Journal of Quantitative Criminology

, Volume 32, Issue 4, pp 531–559 | Cite as

Uncovering the Pathways Between Gang Membership and Violent Victimization

Original Paper

Abstract

Objectives

Gang members are more likely to be victimized violently than non-gang youth, but the extent to which this relationship is confounded, direct, or mediated remains unclear. This study responds to recent calls by scholars for more methodologically sound research in this area with the goal of uncovering the pathways between gang membership and violent victimization.

Methods

Using a school-based longitudinal sample of adolescents, the current study uses Preacher and Hayes multiple mediator structural equation modeling and counterfactual methodology to test whether and which theoretical pathways—self-control, cultural orientations, routine activity, and lifestyle theory—mediate the contemporaneous and prospective effects of gang membership on violent victimization.

Results

The results indicate that 27 % of the contemporaneous effects of gang membership on victimization is attributable to selection, with the remaining 73 % endogenous to gang membership, supporting Thornberry et al.’s (J Res Crime Delinquency 30:55–87, 1993) enhancement model. Entry into a gang increases risk taking, temper, self-centeredness, negative peer commitment, neutralization of violence, aggressive conflict resolution, unstructured socializing, and delinquency, and decreases empathy and positive peer commitment. The contemporaneous gang membership-victimization link was fully mediated, due almost entirely to delinquency. Prospective models reveal a gang membership-victimization link that is fully confounded by selection, although attrition and desistance from gangs may be responsible for this finding.

Conclusions

The existing risky attitudes and behaviors of youth who select into gangs matters a great deal for understanding the gang membership-victimization link, but these very risks are exacerbated upon entry into a gang. Our mediation findings suggest that interventions targeting highly delinquent gang members should pay dual dividends of reducing delinquency and victimization.

Keywords

Gang Violent victimization Mediation Selection Preacher and Hayes 

Introduction

Youth violence leads to over 700,000 injuries annually and is the second leading cause of death among young people between ages 15 and 24 (National Center for Injury Prevention and Control 2012). Understanding the sources of youth violence is a high priority for researchers, policymakers, public health officials, and the criminal justice system. It is well established that one major source of youth violence in the US and abroad is street gangs (Block and Block 1993; Decker 1996; Klein and Maxson 2006; Papachristos 2009; Melde and Esbensen 2013; Pyrooz and Decker 2013). For example, gang-related homicides account for 13 % of all homicides in the United States and over 20 % in cities with populations exceeding 200,000 (Howell et al. 2011; National Gang Center 2014; Pyrooz 2012). Studies consistently find that gang members are not only disproportionately perpetrators of violence, but also victims of violence. While researchers have expended considerable effort uncovering the mechanisms and patterns of the link between gang membership and offending, the study of the relationship between gang membership and victimization is comparatively limited.

It is indisputable that gang members are disproportionately victims of violence compared to those who are not in gangs, but the mechanisms that bring about this relationship remain unclear. Indeed, it may appear that gang membership might increase risk for victimization, but it is equally plausible that gangs attract individuals who are prone to victimization. This is why DeLisi et al. (2009) adopted aThornberry et al.’s (1993) theoretical models, originally applied to offending, to test whether gangs select high-risk individuals, facilitate high-risk environments, or a blend of the two, that is, enhancement. Even after accounting for selection bias using counterfactual methods, DeLisi et al. (2009) found evidence of a positive link between gang membership and violent victimization. Yet serious questions were raised about the theoretical and empirical basis of this relationship in an exchange between Gibson et al. (2009, 2012) and Ozer and Engel (2011), where conflicting findings were observed—selection for the former, facilitation for the latter—using the same source of data (the first evaluation of the Gang Resistance Education and Training program). Accordingly, two recent reviews of the literature emphasized key directions forward, with Fox (2013: 1016) holding that it is necessary to understand the mechanisms underlying “how and why gang members are victimized at higher rates than their non-gang counterparts” and Gibson et al. (2012) highlighting a series of methodological requirements for future research, notably, the need for longitudinal data, proper time order, and appropriate extraneous controls. We respond to both calls by examining the direct and indirect pathways between gang membership and violent victimization.

This study addresses the following research questions: (1) Does joining a gang facilitate violent victimization, even after controlling for selection effects? (2) If so, what are the mechanisms associated with gang membership that promote this relationship? (3) Is there any residual relationship between gang membership and violent victimization after accounting for selection confounders and theoretical mediators? We answer these questions using three waves of data from the Outcome Evaluation of the Teens, Crime, and the Community/Community Works (TCC/CW) Training Program 2004–2005 (Esbensen 2005) and Preacher and Hayes (2008) multiple mediator structural equation modeling. Together, our methodology and our focus on mediating mechanisms go a long way to address the shortcomings found in the extant literature. Our overall goal is to use appropriate and rigorous methodology to understand the social reality underpinning the relationship between gang membership and violence.

Gang Membership and Violent Victimization in Context

A growing list of empirical studies have established a link between gang membership and violent victimization (Barnes et al. 2012; Childs et al. 2009; Decker and Pyrooz 2010; DeLisi et al. 2009; Fox et al. 2013; Gibson et al. 2009; Gover et al. 2009; Katz et al. 2011; Ozer and Engel 2011; Pyrooz et al. 2014; Rufino et al. 2012; Sweeten et al. 2013; Spano et al. 2008; Taylor et al. 2007, 2008; Zavala and Spohn 2013). Drawn from wide and diverse research designs that span a range of demographic groups and sampling frames, this body of evidence is large enough that few would dispute the fact that gang members are victims of violence at a rate that is much higher than their peers who are not in gangs.

But there are questions about whether the relationship between gang membership and violent victimization is a product of the group processes of street gangs or is an artifact of selection into street gangs. That is, the very individuals who self-select into more risky environments generally (Berg 2012; Pratt et al. 2014) may be the same individuals who self-select into gangs particularly (Curry et al. 2014; Klein and Maxson 2006). Indeed, selection on propensity toward violence and aggression is built into the DNA of gangs. In Densley’s (2012, 2013) studies of street gang recruitment, he found that gangs look for clusters of hard-to-fake signals that will solve the trust dilemmas that gangs face in sorting good from bad recruits. These signals not only included criminal credentials and violence capital, but also a willingness sacrifice one’s self for the benefit of the gang (e.g., take a beating, holding weapons, not snitching). Likewise, it is not uncommon that the victims of bullies in school yards and neighborhoods seek the association of gangs who can offer at least the appearance of protection (Decker and Van Winkle 1996; Jankowski 1991; Melde et al. 2009; White and Mason 2012). Therefore, the observed linkages that this body of research has produced over the last decade could be a product of the self-selection processes as opposed to gang-related group processes.

It is only fitting, then, that in their discussion of their longitudinal findings on gang membership and victimization, Peterson et al. (2004) aligned their results with Thornberry et al.’s (1993) selection, facilitation, and enhancement models which were introduced originally to explain the relationship between gang membership and offending. DeLisi et al. (2009) formally extended this theoretical model to gang membership and victimization in framing their national study. The selection model posits that high-risk youth select into gangs, which means the relationship between gang membership and violent victimization is spurious. Under this model, the reason why gang members experience more violent victimization than non-gang peers is due to pre-existing risk characteristics. The social facilitation model is aligned closely with social learning theory (Thornberry et al. 1993), holding that gangs change the attitudes and behavior patterns of their members and facilitate high-risk environments as a consequence of gang-related group processes (Decker et al. 2008), which increases the chances of violent victimization. The enhancement model is the combination of the selection and the social facilitation models, holding that the risks youth bring with them into gangs and the environments that gangs construct work together to explain increases in violent victimization among gang members

Both facilitation and enhancement models hold that there is a “black box” of factors endogenous to gangs that increases the risk of victimization. These mechanisms operate at multiple levels of explanation (Short 1998), including micro-level gang processes and individual-level personal changes. Pyrooz et al. (2014) provided a theoretical elaboration of endogenous gang processes by arguing that after adjusting for individual-level factors any residual relationship between gang membership and violent victimization was a product of gang-related group processes, such as collective identity, normative orientations toward criminal involvement, status acquisition and maintenance, and extra-individual liabilities (see also Decker 1996; Hughes 2013; McGloin and Decker 2010; Short and Strodtbeck 1965). Papachristos (2009) and Papachristos et al. (2013) have modeled the spatial and network processes of violence between rival gangs in Boston and Chicago, finding that symbolic threats, geographic proximity, and ongoing network conflicts between rival gangs triggered violence and reciprocity in violence, all of which operates above and beyond individual-level motivations.

Individual-level longitudinal studies have been effective in uncovering the changes associated with joining and leaving gangs. Using multi-site school-based samples of youth, Melde and Esbensen (2011, 2012, 2014) demonstrated over a series of studies that gang membership was associated with changes in attitudes, peer networks, and unstructured activities, which facilitated offending (see also Matsuda et al. 2013). Based on an adjudicated high-risk sample of youth and young adult gang members, Sweeten et al. (2013) found that leaving gangs decreased levels of offending, but the mechanisms accounting for these changes included shifts in attitudes, peer associations, routine activities, and victimization. These findings are not limited to the United States. Recent research in the Netherlands (Weerman et al. 2015) and a representative study that examined youth in England and Wales (Medina-Ariza et al. 2014) have collectively shown that joining a gang resulted in greater commitment to delinquent peers, perceptions of peer delinquency, time spent with peers, hanging out in the street, and weaker social controls.

These studies have gone to great lengths to better understand the changes associated with status transitions in and out of gangs, yet have focused their theoretical and empirical efforts exclusively on offending and not victimization. Indeed, the body of literature examining the relationship between gang membership and victimization pales in comparison to the analytic sophistication and the overall volume of studies examining gang membership and offending. It is known that there is a relationship between gang membership and victimization, but the goal of this study is to better understand why this relationship exists.

