Neighborhood Exposure to Violence and Delinquency in the Kingdom of Saudi Arabia: A Partial Test of Social Structure and Social Learning Theory

Abstract

The social structure and social learning (SSSL) model for crime and deviance has received an impressive amount of empirical support in the United States and other Western industrialized countries. Comparatively, less research, however, has examined whether the SSSL model offers a viable framework for explaining variation in delinquent behavior in other geographic contexts, particularly, countries that place a stronger emphasis on social control stemming from both formal and religious sources. The current study addresses this void in the literature by examining a sample of youth from Saudi Arabia, a Middle Eastern country that enforces Sharia (a set of laws based in Islamic tradition) and strict gender roles. The association between neighborhood exposure to violence and risk for violent and nonviolent delinquent behavior was examined using structural equation models. Subsequent models were aimed at more closely examining the mediating role of delinquent peer association between neighborhood exposure to violence and violent and nonviolent delinquent behavior. Results indicate that males exposed to neighborhood violence are more likely to engage in violent and nonviolent delinquent behavior, whereas females are more likely to engage in violent, but not nonviolent, delinquent behavior. In line with the SSSL model, delinquent peer association fully mediates the direct effect of neighborhood exposure to violence on delinquent behavior in both males and females. Findings from the current study suggest that the SSSL model may provide a useful framework for explaining individual differences in delinquent behavior in Saudi Arabia.

Keywords

cultural contexts criminology violent offenders

Social learning theory has been subjected to a wide variety of empirical tests across different samples, geographic locations, and cultural contexts (Akers & Jennings, 2015). Results from this line of research suggest that social learning theory offers one of the most useful theoretical frameworks for explaining variation in crime and deviance (Pratt et al., 2010). Some of the strongest evidence for social learning theory has been generated by cross-cultural tests where delinquent peer association has been found to be one of strongest predictors of adolescent substance use (Meneses & Akers, 2011; Miller, Jennings, Alvarez-Rivera, & Miller, 2008) and violent criminal behavior (Jennings, Park, Tomsich, Gover, & Akers, 2011), over and above the influence of other theoretical constructs including low self-control (Gottfredson & Hirschi, 1990) and strain (Agnew, 1992). These results have been reported in samples across several different countries that enforce different laws and societal norms, thus suggesting that the theoretical elements of social learning theory (i.e., differential association, definitions favorable to crime and deviance, differential reinforcement, and imitation) are generalizable across social context and important to consider when examining crime and delinquency.

More recently, Akers and Sellers (2004) have extended social learning theory and proposed a cross-level explanatory model of social learning that they call the social structure and social learning (SSSL) model for crime and deviance. The SSSL model holds that social structural factors (i.e., neighborhood violence, social disorganization, and socioeconomic disadvantage) are important distal social influences on criminal and delinquent behavior, but proximal social learning factors (i.e., delinquent peer association) are more directly involved in explaining why some commit criminal and delinquent acts and others do not. As such, the SSSL model posits that the main effects of social structural factors on criminal and delinquent behavior will be entirely or substantially mediated by social learning factors. Similar to results from prior research on social learning theory, a number of studies have found support for the SSSL model (Gibson, Poles, & Akers, 2010; Whaley, Smith, & Hayes-Smith, 2011; Wu, Eschbach, & Grady, 2008) with some starting to test the cross-cultural generalizability of the SSSL model for substance use in Western countries (i.e., countries that are equated with Western civilization that have the origins of their social norms, customs, and values in European culture characterized by democracy and individualism) outside of the United States (Kim, Akers, & Yun, 2013). Although this is an important step in evaluating whether the SSSL model provides a general theory of crime and deviance across populations, many of the Western societies examined are similar in cultural values and legal guidelines for crime and delinquency. Much is unknown about whether the SSSL model provides a useful framework for explaining crime and delinquency in a cultural context with considerably different value systems and norms. Middle Eastern countries appear to be a prime candidate for such inquiry, as many of these countries differentially emphasize agents of social control, especially religion, in the development of their legal traditions and enforcement of legal codes. For example, some Middle Eastern countries enforce Sharia, a set of legal statues derived directly from Islam. Sharia exists alongside more traditional legal systems based more directly on European legal traditions. The enforcement of both sets of laws produces a unique situation in which social control stems from a more diverse set of influences and may resonate differently with residents. Such countries also enforce different social norms, particularly those surrounding gender. Based on these observations, Middle Eastern countries appear to be prime candidates for examining the generalizability of the SSSL model.

The Social Structure and Social Learning (SSSL) Model

The SSSL model of crime and deviance holds that social structural influences, such as characteristics of a community, society, or culture in general, will affect an individual’s likelihood of exposure to different levels of social learning (Akers, 1998). Exposure to varying levels of social learning will, in turn, influence the probability of an individual beginning, continuing, and desisting from criminal and delinquent behavior. The unique contribution of the SSSL model is its cross-level, macro–micro integration between social structural risk factors and social learning risk factors. Although previous theoretical frameworks have focused on explaining individual differences in criminal and delinquent behavior from a macro- or micro-perspective, the SSSL model bridges both levels of analysis to argue that the social learning process is the most important mechanism involved in explaining the effects of social structure, culture, community, and neighborhood on variation in antisocial behavior. Several studies have tested the main arguments of the SSSL model and found support for the proposition that social learning partially mediates the direct effects of social structural factors on delinquent behavior and substance use (Bellair & McNulty, 2009; Bellair, Roscigno, & McNulty, 2003; Gibson et al., 2010; Haynie, Silver, & Teasdale, 2006; Hwang & Akers, 2003; Capece & Lanza-Kaduce, 2013; Lanza-Kaduce, Capece, & Allen, 2006; Verrill, 2008; Whaley et al., 2011; Wu et al., 2008). Findings from this growing body of research analyzing samples of U.S. youth have demonstrated that neighborhood disadvantage is one of the strongest correlates of delinquent behavior (Bellair & McNulty, 2009), and delinquent peer association is the strongest social learning variables associated with deviant behavior (Hwang & Akers, 2003; Pratt et al., 2010). However, very few studies have used samples drawn from other societies with different cultures, social values, and legal traditions to test key propositions from the SSSL model.

