Assessing Youth Motivations for Joining a Peer Group as Risk Factors for Delinquent and Gang Behavior

Abstract

This article expands upon the limited literature on the relationship between youth motivations for joining a peer group and delinquent behavior. Using network survey data from 200 youth who self-identified as group members, we conducted factor analysis to develop measures of motivation to join a group, and then describe these motivations and the differences between delinquent and nondelinquent group members. Using negative binomial and binary logistic regression models, we examined the relationship between these motivations and delinquent behavior. The results indicated that youth who join a group for instrumental purposes exhibit more delinquent behavior than those who do not, while joining a group for reasons associated with filling a void and/or belonging has a weaker relationship to delinquency. The findings suggested that certain motivations for joining groups could serve as important risk factors that could be targeted by agencies and organizations seeking to prevent youth delinquent and gang behavior.

Keywords

peer influences risk factors gang activity delinquency

Introduction

Research from across the social and behavioral sciences has long agreed that peers and peer relations are developmentally significant in the lives of youth and young adults (see Sussman, Pokhrel, Ashmore, & Brown, 2007). A large body of criminological research has established that one’s delinquent peer association is a key influence on one’s own delinquency. Regardless of whether the mechanisms of delinquent peer association are rooted in selection (where delinquent youth choose other delinquent youth as peers) or influence (where delinquent peers influence nondelinquent peers to become delinquent), research has shown that a youth’s developmental trajectory may be explained by his or her peer group (Crosnoe & McNeely, 2008). However, few studies have examined the reasons why youth join groups, and how these reasons might be associated with delinquency and violence.

There are both tangible and perceived benefits for youth in peer groups—peer groups can offer youth friendship and emotional and social support, as well as an escape from other aspects of their lives (Crosnoe, 2000). These factors can have both prosocial and antisocial influences on youth, depending on the circumstances; some youth may seek out prosocial peers as a way to avoid negative influences in their homes, while others may associate with antisocial peers despite a positive home environment (Schwartz, Dodge, Pettit, & Bates, 2000). Furthermore, research has shown that once in a group, whether via selection or influence, youth are likely to conform to the behaviors of peers of that group (Brown, Eicher, & Petrie, 1986). Research findings on peer influence are consistent across different types of negative or antisocial behaviors including aggression (Farmer et al., 2003), smoking, alcohol, and other drug abuse (La Greca, Prinstein, & Fetter, 2001; Sussman, et al., 1999; Urberg, Degirmencioglu, Tolson, & Halliday-Scher, 2000), and delinquency (Demuth, 2004; Tolone & Tieman, 1990; Urberg et al., 2000).

With regard to delinquency, the influence of delinquent peers on individual delinquent activity remains one of the strongest and most consistent findings in the youth offending literature (Battin-Pearson, Thornberry, Hawkins, & Krohn, 1998; Haynie, 2002, 2001; Haynie & Osgood, 2005; Lahey, Loeber, Stouthamer-Loeber, & Farrington, 1999; Loeber & Dishion, 1987; Maxson, Whitlock, & Klein, 1998; Shoemaker, 2005; Warr, 2002). Within this literature, many research efforts have focused on understanding the mechanisms that affect this relationship, mainly whether youth select negative peers or are simply influenced by delinquent peers once they become part of a group. Despite this focus, few studies have examined self-reported motivations for joining a group and whether motivations for joining a group are associated with delinquent behavior. More specifically, criminological research has yet to examine the influence of these motivations in the context of other risk factors for delinquency. Understanding the importance of motivations for joining a group relative to other risk factors for delinquency (e.g., parental involvement, school attachment, etc.) could aid policy makers who want to prevent delinquent activity among youth. If particular types of motivations for joining a group are strong predictors of delinquent behavior, identifying those motivations could aid resource targeting or service provision, and more importantly, reduce the risk of delinquency by directly addressing certain motivations for group joining.

This article, part of a larger multisite social network study of the influence of norms and networks on youth delinquency (Cahill, Lowry, & Roman, 2011; Roman et al., 2012), focuses on the motivations that youth have to join a group of friends—defined as people they hang out with or see on a regular basis. Prior to discussing the current study and explicitly examining the relationship between a youth’s motivation to join a group and his or her delinquent behavior, we review the literature related to motivation to join peer groups—in particular, gangs and other delinquent groups.

Background Literature

Although the literature dissecting motivations for a joining peer groups is limited, the majority of studies suggest that an adolescent enters a peer group because she or he is attracted to the norms, values, behaviors, and activities of a particular group (Crosnoe & McNeely, 2008). The attraction to a group may be related to homophily in that individuals will befriend people like themselves; research by Cairns and Cairns (1994) suggests that friendship ties in youth are most likely to form when two people have similar characteristics, such as age, sex, race, and social class. This is related to a youth’s desire to easily coordinate needs and activities. It is also related to propinquity or nearness (Warr, 2002). For example, having friends with similar interests can help relieve boredom, and researchers have suggested that the very nature of boredom can be relieved by the presence of others (Csikszentmihalyi & Larson, 1984). Other research suggests that some youth choose to associate with peers who have complementary skills so that each youth can bring something to the group that is of benefit to other individuals; this helps youth create mutually dependent relationships that can help sustain their own social positions (Farmer, Xie, Cairns, & Hutchins, 2007).

Some youth are motivated to join a peer group simply because they already have friends who are members of that peer group (Farmer et al., 2003; Ryan, 2001). Research has also shown that certain components of peer groups, such as the amount of time youth spend together, is relevant to understanding how youth form peer groups and the group’s relationship to individuals’ delinquent behavior (Greene & Banerjee, 2008; Haynie & Osgood, 2005; Osgood, Wilson, O’Malley, Bachman, & Johnson, 1996). These circumstances may account for why youth choose to join a peer group (e.g., to fill up empty time). Routine activities for youth often include a large portion of unstructured and unsupervised time during which they are simply hanging out with other youth (Felson & Boba, 2010; Osgood et al., 1996). During this time, they are learning and/or reinforcing different types of behavior—whether prosocial, deviant, or criminal (Akers, 1985; Dishion, 2000; Osgood et al., 1996). Other youth may be manipulated into joining groups; Warr (1996) found that there are group leaders (called instigators) who recruit weaker individuals, initiate delinquent behavior, and motivate others to join the group.

