What Does Geography Have To Do With It?

Abstract

In the first decade of the 21st century, the Eurogang Program of Research (hereafter referred to as the Eurogang Network), a group of U.S. and European criminologists, began a systematic effort to examine the European gang phenomenon. However, the notion that American-style street gangs represent a real threat to European nations has not been universally accepted, as critics warn about the dangers of making gangs and gang members into “folk devils” for a broad range of societal ills. In order to provide insights into the status of troublesome youth groups in the European context, the present study examines three separate surveys of school children aged 13 to 16 years who resided in Bosnia and Herzegovina, Germany, and the Netherlands. Using definitional and research protocols developed by the Eurogang Network, this article explores correlates of individual delinquency reported by street gang members and members of other delinquent youth groups that reportedly engaged in criminal activities but were not defined as street gangs using the Eurogang Network definition. The study turns on the extent to which the self-reported individual and group misbehavior can be understood in terms of personal-biographical characteristics—but especially citizenship status, specific youth-group membership, and nation of current residence—and the influence of putative risk and protective measures. The implications for youth gang intervention efforts are discussed, along with what the findings suggest about comparative efforts to explain misbehavior by members of two types of troublesome youth groups.

Keywords

crime/delinquency theory criminal organizations/gangs other social constructions of crime/justice

Introduction

Youth gangs have been a staple of American delinquency studies and general criminological research since the early 20th century (Curry & Decker, 2002). Contemporary U.S. gangs are characterized as male-dominated groups of minority youth dedicated to the commission of criminal acts, usually involving drugs, violence, or both (Curry & Decker, 2002; Klein & Maxson, 2006). These characterizations have changed little since the initial forays into street gangs conducted in early 1900s Chicago and elsewhere (see Shaw & McKay, 1942; Thrasher, 1927). We now have reason to question the racial and gender stereotypes promulgated by these early gang researchers and even contemporary police-based studies of youth gangs (Curry & Decker, 2002; Howell, 2007). Nonetheless, gender, race, and violence continue to be stereotypical elements embedded in many U.S.-based gang studies.

While there is a broad and deep well of knowledge about U.S. gangs, the same cannot be said of gang studies in other nations. Most international gang researchers take a rather piecemeal approach, describing the phenomenon in specific cities or regions of their respective nations. Even when the research goals are more ambitious, such single-nation studies leave much to be desired. Consider, by way of illustration, Bennett and Holloway’s (2004) study of gangs, drugs, and crime in the United Kingdom (UK). After acknowledging that their work focused on U.S.-style street gangs (see also Klein, Kerner, Maxson, & Weitekamp, 2001), Bennett and Holloway go on to describe typical gangsters in the United Kingdom as criminally active males who carry weapons, including guns, and commit robberies and illegal drug-supply crimes, a summary similar to that which might be provided for any study of gangs in the United States during the same time frame. Unlike the United States, the majority of the U.K. gang members were White, with Caribbean and Bangladeshi youth constituting the largest minority elements. These findings must be tempered by the limitations associated with the data source, the New English and Welsh Arrestee Drug Abuse Monitoring program. In some ways, this study resembles the law enforcement-centered work on gangs that was common in the United States during the 1990s and early 2000s, an approach that can yield skewed views of gang members, relying as they do on arrested or the police-identified gang members (see Bjerregaard, 2002; Esbensen, Winfree, He, & Taylor, 2001). Nation-based studies can be insightful, but given the use of idiosyncratic definitions and one-off data sets, they often do not readily yield results that can be compared cross-nationally. Gang data from international and intranational sources that try to make sense of the resulting multiple-frame views represent another alternative. For example, consider Duffy and Gillig’s (2004) Teen Gangs: A Global View, a compilation of individual studies that collectively provides an international look at gangs in 14 nations around the world; however, even in this work, cross-national comparisons are difficult, as the sophistication of data and methods of study about gangs across nations is highly variable.

Beginning in 1998, a group of United States and European researchers, policy makers and practitioners—the Eurogang Network—began working on a common strategy to explore the European gang phenomenon. Consistent with the “moral panic” perspective, there was considerable divergence of opinion among members about not only whether Europe had a gang problem or but also whether U.S.-style gangs existed in Europe (see Aldridge, Medina, & Ralphs, 2012; Medina, Aldridge, Shute, & Ross, 2013). A concurrent concern was that the search for blame would end at the gang’s turf, turning them into a kind of “folk devil,” something that has already happened with sports hooligans in Europe (Cohen, 1980; St. Cyr, 2003; see also Pearton & Gaskell, 1981). As Maxson and Klein (1995) have shown, even the majority of U.S. gangs do not fit widely held gang stereotypes (see also Klein & Maxson, 2006). Maxson and Klein, and other members of the Eurogang Network (e.g., Decker & Weerman, 2005; Klein, 2005; Klein et al., 2001) have catalogued these false impressions. The Eurogang Network has prepared a series of protocols for studying the European gang phenomenon (Weerman et al., 2009) and promulgated the following definition: A youth gang is a durable, street-oriented youth group whose involvement in illegal activity is part of their group identity (Weerman et al., 2009; see too Medina et al., 2013; Webb et al., 2011).

The present study builds on the Eurogang Network’s work by providing a three-nation comparison of the nature and extent of gang attitudes, orientations, and perspectives. In their landmark gang study of adolescents in the United States and the Netherlands, Esbensen and Weerman (2005) argued persuasively for the validity of such comparisons. We refocus this type of comparative analysis by looking solely at those individuals living in Bosnia and Herzegovina (BiH), Germany, and the Netherlands that claim membership in either street gangs or other delinquent youth groups (DYGs).

Gaining Insights Into Gang Members: Issues and Concerns

Three clusters of putative gang correlates provide the analytical framework for this study. First, U.S. and European researchers often point to the existence of risk factors, including deviant peer commitments, and protective factors, such as commitment to schools and parental oversight, that play varying roles in the movement toward and away from youth gangs (Esbensen, Peterson, Taylor, & Freng, 2010; Esbensen & Weerman, 2005; Nash, Mujanovic & Winfree, 2011; Weerman & Esbensen, 2005; Winfree, 2012). Moreover, many risk factors have close ties to criminological theories such as differential association and social learning theory (Akers, 1985), while the protective factors are often linked to either or both social control and self-control theories (Hirschi, 1969; Gottfredson & Hirschi, 1990). Even among adolescents involved in youth gangs and other crime-involved social groups, it is likely that some of these factors may still be at work, protecting some and endangering others. It also remains an empirical question as to whether such factors work in the same or similar ways in nations other than the United States.

Second, researchers have suggested that to understand street gangs we must gain better insights into serious offending, as the two often go hand-in-hand (Esbensen et al., 2010; Thornberry, Huizinga & Loeber, 2004). Battin-Pearson, Thornberry, Hawkins, and Krohn (1998), for example, addressed this issue in their summary of the Program of Research on the Causes and Correlates of Delinquency. Risk and protective features, they maintained, are important; however, gang membership has an independent contributing role in the etiology of delinquency. Gang members engage in more serious delinquency than nongang members, and even after leaving the gang they continue to reoffend at higher rates than non-gang members.