Pathways Between Gang Membership and Victimization

Three theoretical pathways linking gang membership to changes in individual risk for victimization are posited. First, gang membership may exacerbate levels of low self-control, which is an established correlate of victimization (Pratt et al. 2014; Schreck 1999). Individuals with low self-control maintain qualities—such as being self-centered, impulsive, risk-taking, short-sighted, and lacking empathy—that induce vulnerability to victimization (Gottfredson and Hirschi 1990; Schreck 1999). As a result of taking spontaneous risks, acting without thinking through the consequences, provoking rather than calming conflicts, and ignoring the feelings of others, individuals with low self-control situate themselves among persons or places that elevate their risk of victimization. It has been established that self-control is a risk factor of gang membership (Kissner and Pyrooz 2009; Lynskey et al. 2000), but it is also malleable over time (Burt et al. 2006, 2014; Muraven and Baumeister 2000; Na and Paternoster 2012). In a high-risk gang environment where self-control is constantly exercised and tested—by fellow gang members, rival gangs, and law enforcement—it is possible that self-control could eventually deplete and make individuals more susceptible to behaviors that s/he would not normally engage in.

Second, consistent with theories that emphasize the cultural orientations, gang members may be socialized to beliefs, norms, practices, and rituals that place value on violence. Differential normative orientations toward status attainment, defense of honor, and the resolution of disputes could lead to victimization (Anderson 1999; Black 1983; Jacobs and Wright 2006; Kubrin and Weitzer 2003; Wolfgang 1957). Indeed, this was the basis for Wolfgang’s (1957) “subculture of violence” these, where many homicides were victim-precipitated by seemingly trivial disagreements. Consistent with social learning theory (Akers 2009) more generally, the group context of the gang can produce “beliefs, behaviors, and perceptions that can put [gang members] at risk of victimization” (Fox et al. 2011: 42). Bullet wounds are flaunted, surviving assaults and shootings become gang folklore, and hiding from rather than fighting gang rivals is scorned among fellow gang members (Decker and Van Winkle 1996; Densley 2013; Fleisher 1998; Garot 2009; Goldman et al. 2014; Klein and Maxson 2006; Ostrander 2008). As an extra-legal group, gangs do not have access to conventional legal channels, which in turn leads to the resolution of disputes through alternative means such as violence, consistent with Anderson’s (1999) code of the street and Tyler’s (1990) model of compliance. It is therefore expected that adherence to the code of the street strengthens upon joining a gang, as demonstrated by Matsuda et al. (2013), and that gang members are less likely to view the law and authority as legitimate (Papachristos et al. 2009). Entry into a gang should mean a greater inclination to accept violence as normative, to be less fearful of victimization (Melde et al. 2009), and to be surrounded by others with shared outlook toward violence.

Third, the link between gang membership and violent victimization may be due to the mechanisms found in routine activity (Cohen and Felson 1979; Felson and Boba 2010) and lifestyle (Hindelang et al. 1978) theories. Both theories place an emphasis on the convergence of offenders, targets, and guardians in time and space. While routine activity theory offers a description of the victimization event itself, lifestyle theory views victimization probabilistically (Pratt and Turanovic 2015). Certain lifestyles are riskier than others by virtue of the persons, places, or times that enhance risk for victimization (Hindelang et al. 1978: 245; see also Weerman et al. 2013). Gang members engage in the very characteristics operating at the heart of these theories (Taylor 2008; Taylor et al. 2008), including spending time hanging out with peers in unstructured social environments (Decker and Van Winkle 1996; Fleisher 1998; Klein 1995; Whyte 1943), engaging in a considerable amount of delinquency (Pyrooz et al. 2015), drinking alcohol, along with using and selling drugs (Howell and Decker 1999; Hunt and Laidler 2001; Swahn et al. 2010), and carrying guns (Bjerregaard and Lizotte 1995; Fagan 1999; Tigri et al. 2015). These activities do not guarantee an incident of victimization, but they may increase the chances for a rival gang to shoot at a gang member, to be cheated by drug addicts, to have one’s belongings stolen while away from home, to be targeted by a stick-up crew, or to be assaulted or robbed while under the influence.

In summary, there is a good reason to believe that the link between gang membership and victimization should be at least partially explained by any of the three pathways we have proposed above.

Advancing Knowledge on the Gang Membership-Victimization Link

Prior investigations into the relationship between gang membership and victimization have concentrated on identifying direct links as opposed to indirect pathways. Some of the mechanisms discussed above have been included in earlier studies,1 but the cross-sectional research design employed in prior works makes it impossible to examine whether these mechanisms carryover from non-gang periods or change upon entering into a gang, which Gibson et al. (2012) identified this as a high priority for research on this topic. In a recent review of this literature, Fox (2013: 1035) argued that the “analysis of the gang-victimization link should include examinations of longitudinal data and controls for delinquency/crime.” Longitudinal studies, alternatively, have focused on the contemporaneous and prospective direct effects of gang membership on victimization (Barnes et al. 2012; DeLisi et al. 2009; Gibson et al. 2009; Ousey et al. 2011; Ozer and Engel 2011; Spano et al. 2008; Sweeten et al. 2013), leaving the specific mechanisms that bring about this relationship unresolved, which Fox (2013: 1016) identified as a key direction in her commentary on the literature.

Part of what makes this line of inquiry—the direct and indirect pathways—criminologically compelling are the mixed findings in the literature, which is perhaps best exemplified by contradictory results produced by Gibson et al. (2009) and Ozer and Engel (2011) using the same source of data (the first evaluation of the Gang Resistance Education and Training program) and similar counterfactual methodology. Gibson et al. (2009: 639) concluded

“In sum, our results showed support for the selection perspective because the relationship between joining a gang and becoming violently victimized was explained by pre-existing differences that perhaps lead gang members to join a gang in the first place.”

Ozer and Engel (2011: 119), alternatively, concluded

“Our reanalysis of these data call into question Gibson et al.’s (2009) reported support for the selection perspective, and demonstrate new findings that are consistent with the growing body of research examining the relationship between gang membership and victimization.”

And with other studies (e.g. Katz et al. 2011) finding this relationship is sensitive to model specification, it is unclear whether the link between gang membership and victimization is best described as confounded, direct, or mediated.
This study aims to contribute to the gang and victimization literatures by assessing the ability of the three previously discussed theoretical pathways to account for the link between gang membership and violent victimization. We examine the model presented in Fig. 1. Consistent with prior literature, we hypothesize that gang membership has a bivariate relationship with victimization. We propose that gang membership changes attitudes, behaviors, and controls, which in turn, should at least partially explain higher rates of victimization among gang youth (i.e., the indirect pathway). Any remaining relationship would be attributed to factors that go unmeasured in the current study, which previous research (Pyrooz et al. 2014) has proposed could be group processes operating above the individual level of explanation. We use appropriate methodology that responds to calls put forth in recent commentary on this topic (Fox 2013; Gibson et al. 2012) about the need to use longitudinal data and uncover the mechanisms that explain this relationship.
Fig. 1

Proposed indirect effects of gang membership on violent victimization

Methods

Data

The data used in the current study are part of the Outcome Evaluation of the Teens, Crime, and the Community/Community Works (TCC/CW) Training Program in fifteen cities across four states (Esbensen 2005).2 The evaluation included three waves of surveys: a pre-test in fall 2004, a post-test in Spring 2005, and a 1-year follow-up survey in Fall 2005. The total number of students in the selected classrooms was 2353. Due to the age of the students (all were younger than 16 years of age), parental consent forms were required prior to their participation in the evaluation. The active parental consent rate was 72 %. Twelve percent of parents did not allow their children to participate in the study (n = 290) and 16 % of students failed to return consent forms (n = 377), resulting in a total of 1686 students taking part in the three-wave surveys. Completion rates of 96, 89, and 72 % were obtained for each wave, respectively. The data analyzed in this study consist of 1185 students in contemporaneous model and 924 students in prospective model without missing information on the key independent and dependent variables.3

Research team staff used group-administered self-report methods to collect survey data by reading questionnaires to students to give better understanding of the questions (Esbensen 2005). In each wave of the survey, students were asked about their gang involvement, violent victimization experiences, attitudes, behaviors, and social controls, which directly address the research questions of the current study. The data have been used widely to study issues related to criminology in general (cf., Melde 2009; Slocum et al. 2010) and gangs in particular (Melde et al. 2009; Melde and Esbensen 2011).

Dependent Variable

The dependent variable is violent victimization. Five forms of violent victimization are studied, including having been (1) attacked or threatened on the way to or from school, (2) attacked or threatened at school, (3) hit by someone, (4) robbed, and (5) attacked with a weapon (see Elliott et al. 1985; Huizinga et al. 1991). Respondents were asked whether they were violently victimized in the last 6 months (0 = never, 1 = once, 2 = two to five times, 3 = six to ten times, and 4 = more than 10 times). The items were summed to construct a scaled frequency score of violent victimization frequency.4

Treatment Variable

The key treatment variable in this study is gang membership. Gang membership was measured as one item, asking that “Do you consider your group of friends to be a gang?” (0 = no and 1 = yes; see Melde and Esbensen 2011). Studies have found that allowing youth to self-nominate as a gang member is a valid way to distinguish between gang members from non-gang members on attitudes and delinquent behaviors (Matsuda et al. 2013), as well as gang-related outcomes such as gang embeddedness (Decker et al. 2014). The treatment group in the current study consists of respondents who self-identify as a gang member for the first time at wave 2 while the control group consists of respondents who have never been in a gang as of wave 2. Consistent with experimental research design, each status did not need to be preserved over the study observation period (i.e., some non-gang youth later join gangs, some gang youth leave gangs).