Research on gangs has also provided insight into peer group formation. Studies that have examined the motivations for joining a gang suggest that youth perceive many personal benefits to being part of a gang. These benefits include: prestige and status (Baccaglini, 1993; Moore, 1991, 1978) protection (Baccaglini, 1993; Thornberry, Krohn, Lizotte, & Chard-Wierschem, 1993), excitement (Miller, 1958; Pennell, Evans, Melton, & Hinson, 1994), avoidance of conflict with parents and family (Moore 1991, 1978), and making money through illegal activities such as selling drugs (Decker & Van Winkle, 1996; Venkatesh, 1999). Youth may also join gangs when their friends and family (specifically siblings and cousins) are already members of the gang (Hill, Howell, Hawkins, & Battin-Pearson, 1999).

In summary, youth motivations to join a group vary widely, but for the most part appear to fall in categories related to establishing a group-based identity based on similarities, access to particular activities, for prestige or protection, and to fill some type of a void. Youth may also be motivated to join a peer group because they are coerced to or feel pressured to join. The research on group formation and gang involvement suggests that youth may be specifically selecting the peers who would offer the most benefits—participation in exciting and illegal activities. However, the link between these motivations, particularly instrumental motivations like participating in illegal activities and delinquent behavior requires further exploration.

This article seeks to draw upon these prior research findings about motivations for joining peer groups and risk factors for delinquency, and look specifically at the relevance of motivations for joining a group to understand this relationship. We build upon the earlier studies on the link between motivations for joining a peer group and delinquency by asking specific questions about motivations to join a group that relate both to risk factors for joining gangs or delinquent groups and to prosocial peer group influences. Our work is not restricted to gang-related behavior, nor is the study restricted by only examining those who join delinquent groups. Rather, the study ties together and expands upon the limited literature on youth motivations for joining a group and the influence of those motivations on delinquency in general.

Current Study

The network survey data described in this article are derived from two social network studies of youth living in high-risk neighborhoods (Cahill et al., 2011; Roman et al., 2012). For these studies, a social network framework was used to understand how the patterns of relationships among youth in two neighborhoods, one predominantly Latino and the other predominantly Black, influence delinquency and violence. The data collected provide a unique opportunity to examine the motivations for joining a group while controlling for an individual’s personal network characteristics and typical risk factors for delinquency. Using data from these network studies, we examine the following research questions in this article: Do the motivations for joining a group predict individual delinquent behavior? Are youth who join groups for prosocial reasons more likely to refrain from delinquent behavior? How important are the different reasons for joining a group relative to other risk factors for delinquency? These questions are important because it may be simpler for youth-serving agencies to ascertain or query youth on their motivations for joining a group than to undertake complex risk assessments or ask youth to report on their own criminal behavior or gang involvement.

To answer these questions, we first take a descriptive look at the common motivations for young people (delinquent and nondelinquent) to join groups. We then examine whether significant differences exist between group and nongroup members in our sample, and between youth who self-identify as being in a group and as being delinquent compared to youth who are in a group but are not delinquent. We then investigate whether the motivations for joining a group predict delinquent behavior, controlling for other risk factors (e.g., parental involvement, delinquent peer association, etc.). The article concludes with a discussion of the research and policy implications of this study, while also addressing the limitations of the current study. The primary limitation is that our data are cross-sectional, and while we use the word predict to describe the results from our regression analyses, we cannot ascertain causal ordering in relation to the respondents’ motivations to join a group and their delinquent behavior.

Research Site and Sample

The design and methodology of the larger study were based on the theoretical and analytical framework of the network approach and involved: (1) selecting appropriate neighborhoods—based on presumed neighborhood cohesion (i.e., closeness and shared value among neighbors) and level of risk for delinquency—to serve as the boundaries to approximate a complete social network of youth (based on geography) and (2) attempting to survey all youth between the ages of 14 and 21 living in the neighborhoods. To determine the specific boundaries of the study areas that would approximate a neighborhood as defined by residents and key informants, we conducted field research that included focus groups, informal interviews, neighborhood ride alongs with representatives from community groups, and examined administrative and census data. The research team partnered with a community-based organization in each site to recruit eligible study participants, and administered the survey at a community center located in each target neighborhood. Though the goal of the study was to conduct a census of eligible youth in each neighborhood (estimated by Census and parcel data), we estimate that we were only able to survey 36% of eligible youth in Site 1 (Montgomery County, MD) and 50% of eligible youth in Site 2 (Washington, DC).

The survey included three parts: (1) questions about the respondent, including whether the respondent identified as being in a peer group (respondents who identified as being a member of group were asked a series of questions related to their reasons for joining the group); (2) the listing of 20 “alters” (individuals named by respondents—they could be friends, parents, siblings, or anyone else the youth knows) followed by a set of questions asked about every alter; and (3) a question that is used to create the structural links among members of the network (do Alter 1 and Alter 2 know each other or hang out regularly?) This approach was designed to capture the networks of youth that comprise people who are important to the youth and might influence the youth—either positively or negatively. However, we do not know if any, some, or all of the youth in the respondent’s personal network are also members of the respondent’s self-identified group.

The full sample for this study included all 303 eligible respondents who completed the survey in both sites—Site 1 (Montgomery County, MD; N = 147) and Site 2 (Washington, DC; N = 156). While prior research suggests that there may be important differences between neighborhood groups and the nature of delinquent and gang activity across neighborhoods because of racial–ethnic factors (Decker, Bynum & Weisel, 1998), prior analyses using these survey data suggest that the underlying structural processes related to social networks are similar (Cahill et al., 2011; Roman et al., 2012). Therefore, we combine the sites' data and control for potential differences by including a dummy site variable and a race or ethnicity variable (Black and Hispanic) in our models.

This article focuses on a subset of the sample—respondents who self-identified as being members of a group (N = 200, or 66% of our total sample).¹ In Site 1, 75.5% (N = 111) of respondents identify as being members of a peer group, while in Site 2 only 57.1% (N = 89) do. The average age of respondents who identify as group members is 17 years old; 55% of these respondents are male, 49% are Black, and 47% are Hispanic–Latino(a). Nineteen percent of respondents are foreign born and 54.5% of respondents have at least one parent who was foreign born. Of respondents in a group who are not born in the United States, the average time spent in the United States is 7.6 years.