Third, the alleged level of organization found within U.S. gangs has also sparked a debate of sorts among both gang researchers and pundits who study gang problems. The media, in both the news and entertainment segments, tend to portray gangs as highly organized and operating at a structural level that allows them to engage in interstate commerce on a level with corporate America. Others question these assumptions and assertions. For instance, in a study of gang homicides, Decker and Curry (2002) questioned the level of organization found among gangs: “Gangs were unable to organize homicides in an effective manner, which reflected the disorganized character of gangs and the neighborhoods in which they reside” (p. 343). Conversely, Ruble and Turner (2000, p. 117) observed the following: “Street gangs usually exhibit a highly complex organization, structure, process, and functionality. All of these aspects intertwine within and around the gang to form a web of interconnectedness and continuity.” Gang organization remains a significant if little understood construct in gang research (Decker, Katz, & Webb, 2008). The impact of these structural and functional characteristics remains largely speculative, especially in a comparative context.

The current state of knowledge about gang members outside the United States remains suspect and somewhat limited. The present study looks into the lives of gang members and members of other crime-involved groups in three different nations, using comparable or, in two cases, identical measures, all with the intent of examining the extent to which they may differ from or resemble each other. Specifically, three questions guide this study. First, what can we learn about the level of the putative risk and protective measures reported by members of crime-involved social groups from their respective personal-biographical characteristics, including their nation of current residence? Second, what can we learn about the level of individual illegal activities reported by such children from their personal-biographical characteristics, including their nation of current residence, and the influence of alleged risk and protective measures? Third, what can we learn about group-level criminal activities reported by the subjects from their demographic features, risk and protective factors, and self-reported illegal activities, along with the structural features attributed to the youth groups? Answers to these questions were sought in three unique school-based surveys, all conducted between 2003 and 2007, when the subject were 13 to 16 years of age. We turn next to an examination of those samples and the variables they contain.

Research Methods

The School-Based Surveys

The work of the Eurogang Network provides a solid methodological and conceptual foundation upon which to base a comparative study of troublesome youth groups. Answers to this study’s three central questions come from a combined data set representing three separate studies.¹ The best way to view these elements is to describe each separate project.

The Netherlands

The Netherlands is a constitutional monarchy; its civil law system incorporates French penal theory. Eight in 10 of its 16.5 million residents are Dutch. More than one half of the rest are of non-Western origins. The primary languages are Dutch and Frisian.

In 2003, the Netherlands Institute for the Study of Crime and Law Enforcement conducted the Study of Crime and Law Enforcement (NSCR) School Project, which surveyed students in the Dutch equivalent of the seventh and ninth grades. The Dutch items were highly compatible with but not identical to the Eurogang school survey. The Dutch sample was drawn from schools in major cities as well as small town and rural areas. This approach represented an attempt to obtain “a relatively ‘high-risk’ sample (i.e., with a substantial proportion of youth involved in illegal activities) while also maintaining adequate inclusion of ‘lower-risk’ students to allow for variation in school contexts and student populations” (Esbensen & Weerman, 2005, p. 14). The 12 participating schools were located in a city with about 500,000 inhabitants, two smaller cities of about 120,000 residents, and a village of roughly 30,000 residents. The Dutch sample is reasonably representative of children residing in the southwestern part of the Netherlands. The schools in the largest city served students considered “at-risk” for joining gangs; the students were also ethnically heterogeneous and largely drawn from lower socioeconomic groups than those residing in the other included cities and the village; that is, the other participating schools had far more ethnically homogenous (i.e., Dutch) populations.

Passive parental consent was employed, but only two parents refused to allow their children to participate. Virtually every eligible child in attendance on the administration day completed a questionnaire. Field researchers obtained completed questionnaires from 1,978 students attending school in four cities across the nation, which represented 83% of all children in attendance. For this age-censored study, the sample was reduced to 1,627 children aged 13 to 16 years.

Germany

A civil law nation with indigenous legal concepts, Germany is a federated republic with 82 million inhabitants. German is the nation’s official language. Ethnic minorities make up about 10% of its population, with Turks being the largest single group, plus significant pockets of Greeks, Italians, Poles, Russians, Serbo-Croatians, and Spanish.

In 2004, researchers at the University of Tübingen’s Institute of Criminology undertook a school-based study of troublesome youth groups and youth violence within southwestern Germany.² The researchers focused on the nature and extent of youth gang problems within local Hauptschulen, the lower-level standard school for adolescents finishing German elementary school education, essentially Years 5 through 9 of the 9 years of compulsory education. Three Hauptschulen participated in the study.

The researchers used the Eurogang school-based survey instrument, which was back-translated and forward-translated into German, and the Eurogang protocol that called for administering the questionnaires en masse in a classroom setting. Prior to its administration, parents were informed of the survey and its purposes, as well as their right to withhold consent for their children to participate (i.e., passive consent). Less than two dozen parents/guardians withheld permission and, once their rights of refusal were explained to them, only three children among the 533 eligible individuals in attendance on the administration day refused to complete a survey. That is, nearly 97% of those students in attendance could have participated; fully 99% of those in attendance and eligible for participation actually completed a survey.

The researchers obtained completed questionnaires from 530 youth across three grade levels (Years 7, 8, and 9) in three school districts. Looking only at adolescents aged 13 to 16 years, when they may be at greatest risk for gang membership, and eliminating 10 cases with excessive missing data, yielded a final German subgroup of 474 individuals. No claim is made as to the representativeness of this sample on any level; at its core it is a convenience sample of children attending school at the three research sites included in the study. However, the researchers selected the three Hauptschulen with the intent of sampling children who were at higher and lower risk of both gang behavior and delinquency.

Bosnia and Herzegovina

BiH is an eastern European nation, previously part of Yugoslavia. In its brief history, this nation of nearly 5 million inhabitants has experienced several major conflicts. About one half of the inhabitants are Bosniak;³ another 37% are Serbian, and the remainder, about 14%, is Croat. BiH is a democratic state; it consists of two entities, the Federation of Bosnia and Herzegovina and the Republica Srpska, and the Brčko district of BiH. The civil law system is used throughout. The primary languages are Bosnian, Serbian, and Croatian.

In 2007, researchers at the University of Sarajevo's Faculty of Criminal Justice Sciences modified the Eurogang instrument slightly, translated the entire Eurogang school survey into the appropriate language, and administered it, again using the Eurogang protocol, to children attending school in five cities throughout BiH.⁴ Twenty schools were visited; within those schools, a total of 58 classrooms at five grade levels were randomly selected. Following a presentation on the purposes of the survey and a review of the students’ right to refuse to participate, more than 98% of the children in attendance completed the 12-page questionnaire.⁵ There is no requirement in BiH for active or passive parental consent, although parents or guardians were informed of the research and the right to withdraw their children from the survey; moreover, after explaining its purpose, all students were afforded the opportunity to opt out of the study, but only a handful exercised this right.

The final BiH sample consisted of 2,206 children; however, 72 respondents were excluded due to excessive missing values. Defining the target sample as school children aged 13 to16 yielded a BiH subgroup of 1,882 individuals. As with the previous two samples, no claims are made as to the representativeness of this sample. At the same time, much like the Dutch sample, the researchers made every effort to include every major population center in the nation, including the cultural, religious, and political variations found in BiH.