Theoretical Mechanisms

Twelve mediators are measured across the theoretical pathways identified in Fig. 1. All of the mediators were constructed by computing the average value across the items represented by the mediators.5 The items, along with the associated descriptive statistics, can be found in Appendix 1.6

Low Self-Control

Risk seeking, temper, empathy, and self-centeredness are tested in the model under self-control theory. The survey questions for risk seeking, temper, and self-centeredness are derived from Grasmick et al. (1993). Respondents were asked to grade a series of statements from strongly disagree (coded as 1) to strongly agree (coded as 5). Risk seeking had four items, for example (for more details, see Appendix 1), “I like to test myself every now and then by doing something a little risky” (α = .80, average inter-item correlation = .49). Temper had four items, for example, “I lose my temper pretty easily” (α = .76, average inter-item correlation = .45). Self-centeredness also had four items, for example, “I try to look out for myself first, even if it means making things difficult for other people” (α = .73, average inter-item correlation = .40). All of the constructs exhibited acceptable levels of internal consistency. Empathy was composed of four items where respondents could answer questions “yes” or “no”, for example, “I would feel sorry for a lonely stranger in a group”. The items comprising each of the constructs were averaged to create a measure where higher scores indicated higher levels of each construct.

Cultural Orientations

Five constructs are measured under cultural orientations, including neutralization of violence, positive peer commitment, negative peer commitment, aggressive conflict resolution, and fear of victimization (Akers and Jensen 2006). The measure of neutralization of violence contains three items (Sykes and Matza 1957), for instance, “it’s okay to beat up someone if they hit you first” (α = .80, average inter-item correlation = .58). Respondents were asked to rate each statement from strongly disagree (coded as 1) to strongly agree (coded as 5). The measures of positive and negative peer commitment are consistent with Esbensen (2003). Positive peer commitment contains two items, for instance, “if your friends told you not to do something because it was wrong, how likely is it that you would listen to them,” (α = .74, average inter-item correlation = .59). The measure of negative peer commitment contains three items, for instance, “if your group of friends was getting you into trouble at home, how likely is it that you would still hang out with them” (α = .82, average inter-item correlation = .60). Respondents were asked to answer each question from not at all likely (coded as 1) to very likely (coded as 5). Aggressive conflict resolution had two items, for instance, “Told the person off or yelled at them to solve conflict” (α = .66, average inter-item correlation = .50). Respondents were asked to answer questions from never (coded as 1) to often (coded as 3). Fear of victimization contained eight items, for example, “how fearful are you of being robbed or mugged” (Melde et al. 2009). Each item ranged from 1 to 5, where 1 represented “not at all afraid” and 5 was “very afraid” (α = .90, average inter-item correlation = .60). All of the constructs have sound psychometric properties and were coded so that higher scores represent greater neutralization of violence, positive and negative peer commitment, aggressive conflict resolution, and fear of victimization, respectively.

Routine Activity/Lifestyle

Unstructured socializing, self-reported delinquency, and parental monitoring measure routine activity and lifestyle theories. Three dummy coded items (0 = “no”, 1 = “yes”) are combined to identify unstructured socializing, for example, “do you ever spend time hanging out with your current friends not doing anything in particular where no adults are present” (Esbensen et al. 2001). The delinquency index includes fourteen types of delinquency within the past 6 months drawn from the Denver Youth Survey (Huizinga et al. 1991) and the National Youth Survey (Elliott et al. 1985). It ranges from minor deviance, such as skipping class without excuses, to serious offending, such as attack with weapons. Each item is coded from 0 to 4 (0 = never, 1 = once, 2 = two to five times, 3 = six to ten times, 4 = more than ten times). Parental monitoring contains four items, for instance, “when I go someplace, I leave a note for my parents or call them to tell them where I am” (α = .73, average inter-item correlation = .40) (see Esbensen 2003; Taylor et al. 2008; Melde and Esbensen 2011), where higher values indicate greater parental monitoring.

Demographic Control Variables

Several demographic characteristics are used in the current study, including age, gender (0 = female, 1 = male), and race and ethnicity. Age at wave 1 was used and it was coded as a truncated continuous variable because the respondents were from 6th to 9th grades and mostly between the ages of 11 and 14. Race and ethnicity is categorized into white, African American, Hispanic, and other races. We used Hispanic (42 %) as the reference group because it was the modal race/ethnicity, followed by white (31 %), other races (15 %), and African American (11 %).

Analytic Strategy

The current study aims to detect the direct and indirect effect of gang membership on violent victimization using a technique established by Preacher and Hayes (2008). This technique is one kind of structural equation model (SEM) and has several advantages, including: (1) testing two or more mediators in a single model simultaneously instead of testing several simple mediator models, which reduces omitted variable bias;7 (2) determining the combined effect of the mediators and whether they eliminate the bivariate effect of the treatment variable; (3) identifying which individual factors mediate the bivariate effect; and (4) comparing the magnitude of indirect effect among all the mediators. Together, the features of the Preacher and Hayes method lend themselves well to the aims of the study: uncovering the direct and indirect pathways between gang membership and victimization.
Fig. 2

Illustration for a multiple mediation model

As Fig. 2a shows, c is the total effect of gang membership on victimization. Figure 2b also shows the direct and indirect paths, where c′ represents the direct effects and an and bn (\({\text{n}} = 1, 2, \ldots , {\text{j}}\)) represent the indirect regression coefficients of each mediator. For each mediator, its indirect effect is calculated as the product of an and bn:
$${\text{c}}_{\text{n}} = {\text{a}}_{\text{n}} {\text{b}}_{\text{n}}$$
(1)
The total indirect effect is the sum of each mediator’s indirect effect:
$${\text{c}}_{\text{ab}} = \mathop \sum \limits_{{{\text{i}} = 1}}^{\text{j}} {\text{c}}_{\text{i}} = \mathop \sum \limits_{{{\text{i}} = 1}}^{\text{j}} {\text{a}}_{\text{i}} {\text{b}}_{\text{i}}$$
(2)
Thus, the total effect of independent variable is the sum of the direct effect and the total indirect effect:
$${\text{c}} = {\text{c}}^{{\prime }} + {\text{c}}_{\text{ab}} = {\text{c}}^{{\prime }} + \mathop \sum \limits_{{{\text{i}} = 1}}^{\text{j}} {\text{a}}_{\text{i}} {\text{b}}_{\text{i}}$$
(3)

The indirect effect of each mediator is computed as the product of regression coefficients and it leads to a problem of violation of normality, which is an assumption in SEM (Preacher and Hayes 2008). Therefore, bootstrapping is adopted in this technique. Bootstrapping is a repeated resampling method with unknown-shape samples to estimate the indirect effects (Singh and Xie 2008) and it is helpful to control Type I error in nonnormality samples (Berkovits et al. 2000). In testing the indirect effects, a subsample is chosen from the original sample to estimate an and bn (\({\text{n}} = 1, 2, \ldots , {\text{j}}\)) and then compute an bn (\({\text{n}} = 1, 2, \ldots , {\text{j}}\)). After k times of resampling in the current study k = 1000, the final indirect effects can be estimated (Preacher and Hayes 2008).

Youth do not join gangs at random. Individuals with certain risk factors are more likely to join gangs (Curry et al. 2014; Klein and Maxson 2006) and these very factors may have an influence on their later violent victimization. Thus, it is important to eliminate selection bias because existing risk factors may confound the results, which is why prior studies examining the effect of gang membership on victimization have employed counterfactual methodology (DeLisi et al. 2009; Gibson et al. 2009; Ozer and Engel 2011). What distinguishes the present study from prior research is that our goal is not to simply identify the direct effect of gang membership on victimization net of selection, but also to uncover the factors that mediate this relationship. We therefore integrate counterfactual methodology into the Preacher and Hayes (2008) method.

A propensity score is defined as the probability that an individual receives the treatment given measured confounders (Rosenbaum and Rubin 1983). It is estimated with a logistic regression model by taking the probability of receiving treatment and using it as a control variable in the mediation analysis to account for selection bias (Coffman 2011), and this technique was extended to the gang literature by Melde and Esbensen (2011). Coffman (2011) compared the bias among different mediation models without confounders, with half confounders, and all confounders in the propensity model. The results suggested that only after controlling for propensity score with all confounders, the estimates of mediation effects were unbiased (Coffman 2011).

Variables used in the propensity score model include demographic factors, wave 1 values of proposed mediators, and victimization at wave 1, because it is a reason for joining gangs (Peterson et al. 2004) and it predicts future victimization (Farrell and Pease 2001; Turanovic and Pratt 2014), along with other community, school, and individual levels predictors of gang joining behavior (see Appendix 3). The propensity score, as wells as demographic characteristics and wave 1 values of proposed mediators, is then used as a control variable in the mediation analysis to test the effect of gang membership at wave 2 on violent victimization in waves 2 and 3 (Melde and Esbensen 2011). This gives us two opportunities to correctly specify the model, a “doubly robust” strategy (see Funk et al. 2010; Ridgeway and McCaffrey 2007). This also ensures that we capture the effects of gang membership net of prior victimization experiences.

We proceed as follows: (1) we compare bivariate results between those who join gangs and those who avoid gangs at wave 2 with and without selection adjustments, and then (2) examine the direct and indirect effects of gang membership on victimization contemporaneously and prospectively. Gang membership is not a stable status (only a quarter of youth reported gang membership in both wave 2 and 3 in the current study), which is why it is appropriate to use contemporaneous and prospective model between gang members, consistent with prior research (e.g., DeLisi et al. 2009; see the discussion in Decker et al. 2013: 378–379).