To date, only one study has tested the SSSL model with youth outside of the United States. Kim et al. (2013) examined a sample of 1,021 high school students from South Korea and found that social learning variables, such as differential peer association, substantially mediated the effect of social structural influences (population density, residential mobility, type of school, and religiosity) on adolescent substance use. Although this study is the first to provide important cross-cultural support for the SSSL model, it is limited in two key ways. First, the study analyzed a sample from a society that has become increasingly Westernized in recent years, especially with regard to social norms, which may not provide the most rigorous test of the generalizability of the SSSL model. Second, Kim et al. (2013) did not include a measure of neighborhood disadvantage or exposure to elements of neighborhood disadvantage, one of the strongest social structural correlates of adolescent delinquency, in their analysis. Without examining a sample of youth exposed to different cultural values, compared with ones prevalent in the United States, and examining whether social learning mediates the effect of one of the most robust structural predictors of adolescent delinquency, there can be very little we can conclude about the cross-cultural generalizability of the SSSL model.

Another hypothesis not yet tested from an SSSL perspective across cultural contexts is whether the role of delinquent peer association differs for males and females across societies. Prior research examining samples from the United States have reported that males are more likely to be exposed to neighborhood violence (Gibson, Morris, & Beaver, 2009), have more delinquent peers (Piquero, Gover, MacDonald, & Piquero, 2005), and engage in higher levels of violent and nonviolent delinquent behavior (Connolly, Schwartz, Jackson, & Beaver, 2018). However, these findings may not generalize to other non-Western countries. One way to rigorously evaluate this hypothesis is to analyze a sample of youth from a society where there are strict gender roles. One society that is a suitable candidate for this type of evaluation is Saudi Arabia. Today, Saudi Arabia still enforces strict gender roles for males and females. Males and females commonly work in same-sex settings, are allowed to be together only in a family setting, and have different legal rights. Although males are allowed to socialize in public settings with other males, females may only travel outside of their household environment with the permission and presence of a male guardian (typically male members of their family). As a result, this gender disparity may drastically shape the opportunities that adolescent females have to offend and the type of peers they are able to socialize with. To our knowledge, no research has examined this hypothesis in a cultural context that exists such as the one in Saudi Arabia.

To begin to address this gap in the existing literature, the current study provides a partial test of the SSSL model with data on youth from Saudi Arabia to evaluate whether delinquent peer association mediates the direct effect of exposure to neighborhood disadvantage on delinquent behavior. Testing the SSSL model with youth from a Middle Eastern cultural context such as Saudi Arabia, with very different norms and values, provides a rigorous test of the cross-cultural applicability of the SSSL model.

The Current Study

The current study is designed to provide partial cross-cultural tests of the SSSL model for delinquent behavior in Saudi Arabia. In line with arguments from the SSSL model, we expect that neighborhood exposure to violence will increase the odds of violent and nonviolent delinquent behavior in both males and females and that delinquent peer association will fully mediate the direct effect of neighborhood exposure to violence on delinquency. Drawing on previous research analyzing adolescent samples from Western societies showing that neighborhood disadvantage is one of the strongest social structural risk factors for delinquent behavior (Bellair & McNulty, 2009) and delinquent peer association is one of the strongest social learning risk factors (Kim et al., 2013; Miller et al., 2008; Pratt et al., 2010), the current study tested the following hypotheses to offer the first test of the SSSL model in Saudi Arabia:

Hypothesis 1: Neighborhood exposure to violence will be positively associated with risk for violent delinquent behavior in males and females.

Hypothesis 2: Neighborhood exposure to violence will be positively associated with risk for nonviolent delinquent behavior in males and females.

Hypothesis 3: Delinquent peer association (social learning factor) will mediate the direct effect of neighborhood exposure to violence on violent delinquent behavior in males and females.

Hypothesis 4: Delinquent peer association (social learning factor) will mediate the direct effect of neighborhood exposure to violence on nonviolent delinquent behavior in males and females.

Method

Data Collection and Sample

The data analyzed in the current study are drawn from a convenience sample of male and female students attending a government-sponsored high school in Jeddah, Saudi Arabia. To be included in the sample, participants had to either be a citizen of Saudi Arabia or currently reside within the city limits of Jeddah at the time of data collection. All participants also had to receive written permission from their parent(s) or guardian(s) to participate in the study. Parents (or guardians) and students were assured of confidentiality and anonymity. Students were informed that their participation was entirely voluntary and that refusal to participate in the study would not result in any form of punishment. Surveys were administered to 800 male and female students. To assure that questions retained their integrity and clarity for Arabic-speaking students, surveys were translated from English to Arabic and then back-translated by bilingual members of the research team with experience creating surveys for youth living in different national contexts. The survey design, content, data collection plans, and overall research project were reviewed and approved by the institutional review board at King Abdulaziz University. Of the 800 male and female students asked to participate, 494 (61.75%) agreed to participate and received permission.

Measures

Neighborhood exposure to violence

Neighborhood exposure to violence was measured by asking students, “While you have lived in your current neighborhood, has anyone ever used violence against you or any member of your household anywhere in your neighborhood?” Previous research has used this measure to capture neighborhood exposure to violence, neighborhood disorganization, and/or a lack of collective efficacy within a respondent’s respective neighborhood (Sampson, Raudenbush, & Earls, 1997). In line with previous research that has utilized this measure, participants were asked to reply to this question with a “yes” or “no” response. Responses were coded such that 0 = no and 1 = yes. Table 1 provides descriptive information about respondents’ exposure to neighborhood violence within the city limits of Jeddah, Saudi Arabia. As presented, approximately 8% of students reported having been violently victimized or having a household member who was victimized in their neighborhood. Male and female students did not significantly differ in their exposure to neighborhood violence, χ²(df) = 1.35(1), p = .24.

Table 1.

Descriptive Statistics.

Variable	Mean/Percent	SD	Minimum	Maximum	χ²	t
Neighborhood violence	8.70%	—	0	1	—	—
Male	7.95%	—	0	1	1.35	—
Female	11.62%	—	0	1	1.35	—
Violent delinquency	36.31%	—	0	1	—	—
Male	75.00%	—	0	1	−10.77**	—
Female	49.22%	—	0	1	−10.77**	—
Nonviolent delinquency	27.29%	—	0	1	—	—
Male	31.04%	—	0	1	−19.41**	—
Female	23.53%	—	0	1	−19.41**	—
Delinquent peers	0.67	1.25	0	9	—	—
Male	0.87	1.45	0	9	—	−3.99**
Female	0.40	0.83	0	4	—	−3.99**
Low self-control	57.23	15.03	24	87	—	—
Male	57.09	10.81	28	84	—	0.27
Female	57.50	10.62	24	87	—	0.27
Psychopathy	58.68	19.26	26	104	—	—
Male	67.11	10.77	26	99	—	−6.51**
Female	54.10	21.13	26	104	—	−6.51**
Saudi Arabian	96.41%	—	0	1	—	—
Male	96.65%	—	0	1	0.28	—
Female	95.63%	—	0	1	0.28	—
Age	17.36	3.42	10	25	—	—
Male	16.61	3.06	10	25	—	5.69**
Female	18.54	3.62	10	25	—	5.69**

p < .01.