Key Variables

This research turned on whether youth claimed membership in one of two types of troublesome youth groups: (1) street gangs (SGs) or (2) other crime-involved youth-oriented social groups, the latter herein designated as DYGs.⁶ Based on the Eurogang definition, a youth was considered to be a street gang member if all of the following conditions existed: (1) the group in which a youth claimed membership had to be “street-oriented”; (2) the group had to have existed at least 3 months or more when the youth was a member; (3) the age range of members had to range between 12 and 25 years of age; (4) illegal activities had to be acceptable to the group; and (5) the group actually had to engage in illegal activities. The present study includes 229 SG members. The BiH sample had the lowest percentage of age-censored children who were SG members (4.7%) or a total of 88 individuals. The German sample contributed 40 SG members to the study; it had the highest percentage of gang members in the study (8.4%), a not entirely surprising fact given the University of Tübingen team purposefully sampled school districts located in geographical areas that allegedly had problems with groups of youth. The sample for the Netherlands consisted of 101 SG members or 6.2% of this age-censored group of Dutch children.

Membership in the second type of troublesome youth group, the DYG, was determined by asking the students if they belonged to an informal group that also engaged in illegal conduct as a group; these groups did not meet the remaining four Eurogang network criteria for a gang. In the BiH sample, there were 83 individuals belonging to a DYG, which was 4.4% of the age-censored sample. Within the German sample, 16.5% (n = 78) claimed membership in a DYG. Finally, the Dutch sample contributed another 126 DYG members, which represented 7.7% of those surveyed.

Table 1 contains a summary of the study’s key variables, as well as significance test comparisons of the “average” responses for the troublesome youth in each sample.⁷ For example, gender is a critical variable in any delinquency study. The total sample was overwhelmingly male (69.2%). The BiH sample had significantly more males (74.9%) than the other two data sets, which more closely resembled each other. The BiH sample aside, the percentages who were female in the German (37.3%) and the Dutch (31.7%) samples were supportive of the emerging view in the United States that youth in troublesome youth groups are far from all-male bastions (Esbensen, Deschenes, & Winfree, 1999; Esbensen et al., 2010; Esbensen & Winfree, 1998). Even in the BiH sample, over one quarter of the troublesome youth group members were females. For the multivariate analyses, males were assigned a value of “1” and females given a value of “0”.

Table 1.

The Variables: Troublesome Youth Group Members by Country and Overall.

	Bosnia–Herzegovina (n = 171)		Germany (n = 118)		Netherlands (n = 227)		Overall (n = 516)
Variables	Mean/%	SD	Mean/%	SD	Mean/%	SD	Mean/%	SD
% Males (0,1)^a,b	74.9%	n.a.	62.7%	n.a.	68.3%	n.a.	69.2%	n.a.
Age: Years (13–16)^a,c	14.80	1.09	14.52	.97	14.74	0.93	14.71	1.00
% Living w/two parents (0,1)^a,b	86.5%	n.a.	82.2%	n.a.	69.2%	n.a.	77.9%	n.a.
% Native (0,1)^a,b	81.3%	n.a.	49.2%	n.a.	68.7%	n.a.	68.4%	n.a.
% Gang member (0,1)^a,c	51.5%	n.a.	33.9%	n.a.	44.5%	n.a	44.4%	n.a.
Parental monitoring (1–5)^b,c	3.44	0.81	3.53	0.82	3.75	0.72	3.60	0.79
School commitment (1–5) ^a,b,c	3.24	0.87	3.51	0.88	2.99	0.92	3.19	0.91
Peer commitment (1–5)^b,c	2.72	1.00	2.81	1.12	4.14	0.95	3.37	1.22
Minor delinquency (1–5)^a,c	2.43	1.15	1.69	0.74	2.23	1.04	2.18	1.06
Property offenses (1–5)^c	1.25	0.59	1.30	0.67	1.16	0.52	1.22	0.58
Personal offenses (1–5)^b,c	1.76	0.88	1.61	0.71	1.41	0.54	1.57	0.73
Group organization (0–5)^c	1.62	1.46	1.33	1.12	1.58	1.04	1.53	1.22
Group offending (0–6)^b,c	2.86	1.88	2.69	1.60	3.63	1.69	3.16	1.78

Note. Significant (p ≤ .05) nation comparisons (using either t tests or chi-square): ^aBiH and Germany, ^bBiH and the Netherlands; ^cGermany and the Netherlands.

The overall mean age was 14.71 years, with few differences by nation of residence. The troublesome youth in the German sample were significantly younger than those in either the BiH or Dutch samples, while the mean age for the latter two was quite similar. The researchers measured the current family structure using several different protocols, represented in this study by a variable called living arrangement. It was possible to collapse the different protocols into two categories: two-parent/adult caregiver households (1) and other living arrangements (0). Nearly 8 in 10 members of the overall sample resided with both parents. Significantly fewer troublesome youth in the Netherlands (69.2%) lived with two adult caregivers than was the case in either BiH (86.5%) or Germany (82.2%); the differences between the latter two were not statistically significant.

Given the study’s emphasis on geography, an important feature of this study was a dichotomy that represented place of birth, a variable called nativity. That is, youth were described as either citizens (i.e., natives) or foreigners. This variable was used for several reasons. First, the crime of a nation’s noncitizen residents has been fodder for criminologists since the 1930s (Sellin, 1938; Van Vechten, 1941). Second, from a policy perspective, Europeans increasingly questioned whether an offender is a citizen or otherwise (Decker, van Gemert, & Pyrooz, 2009; Lynch & Simon, 1999), although such inquiries may mask racist motivations (Ceobanu, 2011). Germany and BiH used the same definition of citizenship: In order to be classified as a citizen, both parents also had to be citizens of the nation in question. The Dutch team used a different definition: To be considered a native, only the mother had to be a Dutch citizen. Over 8 in 10 members of the BiH gang members were natives. Significantly lower percentages of both the German sample (49.2%) and the Dutch sample (68.7%) claimed to be native born than was the case in BiH; however, these latter two figures were not significantly different from one another. While essentially two different definitions were used, it may not be critical to interpreting the data for the simple reason that these are nation-specific legal, social, and cultural definitions of who is or is not a native. Nativity was dummy-coded as follows: Native-born citizens or “natives” were assigned a value of 1, while all other adolescents were given a value of 0.

The surveys included three measures grounded in two criminological theory traditions: differential association/social learning and social control/self-control. Social control and self-control theories contributed two protective measures, starting with parental monitoring, which has close conceptual ties to self-control theory: The greater the level of parental oversight, the less likely the child’s membership in crime-involved youth groups and engagement in delinquency. Four parental monitoring items were included in all three instruments, which formed reasonably reliable scales within each national sample and across the entire study: The higher the scale score, the greater the monitoring of the children by their parents. The Appendix contains the exact items and response sets, along with nation-specific and overall scale scores for all attitudinal measures and behavioral indexes. Significant differences by nation were observed for the mean-level comparisons between the Dutch sample (3.75) and both the BiH (3.44) and Germany (3.53) students, while, once again, the differences between the latter two were not statistically significant.