Measures

Dependent Variables

Dependent variables consist of two measures. First, an additive scale which has been used in previous criminological research (Sweeten, 2011) measures the variety of the respondent’s recent delinquency (i.e., delinquent and criminal behavior over the last 6 months) according to 9 items: (1) avoided paying for things, like a movie, taking bus rides, or anything else; (2) tried to steal or actually stole money or things worth $100 or less; (3) damaged, destroyed, or marked up someone else’s property on purpose; (4) tried to steal or actually stole money or things worth more than $100; (5) tried to steal or actually stole a car or other motor vehicle; (6) was involved in a gang fight; (7) sold illegal drugs such as marijuana, crack, heroine, or methamphetamine; (8) used a weapon or force to get money or things from people; and (9) attacked someone with a weapon or with the idea of seriously hurting or killing them. The recent delinquency scale values range from 0 to 9; scale reliability is good (Cronbach’s α = .802). Second, we assess the respondent’s gang activity by their binary responses (yes = 1) to one of three items: if he or she has ever been a member of a street gang, if he or she considers his or her group to be a gang, or if he or she has ever been in a gang fight. During the focus groups with neighborhood youth at the start of our study, the youth indicated they may be more inclined to admit that they had been in a gang fight as opposed to explicitly identifying as a gang member. Because of the input from the focus group, and because previous studies have used involvement in a gang fight as an indicator of gang membership (e.g., Vaughn, Howard, & Harper-Chang, 2006), we collapse the indicators into one measure representing gang involvement. The descriptive statistics for all dependent variables (and the subcomponents of the gang measure) are shown in Table 1.

Table 1.

Descriptive Statistics for Dependent Variables (N = 200).

	Minimum	Maximum	Mean	SD
Recent Delinquency scale	0	9	0.85	1.645
Gang involvement measures	(%)
Respondent identifies as a member of a street gang	9.5
Respondent been in a gang fight	15.0
Respondent been in a gang fight OR identifies as a member of a street gang	18.0
Respondent been in a gang fight but NOT in gang	6.5

Independent Variables

Group membership

Respondents self-identified as being a member of a group by answering yes to the question, Is there a group of friends that you hang around with a lot? Respondents who identified as being a member of a group were asked 24 questions that were originally designed by Maxson and Whitlock (2004). For each group question, respondents were asked to respond yes or no based on their answer to the question Did you select your group of friends . . . The questions are shown in Table 2 below in the order they were asked in the survey:

Table 2.

Motivations for Joining a Group.

To make friends
To get a reputation
To fill up empty time
To feel important
For support and loyalty
To be noticed
To feel like you belong to something
To avoid home
To keep out of trouble
For protection
For excitement
To get away with illegal activities
To participate in group activities
To have a territory of your own
To get your parents’ respect
Because someone in your family was in the group
To meet guys or girls easily
Because the group is one you can feel proud of
To get money or other things
Because a friend was in the group
To get what you don’t get from family
To be with other people like you
Because the things they do are cool
Because I did not have a choice

We include the 24 group motivation variables in a principal components analysis (with a varimax rotation) to identify separate domains within the set of group motivation items. Following Hedderson (1987), we exclude eight variables from the principal components analysis that have either low correlations with all factors or high correlations to multiple factors (making it difficult to classify the variable with a single factor). From the principal components analysis, we identify three main domains (factors) of motivation to join a group: “filling a void,” “instrumental purposes,” and “belonging.” The rotated component matrix shown in Table 3 below provides the factor loadings for each of the 16 items included in the confirmatory factor analysis. The three extracted factors account for just under 50% of the variance in the included items (49.4%). Instead of directly using factor scores, which are hard to interpret when included in subsequent analyses, we create scale scores to describe these three domains by summing the underlying items associated with each of those factors. The reliability coefficients for each of the three scale scores measuring each domain are α = .78, .71, and .70, respectively.

Table 3.

Rotated Component Matrix.

Group Motivation Item	Component
Group Motivation Item	Fill a Void	Instrumental Purposes	Belonging
To be noticed	0.749	—	—
To feel important	0.722	—	—
To feel like you belong to something	0.686	—	—
To avoid home	0.653	—	—
To fill up empty time	0.64	—	—
To have a territory of your own	0.484	—	—
To make friends	0.405	—	—
Because someone in your family was in the group	—	0.764	—
Because I did not have a choice	—	0.75	—
To get money or other things	—	0.673	—
For protection	—	0.607	—
To be with other people like you	—	—	0.758
Because the group is one you can feel proud of	—	—	0.666
For excitement	—	—	0.652
Because the things they do are cool	—	—	0.649
To participate in group activities	—	—	0.568

Note. Extraction method: principal component analysis.

Rotation method: Varimax with Kaiser Normalization (blank coefficients < .4).

Personal network compositional variables

Our analyses also include variables that provide descriptive information about alter characteristics in aggregate form—these variables describe the composition of an individual’s social network: delinquent alters (a count of the number of alters who the respondent reported as being delinquent²); same-neighborhood alters (a count of the number of alters who live in the same neighborhood as the respondent, as reported by the respondent); the number of alters the respondent would go to for advice; and a homophily Index (the number of alters who were the same gender, race or ethnicity, and nationality, according to the respondent’s report).³ Additionally, the EgoNet software creates a measure of the number of components (or, subgroups) in the respondent’s personal network.⁴

Attachment measures and prosocial influences

We use four measures of cultural attachment and prosocial influences in our analyses. The variable “number of years the respondent has lived in the United States” measures the respondent’s level of assimilation or integration into United States culture. To measure family cohesion, we use the 10-item cohesion subscale of the Family Relationship Characteristics scale (Tolan, Gorman-Smith, Huesmann, & Zelli, 1997) and 1 item from another family scale (Maxson & Whitlock, 2004).⁵ The 11-item scale has high internal consistency (α = .89). Finally, two binary variables (yes = 1) measure support for education in the respondent’s family: parental support for education (whether a parent or guardian regularly insists on school attendance and performance) and educational levels within the family (whether parents or guardians graduated from high school).

Controls. The following control variables are included in the analyses: age, gender (male = 1), race or ethnicity (Black, Hispanic⁶, other), and the number of years lived at current address.