Violent delinquency

Violent delinquency was measured by asking students how often they have committed five different violent delinquent acts in the past 12 months. Students were asked how often in the past 12 months they have (a) hurt someone bad enough to need a doctor, (b) gotten into a fight at school, (c) used force to get money from someone, (d) hit or seriously threaten someone, or (e) attacked someone with the intention to seriously hurt them. These items come from the reliable and well-validated Self-Report of Offending scale (Huizinga, Esbenson, & Weiher, 1991). Item responses were based on a Likert scale where 0 = never, 1 = once, 2 = twice, and 3 = more than twice. Responses to all items were summed to create a variety index of violent delinquency (Cronbach’s α = .81). Inspection of the distribution of scores revealed that the scale was positively skewed. For ease of interpretation with the binary measure of neighborhood violence, the scale was dichotomized such that 0 = did not commit a violent delinquent act in the past 12 months and 1 = committed a violent delinquent act in the past 12 months. Table 1 shows that 36% of Saudi Arabian students reported committing some form of aggressive or violent delinquent behavior in the past 12 months. Males were significantly more likely to report committing a violent delinquent act than were females, χ² = −10.77(1), p < .01.

Nonviolent delinquency

Nonviolent delinquency was measured by self-reports of nonviolent delinquent behavior. Similar to the items used to measure violent delinquency, nine items drawn from the Self-Report of Offending scale (Huizinga et al., 1991) were used to assess involvement in nonviolent delinquent behavior in the past 12 months. Specifically, students were asked how often in the past 12 months they have (a) run away from home, (b) skipped a day of school without permission, (c) damaged someone else’s property on purpose, (d) taken something without paying for it, (e) taken a vehicle without permission, (f) broke into a building or vehicle to steal or look inside, (g) knowingly held or sold stolen goods, (h) lied to parents about something important, or (i) done something that required a parent to come to school. Item responses were coded using a Likert scale where 0 = never, 1 = once, 2 = twice, and 3 = more than twice. Responses to all items were summed to create a variety index of nonviolent delinquency (Cronbach’s α = .78). Similar to the measure of violent delinquency, the distribution of respondent scores on the variety index for nonviolent delinquency was positively skewed. Thus, the measure was transformed into a dichotomous indicator to facilitate interpretation with the binary measure of neighborhood violence where 0 = did not commit a nonviolent delinquent act in the past 12 months and 1 = committed a nonviolent delinquent act in the past 12 months. Table 1 shows that 27% of respondents reported having committed some type of nonviolent delinquent act in the past 12 months with males reporting higher rates of nonviolent delinquent behavior in the past 12 months, χ² = −19.41(1), p < .01.

Delinquent peer association

Students were asked to respond to a series of questions about how many of their close friends smoked cigarettes, used illegal drugs, and/or belonged to a gang. Response categories were coded using a Likert scale where 0 = none of them, 1 = some of them, 2 = most of them, and 3 = all of them. Scores on all three items were summed to create a variety scale of delinquent peer association (Cronbach’s α = .61). Male students reported significantly higher levels of delinquent peer association when compared with female students (t = −3.99, p < .01).

Low self-control

A wealth of research has reported that low self-control is a robust correlate of delinquent offending (Pratt & Cullen, 2000) as well as personal and exposure to victimization (Pratt, Turanovic, Fox, & Wright, 2014). Low self-control was therefore included as a control variable. Low self-control was measured by self-report scores on the self-control scale created by Arneklev, Grasmick, Tittle, and Bursik (1993). Students were asked 24 questions designed to tap different dimensions of self-control such as impulsivity, simple tasks, risk-taking, physical activity, self-centeredness, and overall temperament. Students were asked to indicate how much they agreed or disagreed with each item (i.e., “I often do whatever brings me pleasure here and now, even at the cost of some distant goal,” “Sometimes I will take a risk just for the fun of it,” and “When I have a serious disagreement with someone, it is usually hard for me to talk calmly about it without getting upset”), with responses coded on a 4-point Likert scale: 1 = strongly disagree, 2 = somewhat disagree, 3 = somewhat agree, and 4 = strongly agree. Summed scores of the Grasmick et al. (1993) scale can be used to examine overall low self-control. As such, items were summed together to create a variety scale of low self-control (Cronbach’s α = .93). Responses were reverse coded, so that higher scores on the scale reflect higher levels of low self-control. There were no significant differences in levels of low self-control between male and female respondents (t = .27, p = .39).

Psychopathy

Prior research has found that individuals with higher levels of psychopathy are more likely to report being exposed to violence and higher levels of violent juvenile offending (Baskin-Sommers & Baskin, 2016; Beaver et al., 2016; Connolly et al., 2017). Preliminary analyses also showed that psychopathy was positively and significantly correlated with exposure to neighborhood violence (r = .28, p < .01), violent delinquency (r = .19, p < .001), and nonviolent delinquency (r = 15, p < .01). Psychopathy was measured by responses from the Levenson Self-Report Psychopathy (LSRP) scale (Levenson, Kiehl, & Fitzpatrick, 1995). The LSRP is a 26-item scale that asks respondents to report how closely a range of statements align with their personal beliefs or general personality (i.e., “Success is based on survival of the fittest; I am not concerned with losers,” “People who are stupid enough to get ripped off usually deserve it,” and “Looking out for myself is my top priority”). Responses were coded using a 4-point Likert scale where 1 = strongly disagree, 2 = somewhat disagree, 3 = somewhat agree, and 4 = strongly agree. The LSRP is a well-validated measure of psychopathy and has been shown to demonstrate strong construct validity and reliability (Brinkley, Schmitt, Smith, & Newman, 2001). Following in line with this research, all 26 items were summed to create a cumulative measure of psychopathy (Cronbach’s α = .94). Male respondents reported significantly higher levels of psychopathy compared with female respondents (t = −6.51, p < .01).