School commitment was created from a summation of three school-related items included in all three questionnaires. Again, the items and scale scores are found in the Appendix, along with the response categories: The higher the scale score, the greater the commitment to educational goals and school in particular. High commitment to school is viewed as a protective factor, “inoculating” children from crime-involved youth groups and general delinquency. The school commitment levels reported for all three samples were significantly different from one another: The Dutch mean level (2.99) was significantly lower than those reported for either the German (3.51) or the BiH (3.24) youth; at the same time, the German youth had a significantly higher school mean commitment level than did the BiH youth.

Finally, peer commitment, which is viewed as a risk factor, has conceptual ties to differential association and social learning theories. This report employs a measure of peer commitment created from 3 items included in all three questionnaires: The higher the scale score, the greater the commitment to peers. Only two significant differences were noted, both involving the Dutch students; that is, their mean score for peer commitment (4.14) was significantly higher than was observed for school children in either the BiH (2.72) or German (2.81) samples.

The study included three types of self-reported delinquency, which asked how often in the past year the youth had engaged in a series of behaviors (see the Appendix for response categories). Indexes were created by summing the responses to a series of offenses; then this sum was divided by the number of behaviors in the particular index. For example, minor delinquency included three behaviors, ranging from avoiding fees on subways and buses to vandalism (see the Appendix for the exact wording of each behavior). For the minor delinquency index, the mean was 2.18, close to 1–2 times in the past year. The Germans reported a significantly lower mean level (1.69) than did the members of either the BiH (2.43) or Dutch (2.23) samples. Property offenses consisted of three forms of property theft. The overall mean response was 1.22, or closer to never than 1–2 times in the past year. The sole significant difference in the nation-specific property-offense group mean was between the German (1.3) and Dutch adolescents (1.16). The third self-reported measure was personal offenses, which included three types of crime against persons. The overall mean response was 1.57 or midway between never and 1–2 times in the past year. The personal-offense group mean observed for the Dutch youth (1.41) was significantly lower than the same statistic reported in either the BiH (1.76) or German (1.61) samples.

In order to understand the impact of group membership on offending, the researchers asked a series of questions about the youth’s informal social groups that reflected two aspects of those groups: (1) the level of organizational structure, or group organization, present in the group in which a youth claimed membership and (2) the kinds of illegal actions engaged in by that same group, signified as group offending. With respect to the question of organizational structure, the Eurogang questionnaire included five aspects of leadership and organization that the gang literature suggests should be important, which included such things as having leaders, specific rules and initiation rituals (see the Appendix for a list of all aspects). The subjects in all three nations were asked to indicate which of these organizational elements were present in their informal social group. Possible answers ranged from 0 to 5, with 5 indicating the highest level of organization. The mean response was 1.22, suggesting that on average they saw only a modest amount of organizational structure in their respective groups. Only in the case of the mean comparison between the German (1.33) and Dutch (1.58) samples significant differences were noted. Second, all respondents were also asked whether the groups in which they were or had been members engaged in a series of six specific illegal acts that covered the breadth of behavior often attributed to such misbehaving youth groups (see the Appendix for list of the illegal acts). As importantly, the point of reference for this aggregate measure of criminality was the informal social group itself and not the individual group members. The average number of activities attributed overall to the troublesome youth groups was 3.16 or over three of the illegal behaviors in question. The Dutch school children reported that their social groups engaged in significantly higher levels of group offending (3.63) than the youth in either Germany (2.69) or BiH (2.86).

Findings

Predicting Risk and Protective Measures

Table 2 contains the results of nine separate ordinary least squares regression analyses for the protective/risk measures of parental monitoring, school commitment, and peer commitment. In each case, Model 1 consists of the personal-biographical features alone (e.g., gender, age, living arrangement, nativity, and gang membership); Model 2 explores the impact of geography, in this case nation of residence with BiH serving as the excluded category against which Germany and the Netherlands are compared; and Model 3 reveals the overall influence of all seven variables.⁸

Table 2.

Predicting Protective/Risk Measures From Personal-Biographical and Geographical Measures: Bosnia–Herzegovina, Germany, and the Netherlands.

Independent variables	Parental Monitoring			School Commitment			Peer Commitment
Independent variables	Model 1	Model 2	Model 3	Model 1	Model 2	Model 3	Model 1	Model 2	Model 3
Gender	–.051	—	–.051	.026	—	.031	–.010	—	–.010
Age	–.143	—	–.148	–.209	—	–.201	.057	—	.048
Living Arrangement	.046	—	.082	.176	—	.147	–.079	—	–.082
Nativity	–.056	—	–.061	–.104	—	–.086	.009	—	.001
Gang Member	.033	—	–.032	.042	—	–.036	–.045	—	.048
Netherlands	—	.190	.189	—	–.136	–.123	—	.533	.520
Germany	—	.048	.007	—	.107	.063	—	.023	.031
Adjusted R ²	.021	.026	.051	.077	.040	.099	.029	.270	.275

Note. n = 516. Shaded boxes indicate that the unstandardized coefficient’s contribution was significant at or exceeding the .05 probability level.

In terms of parental monitoring, Model 1 reveals that knowing the gender, age, living arrangement and nativity of a youth tells us very little about the level of parental monitoring he or she reported, as only the contribution of age is statistically significant and in this case older adolescents report less parental monitoring than younger ones. For parental monitoring’s Model 2, youth in the Netherlands reported higher levels of monitoring than did BiH adolescents. The same general pattern held for Model 3, which included all seven variables: Youth who were a resident of the Netherlands as opposed to living in BiH and younger reported higher levels of parental monitoring. Given the relatively low levels of explained variance reported for any of the first three equations in Table 2, it seems safe to observe that among members of crime-involved youth groups, we can learn a little about the level of parental monitoring they experience based on the variables considered.

The results for school commitment were more instructive. For instance, in Model 1 older youth, those residing in other than a two-adult family, and citizens all reported lower levels of school commitment. In terms of geography alone (Model 2), when compared to the BiH children, being a resident of the Netherlands translated into lower levels of school commitment, while German school children had higher levels. In Model 3, the influence of age, living arrangements, nativity, and being a Dutch resident made significant contributions to the equation for school commitment, but living in Germany contrasted with living in BiH was no longer important. This full populated regression equation explains nearly 10% of the variance in school commitment.

The results for the analysis of peer commitment—the sole risk factor—represent a third distinct pattern. First, the individual characteristics by themselves (Model 1) revealed next to nothing about peer commitment. However, living in the Netherlands contrasted with living in BiH was a powerful predictor of peer commitment: Dutch children had higher levels of peer commitment. When the influence of all seven independent variables are included (Model 3) for peer commitment, Dutch children rather than those residing in BiH and children living in less “stable” families reported higher peer commitment; this model explains nearly 28% of the variance in peer commitment.

It is worth noting that membership in a Eurogang Network-defined gang has no influence on the level of parental monitoring, school commitment, or peer commitment observed in this study. Being a resident of the Netherlands as opposed to BiH plays a varying but significant role in the prediction of all three attitudinal complexes: higher levels for both parental monitoring and peer commitment and lower levels for school commitment. Otherwise, no one variable plays a consistently significant role across all three dependent variables. Two come close: Being older portends lower levels of both parental monitoring and school commitment, but has no significant role in the equation for peer commitment; being in a traditional two-parent family is tied to higher school commitment and lower peer commitment. Finally, we learn most about peer commitment and least about parental monitoring from the variables included in this analysis.