Analysis Plan

To accomplish our research objectives, we employ (1) descriptive analyses to assess the motivations for joining a group and the differences between respondents who are in a group and not in group in terms of delinquency and (2) regression analyses to determine if the motivations for joining a group predict delinquent behavior and gang involvement, controlling for additional predictor measures. Using negative binomial regression techniques, we predict the extent of involvement in delinquent behaviors by modeling delinquency as a scale, given a respondent’s motivations for joining a group, his or her network characteristics, and the presence of important risk and protective factors identified in the criminological literature. We then model the likelihood that a youth would be involved in gang activity using binary logistic regression. To determine the appropriate regression model for our scaled dependent variable, we first examined its distribution and skew (see Table 1 for the descriptive statistics for the recent delinquency scale).

The skew and zero-inflated nature of our scaled outcome (count variable) suggest the data violate the normality assumption of ordinary least squares, and that a negative binomial regression model (designed to handle many zero values and high skewness) is most appropriate for our count variable. Comparing the goodness of fit between the negative binomial regression model and the Poisson regression model confirms this assumption; for the recent delinquency outcome variable, the Akaike Information Criterion (AIC), and log-likelihood values are lower for the negative binomial models.⁷

Results

Table 4 shows the percentage of youth who joined a group for each of the 24 reasons provided in the survey.⁸ The table is organized according to the measures of group motivation in addition to eight other items that do not load (i.e., are not correlated with) a single factor. Most youth (70%) report joining a group to belong—particularly “to be with other people like (themselves).” This suggests that youth are choosing to be part of homophilous groups (those made up of individuals who are similar to one another). For respondents who are motivated to join a group to fill a void, the most common reasons are to make friends (46.5%), to fill up empty time (33%), and to feel like you belong to something (28%). These reasons relate to accessing something that is missing in a youth’s life, but some other reasons (e.g., to have a territory of your own) might be more directly linked to delinquent activity or gang involvement. Only 4% of youth feel explicitly coerced to join a group (i.e., they do not have a choice) but 21% are more subtly coerced (i.e., they need protection and turn to a peer group to get it).

Table 4.

Did You Select Your Group of Friends.

	Respondents (%; N = 200)
Fill a void
To make friends	46.5
To fill up empty time	33.0
To feel like you belong to something	28.0
To avoid home	25.0
To feel important	23.5
To be noticed	21.5
To have a territory of your own	15.0
Instrumental purposes
For protection	21.0
To get money or other things	10.0
Because someone in your family was in the group	8.0
Because I did not have a choice	4.0
Belonging
To be with other people like you	70.0
For excitement	65.5
To participate in group activities	52.5
Because the things they do are cool	44.5
Because the group is one you can feel proud of	40.5
Other
For support and loyalty	59.5
To keep out of trouble	51.0
To meet guys or girls easily	34.5
Because a friend was in the group	21.5
To get what you don’t get from family	20.0
To get away with illegal activities	18.0
To get a reputation	15.0
To get your parents’ respect	13.5

Table 5 compares the characteristics of respondents in a group to those not in a group, as well as the characteristics of nondelinquent respondents in a group to delinquent respondents in a group. Respondents in a group report exhibiting significantly more delinquent activity over the last 6 months (M = .85 vs. M = .36), and significantly more respondents in a group report having been involved in gang activity (18.0% vs. 7.8%). The percent of respondents in a group who are Black and Hispanic are roughly equal, while over two thirds of respondents who are not in a group are Black and one third are Hispanic (because more youth in Montgomery County, MD report being in a group than respondents in Washington, DC). However, aside from these measures, there are no significant differences between respondents in a group and respondents not in a group.

Table 5.

Characteristics of Respondents.

	Percentage in a group (N = 200)	Percentage not in a group (N = 103)
Recent delinquency (mean)	0.85*	0.36*
In a gang or gang fight	18.00*	7.80*
Age (mean)	17.23	17.58
Male	55.00	56.30
Black	47.00*	64.10*
Hispanic	49.00*	33.00*
Time at address (mean)	7.78	9.34
Parent support in school	96.50	91.30
Family member graduated high school	79.00	76.70
Family cohesion (mean)	45.28	45.65
Years in US (mean)	15.00	14.42
Number of delinquent alters (mean)	1.71*	0.68*
Number of friends as alters (mean)	12.92	12.03
Number of alters who live in neighborhood (mean)	7.55	6.90
Number of alters and respondents who goes for advice (mean)	10.08	7.81
Homophily index (mean)	43.10	43.66
Number of components (mean)	1.08	1.00
	Percentage in a group, delinquent (N = 86)	Percentage in a group, not delinquent (N = 114)
Age (mean)	16.81*	17.54*
Male	67.44*	45.61*
Black	54.65*	41.23*
Hispanic	43.02*	53.51*
Time at address (mean)	7.84	7.74
Parent support in school	96.51	96.49
Family member graduated high school	79.07	78.95
Family cohesion (mean)	45.76	44.63
Years in US (mean)	14.60	15.52
Number of delinquent alters (mean)	3.26*	0.54*
Number of friends as alters (mean)	4.74*	4.87*
Number of alters who live in neighborhood (mean)	8.17	7.07
Number of alters and respondents who goes for advice (mean)	10.24*	11.91*
Homophily index (mean)	10.62	9.71
Number of components (mean)	0.98*	1.15*

Note. *p < .05.

Comparing delinquent and nondelinquent respondents in a group, we see that nondelinquent youth who are in a group are significantly older (M = 17.54) than delinquent youth in a group (M = 16.81). Delinquent youth in groups are more likely to be male (67.4% vs. 45.6%) and Black (54.7% vs. 41.23%), while nondelinquent group members are significantly more likely to be Hispanic (43% vs. 53.5%). Delinquent group members also have significantly more delinquent alters (M = 3.26 vs. M = .54), fewer alters they go to for advice (M = 10.24 vs. M = 11.91) and fewer subgroups in their personal networks (M = .98 vs. M = 1.15).

Regression Models

The tables below provide the results of negative binomial regression analysis for the scaled dependent variable (recent delinquent activity over the last 6 months) and binary logistic regression results for one categorical dependent variable (ever been in a gang fight or in a gang). Each table shows four models for each dependent variable—one model for each of the motivational scales, included separately, and one model that includes all three motivation scales. The tables provide the exponentiated coefficient multiplied by the standard deviation of the predictor variable—exp(b × SD)—and the standard error for each independent variable in each model.⁹ The fourth model, which has the lowest log likelihood value, shows the results when all three measures of group motivation are included in the same model.¹⁰

Table 6 provides the results of the models predicting respondents’ delinquent activity in the last 6 months. Each of the group motivation scales is a significant and positive predictor of delinquent activity when included in separate models. The instrumental scale (Model 2) has the highest exp(b × SD) value of the three motivational scales. The instrumental purposes scale in Model 2 also has a higher exp(b × SD) value than the number of delinquent alters, meaning that a single standard deviation increase on the instrumental scale represents a 1.273 unit change in the count of recent delinquency, while a single standard deviation increase in the number of delinquent alters represents a 1.246 unit change in the count of recent delinquency. This is an indication that joining groups for instrumental purposes—whether explicit or implicit—may be a stronger determinant of delinquent behavior than associating with delinquent peers. In Model 4, however, which includes all three group motivation scales, none of the scales is significant. This result suggests that the shared variance between the scales controls for the significance of their effects.