Findings

Descriptive and Bivariate Statistics

Table 1 displays the descriptive statistics of the study variables, partitioning them by gang joiners and non-gang youth. The average level of violent victimization at wave 1 is .79, although it was statistically greater for future gang joiners (mean = 1.29) than for non-gang youth (mean = .76). This is consistent with what we would expect from a selection model of gang membership and victimization. These differences remained in place at both waves 2 and 3, although there is a noticeable gain in victimization among gang joiners that is not observed among non-gang youth. There are also statistically significant differences across all of the mediators at waves 1 and wave 2. Future gang members show significantly higher levels of risk taking, temper, self-centeredness, negative peer commitment, neutralization of violence, aggressive conflict resolution, unstructured socialization, and delinquency, and significantly lower levels on empathy, positive peer commitment, and parent monitoring at wave 1. These are the very risk factors that researchers use to model selection into gang membership, which selection proponents (e.g. Gibson et al. 2009) hold account for any relationship between gang membership and victimization.
Table 1

Descriptive statistics of the study variables by gang status and selection unadjusted and adjusted standardized differences (N = 1185)

Variables

Total sample

Gang Status at Wave 2

Standardized differences

Not in a gang

Joined a gang

Unadjusted

Adjusted

Wave 1 constructs and demographics

 Age

12.22 (.93)

12.21 (.92)

12.47 (.96)

.29*

.00

 Male

44.5 %

43.8 %

55.7 %

.24

.00

 White

34.3 %

35.3 %

17.1 %

−.38*

−.00

 African–American

11.1 %

10.4 %

22.9 %

.40*

.00

 Hispanic

41.4 %

41.1 %

45.7 %

.09

−.00

 Other

13.2 %

13.1 %

14.2 %

.04

.00

 Risk taking

2.65 (.88)

2.62 (.87)

3.07 (.81)

.51*

.00

 Temper

2.94 (.94)

2.92 (.93)

3.28 (.93)

.38*

.00

 Empathy

1.75 (.24)

1.76 (.23)

1.64 (.26)

−.49*

−.00

 Self-centeredness

2.28 (.74)

2.26 (.73)

2.60 (.85)

.46*

.00

 Negative peer commitment

1.79 (.85)

1.76 (.83)

2.17 (1.04)

.48*

.00

 Positive peer commitment

4.31 (.97)

4.33 (.96)

3.97 (1.05)

−.37*

−.00

 Neutralization of violence

3.30 (1.04)

3.26 (1.03)

3.79 (1.01)

.51*

.00

 Aggressive conflict resolution

1.64 (.59)

1.62 (.59)

1.94 (.62)

.53*

.00

 Fear of victimization

3.01 (1.04)

3.02 (1.05)

2.88 (1.00)

−.14

−.00

 Unstructured socializing

.42 (.27)

.41 (.27)

.55 (.30)

.49*

.00

 Delinquency

.14 (.25)

.13 (.22)

.33 (.45)

.81*

.00

 Parent monitoring

4.08 (.77)

4.11 (.74)

3.63 (.99)

−.63*

−.01

Wave 2 mediators

 Risk taking

2.74 (.98)

2.69 (.97)

3.57 (.83)

.90*

.49*

 Temper

3.00 (1.00)

2.97 (.99)

3.60 (.97)

.64*

.31*

 Empathy

1.74 (.25)

1.75 (.25)

1.59 (.29)

−.63*

−.38*

 Self-centeredness

2.25 (.78)

2.21 (.75)

2.81 (.97)

.77*

.34*

 Negative peer commitment

1.96 (.93)

1.91 (.91)

2.63 (1.09)

.77*

.39*

 Positive peer commitment

4.24 (.97)

4.28 (.94)

3.57 (1.28)

−.73*

−.41*

 Neutralization of violence

3.42 (1.10)

3.38 (1.09)

4.18 (.84)

.73*

.36*

 Aggressive conflict resolution

1.75 (.62)

1.72 (.60)

2.19 (.67)

.76*

.40*

 Fear of victimization

2.75 (1.02)

2.75 (1.02)

2.65 (1.06)

−.11

.02

 Unstructured socializing

.44 (.29)

.43 (.29)

.65 (.26)

.75*

.45*

 Delinquency

.19 (.30)

.16 (.25)

.64 (.59)

1.60*

1.07*

 Parent monitoring

4.09 (.75)

4.11 (.74)

3.67 (.85)

−.59*

−.16

Violent victimization

 Wave 1

.79 (1.60)

.76 (1.57)

1.29 (1.92)

.33*

.00

 Wave 2

.91 (1.73)

.84 (1.65)

2.00 (2.43)

.67*

.49*

 Wave 3 (N = 925)

.61 (1.26)

.58 (1.23)

1.08 (1.60)

.39*

.20

For dichotomous variables, figures represent percentages, whereas other figures represent means and (standard deviations). The unadjusted standardized difference is computed as the standardized coefficient of gang membership predicting the variable. The adjusted standardized difference is computed as the standardized coefficient of gang membership predicting the variable after controlling for the propensity score

p < .05

Table 1 also assesses the extent to which taking consideration of selection effect satisfies the conditional independence assumption in counterfactual research design. Naïve statistical differences are represented by standardized differences unadjusted for selection; selection adjusted statistical differences are represented by standardized differences after controlling for the propensity score of gang joining behavior at wave 2 and tell us if there is any remaining imbalance in the covariates. All of the wave 1 covariates—including the specified theoretical pathways—are balanced. At waves 2 and 3, the naïve difference in violent victimization between gang and non-gang youth is .67 and .39 standard deviations, respectively. After accounting for selection into gangs, those effects reduced to .49 and .20 standard deviations, respectively.

We can conclude from this exercise that 27 % of the contemporaneous differences in victimization are exogenous to gang membership, while the remaining 73 % of the differences are likely endogenous to gang membership. Likewise, 49 and 51 % of the prospective differences in victimization are exogenous and endogenous to gang membership, respectively. Our research adds to the existing body of evidence that identifies a direct effect of gang membership on victimization, and supports Thornberry’s (1993) enhancement model. It is now key to uncover the theoretical pathways that can account for the relationship between gang membership and victimization, which is what we aim to accomplish in the following analyses. Contemporaneous multiple mediator effects of gang membership on victimization.

Table 2 presents the total, direct, and indirect effects of gang membership on violent victimization at wave 2. The model controls for wave 1 values of proposed mediators, the propensity score for gang joining, demographic characteristics, and whether respondents were in treatment or control group of TCC/CW program. Unstandardized coefficients are presented in the table. Model A1 indicates the effect of gang joining on the potential mediators. It shows that except for fear of victimization, the potential mediators have statistically significant and modest to moderate effects in the expected direction. Joining a gang is positively associated with increased risk taking (b = .48, p < .05), temper (b = .31, p < .05), self-centeredness (b = .27, p < .05), negative peer commitment (b = .37, p < .05), neutralization of violence (b = .40, p < .05), aggressive conflict resolution (b = .24, p < .05), unstructured socializing (b = .13, p < .05), and delinquency (b = .32, p < .05). Also, joining a gang decreases levels of empathy (b = −.10, p < .05) and positive peer commitment (b = −.40, p < .05).
Table 2

The total, direct, and indirect effects of gang membership on the violent victimization at wave 2

Variable

b

SE

t

Model A1 (a1 paths): effects of the onset of gang membership on theoretical mediators

 Risk taking

.48*

.10

4.87

 Temper

.31*

.10

3.05

 Empathy

−10*

.03

−3.37

 Self-centeredness

.27*

.08

3.21

 Negative peer commitment

.37*

.10

3.59

 Positive peer commitment

−.40*

.11

−3.50

 Neutralization of violence

.40*

.11

3.63

 Aggressive conflict resolution

.24*

.65

3.70

 Fear of victimization

.01

.10

.11

 Unstructured socializing

.13*

.03

4.08

 Delinquency

.32*

.03

11.22

 Parent monitoring

−.12

.08

−1.51

Model B1 (b1 paths): effects of factors associated with gang membership on violent victimization

 Risk taking

−.10

.06

−1.46

 Temper

−.05

.06

−.79

 Empathy

−.14

.20

−.70

 Self-centeredness

−.17*

.08

−2.20

 Negative peer commitment

.02

.06

.31

 Positive peer commitment

−.07

.50

−1.44

 Neutralization of violence

.04

.06

.74

 Aggressive conflict resolution

.12

.10

1.22

 Fear of victimization

.04

.06

.71

 Unstructured socializing

−.06

.18

−.36

 Delinquency

2.04*

.21

9.69

 Parent monitoring

−.01

.07

−.14

Model C1 (c1 path): total effect of onset of gang membership on violent victimization

 Onset of gang membership

.84*

.20

4.24

Model D1 (c1′ path): direct effect of onset of gang membership on violent victimization

 Onset of gang membership

.20

.20

.98

Model E1 (a1b1 paths): indirect effects of onset of gang membership on violent victimization

 Total indirect effect

.65*

.17

 

 Risk taking

−.05

.04

 

 Temper

−.01

.02

 

 Empathy

.01

.02

 

 Self-centeredness

−.04*

.03

 

 Negative peer commitment

.01

.03

 

 Positive peer commitment

.03

.03

 

 Neutralization of violence

.02

.03

 

 Aggressive conflict resolution

.03

.03

 

 Fear of victimization

.00

.01

 

 Unstructured socializing

−.01

.02

 

 Delinquency

.67*

.18

 

 Parent monitoring

.00

.00

 

Estimates are based on the Preacher and Hayes technique (2008) using OLS regression, including bootstrap standard errors (1000 replications) for indirect effects. All variables presented above were measured at wave 2, and the model controls for community work treatment, propensity score for joining gangs, demographics, and wave 1 values of proposed mediators. R2 = .30. n = 1185 (70 gang members are compared with 1115 non-gang members)

SE standard error

p < .05

Model B1 displays the effects of the potential mediators on the violent victimization. Only delinquency (b = 2.04, p < .05) and self-centeredness (b = −.17, p < .05), however, are significantly related to violent victimization.

Overall, the total effect of joining a gang on violent victimization is statistically significant (see model C1). More specifically, the violent victimization frequency of gang members is .84 units greater than that of non-gang members (p < .05). After controlling for the indirect path, the direct path becomes non-significant (see Model D1), which means that the significant effect of gang membership on violent victimization operates entirely through the indirect paths. As model E1 shows, there is a statistically positive total indirect effect of gang membership on violent victimization (b = .65, p < .05). Joining a gang is associated with a .65 unit increase in violent victimization indirectly through a combination of all the potential mediators. Among nine potential mediators, delinquency (b = .67, p < .05) and self-centeredness (b = −.04, p < .05) are responsibility for fully mediating the effect of gang membership on violent victimization.

