosity and spirituality tend to be more strongly protective against minor delinquent or “nonvictim” criminal offenses for which society is relatively nonproscriptive (i.e., gambling, sneaking into a movie theater, purchasing stolen goods, etc.; Evans, Cullen, Dunaway, & Burton, 1995). Indeed, there is some empirical support for this general conceptualization in the broader literature as Baier and Wright (2001), in a meta-analysis of more than 60 studies on the relationship between religiosity and delinquency, found significantly stronger deterrent effects among nonviolent crimes as compared with more serious, violent criminal behavior. Nevertheless, this comparative research does not shed light on the differential relationships identified in this study. Perhaps it is sufficient to say that the trends identified merit further investigation as it appears that spirituality may be related to a variety of delinquent behaviors, whereas religious coping, although inexplicably unrelated to minor delinquency, appears to be of unique importance to some of the more severe delinquent behaviors enacted by Salvadoran youth.
Study Assets and Limitations
Among the strengths of this study were the unique character of the sample and the use of standardized, multi-item measures of religiosity and spirituality. The sample is unique inasmuch as respondents include a sizable proportion of active Salvadoran gang members living in their home communities. Few, if any, quantitative studies of Salvadoran high-risk youth and gang members have included substantial proportions of active youth gang members. Second, the measures of religious coping and spirituality used in this study were more psychometrically complex than the single-item measures that are frequently used in research into the relationship between religiosity, spirituality, and problem behaviors. The use of such validated, multi-item measures allows for a potentially greater precision of measurement of these key constructs than is typically the case in the research literature.
Despite these strengths, findings from this study should be interpreted in light of several important limitations. A number of these limitations were a function of the sampling methodology and sample size. First, given that the study data are cross-sectional, causal conclusions regarding the relationship between religious coping, spirituality, and delinquency cannot be drawn. Second, as a nonrandomized purposive sample, the results of this study are limited in terms of generalizability. Third, the relatively limited sample size (N = 290) and the distribution of key demographic variables created limitations in terms of comparisons across groups. Kline (2011) recommended a ratio of cases to free parameters of at least 10:1 and, ideally, 20:1; as such, given the relative complexity of the models examined in this study, comparisons across developmental age groups and gender were not statistically viable. This limitation is especially noteworthy given that the ages of study participants span two distinct developmental time periods.
Additional limitations were primarily a function of the survey instrument. First, while all measures were carefully translated into Spanish, the survey instrument itself is a compilation of measures and scales designed and validated primarily in the United States and in Western Europe. Instruments developed especially for the Central or Latin American national contexts are likely to capture particular sociocultural elements not accounted for by North American or European research instruments (Tran, 2009). Finally, given the fact that many high-risk and gang-involved Salvadoran youth are effectively barred from formal religious involvement, no measure of religious participation or religious service attendance was included in this study. This omission, while necessary due to the nature of the context and population of study, presents an important limitation inasmuch as public religious involvement has been found to be a vital component of the protective effect of religiosity. It also raises questions about the effect of noninstitutional manifestations of religiosity not measured in this study, such as parental involvement in organized religion, and nonorganizational forms of religious expression, such as prayer, meditation, and the reading of religious sources outside of attending formal services.
Conclusion
A robust body of empirical research suggests that religiosity and spirituality are important factors that can serve to protect youth from involvement in delinquent behavior. Notably, however, the relationships between these variables have rarely been examined in developing nations. Overall, the results indicate that spirituality and, to a lesser degree, religious coping have important implications in terms of delinquent behaviors among Salvadoran high-risk and gang-involved youth. When accounting for key social developmental mediating factors, spirituality was found to have important indirect effects on both minor and severe delinquent behaviors. Logistic regression analyses also revealed that spirituality is associated with the decreased likelihood of weapon carrying, property destruction, and public disturbances, but no associations were found for other minor or severe delinquent behaviors. In terms of religious coping, in contrast, no direct or mediated relationships were identified using structural equation modeling to account for mediating factors. However, in examining the associations between religious coping and particular manifestations of delinquent behavior, religious coping was found to be protective against one of the more severe delinquent behaviors: theft. These findings may have important implications for the development of prevention and intervention programs designed to enhance protective factors among high-risk and gang-involved Salvadoran youth. Such programs, for example, might pay special attention to the relationship between spirituality, and to a lesser extent religious coping, and social developmental factors such as peer bonding and pro-/antisocial beliefs. In all, spirituality and religious coping are both of relevance to the involvement of Salvadoran youth in delinquency, but it appears that spirituality is more consistently protective against delinquency in terms of its relationship to key social developmental mediating factors, and its direct associations to particular minor and severe delinquent behaviors.
Footnotes
Authors’ Note:
This research was supported in part by Grant T32 DA016184 from the National Institute on Drug Abuse at the National Institutes of Health.