Table 6.

Negative Binomial Regression—Recent Delinquency (N = 199).

	Model 1		Model 2		Model 3		Model 4
	Exp (b × SD)	SE	Exp (b × SD)	SE	Exp (b × SD)	SE	Exp (b × SD)	SE
Group scales
Fill a void	1.219*	0.048	—	—	—	—	1.066	0.067
Instrumental purposes	—	—	1.273*	0.114	—	—	1.194	0.153
Belonging	—	—	—	—	1.208†	0.060	1.087	0.078
Personal network variables
Number of delinquent alters	1.283*	0.030	1.246*	0.029	1.321**	0.032	1.251*	0.029
Number of alters who live in same neighborhood	1.173	0.023	1.185	0.023	1.170	0.024	1.190	0.023
Number of alters who goes for advice (not delinquent)	0.860	0.026	0.868	0.025	0.838	0.027	0.853	0.027
Number of components	0.885	0.174	0.889	0.177	0.888	0.175	0.891	0.177
Homophily index	0.938	0.015	0.896	0.015	0.900	0.016	0.899	0.016
Controls
Number of years lived in US	1.285	0.032	1.244	0.032	1.257	0.033	1.256	0.032
Age	0.768*	0.050	0.766*	0.049	0.753*	0.052	0.764*	0.048
Male	1.441**	0.236	1.399**	0.234	1.497**	0.241	1.431**	0.242
Black	1.033	0.440	1.048	0.428	1.103	0.412	1.036	0.428
Latino	1.037	0.494	1.008	0.471	1.073	0.466	1.012	0.471
Family member completed high school	1.077	0.320	1.032	0.321	1.050	0.330	1.034	0.321
Parent–school encouragement	0.999	0.473	1.099	0.524	1.012	0.490	1.096	0.538
Family cohesion	0.804†	0.021	0.781	0.022	0.789	0.021	0.788	0.022
Number of years at address	1.294†	0.021	1.262	0.021	1.294†	0.020	1.262	0.021
Survey site	1.137	0.305	1.150	0.273	1.152	0.303	1.122	0.288
Goodness of fit
Log likelihood	−214.424		−213.357		−214.616		−213.126
Deviance value / df	0.958		0.950		0.940		0.956

Note. †p < .10, *p < .05, ** p < .01, ***p < .001.

In all four models, age is a significant negative predictor and being male is a significant positive predictor; the only inconsistency across models is that the number of years spent at the same address is a significant positive predictor only in Models 1 and 3 (although this was approaching significance at p ≤ .105 in Models 2 and 4).¹¹

Table 7 shows the results of the binary logistic regression models predicting a respondent’s gang activity (if she or he has ever been in either a gang fight or a gang). Unlike the results of the negative binomial regression models predicting recent delinquency, only the instrumental scale (Model 2) is a significant positive predictor of gang activity in the individual models; neither the fill a void nor the belonging scales is significant (Models 1 and 3, respectively). In Model 4, the instrumental scale remains significant when all three group motivation measures are included. While the gang activity models are not consistent with the recent delinquency models in terms of the significance of the “fill a void” and “belonging” group motivation measures, “instrumental purposes” emerged as an important factor in the models for both types of outcomes.

Table 7.

Binary Logistic Regression—In a Gang or Gang Fight (N = 199).

	Model 1		Model 2		Model 3		Model 4
	Exp (b × SD)	SE	Exp (b × SD)	SE	Exp (b × SD)	SE	Exp (b × SD)	SE
Group scales
Fill a void	1.359	0.114	—	—	—	—	1.154	0.136
Instrumental purposes	—	—	1.815**	0.275	—	—	1.963*	0.330
Belonging	—	—	—	—	0.910	0.138	0.695	0.157
Personal network variables
Number of delinquent alters	2.602***	0.084	2.404**	0.084	2.789***	0.086	2.367**	0.087
Number of alters who live in same neighborhood	1.083	0.054	1.174	0.056	1.053	0.054	1.166	0.057
Number of alters who goes for advice (not delinquent)	1.219	0.058	1.289	0.060	1.293	0.058	1.401	0.061
Number of components	0.682	0.407	0.695	0.414	0.680	0.403	0.722	0.419
Homophily index	0.395**	0.032	0.366**	0.033	0.397**	0.032	0.386**	0.033
Controls
Number of years lived in United States	3.205*	0.089	3.110**	0.091	3.065*	0.089	3.022*	0.091
Age	0.503*	0.138	0.530*	0.138	0.509*	0.139	0.543*	0.136
Male	1.15	0.514	1.048	0.529	1.177	0.517	1.038	0.538
Black	3.703	2.464	3.388	2.405	4.139	2.591	3.433	2.412
Latino	2.636	2.438	2.780	2.379	3.241	2.558	2.858	2.390
Family member completed high school	0.887	0.566	0.830	0.571	0.871	0.563	0.836	0.584
Parent–school encouragement	0.967	1.515	1.161	1.554	0.977	1.488	1.134	1.558
Family cohesion	0.531*	0.042	0.493*	0.043	0.505*	0.041	0.484*	0.044
Number of years at address	1.141	0.039	1.097	0.039	1.183	0.038	1.109	0.040
Survey site	0.906	0.657	0.937	0.654	0.985	0.641	0.916	0.663
Cox R ²	0.267		0.282		0.261		0.290

Note. †p < .10. *p < .05. **p < .01. ***p < .001.

Another difference from the recent delinquency models is that being male is not significant in determining gang activity and the homophily index is a significant and negative predictor of gang activity. Other results are consistent across binary logistic Models 1–4, and are consistent with the recent delinquency models; the number of delinquent alters and the number of years lived in the United States are significant positive predictors, while age and family cohesion are significant negative predictors.