Prospective Multiple Mediator Effects of Gang Membership on Victimization

Table 3 presents the total, direct, and indirect effects of gang membership on violent victimization at wave 3. The results indicate a similar empirical pattern to the wave 2 victimization findings, although wave 3 missing data reduces the sample size, which produces differences in the point estimates of the coefficients for mediators and treatment variable. All the potential mediators are statistically significant and associated with gang joining in the expected way, with the exception of parental monitoring (see model A2). As model B2 shows, the only mediators statistically associated with violent victimization frequency are positive peer commitment (b = .10, p < .05), aggressive conflict resolution (b = .23, p < .05), and delinquency (b = 55, p < .05). The total effect, however, is not statistically significant (see model C2, b = .29, p > .05), nor are the direct and indirect effects, findings we return to in the following section.
Table 3

The total, direct, and indirect effects of onset of gang membership on the violent victimization frequency at wave 3

Variable

b

SE

t

Model A2(a2 paths): effects of the onset of gang membership on theoretical mediators

 Risk taking

.46*

.11

4.09

 Temper

.26*

.11

2.26

 Empathy

−.09*

.03

−2.75

 Self-centeredness

.17

.09

1.81

 Negative peer commitment

.50*

.12

4.29

 Positive peer commitment

−.33*

.13

−2.60

 Neutralization of violence

.34*

.12

2.80

 Aggressive conflict resolution

.29*

.07

3.84

 Fear of victimization

.08

.11

.69

 Unstructured socializing

.09*

.04

2.38

 Delinquency

.24*

.03

7.68

 Parent monitoring

−.13

.09

−1.42

Model B2 (b2 paths): effects of factors associated with gang membership on violent victimization

 Risk taking

.00

.06

.08

 Temper

−.04

.06

−.71

 Empathy

.10

.18

.55

 Self-centeredness

−.08

.07

−1.14

 Negative peer commitment

.06

.05

−1.09

 Positive peer commitment

.10*

.05

2.23

 Neutralization of violence

.01

.05

.14

 Aggressive conflict resolution

.23*

.09

2.58

 Fear of victimization

.02

.05

.31

 Unstructured socializing

−.01

.16

−.06

 Delinquency

.55*

.20

2.77

 Parent monitoring

−.03

.07

−.46

Model C2 (c2 path): total effect of onset of gang membership on violent victimization

 Onset of gang membership

.29

.17

1.68

Model D2 (c1′ path): direct effect of onset of gang membership on violent victimization

 Onset of gang membership

.18

.18

.99

Model E2 (a2b2 paths): indirect effects of onset of gang membership on violent victimization

 Total indirect effect

.11

.08

 

 Risk taking

.00

.03

 

 Temper

−.01

.02

 

 Empathy

−.01

.02

 

 Self-centeredness

−.01

.02

 

 Negative peer commitment

−.03

.03

 

 Positive peer commitment

−.03*

.02

 

 Neutralization of violence

.00

.02

 

 Aggressive conflict resolution

.07*

.04

 

 Fear of victimization

.00

.01

 

 Unstructured socializing

−.00

.01

 

 Delinquency

.13*

.07

 

 Parent monitoring

−.00

.01

 

Estimates are based on the Preacher and Hayes technique (2008) using OLS regression, including bootstrap standard errors (1000 replications) for indirect effects. All variables presented above were measured at wave 2, and the model controls control for community work treatment, propensity score for joining gangs, demographics, and wave 1 values of proposed mediators. R2 = .20. n = 924 (53 gang members are compared with 871 non-gang members)

SE standard error

p < .05

Discussion

Since Wolfgang (1957) studied victim-precipitated homicide in Philadelphia and observed what he termed a “subculture of violence,” scholars have often invoked differential normative orientations to status attainment, defense of honor, and the resolution of interpersonal disputes to understand violence and victimization (Anderson 1999; Black 1983; Kubrin and Weitzer 2003; Jacobs and Wright 2006). Naturally, the subcultural tradition shares a great deal of overlap with the study of street gangs and gang violence (Cloward and Ohlin 1960; Cohen 1955; Miller 1958). Yet, the life-course framework, with its emphasis on the contours of gang membership, has drawn greater scrutiny to how levels of criminal offending and victimization fluctuate with entry and exit from gangs as scholars have tried to unpack the extent to which high rates of violence were endogenous or exogenous to gang membership. In particular, Thornberry et al. (1993) theoretical models of selection, facilitation and enhancement, which were aligned with general theories of crime, have generated considerable empirical interest. This interest carried over from offending to victimization (DeLisi et al. 2009; Peterson et al. 2004) and has resulted in an emerging debate (Gibson et al. 2009; Ozer and Engel 2011) over whether the effects of gang membership are confounded, direct, or mediated. The present study responded to calls by scholars (Fox 2013; Gibson et al. 2012) to provide a theoretically informed and analytically sophisticated investigation into the pathways between gang membership and victimization. In this discussion we concentrate on three major points emerging from this research that merit further consideration.

First, the risks that youth bring with them into gangs partially, but not entirely, confound the relationship between gang membership and victimization. In other words, high rates of victimization are not solely an artifact of selection nor are they a product of gang-related group processes, which is consistent with the enhancement model. What this means is that proponents of selection and facilitation perspectives must cede any allegiance to a singular model and embrace the fact that a non-trivial portion of the variance is attributable to the baggage youth bring into gangs and the environments that gangs construct. Ours is not the first study of gang membership to make this point, and is consistent with a growing body of work using comparable methodology and research design (e.g. DeLisi et al. 2009). Given the emerging consensus that gang membership increases victimization beyond sources of selection, we would encourage future research to report the share attributable to selection and facilitation as merely a descriptive statistic (e.g., effects were 27 % exogenous, 73 % endogenous) while concentrating instead on a host of important factors—functional form, magnitude, interactions—related to the mechanisms of change, which leads to our next point.

Second, entry into a gang corresponds with an onslaught of changes in risk factors that were found to mediate the gang membership-victimization link. While not all of these changes were related to victimization, we add to a growing chorus of studies that have found that there is a range of changes associated with entrance into a gang (Matsuda et al. 2013; Medina-Ariza et al. 2014; Melde and Esbensen 2011; Sweeten et al. 2013; Thornberry et al. 2003; Weerman et al. 2015). Consistent with prior empirical work, we find that changes operate in the theoretically riskiest direction upon gang joining: increases in risk taking, temper, negative peer commitment, neutralization of violence, unstructured socializing, and delinquency, and decreases in positive peer commitment and school commitment. These factors operate at the heart of core theories of offending and victimization, including self-control, cultural orientations, routine activity, and lifestyle theories. Indeed, it would defy convention if we were to observe increased risk across the theoretical landscape yet not observe increases in victimization. However, the fact that the proposed pathways fully mediated the effects of gang membership on victimization raises serious questions about levels of explanations and the etiology of victimization in the gang context.

When Gibson et al. (2009) found evidence in support of the selection perspective it challenged, in Ozer and Engel’s (2011: 105, 117, 119) words, “conventional wisdom” about the gang membership-victimization link. What is most interesting, however, is the fact that studies almost never confound or mediate fully the effects of gang membership on criminal offending. Some of the most rigorous evidence is found in works using sophisticated multi-level analyses modeling within-individual changes simultaneous to gang status transitions—which almost certainly bias gang effects downward—yet cannot account for the relationship between gang membership and offending (the best examples are found in Bjerk 2009; Melde and Esbensen 2014; Ousey et al. 2011; Sweeten et al. 2013). The same cannot be said for gang membership and victimization. In fact, the literature could best be described as a mixed bag of confounding (e.g. Gibson et al. 2009), full mediation (e.g. Katz et al. 2011), and partial mediation (e.g. Pyrooz et al. 2014). Our findings fall in the camp of full mediation. This raises the question: why are the effects of gang membership on offending far more sturdy and robust to numerous model specifications than the effects of gang membership on victimization? The sensitivity of the gang membership-victimization link is something that must be unpacked in future research to determine “how much” and “when” gang membership matters to victimization. Among the mediators included in this study, there is one pathway in particular that warrants further consideration.

Third, our findings suggest that gang membership is linked victimization to the extent that gang members engage in delinquency. Criminal offending and victimization are closely associated with one another (Berg 2012). Indeed, the victim-offender overlap finds itself on the short list of criminological facts. This observation is especially strong within the context of gangs, which Pyrooz et al. (2014) aligned with the gang-related group process perspective. We expanded on this observation with a supplementary model estimated without delinquency as a mediator. The results revealed statistically significant total, direct, and indirect effects of gang membership on victimization, which confirmed delinquency acted as the primary mediator of the gang membership-victimization link. The fact that delinquency accounted entirely for high levels of victimization among gang members reveals several important points, particularly that victimization is not simply a dependent variable that is interchangeable with offending (see Esbensen and Huizinga 1993; Pratt et al. 2014). Much like offending, it remains theoretically profitable to rely on gang-related group processes as the driver of the mechanisms of change for victimization among gang members. Unlike offending, however, our evidence would lead us to suggest that at least explaining the gang membership-victimization link may be an individual-level enterprise. That is, variation in victimization risk is not operating at some higher level of explanation beyond the individual, but instead involves taking active steps to put oneself at risk (i.e., self-selection into risky environments). While this seems at odds with the nature of gang violence (e.g. Decker 1996; Melde and Esbensen 2013; Papachristos et al. 2013), particularly for victimization, this conclusion needs to be sorted out more systematically across a wider variation of specifications in research design and analytic methodology. It is premature to conclude that the continued relevance of group process to gang member victimization matters only to the extent that it exacerbates risk factors.