Predicting Self-Reported Misconduct

Table 3 summarizes the results of 12 regression equations: Self-reported offenses are the dependent variables. Each offense type has four models: Model 1 includes only the personal-biographical information; Model 2 examines the influence of the three attitudinal measures; Model 3 reveals the influence of geography; and Model 4 summarizes the full populated regression equation.

Table 3.

Predicting Self-Reported Misconduct From Personal-Biographical, Geographical, and Protective/Risk measures: Bosnia-Herzegovina, Germany, and the Netherlands.

Independent variables	Minor Delinquency				Property Offenses				Personal Offenses
Independent variables	Model 1	Model 2	Model 3	Model 4	Model 1	Model 2	Model 3	Model 4	Model 1	Model 2	Model 3	Model 4
Gender	.131	—	—	.110	.086	—	—	.089	.197	—	—	.190
Age	.104	—	—	.027	.044	—	—	.012	.043	—	—	–.012
Living Arrangement	–.096	—	—	–.040	–.024	—	—	–.027	.029	—	—	–.044
Nativity	–.013	—	—	–.090	–.058	—	—	–.057	–.022	—	—	.030
Gang Member	.287			.248	.041	—	—	.037	.129	—	—	.106
Parental Monitoring	—	–.163	—	–.138	—	–.140	—	–.112	—	–.247	—	–.182
School Commitment	—	–.177	—	–.151	—	–.088	—	–.114	—	–.055	—	–.117
Peer Pressure	—	.134	—	.128	—	–.066	—	–.118	—	.050	—	.185
Netherlands	—	—	–.075	–.187	—	—	–.095	–.083	—	—	–.232	–.289
Germany	—	—	–.288	–.406	—	—	.045	.059	—	—	–.065	–.025
Adjusted R-square	.122	.093	.064	.225	.006	.034	.011	.046	.057	.068	.040	.169

Note. n = 516. Shaded boxes indicate that the unstandardized coefficient’s contribution was significant at or beyond .05 probability level.

In the case of minor delinquency, four of the five personal-biographical variables (Model 1) explain a respectable 12.2% of the variance, which owes much to being male, somewhat older and an SG member who is living in something other than a two-parent household All three attitudinal measures made significant contributions to Model 2: Lower parental monitoring and school commitment and higher peer pressure are linked to higher minor offending (R ² = .093). Geography is less important in its own right, as Germans compared to youth in BiH have lower levels of minor offending. Model 4 includes all the variables from the previous models. The influences of age and living arrangement, which made significant contributions when considered along with the other personal-biographical variables, disappeared when they were considered simultaneously with the attitudinal and self-reported delinquency measures. Being a male (rather than female) and an SG member (as opposed to a member of DYG) continue to make significant contributions in Model 4. It is also interesting to note that while being a citizen played no role in Model 1’s equation, natives report significantly lower levels of minor offending than non-natives in the fully populated Model 4. Parental monitoring, school commitment, and peer commitment made significant contributions to both Models 2 and 4: Those who reported higher levels of parental monitoring and school commitment engaged in lower levels of self-reported minor delinquency, while youth reporting higher levels of commitment to peers engaged in higher levels of minor delinquency. In Model 4, both the German and Dutch youth reported lower levels of minor delinquency than those living in BiH. Model 4 explains 22.5% of the variance in minor delinquency, much of it coming from geography.

The results for property offenses were different. First, the personal-biographical variables (Model 1) revealed almost nothing about property offending. Second, similar results were observed for the attitudinal measures, except that youth with low levels of parental monitoring committed more property offenses. Geography’s influence (Model 3) was also negligible. In Model 5, we learn that males compared to females and those children with lower levels of both parental monitoring and school commitment report higher levels of property offending; however, the explained variance is rather anemic for this model (R ² = .046). As a general observation, none of the selected variables reveals much at all about the property offending of this sample of troublesome Dutch, BiH, and German youth.

The equations for personal offenses have more in common with those described for minor delinquency than for property offending. Gender, whose influence is seen in the Model 1, continues to play a significant role in Model 4, as males were significantly more likely than females to report higher levels of personal offending. Also, being an SG member (as opposed to being in a DYG) was also associated with higher personal offending. Parental monitoring and school commitment are both important protective factors in the final fully explicated equation (Model 4): Higher parental monitoring and school commitment was associated with lower personal offending levels, whereas, on their own, only parental monitoring made a significant contribution (Model 2). Being Dutch versus being a resident of BiH was associated with lower offending levels in both Models 3 and 4, but living in Germany as opposed to a BiH was unimportant in either equation. Overall, the explained variance (R ² = .169) is midway between that reported for property offenses and minor delinquency.

The most consistent performers across all three types of self-reported delinquency were gender, parental monitoring, and school commitment; three other variables—SG membership, peer commitment, and being a resident of the Netherland—were important in two of the three equations. Higher levels of self-reported delinquency would appear to be associated with who they are (i.e., being male and gang members), where they live (i.e., principally being residents of BiH vs. Netherlands), and their attitudinal complexes (i.e., lower parental monitoring and school commitment and, to a slightly lesser extent, higher peer commitment). The combined models explain most about minor offending, followed by personal and property offenses.

Predicting Group Misconduct

The students’ questionnaires contained a series of inquiries about their group’s illegal behaviors. Specifically, they were asked if their groups engaged in any of the six types of activities, which provided a congregate measure of group offending. Table 4’s Model 1, which includes the influence of the personal-biographical variables, explained 11.6% of the variance: SG members, older adolescents, males, and those individuals living in a home without two adults reported higher levels of group offending than was the case for members of DYGs, younger adolescents, females, and those living with two adults. However, when considered along with all of the other independent variables (Model 6), only age and gang membership from the selected personal-biographical variables continued to contribute to the regression equation in the same fashion.

Table 4.

Predicting Attributed Group Offending: Bosnia–Herzegovina, Germany, and the Netherlands.

Independent Variables	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Gender	.082	—	—	—	—	.017
Age	.173	—	—	—	—	.116
Living Arrangement	–.097	—	—	—	—	–.012
Nativity	.025	—	—	—	—	.042
Gang Member	.256	—	—	—	—	.159
Parental Monitoring	—	–.100	—	—	—	–.024
School Commitment	—	–.242	—	—	—	–.132
Peer Pressure	—	.206	—	—	—	.061
Germany	—	—	–.025	—	—	.111
Netherlands	—	—	.221	—	—	.251
Minor Delinquency	—	—	—	.351	—	.223
Property Crime	—	—	—	.026	—	.046
Personal Crime	—	—	—	.154	—	.185
Group Organization	—	—	—	—	.129	.049
Adjusted R-square	.116	.134	.051	.203	.015	.328

Note. n = 516. Shaded boxes indicate that the unstandardized coefficient’s contribution was significant at or beyond .05 probability level.

The protective and risk factors (Model 2) revealed lower group offending reported by those youth with higher levels of school commitment and parental monitoring. Youth with higher peer commitments were also significantly more likely to report higher levels of group offending. These variables combined to explain 13.4% of the variance. However, only school commitment continued this pattern once all of the other independent variables were included in the equation (Model 6).