Saudi Arabian nationality

Participants were asked to select which of the following ethnicities best described their nationality: Saudi Arabian, Pakistani, Egyptian, Yemeni, Bangladeshi, Filipino, Jordanian or Palestinian, Indonesian, Sri Lankan, Sudanese, Syrian, or Turkish. Respondents who selected “Saudi Arabian” were coded “1,” while respondents who selected another nationality were coded “0.” Over 96% of the sample reported that Saudi Arabian ethnicity best described their nationality. No significant differences between males and females were detected (χ² = .28, p = .59).

Age

A single item was used to measure age that asked respondents to report how old they were in years. As shown in Table 1, the average age for the overall sample was 17 years, 16 years for males, and 18 years for females. The difference in age between male and female students was statistically significant indicating that, on average, female respondents in the sample were older than male respondents (t = 5.69, p < .01).

Analytic plan

To test the hypotheses, we analyzed the data in four steps. First, a series of tetrachoric and polychoric correlations were estimated to examine the bivariate association between neighborhood exposure to violence, violent delinquency, nonviolent delinquency, and delinquent peers for males and females. Tetrachoric correlations were estimated to assess the strength of association between binary variables such as neighborhood exposure to violence, violent delinquency, and nonviolent delinquency. Polychoric correlations were estimated to assess the strength of association between a binary variable and an ordered categorical variable, such as delinquent peers. Correlation estimates were examined to evaluate the extent to which key elements of social structure, such as neighborhood exposure to violence, and social learning, such as delinquent peers, were associated with violent and nonviolent delinquency in males and females.

Second, a series of sex-specific binary logistic regression models were estimated to examine the effect of neighborhood violence on violent and nonviolent delinquency among male and female respondents. Models were estimated once controlling for the confounding effects of low self-control, psychopathy, nationality, and age and then a second time including delinquent peers as an additional control. The first set of models were estimated to assess whether neighborhood violence predicted violent and nonviolent delinquency among males and females after controlling for a range of factors commonly associated with both neighborhood violence and delinquency. The second set of logistic regression models were estimated to examine whether the inclusion of delinquent peers in the logistic regression equation resulted in a nonsignificant association between neighborhood violence and delinquency in males and females. If after controlling for delinquent peers, neighborhood exposure to violence was no longer significantly associated with delinquency, then this could be interpreted as preliminary evidence suggesting that delinquent peers fully mediate the direct association between neighborhood violence and delinquent behavior. Regression coefficients generated from the estimated binary logistic models should be interpreted as odds ratios (ORs) where values over 1.00 indicate a positive association between an independent variable and a dependent variable and values under 1.00 indicate a negative association between an independent variable and a dependent variable. All binary logistic regression models were estimated using Stata 15.1 (StataCorp, 2017). To retain as many cases as possible, multiple imputation using chained equations was used in Stata to generate 20 multiple imputed data sets. This resulted in a final analytic sample of 494 respondents.

Third, latent path models were estimated to examine the underlying associations between neighborhood exposure to violence, delinquent peers, and violent and nonviolent delinquency in males and females. Path modeling was used over traditional regression techniques as it allows for the estimation of multiple pathways between independent variables, mediators, and dependent variables. This advantage over traditional regression strategies also reveals more information about mediating process where both direct and indirect effects are simultaneously estimated (James, Mulaik, & Brett, 2006). Latent path models were estimated only when there was preliminary evidence of mediation generated from the binary logistic regression analyses. All latent path models were estimated using the statistical software program Mplus version 8.1 (Muthén & Muthén, 1998-2011) using a weighted least squares estimator to account for missing data. Standard errors for the indirect effects for each model were bootstrapped and bias corrected with 500 replications with replacement (Preacher & Hayes, 2008). Model fit was assessed using the comparative fit index (CFI), the Tucker–Lewis index (TLI), and the root mean square error of approximation (RMSEA). As recommended, acceptable model fit was assessed using the following cut-off values: CFI ≥ .95, TLI ≥ .95, and RMSEA ≤ .06 (Hu & Bentler, 1999).

Results

The first step of the analysis included examining the bivariate associations between key theoretical constructs from Akers’ SSSL model. Table 2 presents the correlations for Saudi Arabian male students above the diagonal and correlations for Saudi Arabian female students below the diagonal. As can be seen, self-reported exposure to neighborhood violence was moderately correlated with violent delinquency (r = .34), nonviolent delinquency (r = .49), and delinquent peers (r = .50) in males. Delinquent peer association was strongly correlated with violent delinquency (r = .78) and nonviolent delinquency (r = .85) in males. A slightly different picture emerged for females. Although neighborhood exposure to violence was moderately correlated with violent delinquency (r = .28) and delinquent peers (r = .31), exposure to neighborhood violence was only weakly correlated with female nonviolent delinquency (r = .10). Similar to male students, however, delinquent peer association was strongly correlated with violent delinquency (r = .66) and nonviolent delinquency (r = .52).

Table 2.

Tetrachoric and Polychoric Bivariate Correlations for Exposure to Neighborhood Violence, Delinquency, and Delinquent Peers.

Variable	1	2	3	4
1. Neighborhood violence	—	.34 (.14)	.49 (.13)	.50 (.12)
2. Violent delinquency	.28 (.15)	—	.71 (.03)	.78 (.04)
3. Nonviolent delinquency	.10 (.07)	.79 (.06)	—	.85 (.04)
4. Delinquent peers	.31 (.10)	.66 (.07)	.52 (.10)	—

Notes. Male correlations presented above the mid-section and female correlations presented below the diagonal.

The second step in the analysis focused on examining the effect of neighborhood exposure to violence on violent and nonviolent delinquency while controlling for low self-control, psychopathy, nationality, and age. Table 3 presents the results from a series of binary logistic regression models examining violent and nonviolent delinquency in males. The results from the baseline model for violent delinquent behavior show that male students who reported being the victim or knowing of a victim of violence in their neighborhood were three times more likely (OR = 3.17, p < .01) to report having committed a violent delinquent act in the past 12 months compared with males who did not report being the victim or knowing of a victim of violence in their neighborhood. Neighborhood exposure to violence was also significantly associated with nonviolent delinquent behavior in the baseline model, where males who reported being the victim or knowing of a victim of neighborhood violence were over six times as likely (OR = 6.84, p < .01) to report having committed a nonviolent delinquent act. The second set of binary logistic regression models presented in Table 3 included a measure of delinquent peer association. The results from these models revealed that after controlling for the influence of delinquent peers, all previous associations between neighborhood exposure to violence and delinquent behavior were rendered nonsignificant. Instead, males who reported higher levels of delinquent peer affiliation were significantly more likely to commit a violent (OR = 4.55, p < .01) and nonviolent (OR = 3.13, p < .01) delinquent act over and above the influence of neighborhood exposure to violence. The inclusion of delinquent peers into the regression equations also increased the amount of explained variance. Taken together, results from this stage of the analysis provide preliminary evidence in line with Akers’ SSSL model that delinquent peer association fully mediates the association between exposure to a social structural factor such as neighborhood violence and delinquent behavior among Saudi Arabian males.