The endogeneity of victimization to gang membership offers some preliminary policy implications that align themselves well with recent policing efforts to concentrate on the “hot people” at risk of violent victimization (Kennedy 2011; Papachristos et al. 2015). The victim-offender overlap has taught us that it is inappropriate to treat victims and offenders as independent groups; criminal justice practices and policies built around such an assumption are a source of injustice (Baron 2003; Kennedy 2011; Lauritsen and Laub 2007). The risk of victimization is not evenly distributed across gang members. There is considerable variation in the victimizations reported by the gang joining respondents in this study (see Table 1). This means that resources should be directed toward the highest-risk gang members, that is, the individuals involved in the greatest amount of criminal offending. Likewise, given that there is nothing criminal about being a gang member (Sweeten et al. 2013: 491), the goal should be to reduce the likelihood of offending in this population. Such a strategy should produce dual dividends: reducing offending while also preventing victimization.

Future inquiry into the gang membership-victimization link should aim to expand on the limitations of this study. Confidence in our null prospective findings must be tempered by turnover in gang membership and study attrition. Longitudinal research shows that membership in a gang is brief, lasting <2 years for the majority of gang members in the U.S. (Pyrooz 2014). Among our wave 2 gang joiners, only 30 % remained in a gang at wave 3. While treatment need not be preserved in quasi-experimental research, it is highly likely that our prospective findings are capturing periods of former gang membership for a large number of gang joiners (see the discussions in Decker et al. 2013: 378–379; Gibson et al. 2012: 491).

In addition, although gang and non-gang joiners were statistically just as likely to not participate in the wave 3 interview, gang members who persisted in the study differed in important ways from gang members who dropped out of the study. The former engaged in statistically less delinquency at wave 2—our primary mediator—than the latter (see Appendix 2). To satisfy the calls by Gibson et al. (2012) to establish causal ordering, turnover in gang membership and study attrition will require researchers to employ methods such as space–time budgets, life history calendars, or shorter intervals between waves to study the gang membership-victimization link (see Pyrooz et al. 2014: 337).

Empirical research on this topic to date is derived heavily from school-based samples, which could offer an alternative social reality the victimization of gang members. Moreover, this line of study would benefit from greater analytic sophistication, subjecting the link between gang membership and victimization to different modeling strategies (e.g., within-individual fixed effects) and alternative strategies to account for selection bias. Indeed, there is the chance that selection bias was not eliminated fully by the observed confounders in the present study. It would also be beneficial to better unpack the forms and sources of victimization among gang members. Does gang membership raise the risk of all forms of victimization or is it specific to interpersonal forms of victimization? Further, is the victimization gang members experience gang-related or gang-motivated? This is an important distinction, and would lead to a more complete understanding of the gang membership-victimization link. Any of these efforts should be mindful of temporal ordering and continuity and change in gang membership.

In summary, this study contributed to the extant gang and victimization literature by uncovering the pathways leading to high victimization rates among members of gangs, a major source of youth violence in the U.S. We reached three primary conclusions based on our multiple mediator analysis: (1) while gang members are more likely to be victims of violence, factors exogenous and endogenous to gangs are jointly responsible for this observation, (2) transitioning into a gang corresponds with significant negative changes across a spectrum of risky behavioral or attitudinal characteristics, and (3) the changes mechanisms linked to joining a gang fully accounted for the gang membership-victimization link, with delinquency functioning as the primary mediating mechanism. This study contributed several important pieces of knowledge to this growing body of research, but there is much more to be learned about the gang membership-victimization link.

Footnotes

  1. 1.

    Examples include self-control (Childs et al. 2009; Fox et al. 2013; Pyrooz et al. 2014; Taylor et al. 2007, 2008), neutralizations and peer commitment (Pyrooz et al. 2014; Taylor et al. 2007, 2008), routine activities and lifestyles (Katz et al. 2011; Pyrooz et al. 2014; Taylor et al. 2007, 2008; Zavala and Spohn 2013), and social controls (Taylor et al. 2007, 2008; Zavala and Spohn 2013).

  2. 2.

    The purpose of this training program was to address issues of guns, violence, hate crimes, substance abuse, conflict management, and reducing victimization (Esbensen 2005). Thirty-one classes on these topics were delivered to the students by community resource officials. The evaluation employed a purposive sampling method that only schools offering the TCC/CW program were included in the initial sample. Among the initial 250 schools, only 18 met the evaluation criteria, including (1) whether the program was operated completely (2) whether there was enough classes for treatment and comparison group assignments, and (3) whether the school would agree to provide information to the evaluation (Esbensen 2005). Three schools rejected the evaluation offer; the final evaluation included 15 schools in four states. Because only certain teachers were trained for the TCC/CW program, a random assignment of treatment and comparison groups was not applicable. Therefore, 49 comparison classrooms (from 6th grade to 9th grade) were selected to match on the grade and subject with 48 treatment classrooms (Esbensen 2005). The training program failed to reduce adolescent violent victimization (Esbensen 2005).

  3. 3.

    Appendix 2 examines differences among four groups of the respondents, including non-gang and gang members who participated in the wave three survey, and non-gang and gang members who dropped out, a point we return to in the discussion section.

  4. 4.

    Alternative measures of violent victimization were examined, including a pooled prevalence score and a variety score. A variety score is a valid measure of criminality (Sweeten 2012) and has been used to measure victimization (Doherty et al. 2012; Pyrooz et al. 2014). The results from models using alternative measures of victimization do not change the substantive interpretation of the findings. Results are available upon request from the lead author.

  5. 5.

    All constructs were computed based on mean rather the summated values in order to reduce missing cases. A case was deleted only if more than half the items were missing for any given construct. Listwise deleting based on one item would have produced 300 missing cases, compared to only 48 missing cases from the procedure employed.

  6. 6.

    The statistics showed in Appendix 1 are computed based on a pooled sample combing wave 1 and 2.

  7. 7.

    The risk of omitted variable bias is elevated when each mediator is tested separately because other tested and omitted mediators are not considered simultaneously. The estimated effect of each mediator will likely be greater than that in a multiple mediator model.

Notes

Acknowledgments

We thank Cortney Franklin, Yan Zhang, and Matthias Woeckener, as well as JQC Associate Editor McGloin and the anonymous reviewers, for their helpful comments on earlier versions of this manuscript.