Discussion and Conclusions

This article sought to fill an important gap in the delinquency and peer association literature by identifying which group-joining motivations relate to one other, how those motivations may vary for delinquent and nondelinquent group members, and how those motivations fit into the broader delinquency risk factor framework. The results of the regression models show that instrumental reasons are the strongest motivation-related predictor of recent delinquent activity. The instrumental purposes scale is significant in both models, and for the gang activity model it remains a significant predictor even when controlling for other motivations for joining a group (see binary logistic Model 4 results). For Model 2 predicting recent delinquency, the instrumental purposes scale is a stronger predictor than the respondent’s number of delinquent alters, which research has shown is one of the strongest predictors of delinquency (Battin-Pearson et al., 1998; Haynie, 2002, 2001; Haynie & Osgood, 2005; Shoemaker, 2005; Lahey et al., 1999; Loeber & Dishion, 1987; Maxson et al., 1998; Warr, 2002). These findings suggest that instrumental purposes should be further investigated as a potential indicator of youth delinquency.

While the instrumental purposes scale is a strong positive predictor of being in a gang and/or gang fight, we recognize that it is possible that a youth joined a group without knowing there were delinquent group members or that group members engaged in delinquent acts together, especially because one of the key items that make up the “instrumental purposes” factor is joining a group for protection (see Table 4). In other words, it is possible that youth do not seek a group that will engage in delinquent acts, but feel pressure to join for protection, and after joining they engage in delinquent acts through affiliation with delinquent peers and opportunities to offend.

The other measures of motivation for joining a group (to fill a void and for belonging) are not as strong or consistent predictors of delinquency as instrumental purposes. In Model 1 for recent delinquency, the “fill a void” scale is a positive significant predictor, but in Model 1 for being in a gang or gang fight it is not. In Model 3 for recent delinquency, the belonging scale is a positive predictor of delinquent activity, but it only approaches significance at p < .10. However, in binary logistic Model 3, it is not significant. This suggests, and longitudinal data would need to confirm, that instrumental purposes may drive delinquent behavior, regardless of other motivations youth have for joining a group.

The descriptive analyses also offer some context for interpreting these results. In our sample, youth in groups are more delinquent than youth who are not in groups, supporting prior research that youth in groups are at greater risk for exhibiting delinquent activity (Adler & Adler, 1995; Warr, 1996, 2002). Our descriptive analyses also show that delinquent group members are significantly younger and not as close to their alters as nondelinquent group members (see Table 5). Even though we do not know whether the alters the respondents named in the personal networks are the same people who are in their self-identified group, the evidence that delinquent group members are less close to their alters supports the finding that youth who join a group for instrumental purposes exhibit more delinquent activity.

To understand the results of the regression models and the influence of the group motivation scales on delinquent activity, it is important to revisit the results of the factor analysis and examine which items are not included in the group motivation measures. The factor analysis identified three factors measuring separate types of motivations for joining a group. However, eight specific motivations were not highly correlated with one specific factor. Some of these motivations, including to get what you don’t get from family and to get your parents’ respect, could be prosocial or delinquent motivations. This would depend on whether the respondent’s family members are delinquent, as prior research indicates that negative parent involvement can have a negative impact on youth and positive parent involvement can have a positive impact on youth (Durbin, Darling, Steinberg, & Brown, 1993; Farrington, 1978; McCord, McCord, & Howard, 1963; Patterson, 1986; Warr 2002). The unknown family characteristics may explain why these items do not load with (i.e., are not highly correlated to) the other “fill a void” measures, which are more explicitly about joining a group to avoid home and have space of your own.

Other group motivation items that do not load onto factors, specifically “to get a reputation” and “to get away with illegal activities,” were identified in prior studies as reasons for joining a gang (Baccaglini, 1993; Decker & Van Winkle, 1994). These items do not load onto the instrumental purposes scale—which we anticipated they would—possibly because very few youth in our sample identified as being gang members. We also anticipated that the four other motivations (for support and loyalty, to meet guys or girls easily, because a friend was in the group, and to keep out of trouble) would be part of the belonging scale, but the items do not group together in that factor.¹²

Though there are meaningful findings that emerged from this article, there are limitations to our research that must be considered in interpreting them. First, our findings may have limited generalizability, given that the study sampled a relatively small number (N = 303) of youth from a predominantly Black neighborhood in the District of Columbia and a predominantly Hispanic neighborhood in Montgomery County, Maryland. Related to the sample, we set out to survey all youth between certain ages living in both neighborhoods, but estimate we surveyed roughly half, leaving the possibility that the surveyed sample is not representative of all youth in the neighborhood.

Additionally, there are limitations that stem from our broader data collection strategy and use of network survey data. We cannot confirm the level of delinquency or criminal behavior of alters in a respondent’s personal network because we do not have self-report data from all alters listed. An alter’s delinquency is reported by the respondent who named him or her, and therefore measures the respondent’s perception of each alter’s delinquency, and not necessarily the alter’s actual delinquent behavior.

Relatedly, under-reporting bias may be potential problem in this study. Although the self-report method has generally proven to be valid and reliable for youth who report their own delinquent and criminal behavior (Mulvey, 2011; Thornberry & Krohn, 2000), given the nature of our network questions (naming friends and important relationships and then describing their behavior), our respondents may have purposely under-reported the criminal behavior of people important to them. This may or may not have extended to the group motivation questions, where youth may have under-reported being coerced into joining a group of which they are still part.

The cross-sectional nature of the data also poses limitations, particularly with regard to the group motivation questions. Although, we do know that the youth who answered the group-joining motivation questions were in a group at the time they were surveyed, we did not ask when the respondent joined the group. Therefore, we do not know if the respondents’ delinquent activity predated their group membership, or if they only started exhibiting delinquent behavior after they joined the group. This may help account for some inconsistencies between the recent delinquency and gang activity models; although the series of recent delinquency questions assesses behavior in the last 6 months, the gang involvement questions measure gang activity over one’s lifetime. Therefore, we still cannot ascertain causal ordering due to the fact that we do not know if the youth joined the group more or less than 6 months prior to being surveyed.