Table 3.

Binary Logistic Regression Models Predicting Violent and Nonviolent Delinquency in Males.

	Violent Delinquency				Nonviolent Delinquency
	Model 1		Model 2		Model 1		Model 2
	OR (95% CI)	SE	OR (95% CI)	SE	OR (95% CI)	SE	OR (95% CI)	SE
Neighborhood violence	3.17** [1.24, 6.38]	1.83	0.67 [0.09, 5.02]	0.69	6.84** [1.66, 28.14]	4.93	2.60 [0.42, 15.89]	0.85
Delinquent peers	—	—	4.55** [2.61, 7.92]	1.28	—	—	3.13** [1.90, 5.16]	0.79
Low self-control	0.97* [0.86, 0.99]	0.01	0.95* [0.91, 0.99]	0.02	0.97 [0.93, 1.00]	0.01	0.95* [0.90, 0.99]	0.02
Psychopathy	1.05** [1.02, 1.08]	0.01	1.06** [1.02, 1.10]	0.02	1.04* [1.01, 1.08]	0.01	1.05* [1.01, 1.10]	0.02
Saudi Arabian	0.14 [0.01, 1.25]	0.16	0.12 [0.01, 1.17]	0.14	0.10 [0.01, 2.48]	0.09	0.12 [0.001, 0.394]	0.07
Age	1.02 [0.91, 1.15]	0.06	0.82* [0.70, 0.97]	0.06	1.15* [1.00, 1.33]	0.08	0.98 [0.83, 1.15]	0.08
Pseudo R²	.21		.42		.19		.39

Notes. OR = odds ratio; CI = confidence interval.

p < .01. *p < .05.

Table 4 presents the results from a series of binary logistic regression models examining the association between neighborhood exposure to violence, violent delinquency, and nonviolent delinquency in females. Results from the baseline models showed that females who reported being a victim or knowing of a victim of violence in their neighborhood were more than three times more likely (OR = 3.65, p < .05) to report having committed a violent delinquent act in the past 12 months compared with females reporting not being a victim or knowing of a victim of violence in their neighborhood. In contrast to males, however, Saudi Arabian females exposed to neighborhood violence were not at a significantly higher risk of engaging in nonviolent delinquency (OR = 2.78, p = .17). The second set of models introduced the measure of delinquent peer association into the regression equations. As presented in Table 4, once the measure of delinquent peer association was introduced into the regression equation examining violent delinquency, the association between neighborhood exposure to violence and violent delinquent behavior was no longer statistically significant (OR = 2.57, p = .12). Saudi Arabian females who reported higher levels of delinquent peer association were significantly more likely to report having committed a violent delinquent act in the past 12 months (OR = 4.24, p < .01). Although the baseline regression model for female nonviolent delinquency revealed that neighborhood exposure to violence was not significantly associated with risk for nonviolent delinquent behavior, delinquent peers was still introduced into the regression equation. Results from this second model revealed that females who associated with delinquent peers were more likely to report committing a nonviolent delinquent act in the past 12 months (OR = 3.44, p < .01). The inclusion of this measure increased the amount of explained variance in both binary logistic regression equations.¹ Taken together, evidence from this stage of the analysis offers mixed support for Akers’ SSSL model.

Table 4.

Binary Logistic Regression Models Predicting Violent and Nonviolent Delinquency in Females.

	Violent Delinquency				Nonviolent Delinquency
	Model 1		Model 2		Model 1		Model 2
	OR (95% CI)	SE	OR (95% CI)	SE	OR (95% CI)	SE	OR (95% CI)	SE
Neighborhood violence	3.65* [1.97, 7.45]	1.62	2.57 [0.77, 13.64]	1.91	2.78 [0.78, 9.89]	1.80	2.01 [0.92, 10.29]	2.55
Delinquent peers	—	—	4.20** [1.63, 9.82]	2.02	—	—	3.44** [1.73, 6.82]	1.20
Low self-control	1.13** [1.05, 1.22]	0.04	1.12** [1.03, 1.21]	0.04	1.06* [1.02, 1.13]	0.03	1.04 [0.96, 1.12]	0.03
Psychopathy	0.94 [0.88, 1.00]	0.03	0.96 [0.90, 1.03]	0.03	1.00 [0.93, 1.07]	0.03	1.01 [0.94, 1.09]	0.03
Saudi Arabian	0.15 [0.03, 3.17]	0.10	0.13 [0.02, 2.63]	0.09	0.17 [0.08, 2.38]	0.12	0.10 [0.01, 1.95]	0.08
Age	0.81** [0.70, 0.94]	0.06	0.83* [0.71, 0.96]	0.06	0.85* [0.73, 0.98]	0.06	0.86 [0.72, 1.02]	0.07
Pseudo R²	.23		.33		.14		.29

Notes. OR = odds ratio; CI = confidence interval.

p < .01. *p < .05.

The last step in the analysis involved estimating three latent path models to further examine the association between exposure to neighborhood violence, delinquent peer association, and delinquent behavior in Saudi Arabian males and females. Based on the results from the binary logistic regression models, only latent path models examining the indirect effects of neighborhood exposure to violence on violent and nonviolent delinquency in males and nonviolent delinquency in females via delinquent peer association were examined. Figure 1 presents the standardized parameter estimates from the first path model examining male violent delinquent behavior. In addition to examining whether and to what extent delinquent peers mediate the pathway from neighborhood violence to violent delinquent behavior, the model also included direct pathways from low self-control, psychopathy, nationality, and age to each key theoretical element. Model fit was acceptable for the path model that included all significant and nonsignificant pathways (CFI = .93, TLI = .91, and RMSEA = .05). As shown, after modeling an indirect path from neighborhood exposure to violence to violent delinquent behavior via delinquent peers, neighborhood violence was no longer significantly associated with violent delinquency (b = .00, p = .21). The bootstrap estimate of the standardized indirect effect was significant, b = .33, SE = .02, 95% confidence interval (CI) = [.13, .52], p < .01, indicating that the effect of neighborhood exposure to violence on violent delinquency in males occurred entirely through delinquent peer association.