Looking at geography’s influence alone (Model 3), only living in Holland versus BiH made a significant contribution, with those living in the former nation reporting that their groups engaged in higher levels of offending than those in the latter. However, the explained variance for this equation was low (5.1%). Interestingly, once all of the variables were included in the regression equation (Model 6), geography’s influence increased, as both German and Dutch youth versus those in BiH reported higher levels of group offending.

Individual offending (Model 4) exhibited a consistent and significant influence on group offending, but only in the cases of minor delinquency and personal crime: Higher levels of personal offending in these categories were related to higher levels of group offending. The explained variance was 20.3%, essentially all of it provided by self-reported minor delinquency and personal offenses. The influence of these two variables was the same in Model 6.

Model 5 examines the influence of the group’s perceived organizational level on group offending. In this instance, the equation that includes only group organization yields an explained variance of only 1.5%. Once this variable is included with the other independent variables in Model 6, group organization no longer plays a significant role. To summarize, Model 6 explained 32.8% of the variance in group-level offending. The geographic measures continue to pay a significant role, as the youth in either Holland or Germany versus those in BiH both report significantly higher levels of group offending. Being older and SG members contrasted with DYG members are both important to our understanding of group offending. Exhibiting low commitment to school also plays a strong and consistent role across both equations that include this variable (Models 2 and 6). Finally, the delinquency of individuals who engage in higher levels of minor delinquency and personal offending also report higher levels of group offending.

Summary, Limitations, and Conclusions

The overriding goal of the current study was to shed light on troublesome youth groups in the European context, a subject about which there is relatively little extant research of a truly comparative nature. Three similar but somewhat different school-based surveys of students in three nations provided the data, while three questions framed the study. The first question asked what we can learn about the level of the putative risk and protective measures reported by members of crime-involved youth groups from their respective personal-biographical characteristics, including nation of current residence. All three risk and protective measures were best understood in terms of the nation of residence. For example, the level of parental monitoring, a protective factor, reported by members of both SGs and DYGs in the Netherlands was significantly higher than that reported by troublesome youth group members in BiH. School commitment, a second protective factor, presented a different picture, but also one related to geography, albeit to a lesser degree than was the case for parental monitoring. That is, crime-group-involved children living in the Netherlands reported significantly lower school commitment than was found in the BiH sample. However, lower school commitment was also much indebted to being younger, living in a family without two adult role models, and being a citizen of that country. Finally, SG and DYG adolescents in the Netherlands reported significantly higher peer pressure than did the BiH youngsters. Children living with both parents reported lower levels of peer pressure. Gender and gang membership played no significant roles in our understanding of any of the protective/risk measures. It is also worth noting that the explained variance figures for the protective measures were rather low, peaking at 10% for school commitment, while the final equation for peer pressure explained a substantial amount of variance, largely attributable to geography (i.e., living in the Netherlands).

The second question was as follows: What can we learn about the level of individual illegal activities reported from an individual’s personal-biographical characteristics, including his or her nation of current residence, and the influence of alleged risk and protective measures? The answer depended on the type of behavior. For example, nearly all of the variables contributed in significant ways to the equation for minor delinquency, while these same variables yielded little insight into property offending. The analyses for personal offending look more like minor delinquency than property offending. In general, it would appear that we know more about youth who engage in minor delinquency and personal offending than those who admit to property offending. Moreover, being an SG member in also far more important for these two offense types. This interpretation is also strengthened by the positive role played by peer commitment for both types of behavior. A consistent factor across all three offense types, however, is the absence of both a commitment to school and parental monitoring. It would appear that the control theorists of the mid-20th century were correct in their emphasis on strong parenting and school engagement as guards against delinquency. Finally, the Dutch school children exhibited lower minor delinquency and personal offending than did those in BiH; residents of Germany also reported lower levels of minor offending than was the case among the sample from BiH. Thus, it seems reasonable to conclude that geography plays both an indirect role in youthful offending through its impact on parental monitoring, school commitment, and peer commitment, and a direct role in its own right.

The third question asked what we can learn about group-level criminal activities reported by the subjects from their demographic features, risk and protective factors, and self-reported illegal activities, along with the structural features attributed to their respective groups. The analyses revealed that the SG and DYG members did not simply transfer their own criminality in wholesale fashion to their groups. The two forms of individual youthful misconduct—minor delinquency and personal crime—did play major roles in predicting the level of attributed group offending; however, geography—as measured by residence in Germany or Holland versus BiH—was also important, as were age, SG membership as opposed to membership in a TYG, and parental monitoring. The group offending attributed to SGs and TYGs is clearly more than the sum of the individual offending levels of their members.

Overall, this study revealed that geography is an important predictor of protective/risk measures, the level of self-reported misconduct, and group offending. Moreover, gender, family situation, and nativity—variables that both popular belief and prior studies suggest impinge on both delinquency and SG behavior—were relatively unimportant when we examined youngsters involved with informal groups that routinely engage in criminal conduct. At the same time, control theory’s school commitment was a consistent performer across all of the equations. To reiterate, Nye (1958), Hirschi (1969), and other control theorists were correct in suggesting that strengthening an individual child’s commitment to school can also serve as a bulwark against group delinquency.

This work is not without its limitations. First, the measures of peer and school commitment were based on conceptually similar but objectively different items. While Weerman and Esbensen (2005; see also Esbensen & Weerman, 2005) also included the same proxy measures with positive results, where they are used in the present study, the results must be interpreted with caution, as several measures exhibited low within-nation reliability scores. Also, several sets of items from the Netherlands, while grounded in the proper context, were different from those used by researchers in Germany and BiH (see the Appendix). Still, the measures of protective and risk factors largely performed as expected, in spite of the fact that the overall sample was homogenous in at least one important respect (i.e., each child was a member of either an SG or TYG). Second, the samples employed for this study were not fully representative of any of the three nations, although the Dutch sample was fairly representative of a region in the Netherlands and the majority of the regions in BiH were found in that nation’s sample. Still, absent more representative samples, the findings respective to each nation must be used with caution. Finally, this is a study of in-school adolescents, as that activity was central to the Eurogang protocol; hence, a valid criticism of this and any school-based survey is that troublesome youth who are absent from school are also absent from the study. There is no way to know if we would achieve the same or even similar results if school dropouts were included in the analyses, a criticism that is often leveled at U.S.-based research that employs school-based surveys. Perhaps future researcher can use a method similar to that employed to capture homeless persons with the goal of gaining insights into free-world SG and TYG members (see, e.g., Herrera & Colletti, 2010).

Even with these limitations, the current study yields several implications for gang theory and gang prevention efforts. First, the protective and risk factors drawn from social learning and social control theories performed largely as their respective theories would predict, even given the sample’s high degree of behavioral homogeneity. Peer commitment, part of the social learning/differential association theory perspective, would appear least flexible among the three measures. As Akers (1985, 1998) might suggest, among the SG- and DYG-involved children, crime-supportive definitions and other discriminative stimuli related to peers influences have done their work and are relatively immutable without the infusion of oppositional definitions and reinforcing stimuli.