Future research should address the causal ordering issues present in this study by either asking specific questions about the time frame for joining a group in a cross-sectional study or through a longitudinal study that could capture movement in and out of groups and test whether motivations predate group joining. It is also important to focus on how motivations for joining a group could help practitioners and programmers identify and target high-risk youth. Regardless of the data limitations, our findings may be useful for social service providers, who may find it useful to ask about motivations for joining a peer group as a separate set of questions or as part of a risk assessment protocol. Validated risk or need assessment tools such as the Youth Level of Service or Case Management Inventory (YLSI–CMI) and the Structured Assessment of Violence Risk in Youth (SAVRY) have shown to be accurate predictors of juvenile recidivism, but may not accurately reflect youth offending patterns (Welsh, Schmidt, McKinnon, Chattha, & Meyers, 2008). Measures of group-joining motivations may prove useful for assessing youth risk of offending, considering how nonintrusive these motivation questions are and how revealing they may be. Our results suggest that with regard to delinquency, instrumental motivations for joining a group may be a stronger predictor than having delinquent peers. In addition to being less intrusive, asking a youth about their motivations for joining a group may be more likely to elicit a truthful response than asking youth to report on their friends’ behavior. Although we asked these questions about youth who were already part of a group, these questions may have particular salience for younger adolescents who might not yet be part of high-risk groups or for immigrant youth who have not yet become part of local peer group. Service providers working in the youth prevention field could integrate questions about group motivations into existing assessments and compare these questions with the other items on the assessment to assess their power of predicting youth offending. Should these items have predictive power, they could serve as flags for providers to identify high-risk youth to target for service provision.

Footnotes

Acknowledgments

The authors wish to thank Janine Zweig and Julie Samuels, Senior Fellows at the Urban Institute, and Nancy La Vigne, Director of the Urban Institute’s Justice Policy Center, for their review of this article. We also thank the two anonymous reviewers for their thoughtful comments on the first submission of this article.

Authors’ Note

Points of view or opinions in this document are those of the authors and do not represent the official position or policies of the Urban Institute, Temple University, the United States Department of Justice, Office of Justice Programs, Office of Juvenile Justice and Delinquency Prevention, or the District of Columbia Executive Office of the Mayor.

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 article uses data collected for two Urban Institute and Temple University projects, one funded by the Office of Juvenile Justice and Delinquency Prevention, Office of Justice Programs, U.S. Department of Justice (Grant No. 2007-MU-FX-0001), and the other funded by the Justice Grants Administration, Executive Office of the Mayor, District of Columbia (Grant No. 2009-JAGR-1114).

Notes

References

Adler

P. A.

Adler

(1995). Dynamics of inclusion and exclusion in preadolescent cliques. Social Psychology Quarterly, 58, 145–162.

Akers

R. L.

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

Baccaglini

W. F.

(1993). Project youth gang-drug prevention: A statewide research study. Rensselaer, NY: New York State Division for Youth.

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. Retrieved from http://www.ncjrs.gov/pdffiles/171119.pdf.

Brown

B. B.

Eicher

S. A.

Petrie

(1986). The importance of peer group (“crowd”) affiliation in adolescence. Journal of Adolescence, 9, 73–96.

Cahill

Lowry

S. S.

Roman

C. G.

(2011). Social networks and the behaviors of youth in the district of columbia: An interim research report. Washington, DC: The Urban Institute.

Cairns

R. B.

Cairns

B. D.

(1994). Lifelines and risks: Pathways of youth in our time. New York, NY: Harvester Wheatsheaf.

Crosnoe

(2000). Friendships in childhood and adolescence: The life course and new directions. Social Psychology Quarterly, 63, 377–391.

Crosnoe

McNeely

C. A.

(2008). Peer relations, adolescent behavior, and public health research and practice. Family & Community Health, 31, 71–80.

10.

Csikszentmihalyi

Larson

(1984). Being adolescent: Conflict and growth in the teenage years. New York, NY: Basic Books.

11.

Decker

S. H.

Bynum

Weisel

(1998). Tale of two cities: Gangs as organized crime groups. Justice Quarterly, 15, 395–425.

12.

Decker

S. H.

Van Winkle

(1994). Slinging dope: The role of gangs and gang members in drug sales. Justice Quarterly, 11, 583–604.

13.

Decker

S. H.

Van Winkle

B. V.

(1996). Life in the gang: Family, friends, and violence. New York, NY: Cambridge University Press.

14.

Demuth

(2004). Understanding the delinquency and social relationships of loners. Youth & Society, 35, 366–392.

15.

Dishion

T. J.

(2000). Cross-setting consistency in early adolescent psychopathology: Deviant friendships and problem behavior sequelae. Journal of Personality, 68, 1109–1127.

16.

Durbin

D. L.

Darling

Steinberg

Brown

B. B.

(1993). Parenting style and peer group membership among European-American adolescents. Journal of Research on Adolescence, 3, 87–100.

17.

Farmer

T. W.

Estell

D. B.

Leung

M. C.

Trott

Bishop

Cairns

B. D.

(2003). Individual characteristics, early adolescent peer affiliations, and school dropout: An examination of aggressive and popular group types. Journal of School Psychology, 41, 217−232.

18.

Farmer

T. W.

Xie

Cairns

B. D.

Hutchins

B. C.

(2007). Social synchrony, peer networks, and aggression in school. In Hawley

P. H.

Little

T. D.

Rodkin

P. C.

(Eds.), Aggression and adaptation: The bright side to bad behavior. Mahwah, NJ: Lawrence Erlbaum.

19.

Farrington

D. P.

(1978). The family background of aggressive youths. In Hersov

L. A.

Berger

(Eds.), Aggression and anti-social behaviour in childhood and adolescence. New York, NY: Pergamon Pressork.

20.

Felson

Boba

(2010). Crime and everyday life. (4th ed.). Thousand Oakes, CA. Sage.

21.

Greene

Banerjee

S. C.

(2008). Adolescents’ responses to peer smoking offers: The role of sensation seeking and self-esteem. Journal of Health Communication: International Perspectives, 13, 267–286.

22.

Haynie

D. L.

(2001). Delinquent peers revisited: A network approach for understanding adolescent delinquency. American Journal of Sociology, 106, 1013–1057.

23.

Haynie

D. L.

(2002). Friendship networks and delinquency: The relative nature of peer delinquency. Journal of Quantitative Criminology, 18, 99–134.

24.

Haynie

D. L.

Osgood

D. W.

(2005). Reconsidering peers and delinquency: How do peers matter? Social Forces, 84, 1109–1130.

25.

Hedderson

(1987). SPSS made simple. Belmont, CA: Wadsworth.

26.