Figure 1.

Structural equation mediation model for violent delinquency in males.

Figure 2 shows the standardized parameter esitmates from the path model for male nonviolent delinquent behavior that included direct pathways from each covariate to observable and latent measures of neighborhood exposure to violence, delinquent peers, and nonviolent delinquent behavior. Model fit was good (CFI = .95, TLI = .92, and RMSEA = .05). Results from the estimated model revealed a similar pattern of results compared with the path model for violent delinquency. Specifically, the standardized direct effect of neighborhood violence on nonviolent delinquency was not statistically signficant (b = .00, p = .26), whereas the indirect effect of neighborhood violence on nonviolent delinquency was significant (b = .38, SE = .03, 95% CI = [.17, .58], p < .01). Thus, results from both latent path models examining the indirect effect of neighborhood exposure to violence on male violent and nonviolent delinquency suggest that the influence of neighborhood violence on both forms of delinqent behavior operate via self-reported levels of delinquent peer association.

Figure 2.

Structural equation mediation model for nonviolent delinquency in males.

The final latent path model focused on evaluating the effect of neighborhood exposure to violence on violent delinquent behavior through delinquent peer affiliation in Saudi Arabian females. Model fit statistics for the saturated model that included all significant and nonsignificant direct pathways from each covariate to each observable variable and latent factor indicated that the model was acceptable (CFI = .92, TLI = .89, and RMSEA = .06). Parameter estimates from the path model are presented in Figure 3. As can be seen, the direct pathway from neighborhood exposure to violence to violent delinquency was no longer statistically significant once the indirect pathway from neighborhood violence to delinquent peers to violent delinquency was specified. The bootstrap estimate of the standardized indirect effect of neighborhood violence on violent delinquency via delinquent peers was significant and accounted for all the total effect of neighborhood violence on violent delinquent behavior (b = .17, SE = .01, 95% CI = [.08, .25], p < .01).

Figure 3.

Structural equation mediation model for violent delinquency in females.

Discussion

This study investigated whether the SSSL model provides a viable theoretical framework for explaining individual differences in violent and nonviolent delinquent behavior in Saudi Arabia. We examined (a) the extent to which neighborhood exposure to violence was directly associated with violent and nonviolent delinquent behavior in males and females and (b) whether delinquent peer association mediated the direct effect of neighborhood exposure to violence on violent and nonviolent delinquent behavior in males and females. Overall, three key findings emerged from the anlaysis in support of the SSSL model that deserve further discussion.

First, neighborhood exposure to violence was found to increase the odds of committing a violent delinquent behavior in both males and females and provided support for Hypothesis 1. These findings are consisent with neighborhood research analyzing youth from the United States (Goodnight et al., 2012; Leventhal & Brooks-Gunn, 2000; Warner & Fowler, 2003) and other Western countries (Mazerolle, Wickes, & McBroom, 2010), showing that exposure to violence within neighborhood settings increase the risk for youth violence. One reason for this association could be because youth who are consistently exposed to violence may begin to imitate such aggressive behaviors and use them for goal-directed purposes. Ethnographic research from the United States has found support for this process, whereby youth in neighborhoods characterized by high levels of violent crime begin to adopt specific “street code” values that reward aggressive and violent behavior with respect from peers and local community members (Anderson, 2000; Stewart & Simons, 2010). Future research would benefit from further investigating the prevelance of street-code values in low socioeconomic neighborhoods in non-Western societies such as Saudi Arabia. Moreover, it would be interesting and important to assess whether these street-code values condition the association between social structural and social learning factors.

Second, neighborhood exposure to violence significantly increased the odds for engaging in nonviolent delinquent behavior in males, but not in females. Based on these results, partial support is found for Hypothesis 2 as neighborhood violence was significantly associated with nonviolent delinquent behavior, but only for male students. One factor that may account for this finding is the difference in socially appropriate roles for males and females in Saudi Arabian society. Although males are typically given authority to socialize outside of the household environment unsupervised, but, in accordance with Sharia, females are closely supervised by parents and guardians, and given clear restrictions as to who they can socialize with, when, and where. This difference may explain why males in neighborhoods characterized by high levels of violent crime social disorganization may be more likely to report status delinquent offenses such as running away from home, skipping school, and lying to parents about something important as they are presented with more opportunities to engage in such behaviors. Differences in autonomy and opportunity may also explain why males growing up in neighborhoods with violence are more likely to report committing property crimes such as taking a vehicle without permission, breaking into a vehicle, and taking property without paying for it. One reason why this may be the case is because when the data were collected for this study, females were prevented from legally operating any type of motor vehicle and may not have perceived any real benefit from stealing a vehicle compared with males due to the differential application of formal social controls and the additional risk of detection by authorities for operating a vehicle, stolen or otherwise. It is also possible that males may have many other unmeasured risk factors not accounted for in the anlysis that increase their exposure to violence, delinquent peer associations, such as prior delinquent behavior (see, for example, Gibson et al., 2009). Future research should collect additional data on Saudi youth to explore the links between sex differences in parental supervision, unstructered socializing, and prior levels of delinquent behavior.

Third, delinquent peer association entirely mediated the direct effect of neighborhood exposure to violence on risk for violent delinquent behavior in males and females. Evidence from this segment of the analysis provides support for the SSSL model with regard to violent delinquent behavior in Saudi Arabia and offers support for Hypothesis 3. Of primary importance to the central goal of the current study, the results generated from the path models offer some of the first empirical evidence supportive of the SSSL model for delinquent behavior in a Middle Eastern cultural context. These findings, despite the difference in social roles and cultural expectations for males and females in Saudi Arabia and Western societies, highlight the generalizability of social learning influences on violent delinquent behavior during adolescence across context.