Finally, given geographical location’s influence throughout most of the analyses, any gang intervention program or inquiry into gang prevention or desistance that fails to appreciate a nation’s unique sociolegal culture may experience poor results. This qualifying statement owes much to the observed relationship between level of organization and misconduct attributed to the gangs in which sample members claimed membership. The good news for gang desistance programs is, as revealed by Pyrooz and Decker (2011, p. 421), that organized gangs are “no more likely to produce long lasting allegiance to the gang than are less organized gangs.” Their observation, however, does not lessen the importance of the purported relationship between group offending and organization. Programs intended to reduce the organizational structure of an SG or DYG may not be fruitful, as this construct—at least in the context of the adolescents included in this study—exhibited a weak relationship with group offending. Given the relatively low level of organization reported in this study, especially when contrasted with that reported for U.S. gangs (see, e.g., Decker et al., 2008; Ruble & Turner, 2000), this is a factor that may bear closer scrutiny by future European gang researchers.

Footnotes

Appendix

Scale/Index	Items (Including, Where Appropriate, Proxy Items Used in the Netherlands)	Reliability Analysis (Where Applicable)
		BiH	German	Dutch	Total
		α	α	α	α
Parental monitoring	BiH and German version:^a(1) When I go someplace, I leave a note for my parents or call them to tell them where I am going; (2) My parents know where I am when they are not at home; (3) I know how to get in touch with my parents if they are not at home; (4) My parents know who I am with if I am not at home.	.74	.50	.55	.61
Parental monitoring	Dutch version:^b At home, I have to do what my parents say; (2) I know what is and what is not allowed for me at home; (3) My parents know where I go to outside home; (4) My parents allow me everything.	.74	.50	.55	.61
School commitment	BiH and German version:^a (1) I try hard in school; (2) Grades are very important to me; (3) I usually finish my homework.	.73	.58	.70	.64
School commitment	Dutch version:^b (1) I pay attention in class; (2) I work hard to get good grades; (3) I do as much as I can for school.	.73	.58	.70	.64
Peer commitment	BiH and German version:^c (1) If your group of friends was getting into trouble at home, how likely is it that you would still hang out with them? (2) If your group of friends was getting into trouble at school, how likely is it that you would still hang out with them? (3) If your group of friends was getting into trouble with the police, how likely is it that you would still hang out with them?	.68	.78	.66	.79
Peer commitment	Dutch version:^d (1) My friends sometimes make me do things I actually don’t want to do; (2) My friends would think it’s stupid when I don’t dare to do something; (3) My friends laugh at me when I am afraid of something.
Minor delinquency	Version for all:^e (1) Avoided paying fees at buses or subways (fare dodging); (2) Damaged or destroyed property; (3) Spray painted a building or wall (illegal graffiti).	n.a.	n.a.	n.a.	n.a.
Property offenses	Version for all:^e (1) Stole or tried to steal something worth more than (equivalent to U.S.$50); (2) Entered a building or other structure to steal something; (3) Stole or tried to steal a motor vehicle theft.	n.a.	n.a.	n.a.	n.a.
Personal offenses	Version for all:^e (1) Hit someone with the intention of hurting them; (2) Attacked someone with a weapon; (3) Used a weapon or fore in order to get money from person.	n.a.	n.a.	n.a.	n.a.
Group organization	Version for all: (1) Recognized leaders; (2) Symbols; (3) Regular meetings; (4) Specific rules; (5) Initiation rituals.	n.a.	n.a.	n.a.	n.a.
Group offending	Version for all: (1) Steal things; (2) Rob other people; (3) Steal cars; (4) Use alcohol; (5) Use drugs; (6) Damage or destroy property.	n.a.	n.a.	n.a.	n.a.

^aResponse set for this item was coded as follows: (1) strongly disagree, (2) disagree, (3) neither agree nor disagree, (4) agree, and (5) strongly agree. ^bResponse set for this item was coded as follows: (1) totally disagree, (2) disagree, (3) neither agree nor disagree, (4) agree, and (5) totally agree. ^cResponse set for this item was codes as follows: (1) would not hang out, (2) not likely to hang out, (3) somewhat likely to hang out, (4) very likely to hang out, and (5) very likely to hang out. ^dResponse set for this item was coded as follows: (1) totally agree, (2) agree, (3) neither agree nor disagree, (4) disagree, and (5) totally disagree. ^eResponse set for this index was coded as follows: (1) never; (2) 1–2 times, (3) 3–5 times, (4) 6–10 times, and (5) more than 10 times.

Acknowledgment

The author acknowledges the data collection efforts of the following individuals: Muhamed Budimlic, Almir Maljevic, Eldan Mujanovic, and Elmedin Muratbegovic (University of Sarajevo, Sarajevo, Bosnia and Herzegovina); Klaus Bott, Hans-Juergen Kerner, Kerstin Reich, and Elmar G. M. Weitekamp (University of Tübingen, Tübingen, Germany); and Frank Weerman (Netherlands Institute for the Study of Crime and Law Enforcement, Leiden, Netherlands). The author also acknowledges the comments and suggestions of T. J. Taylor, Dean Dabney, Finn Esbensen, Cheryl Maxson, and the anonymous reviewers. Any mistakes in interpretation or fact are those of the author.

Author’s Note

An earlier version was a plenary session presentation for Eurogang XI held at Grundtvigs Højskole, Hillerød, Denmark, September 22–24, 2011.

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 author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported in part by research award 1704-257 from the International Center, National Institute of Justice, Office of Justice Programs, U.S. Department of Justice. Points of view expressed are the author's and do not necessarily reflect the official position of the U.S. Department of Justice.

Notes

References

Akers

R. L.

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

Akers

R. L.

(1985). Deviant behavior: A social learning approach (3rd ed.). Belmont, CA: Wadsworth.

Aldridge

Medina

Ralphs

(2012). Counting gangs: Conceptual and validity problems with the Eurogang definition. In Esbensen

F.-A.

Maxson

C. L.

(Eds.), Youth gangs in international perspective (pp. 35–52). New York, NY: Springer.

Battin-Pearson, S. R., Thornberry, T. P., Hawkins, J. D., & Krohn, M. D. (1998). Gang membership, delinquent peers, and delinquent behavior. Bulletin. Youth Gang Series. Washington, DC: U.S. Department of Justice, Office of Juvenile Justice and Delinquency Prevention.

Bennett

Holloway

(2004). Gang membership, drugs, crime in the UK, British Journal of Criminology, 44, 305–323.

Bjerregaard

. (2002). Self-definitions of gang membership and involvement in delinquent activities. Youth & Society, 34, 31–54.

Ceobanu

A. M.

(2011). Usual suspects? Public views about immigrants’ impact on crime in European countries. International Journal of Comparative Sociology, 52, 114–131.

Cohen

(1980). Folk devils and moral panics: The creation of mods and rockers (2nd ed.). New York, NY: St. Martin’s Press.

Curry

G. D.

Decker

S. H.

(2002). Confronting gangs: Crime and community. New York, NY: Oxford Press.

10.

Decker, S. H., & Curry G. D. (2002). Gangs, gang homicides, and gang loyalty: Organized crimes or disorganized criminals. Journal of Criminal Justice, 30, 343–352.

11.

Decker

S. H.

Katz

Webb

V. J.

(2008). Understanding the black box of gang organization: Implications for involvement in violent crime, drug sales, and violent victimization. Crime and Delinquency, 54, 153–172.