References

  1. Akers RL (2009) Social learning and local structure: a general theory of crime and deviance. Transaction, New BrunswickGoogle Scholar
  2. Akers RL, Jensen GF (2006) The empirical status of social learning theory of crime and deviance: the past, present, and future. In: Cullen FG, Wright JP, Blevins KR (eds) Taking stock: the status of criminological theory. Transaction Publishers, New Brunswick, New Jersey, pp 37–76Google Scholar
  3. Anderson E (1999) Code of the street. Norton, New YorkGoogle Scholar
  4. Barnes JC, Boutwell BB, Fox KA (2012) The effect of gang membership on victimization: a behavioral genetic explanation. Youth Violence Juv Justice 10:227–244CrossRefGoogle Scholar
  5. Baron SW (2003) Street youth violence and victimization. Trauma Violence Abuse 4:22–44CrossRefGoogle Scholar
  6. Berg MT (2012) The overlap of violent offending and violent victimization: Assessing the evidence and explanations. In: DeLisi M, Conis PJ (eds) Violent Offenders: theory, research, policy, and practice, 2nd edn. Jones and Bartlett Learning, Burlingtong, pp 17–38Google Scholar
  7. Berkovits I, Hancock GR, Nevitt J (2000) Bootstrap resampling approaches for repeated measure designs: relative robustness to sphericity and normality violations. Educ Psychol Meas 60:877–892CrossRefGoogle Scholar
  8. Bjerk D (2009) How much can we trust causal interpretations of fixed-effects estimators in the context of criminality? J Quant Criminol 25:391–417CrossRefGoogle Scholar
  9. Bjerregaard B, Lizotte AJ (1995) Gun ownership and gang membership. J Crim Law Criminol 86:37–58CrossRefGoogle Scholar
  10. Black D (1983) Crime as social control. Am Sociol Rev 48:34–45CrossRefGoogle Scholar
  11. Block CR, Block R (1993) Street gang crime in Chicago. US Department of Justice, Office of Justice Programs, National Institute of Justice, Washington DCGoogle Scholar
  12. Burt CH, Simons RL, Simons LG (2006) Longitudinal test of the effects of parenting and the stability of self-control: negative evidence for the general theory of crime. Criminology 44:353–396CrossRefGoogle Scholar
  13. Burt CH, Sweeten G, Simons RL (2014) Self-control through emerging adulthood: instability, multidimensionality, and criminological significance. Criminology 52:450–487CrossRefGoogle Scholar
  14. Childs KK, Cochran JK, Gibson CL (2009) Self-control, gang membership, and victimization: an integrated approach. J Crime Justice 32:35–60CrossRefGoogle Scholar
  15. Cloward R, Ohlin LE (1960) Delinquency and opportunity: A theory of delinquent gangs. Free Press, New YorkGoogle Scholar
  16. Coffman DL (2011) Estimating causal effects in mediation analysis using propensity scores. Struct Equ Model Multidiscip J 18:357–369MathSciNetCrossRefGoogle Scholar
  17. Cohen AK (1955) Delinquent boys: the culture of the gang. Free Press, GlencoeGoogle Scholar
  18. Cohen LE, Felson M (1979) Social change and crime rate trends: a routine activity approach. Am Sociol Rev 44:588–608CrossRefGoogle Scholar
  19. Curry GD, Decker SH, Pyrooz DC (2014) Confronting gangs: crime and community, 3rd edn. Oxford University Press, New YorkGoogle Scholar
  20. Decker SH (1996) Collective and normative features of gang violence. Justice Q 13:243–264CrossRefGoogle Scholar
  21. Decker SH, Pyrooz DC (2010) Gang violence worldwide: context, culture, and country. Small arms survey. Cambridge, Cambridge, pp 128–155Google Scholar
  22. Decker SH, Van Winkle B (1996) Life in the gang: family friends and violence. Cambridge University Press, New YorkCrossRefGoogle Scholar
  23. Decker SH, Katz CM, Webb VJ (2008) Understanding the black box of gang rrganization: implications for involvement in violent crime, drug sales, and violent victimization. Crime Delinq 54:153–172CrossRefGoogle Scholar
  24. Decker SH, Melde C, Pyrooz DC (2013) What do we know about gangs and gang members and where do we go from here? Justice Q 30:369–402CrossRefGoogle Scholar
  25. Decker SH, Pyrooz DC, Sweeten G, Moule RK Jr (2014) Validating self-nomination in gang research: assessing differences in gang embeddedness across non-, current, and former gang members. J Quant Criminol. doi:10.1007/s10940-014-9215-8 Google Scholar
  26. DeLisi M, Barnes JC, Beaver KM, Gibson CL (2009) Delinquent gangs and adolescent victimization revisited a propensity score matching approach. Crim Justice Behav 36:808–823CrossRefGoogle Scholar
  27. Densley JA (2012) Street gang recruitment: signaling, screening, and selection. Soc Probl 59:301–321CrossRefGoogle Scholar
  28. Densley JA (2013) How gangs work: an ethnography of youth violence. Palgrave, LondonCrossRefGoogle Scholar
  29. Doherty EE, Robertson JA, Green KM et al (2012) A longitudinal study of substance use and violent victimization in adulthood among a cohort of urban African Americans. Addiction 107:339–348PubMedPubMedCentralCrossRefGoogle Scholar
  30. Elliott DS, Huizinga D, Ageton SS (1985) Explaining delinquency and drug use. Sage, Beverly HillsGoogle Scholar
  31. Esbensen F-A (2003) Evaluation of the gang resistance education and training (GREAT) program in the United States, 1995–1999. Inter-University Consortium for Political and Social Research, Ann ArborGoogle Scholar
  32. Esbensen F-A (2005) Evaluation of the teens, crime, and the community and community works program. U.S. Department of Justice, National Institute of Justice, Washington DCGoogle Scholar
  33. Esbensen F-A, Huizinga D (1993) Gangs, drugs, and delinquency in a survey of urban youth. Criminology 31:565–589CrossRefGoogle Scholar
  34. Esbensen F-A, Winfree LT, He N, Taylor TJ (2001) Youth gangs and definitional issues: when is a gang a gang, and why does it matter? Crime Delinq 47:105–130CrossRefGoogle Scholar
  35. Fagan J (1999) Youth gangs, drugs, and socioeconomic isolation. In: Flannery DJ, Huff CR (eds) Youth violence: prevention, intervention, and social policy. American Psychiatric Publishing, Washington DC, pp 145–170Google Scholar
  36. Farrell G, Pease K (2001) Repeat victimization. Criminal Justice Press, MonseyGoogle Scholar
  37. Felson M, Boba R (2010) Crime and everyday life, 4th edn. Sage, Thousand OaksCrossRefGoogle Scholar
  38. Fleisher MS (1998) Dead end kids: gang girls and the boys they know. University of Wisconsin Press, MadisonGoogle Scholar
  39. Fox KA (2013) New developments and implications for understanding the victimization of gang members. Violence Vict 28:1015–1040PubMedCrossRefGoogle Scholar
  40. Fox KA, Nobles MR, Akers RL (2011) Is stalking a learned phenomenon? An empirical test of social learning theory. J Crim Justice 39:39–47CrossRefGoogle Scholar
  41. Fox KA, Lane J, Akers RL (2013) Understanding gang membership and crime victimization among jail inmates: testing the effects of self-control. Crime Delinq 59:764–787CrossRefGoogle Scholar
  42. Funk MJ, Westreich D, Weisen C, Davidian M (2010) Doubly robust estimation of treatment effects. In: Faries DE, Obenchain R, Haro JM, Leon AC (eds) Analysis of Observational Health Care Data Using SAS. SAS Institute, Cary, North Carolina, pp 85–103Google Scholar
  43. Garot R (2009) Reconsidering retaliation structural inhibitions, emotive dissonance, and the acceptance of ambivalence among inner-city young men. Ethnography 10:63–90CrossRefGoogle Scholar
  44. Gibson CL, Miller JM, Jennings WG et al (2009) Using propensity score matching to understand the relationship between gang membership and violent victimization: a research note. Justice Q 26:625–643CrossRefGoogle Scholar
  45. Gibson CL, Swatt ML, Miller JM et al (2012) The causal relationship between gang joining and violent victimization: a critical review and directions for future research. J Crim Justice 40:490–501CrossRefGoogle Scholar
  46. Goldman L, Giles H, Hogg MA (2014) Going to extremes: social identity and communication processes associated with gang membership. Group Process Intergroup Relat 17:813–832CrossRefGoogle Scholar
  47. Gottfredson MR, Hirschi T (1990) A general theory of crime. Stanford University Press, StanfordGoogle Scholar
  48. Gover AR, Jennings WG, Tewksbury R (2009) Adolescent male and female gang members’ experiences with violent victimization, dating violence, and sexual assault. Am J Crim Justice 34:103–115CrossRefGoogle Scholar
  49. Grasmick HG, Tittle CR, Bursik RJ, Arneklev BJ (1993) Testing the core empirical implications of Gottfredson and Hirschi’s general theory of crime. J Res Crime Delinq 30:5–29CrossRefGoogle Scholar
  50. Hindelang MJ, Gottfredson MR, Garofalo J (1978) Victims of personal crime: an empirical foundation for a theory of personal victimization. Ballinger, CambridgeGoogle Scholar
  51. Howell JC, Decker SH (1999) The youth gangs, drugs, and violence connection. US Department of Justice, Office of Justice Programs, Office of Juvenile Justice and Delinquency Prevention, Washington DCCrossRefGoogle Scholar
  52. Howell JC, Egley A, Tita GE, Griffiths E (2011) US gang problem trends and seriousness, 1996–2009. US Department of Justice, Bureau of Justice Assistance, Office of Juvenile Justice and Delinquency Prevention, Washington DCGoogle Scholar
  53. Hughes LA (2013) Group cohesiveness, gang member prestige, and delinquency and violence in Chicago, 1959–1962. Criminology 51:795–832CrossRefGoogle Scholar
  54. Huizinga D, Esbensen F-A, Weiher AW (1991) Are there multiple paths to delinquency? J Crim Law Criminol 1973–82:83–118Google Scholar
  55. Hunt GP, Laidler KJ (2001) Alcohol and violence in the lives of gang members. Alcohol Res Health 25:66–71PubMedGoogle Scholar
  56. Jacobs BA, Wright R (2006) Street justice: retaliation in the criminal underworld. Cambridge University Press, New YorkCrossRefGoogle Scholar
  57. Jankowski MS (1991) Islands in the street: gangs and American urban society. University of California Press, OaklandGoogle Scholar
  58. Katz CM, Webb VJ, Fox K, Shaffer JN (2011) Understanding the relationship between violent victimization and gang membership. J Crim Justice 39:48–59CrossRefGoogle Scholar
  59. Kennedy DM (2011) Don’t shoot: one man, a street fellowship, and the end of violence in inner-city America. Bloomsbury Publishing, New YorkGoogle Scholar
  60. Kissner J, Pyrooz DC (2009) Self-control, differential association, and gang membership: a theoretical and empirical extension of the literature. J Crim Justice 37:478–487CrossRefGoogle Scholar
  61. Klein MW (1995) The American street gang: its nature, prevalence, and control. Oxford University Press, New YorkGoogle Scholar
  62. Klein MW, Maxson CL (2006) Street gang patterns and policies. Oxford University Press, New YorkCrossRefGoogle Scholar
  63. Kubrin CE, Weitzer R (2003) Retaliatory homicide: concentrated disadvantage and neighborhood culture. Soc Probl 50:157–180CrossRefGoogle Scholar