Although the present study does provide support for the SSSL model in a non-Western cultural context, there are several limitations that should be mentioned and addressed with future research. First, the analyses are based on cross-sectional data which make it difficult to establish temporal order between neighborhood violence, delinquent peer association, and delinquent behavior. As temporal order is required to establish causality and is essential to any theoretical model, future research should attempt to replicate and extend the reported findings with longitudinal data to strengthen causal inference between the primary theoretical social structural and social learning elements. Second, the analysis was based on a sample of youth from only one major metropolitan city in Saudi Arabia. Although this most likely produced a stronger measure of exposure to neighborhood violence as participants from highly populated inner city neighborhoods may be more knowledgable of violent victimizations occuring in their neighorhood, the reported findings cannot be generalized to youth in other major cities or rural areas of Saudi Arabia. Future research should aim to collect data on Saudi youth in different regions of the country and provide additional tests of the SSSL model in a non-Western cultural context. Third, although the employed analysis is the first to present supportive evidence of the SSSL model for delinquent behavior in a Middle Eastern context, the present study only found supportive evidence for the major proposition of the theory, that is, that social learning influences mediate the direct effect of structural influences on delinquent behavior. Future studies should extend these findings by including measures of other structural dimensions of the SSSL model such as levels of social disorganization including collective efficacy and sociodemographic correlates including socioeconomic status, parental employment, and government assistance (Akers, 1998).

Conclusion

Social learning theory and the SSSL model for crime and deviance have become mainstays of Western criminological theory. As theoretical tests have accumulated, theorists and researchers have called for more cross-cultural tests of these explanatory models in non-Western contexts (Akers & Jennings, 2015; Krohn, 1999; Sampson, 1999). The current study analyzed a sample of youth from a major metropolitan city in Saudi Arabia to test the central proposition of the SSSL model that delinquent peer association fully mediates the direct effect of a social structural influence, such as neighborhood exposure to violence, on violent and nonviolent delinquent behavior in males and females. Our results extend and replicate prior research in three ways. First, exposure to delinquent peers was found to fully mediate the direct effect of neighborhood exposure to violence on violent delinquency in males and females. This finding provides cross-cultural support for the SSSL model for violent delinquent behavior and replicates prior research that has tested the SSSL model in Western countries. Second, exposure to delinquent peers was found to fully mediate the direct effect of neighborhood violence on nonviolent delinquent behavior in males. Third, neighborhood violence was not significantly associated with nonviolent delinquent behavior in females, but delinquent peer association significantly increased the odds of nonviolent delinquent behavior. Future research should seek to explore the social and cultural factors that differentially impact the likelihood of nonviolent offending in males and females growing up in Middle Eastern contexts.

Footnotes

Declaration of Conflicting Interests

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

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This project was funded by the Dean of Scientific Research (DSR), King Abdulaziz University, Jeddah, under Grant 1-125-1433/HiCi.

ORCID iDs

Eric J. Connolly

Kevin M. Beaver

Notes

Author biographies

Eric J. Connolly, PhD, is an assistant professor in the Department of Criminal Justice and Criminology at Sam Houston State University. His research interests include biosocial criminology, criminological theory, and developmental/life-course criminology.

Mohammed Said Al-Ghamdi is a professor in the Center for Social and Humanities Research at King Abdulaziz University.

Ahmed Nezar Kobeisy, PhD, is a researcher in the Center for Social and Humanities Research at King Abdulaziz University.

Fathiyah H. Alqurashi is a researcher in the Center for Social and Humanities Research at King Abdulaziz University.

Joseph A. Schwartz, PhD, is an assistant professor in the School of Criminology and Criminal Justice at the University of Nebraska Omaha. His research interests include behavior genetics, biosocial criminology, and the interplay between biological and environmental influences on behavior.

Kevin M. Beaver, PhD, is a professor in the College of Criminology and Criminal Justice at Florida State University and a visiting distinguished professor in the Center for Social and Humanities Research at King Abdulaziz University.

References

Akers

R. L.

(1998). Social learning and and social structure: A general theory of crime and deviance. Boston: Northeastern University Press.

Akers

R. L.

Jennings

W. G.

(2015). Social learning theory. The Handbook of Criminological Theory, 4, 230-240.

Akers

R. L.

Sellers

C. S.

(2004). Criminolgical theories: Introduction, evaluation, and application (4th ed.). Los Angeles, CA: Roxbury.

Agnew

(1992). Foundation for a general strain theory of crime and delinquency. Criminology, 30, 47-88.

Allison

(2012, September 10). When can you safely ignore multicollinearity. Statistical Horizons. Retrieved from http://statisticalhorizons.com/multicollinearity

Anderson

(2000). Code of the street: Decency, violence, and the moral life of the inner city. New York, NY: W.W. Norton.

Arneklev

B. J.

Grasmick

H. G.

Tittle

C. R.

Bursik

R. J.

(1993). Low self-control and imprudent behavior. Journal of Quantitative Criminology, 9, 225-247.

Baskin-Sommers

A. R.

Baskin

(2016). Psychopathic traits mediate the relationship between exposure to violence and violent juvenile offending. Journal of Psychopathology and Behavioral Assessment, 38, 341-349.

Beaver

K. M.

Al-Ghamdi

M. S.

Kobeisy

A. N.

Alqurashi

F. H.

Connolly

E. J.

Schwartz

J. A.

(2016). The association between psychopathic personality traits and victimization and exposure to violence in a sample of Saudi Arabian youth. Psychiatric Quarterly, 87, 217-228.

10.

Bellair

P. E.

McNulty

T. L.

(2009). Gang membership, drug selling, and violence in neighborhood context. Justice Quarterly, 26, 644-669.

11.

Bellair

P. E.

Roscigno

V. J.

McNulty

T. L.

(2003). Linking local labor market opportunity to violent adolescent delinquency. Journal of Research in Crime and Delinquency, 40, 6-33.

12.

Brinkley

C. A.

Schmitt

W. A.

Smith

S. S.

Newman

J. P.

(2001). Construct validation of a self-report psychopathy scale: Does Levenson’s self-report psychopathy scale measure the same constructs as Hare’s psychopathy checklist-revised? Personality and Individual Differences, 31, 1021-1038.

13.

Capece

Lanza-Kaduce

(2013). Binge drinking among college students: A partial test of Akers’ social structure-social learning theory. American Journal of Criminal Justice, 38, 503-519.

14.

Connolly

E. J.

Al-Ghamdi

M. S.

Kobeisy

A. N.

Alqurashi

Schwartz

J. A.

Beaver

K. M.

(2017). Identifying latent classes of antisocial behavior among youth from Saudi Arabia: An assessment of the co-occurrence between aggression, psychopathy, low self-control, and delinquent behavior. Youth Violence and Juvenile Justice, 15, 219-239.

15.

Connolly

E. J.

Schwartz

J. A.

Jackson

D. B.

Beaver

K. M.

(2018). How far does the apple fall from the tree? Maternal delinquency and sex-specific patterns of offspring delinquent behavior. Journal of Criminal Justice, 54, 50-61.