Hill

K. G.

Howell

J. C.

Hawkins

J. D.

Battin-Pearson

S. R.

(1999). Childhood risk factors for adolescent gang membership: Results from the seattle social development project. Journal of Research in Crime and Delinquency, 36, 300–322.

27.

Krackhardt

(1999). The ties that torture: Simmelian tie analysis in organizations. Research in the Sociology of Organizations, 16, 183–210.

28.

Krohn

M. D.

(1986). The web of conformity: A network approach to the explanation of delinquent behavior. Social Problems, 33, 601–613.

29.

La Greca

A. M.

Prinstein

M. J.

Fetter

M. D.

(2001). Adolescent peer crowd affiliation: Linkages with health-risk behaviors and close friendships. Journal of Pediatric Psychology, 26, 131−143.

30.

Lahey

B. B.

Loeber

R. A.

Stouthamer-Loeber

Farrington

D. P.

(1999). Boys who join gangs: A prospective study of predictors of first gang entry. Journal of Abnormal Child Psychology, 27, 261–276.

31.

Loeber

Dishion

T. J.

(1987). Anti-social and delinquent youths: Methods for their early identification. In Burchard

J. D.

Burchard

(Eds.), Prevention of delinquent behavior. Newbury Park, CA: Sage.

32.

Maxson

C. L.

Whitlock

M. L.

(2004). School-based protection of youth at-risk for joining gangs. Report to U.S. Department of Education: Institute of Education Sciences.

33.

Maxson

C. L.

Whitlock

M. L.

Klein

M. K.

(1998). Vulnerability to street gang membership: Implications for prevention. Social Service Review, 72, 70–91.

34.

McCord

Howard

(1963). Family interaction as antecedent to the direction of male aggressiveness. Journal of Abnormal and Social Psychology, 67, 239–242.

35.

Miller

W. B.

(1958). Lower class culture as a generating milieu of gang delinquency. Journal of Social Issues, 14, 5–19.

36.

Moore

J. W.

(1978). Homeboys: Gangs, drugs and prison in the barrios of Los Angeles. Philadelphia, PA: Temple University Press.

37.

Moore

J. W.

(1991). Going down to the barrio: Homeboys and homegirls in change. Philadelphia, PA: Temple University Press.

38.

Mulvey

(2011). Highlights from pathways to desistance: A longitudinal study of serious adolescent offenders. Washington, DC: U.S. Department of Justice, Office of Justice Programs, Office of Juvenile Justice and Delinquency Prevention.

39.

Osgood

D. W.

Wilson

J. K.

O’Malley

P. M.

Bachman

J. G.

Johnston

L. D.

(1996). Routine activities and individual deviant behavior. American Sociological Review, 61, 635–55.

40.

Patterson

G. R.

(1986). Performance models for anti-social boys. American Psychologist, 41, 432–444.

41.

Pennell

Evans

Melton

Hinson

(1994). Down for the set: Describing and defining gangs in San Diego. San Diego, CA: Criminal Justice Research Division, Association of Governments.

42.

Phinney

J. S.

(1992). The multigroup ethnic identity measure: A new scale for use with diverse groups. Journal of Adolescent Research, 7, 156–176.

43.

Roman

C. G.

Cahill

Lachman

Lowry

S. S.

Orosco

McCarthy

(2012). Social Networks, Delinquency, and Gang Membership: Using a Neighborhood Framework to Examine the Influence of Network Composition and Structure in a Latino Community. Washington, DC: The Urban Institute.

44.

Ryan

A. M.

(2001). The peer group as a context for the development of young adolescent motivation and achievement. Child Development, 72, 1135–1150.

45.

Schwartz

Dodge

K. A.

Pettit

G. S.

Bates

J. E.

(2000). Friendship as a moderating factor in the pathway between early harsh home environment and later victimization in the peer group. Developmental Psychology, 36, 646–662.

46.

Shoemaker

D. J.

(2005). Theories of delinquency: An examination of explanations of delinquent behavior (5th ed.). New York, NY: Oxford.

47.

Sussman

Pokhrel

Ashmore

R. D.

Brown

B. B.

(2007). Adolescent peer group identification and characteristics: A review of the literature. Addictive Behaviors, 32, 1602–1627.

48.

Sussman

Simon

T. R.

Stacy

A. W.

Dent

C. W.

Ritt

Kipke

M. D.

Flay

B. R.

(1999). The association of group self-identification and adolescent drug use in three samples varying in risk. Journal of Applied Social Psychology, 29, 1555−1581.

49.

Sweeten

(2011). Scaling Criminal Offending. Journal of Quantitative Criminology, 28 (3), 533–557.

50.

Thornberry

T. P.

Krohn

M. D.

(2000). The Self-Report Method for Measuring Delinquency and Crime. Criminal Justice, 4, 33–83.

51.

Thornberry

T. P.

Krohn

M. D.

Lizotte

A. J.

Chard-Wierschem

(1993). The role of juvenile gangs in facilitating delinquent behavior. Journal of Research in Crime and Delinquency, 30, 55–87.

52.

Tolan

P. H.

Gorman-Smith

Huesmann

Zelli

(1997). Assessment of family relationship characteristics: A measure to explain risk for anti-social behavior and depression among urban youth. Psychological Assessment, 9, 212–223.

53.

Tolone

W. L.

Tieman

C. R.

(1990) Drugs, delinquency, and “nerds”: Are loners deviant? Journal of Drug Education, 20, 153–162.

54.

Urberg

K. A.

Degirmencioglu

S. M.

Tolson

J. M.

Halliday-Scher

(2000). Adolescent social crowds: Measurement and relationship to friendships. Journal of Adolescence Research, 15, 427−445.

55.

Vaughn

M. G.

Howard

M. O.

Harper-Chang

(2006). Do prior trauma and victimization predict weapon carrying among delinquent youth? Youth Violence and Juvenile Justice, 4, 314–327.

56.

Venkatesh

(1999). Looking at crime from the street level. Washington, DC: National Institute of Justice Research Forum.

57.

Warr

(1996). Organization and Instigation in Delinquent Groups. Criminology, 34(1), 11–37.

58.

Warr

(2002). Companions in crime. New York, NY: Cambridge University Press.

59.

Welsh

J. L.

Schmidt

McKinnon, Chattha

H. K.

Meyers

J. R.

(2008). A comparative study of adolescent risk assessment instruments. Assessment, 15, 104–115.