12.

Decker

S. H.

van Gemert

Pyrooz

(2009). Gangs, migration and crime: The changing landscape in Europe and the United States. Journal of International Migration and Immigration, 10, 393–408.

13.

Decker

S. H.

Weerman

(2005). European street gangs and troublesome youth groups. Walnut Creek, CA: Altamira Press.

14.

Duffy

M. P.

Gillig

S. E.

(2004). Teen gangs: A global view. Westport, CT: Greenwood.

15.

Esbensen

F.-A.

(2004). Evaluating G.R.E.A.T.: A School-Based gang prevention program. NIJ: Research for Policy. Washington, DC: U.S. Department of Justice.

16.

Esbensen

F.-A.

Deschenes

E. P.

Winfree

L. T.

Jr (1999). Differences between gang girls and gang boys: Results from a multi-site survey. Youth and Society, 31, 27–53.

17.

Esbensen

F.-A.

Peterson

Taylor

T. J.

Freng

(2010). Youth violence: Sex and race differences in offending, victimization, and gang membership. Philadelphia, PA: Temple University Press.

18.

Esbensen

F.-A.

Weerman

F. W.

(2005). Youth gangs and troublesome youth groups in the United States and the Netherlands. European Journal of Criminology, 2, 5–37.

19.

Esbensen

F.-A.

Winfree

Jr.

L. T. Jr . (1998). Gender and race differences between gang and nongang youths: Results from a multisite survey. Justice Quarterly, 15, 505–525.

20.

Esbensen

F.-A.

Winfree

L. T.

Jr He

Taylor

T. J.

(2001). Youth gangs and definitional issues: When is a gang a gang, and why does it matter? Crime & Delinquency, 47, 105–130.

21.

Gatti

Haymoz

Schadee

H. M. A.

(2011). Deviant youth groups in 30 countries: Results from the Second International Self-Report Delinquency study. International Criminal Justice Review, 2, 208–224.

22.

Gottfredson

M. R.

Hirschi

(1990). A general theory of crime. Stanford, CA: Stanford University Press.

23.

Herrera

Colletti

. (2010). The prevalence of smoking among two homeless populations in Southern California. Los Angeles, CA: Urban Initiatives. Retrieved January 9, 2012, from http://www.urban-initiatives.org/images/PDFs/PrevalenceOfSmoking.pdf

24.

Hirschi

(1969). Causes of delinquency. Berkley, CA: University of California Press.

25.

Howell

D. C

. (2011). Treatment of missing data. Retrieved July 8, 2011, from http://www.uvm.edu/~dhowell/StatPages/More_Stuff/Missing.html

26.

Howell

J. C

. (2007). Menacing or mimicking? Realities of youth gangs. Juvenile and Family Court Journal, 58, 39–49.

27.

Junger-Tas

Marshall

Enzmann

Kilias

Steketee

Gruszczynska

(Eds.). (2010). Juvenile delinquency in Europe and Beyond: Results of Second International Self-Report Delinquency Study. Berlin, Germany: Springer.

28.

Junger-Tas

Marshall

Enzmann

Kilias

Steketee

Gruszczynska

(2012). The many faces of youth crime: Contrasting theoretical perspectives on juvenile delinquency across countries and cultures. New York, NY: Springer.

29.

Klein

M. W.

(2005). The value of comparisons in street gang research. Journal of Contemporary Criminal Justice, 21, 135–205.

30.

Klein

M. W.

Maxson

C. L.

(2006). Street gang patterns and policies. New York, NY: Oxford University Press.

31.

Klein

M. W.

Kerner

H.-J.

Maxson

C. L.

Weitekamp

E. G. M.

(2001). The Eurogang paradox: Street gangs and youth groups in the U.S. and Europe. Dordrecht, The Netherlands: Kluwer.

32.

Lynch

J. P.

Simon

R. J.

(1999). A comparative assessment of criminal involvement among immigrants and natives across seven nations. International Criminal Justice Review, 9, 1–17.

33.

Maxson

C. L.

Klein

M. W.

(1995). Investigating gang structures. Journal of Gang Research, 3, 33–40.

34.

Medina

Aldridge

Shute

Ross

. (2013). Measuring gang membership in England and Wales: A latent class analysis with Eurogang survey questions. European Journal of Criminology, 10, 591–605.

35.

Nash

J. K.

Mujanovic

Winfree

L. T.

Jr (2011). Protective effects of parental monitoring on offending in victimized youth in Bosnia and Herzegovina. Child and Youth Services. 32, 224–242.

36.

Nye, F. I. (1958). Family-risk relationships and delinquent behavior. New York, NY: Wiley.

37.

Pearton

R. E.

Gaskell

(1981). Youth and social conflict: Sport and spectator violence. International Review for Sociology of Sport, 16, 57–67.

38.

Posick

(2013). The overlap between offending and victimization among adolescents: Results from the Second International Self-Report Delinquency study. Journal of Contemporary Criminal Justice, 29, 106–124.

39.

Pyrooz

D. C.

Decker

S. M.

(2011). Motives and methods for leaving the gang: Understanding the process of gang desistance. Journal of Criminal Justice, 39, 417–425.

40.

Ruble

Turner

W. L.

(2000). A systematic analysis of the dynamics and organization of urban street gangs. The American Journal of Family Therapy, 28, 117–132.

41.

Sellin

(1938). Culture conflict and crime. New York, NY: Social Science Research Council.

42.

Shaw, C. R., & McKay, H. D. (1942). Juvenile delinquency and urban areas. Chicago, IL: University of Chicago Press.

43.

St. Cyr

J. L

. (2003). The folk devils react: Gangs and moral panic. Criminal Justice Review, 28, 26–46.

44.

Thornberry

T. P.

Huizinga

Loeber

(2004). The causes and correlates studies: Findings and policy implications. Juvenile Justice Journal, 9, 3–19.

45.

Thrasher, F. M. (1927). The gang: A study of 1,313 gangs in Chicago. Chicago, IL: University of Chicago Press.

46.

Van Vechten

C. C.

(1941). The criminality of the foreign born. Journal of Criminal Law and Criminology, 32, 139–147.

47.

Webb

V. J.

Ren

Zhao

J. S.

N. P.

Haen

Marshall, I.

(2011). A comparative study of youth gangs in China and the United States: Definition, offending, and victimization. International Criminal Justice Review, 21, 225–242.

48.

Weerman

F. M.

Esbensen

F.-A.

(2005). A cross-national comparison of gangs: The Netherlands and the United States. In Decker

Scott H.

Weerman

Frank W.

(Eds.), European street gangs and troublesome youth groups (pp. 219–255). Walnut Creek, CA: Altamira Press.

49.

Weerman

F. M.

Maxson

C. L.

Esbensen

F.-E.

Aldridge

Medina

van

Gemert. F.

(2009). Eurogang program manual: Background, development, and use of the Eurogang instruments in multi-site, multi-method comparative research. St. Louis, MO: Department of Criminology and Criminal Justice, University of Missouri, St. Louis.

50.

Winfree

L. T.

Jr (2012). A comparative theoretical examination of troublesome adolescents in Germany and Bosnia-Herzegovina. Journal of Contemporary Criminal Justice, 28, 406–425.