  64. Lauritsen JL, Laub JH (2007) Understanding the link between victimization and offending: new reflections on an old idea. Surveying crime in the 21st century: commemorating the 25th anniversary of the British crime survey. Criminal Justice Press, Monsey, pp 55–75Google Scholar
  65. Lynskey DP, Winfree LT, Esbensen F-A, Clason DL (2000) Linking gender, minority group status and family matters to self-control theory: a multivariate analysis of key self-control concepts in a youth-gang context. Juv Fam Court J 51:1–19CrossRefGoogle Scholar
  66. Matsuda KN, Melde C, Taylor TJ et al (2013) Gang membership and adherence to the “Code of the Street”. Justice Q 30:440–468CrossRefGoogle Scholar
  67. McGloin JM, Decker SH (2010) Theories of gang behavior and public policy. Criminology and public policy: putting theory to work. Temple University Press, Philadelphia, pp 150–165Google Scholar
  68. Medina-Ariza JJ, Cebulla A, Aldridge J et al (2014) Proximal adolescent outcomes of gang membership in England and Wales. J Res Crime Delinq 51:168–199CrossRefGoogle Scholar
  69. Melde C (2009) Lifestyle, rational choice, and adolescent fear: a test of a risk-assessment framework. Criminology 47:781–812CrossRefGoogle Scholar
  70. Melde C, Esbensen F-A (2011) Gang membership as a turning point in the life course. Criminology 49:513–552CrossRefGoogle Scholar
  71. Melde C, Esbensen F-A (2012) The onset of (Euro) gang membership as a turning point in the life course. Youth gangs in international perspective. Springer, New York, pp 169–187CrossRefGoogle Scholar
  72. Melde C, Esbensen F-A (2013) Gangs and violence: disentangling the impact of gang membership on the level and nature of offending. J Quant Criminol 29:143–166CrossRefGoogle Scholar
  73. Melde C, Esbensen F-A (2014) The relative impact of gang status transitions: identifying the mechanisms of change in delinquency. J Res Crime Delinq 51:349–376CrossRefGoogle Scholar
  74. Melde C, Taylor TJ, Esbensen F-A (2009) “I got your back”: an examination of the protective function of gang membership in adolescence. Criminology 47:565–594CrossRefGoogle Scholar
  75. Miller WB (1958) Lower class culture as a generating milieu of gang delinquency. J Soc Issues 14:5–19CrossRefGoogle Scholar
  76. Muraven M, Baumeister RF (2000) Self-regulation and depletion of limited resources: does self-control resemble a muscle? Psychol Bull 126:247–259. doi:10.1037/0033-2909.126.2.247 PubMedCrossRefGoogle Scholar
  77. Na C, Paternoster R (2012) Can self-control change substantially over time? Rethinking the relationship between self- and social control. Criminology 50:427–462CrossRefGoogle Scholar
  78. National Center for Injury Prevention and Control (2012) Youth violence: Facts at a glance. Retrieved from https://ophp.sph.rutgers.edu/njphtc/AP_-_Public_Health_Policy_files/YV-DataSheet-a.pdf
  79. National Gang Center (2014) National youth gang survey analysis. Retrieved from http://www.nationalgangcenter.gov/Survey-Analysis
  80. Ostrander RN (2008) Meditations on a bullet: violently injured young men discuss masculinity, disability and blame. Child Adolesc Soc Work J 25:71–84CrossRefGoogle Scholar
  81. Ousey G, Wilcox P, Fisher B (2011) Something old, something new: revisiting competing hypotheses of the victimization-offending relationship among adolescents. J Quant Criminol 27:53–84CrossRefGoogle Scholar
  82. Ozer MM, Engel RS (2011) Revisiting the use of propensity score matching to understand the relationship between gang membership and violent victimization: a cautionary note. Justice Q 29:105–124CrossRefGoogle Scholar
  83. Papachristos AV (2009) Murder by structure: dominance relations and the social structure of gang homicide. Am J Sociol 115:74–128PubMedCrossRefGoogle Scholar
  84. Papachristos AV, Meares TL, Fagan J (2009) Why do criminals obey the law? The influence of legitimacy and social networks on active gun offenders. In: Influence of legitimacy and social networks on active gun offenders January 12 2009. Yale Law and Economics Research Paper, pp 09–199Google Scholar
  85. Papachristos AV, Hureau DM, Braga AA (2013) The corner and the crew: the influence of geography and social networks on gang violence. Am Sociol Rev 78:417–447CrossRefGoogle Scholar
  86. Papachristos AV, Wildeman C, Roberto E (2015) Tragic, but not random: the social contagion of nonfatal gunshot injuries. Soc Sci Med 125:139–150PubMedCrossRefGoogle Scholar
  87. Peterson D, Taylor TJ, Esbensen F-A (2004) Gang membership and violent victimization. Justice Q 21:793–815CrossRefGoogle Scholar
  88. Pratt TC, Turanovic JJ (2015) Lifestyle and routine activity theories revisited: the importance of “risk” to the study of victimization. Victims and Offenders. doi:10.1080/15564886.2015.1057351
  89. Pratt TC, Turanovic JJ, Fox KA, Wright KA (2014) Self-control and victimization: a meta-analysis. Criminology 52:87–116CrossRefGoogle Scholar
  90. Preacher KJ, Hayes AF (2008) Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behav Res Methods 40:879–891PubMedCrossRefGoogle Scholar
  91. Pyrooz DC (2012) Structural covariates of gang homicide in large U.S. cities. J Res Crime Delinquency 49:489–518CrossRefGoogle Scholar
  92. Pyrooz DC (2014) “From your first cigarette to your last dyin” day’: the patterning of gang membership in the life-course. J Quant Criminol 30:349–372CrossRefGoogle Scholar
  93. Pyrooz DC, Decker SH (2013) Delinquent behavior, violence, and gang involvement in China. J Quant Criminol 29:251–272CrossRefGoogle Scholar
  94. Pyrooz DC, Moule RK, Decker SH (2014) The contribution of gang membership to the victim–offender overlap. J Res Crime Delinq 51:315–348CrossRefGoogle Scholar
  95. Pyrooz DC, Turanovic JJ, Decker SH, Wu J (2015) Taking stock of the relationship between gang membership and offending: a meta-analysis. Crim Justice Behav. doi:10.1177/0093854815605528 Google Scholar
  96. Ridgeway G, McCaffrey DF (2007) Demystifying double robustness: a comparison of alternative strategies for estimating a population mean from incomplete data. Stat Sci 22:540–543. doi:10.1214/07-STS227C MathSciNetMATHCrossRefGoogle Scholar
  97. Rosenbaum PR, Rubin DB (1983) The central role of the propensity score in observational studies for causal effects. Biometrika 70:41–55MathSciNetMATHCrossRefGoogle Scholar
  98. Rufino KA, Fox KA, Cramer RJ, Kercher GA (2012) The gang–victimization link: considering the effects of ethnicity and protective behaviors among prison inmates. Deviant Behav 34:25–37CrossRefGoogle Scholar
  99. Schreck CJ (1999) Criminal victimization and low self-control: an extension and test of a general theory of crime. Justice Q 16:633–654CrossRefGoogle Scholar
  100. Short J Jr (1998) The level of explanation problem revisited—the American Society of Criminology 1997 presidential address. Criminology 36:3–36ADSCrossRefGoogle Scholar
  101. Short J Jr, Strodtbeck FL (1965) Group process and gang delinquency. University of Chicago Press, ChicagoGoogle Scholar
  102. Singh K, Xie M (2008) Bootstrap: a statistical method. Retrieved from http://www.stat.rutgers.edu/home/mxie/rcpapers/bootstrap.pdf
  103. Slocum LA, Taylor TJ, Brick BT, Esbensen F-A (2010) Neighborhood structural characteristics, individual-level attitudes, and youths’ crime reporting intentions. Criminology 48:1063–1100CrossRefGoogle Scholar
  104. Spano R, Freilich JD, Bolland J (2008) Gang membership, gun carrying, and employment: applying routine activities theory to explain violent victimization among inner city, minority youth living in extreme poverty. Justice Q 25:381–410CrossRefGoogle Scholar
  105. Swahn MH, Bossarte RM, West B, Topalli V (2010) Alcohol and drug use among gang members: experiences of adolescents who attend school. J Sch Health 80:353–360PubMedCrossRefGoogle Scholar
  106. Sweeten G (2012) Scaling criminal offending. J Quant Criminol 28:533–557Google Scholar
  107. Sweeten G, Pyrooz DC, Piquero AR (2013) Disengaging from gangs and desistance from crime. Justice Q 30:469–500CrossRefGoogle Scholar
  108. Sykes GM, Matza D (1957) Techniques of neutralization: a theory of delinquency. Am Sociol Rev 22:664–670CrossRefGoogle Scholar
  109. Taylor TJ (2008) The Boulevard Ain’t Safe for Your kids.1 youth gang membership and violent victimization. J Contemp Crim Justice 24:125–136CrossRefGoogle Scholar
  110. Taylor TJ, Peterson D, Esbensen F-A, Freng A (2007) Gang membership as a risk factor for adolescent violent victimization. J Res Crime Delinquency 44:351–380CrossRefGoogle Scholar
  111. Taylor TJ, Freng A, Esbensen F-A, Peterson D (2008) Youth gang membership and serious violent victimization: the importance of lifestyles and routine activities. J Interpers Violence 24:381–410Google Scholar
  112. Thornberry TP, Krohn MD, Lizotte AJ, Chard-Wierschem D (1993) The role of juvenile gangs in facilitating delinquent behavior. J Res Crime Delinquency 30:55–87CrossRefGoogle Scholar
  113. Thornberry TP, Krohn MD, Lizotte AJ et al (2003) Gangs and delinquency in developmental perspective. Cambridge University Press, New YorkGoogle Scholar
  114. Tigri HB, Reid S, Turner MG, Devinney JM (2015) Investigating the relationship between gang membership and carrying a firearm: results from a national sample. Am J Crim Justice. doi:10.1007/s12103-015-9297-3
  115. Turanovic JJ, Pratt TC (2014) “Can’t stop, won’t stop”: self-control, risky lifestyles, and repeat victimization. J Quant Criminol 30:29–56CrossRefGoogle Scholar
  116. Tyler BM (1990) The rise and decline of the watts summer festival, 1965–1986. Am Stud 31:61–81Google Scholar
  117. Weerman FM, Bernasco W, Bruinsma GJN, Pauwels LJR (2013) When is spending time with peers related to delinquency? The importance of where, what, and with whom. Crime Delinquency. doi:10.1177/0011128713478129 Google Scholar
  118. Weerman FM, Lovegrove PJ, Thornberry T (2015) Gang membership transitions and its consequences: exploring changes related to joining and leaving gangs in two countries. Eur J Criminol 12:70–91CrossRefGoogle Scholar
  119. White R, Mason R (2012) Bullying and gangs. Int J Adolesc Med Health 24:57–62PubMedCrossRefGoogle Scholar
  120. Whyte WF (1943) Social organization in the slums. Am Sociol Rev 8:34–39CrossRefGoogle Scholar
  121. Wolfgang MF (1957) Victim precipitated criminal homicide. J Crim Criminol Police Sci 48:1–11CrossRefGoogle Scholar
  122. Zavala E, Spohn RE (2013) The role of vicarious and anticipated strain on the overlap of violent perpetration and victimization: a test of general strain theory. Am J Crim Justice 38:119–140CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media New York 2015

Authors and Affiliations

  1. 1.Department of Criminal Justice and CriminologySam Houston State UniversityHuntsvilleUSA
  2. 2.Department of SociologyUniversity of Colorado BoulderBoulderUSA

Personalised recommendations