16.

Gibson

C. L.

Morris

S. Z.

Beaver

K. M.

(2009). Secondary exposure to violence during childhood and adolesence: Does neighborhood context matter? Justice Quarterly, 26, 30-57.

17.

Gibson

C. L.

Poles

T. B.

Akers

R. L.

(2010). A partial test of social structure social learning: Neighborhood disadvantage, differential association with delinquent peers, and delinquency. In Criminological theory: A life-course approach (pp. 133-148). New York, NY.

18.

Goodnight

J. A.

Lahey

B. B.

Van Hulle

C. A.

Rodgers

J. L.

Rathouz

P. J.

Waldman

I. D.

D’onofrio

B. M

. (2012). A quasi-experimental analysis of the influence of neighborhood disadvantage on child and adolescent conduct problems. Journal of Abnormal Psychology, 121, 95-108.

19.

Gottfredson

M. R.

Hirschi

(1990). A general theory of crime. Palo Alto: Stanford University Press.

20.

Haynie

D. L.

Silver

Teasdale

(2006). Nieghborhood characateristics, peer networks, and adolescent violence. Journal of Quantitative Criminology, 22, 147-169.

21.

L. T.

Bentler

P. M.

(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal, 6, 1-55.

22.

Huizinga

Esebensen

F. A.

Weiher

A. W.

(1991). Are there multiple paths to delinquency. The Journal of Criminal Law & Criminology, 82, 83-118.

23.

Hwang

Akers

R. L.

(2003). Substance use by Korean adolescents: A crosscultural test of social learning, social bonding, and self-control theories. Social Learning Theory and The Explanation of Crime, 11, 39-64.

24.

James

L. R.

Mulaik

S. A.

Brett

J. M.

(2006). A tale of two methods. Organizational Research Methods, 9, 233-244.

25.

Jennings

W. G.

Park

Tomsich

E. A.

Gover

A. R.

Akers

R. L.

(2011). Assessing the overlap in dating violence perpetration and victimization among South Korean college students: The influence of social learning and self-control. American Journal of Criminal Justice, 36, 188-206.

26.

Kim

Akers

R. L.

Yun

(2013). A cross-cultural test of social structure and social learning: Alcohol use among South Korean adolescents. Deviant Behavior, 34, 895-915.

27.

Krohn

M. D.

(1999). Social learning theory: The continuing development of a perspective. Theoretical Criminology, 3, 462-476.

28.

Lanza-Kaduce

Capece

Alden

(2006). Liquor is quicker: Gender and social learning among college students. Criminal Justice Policy Review, 17, 127-143.

29.

Levenson

M. R.

Kiehl

K. A.

Fitzpatrick

C. M.

(1995). Assessing psychopathic attributes in a noninstitutionalized population. Journal of Personality and Social Psychology, 68, 151-158.

30.

Leventhal

Brooks-Gunn

(2000). The neighborhoods they live in: The effects of neighborhood residence on child and adolescent outcomes. Psychological Bulletin, 126, 309-337.

31.

Mazerolle

Wickes

McBroom

(2010). Community variations in violence: The role of social ties and collective efficacy in comparative context. Journal of Research in Crime and Delinquency, 47, 3-30.

32.

Meneses

R. A.

Akers

R. L.

(2011). A comparison of four general theories of crime and deviance: Marijuana use among American and Bolivian university students. International Criminal Justice Review, 21, 333-352.

33.

Miller

H. V.

Jennings

W. G.

Alvarez-Rivera

L. L.

Miller

J. M.

(2008). Explaining substance use among Puerto Rican adolescents: A partial test of social learning theory. Journal of Drug Issues, 38, 261-283.

34.

Muthén

L. K.

Muthén

B. O.

(1998-2012). Mplus user’s guide. Los Angeles, CA: Author.

35.

Piquero

N. L.

Gover

A. R.

MacDonald

J. M.

Piquero

A. R.

(2005). The influence of delinquen peers on delinquency: Does gender matter? Youth & Society, 36, 251-275.

36.

Pratt

T. C.

Cullen

F. T.

(2000). The empirical status of Gottfredson and Hirschi’s general theory of crime: A meta-analysis. Criminology, 38, 931-964.

37.

Pratt

T. C.

Cullen

F. T.

Sellers

C. S.

Thomas Winfree

Jr Madensen

T. D.

Daigle

L. E.

… Gau

J. M.

(2010). The empirical status of social learning theory: A meta-analysis. Justice Quarterly, 27, 765-802.

38.

Pratt

T. C.

Cullen

F. T.

Sellers

C. S.

Thomas Winfree

Madensen

T. D.

Jr. Daigle

L. E.

Gau

J. M.

(2010). The empirical status of social learning theory: A meta-analysis. Justice Quarterly, 27, 765-802.

39.

Pratt

T. C.

Turanovic

J. J.

Fox

K. A.

Wright

K. A.

(2014). Self-control and victimization: A meta-analysis. Criminology, 52, 87-116.

40.

Preacher

K. J.

Hayes

A. F.

(2008). Assessing mediation in communication research. The Sage Sourcebook of Advanced Data Analysis Methods for Communication Research, 13-54.

41.

Sampson

R. J.

(1999). Techniques of research neutralization. Theoretical Criminology, 3, 438-451.

42.

Sampson

R. J.

Raudenbush

S. W.

Earls

(1997). Neighborhoods and violent crime: A multilevel study of collective efficacy. Science, 277, 918-924.

43.

StataCorp. (2017). Stata statistical software: Release 15. College Station, TX: Author.

44.

Stewart

E. A.

Simons

R. L.

(2010). Race, code of the street, and violent delinquency: A multilevel investigation of neighborhood street culture and individual norms of violence. Criminology, 48, 569-605.

45.

Verrill

S. W.

(2008). Social structure-social Learning and Delinquency: Mediation or Moderation? New York, NY: LFB Scholarly Pub.

46.

Warner

B. D.

Fowler

S. K.

(2003). Strain and violence: Testing a general strain theory model of community violence. Journal of Criminal Justice, 31, 511-521.

47.

Whaley

R. B.

Smith

J. M.

Hayes-Smith

(2011). Teenage drug and alcohol use: Comparing individual and contextual effects. Deviant Behavior, 32, 818-845.

48.

Z. H.

Eschbach

Grady

J. J.

(2008). Contextual influences on polydrug use among young, low-income women: Effects of neighborhood and personal networks. The American Journal on Addictions, 17, 135-144.