School Transitions,Peer Processes,and Delinquency: A Social Network Approach to Turning Points in Adolescence

Abstract

Objectives: We examine how normative school transitions (e.g., moves from elementary to middle school) shape adolescents’ experiences with three network processes that inform delinquency: delinquent popularity, delinquent sociability, and friend selection on shared delinquency participation. Methods: By applying stochastic actor-oriented models to a sample of panel data on 13,752 students from 26 school districts in the PROSPER study, we compare outcomes for students who change schools between 6th and 7th grade to those who remain in the same building. Results: We find that adolescents who transition schools between these grades have significantly different experiences with delinquency-related network processes when compared to their peers who do not make this change. For instance, in schools that merge students from multiple elementary schools to a single middle school, delinquent youth experience a reduction in their popularity and sociability following the school transition. These declines do not characterize the social experiences of delinquent adolescents who do not change schools during this period. Conclusions: Our findings suggest that school districts can organize transition patterns to provide youth a chance to sever harmful connections, start anew, and reduce their participation in delinquency.

Keywords

peer networks youth crime school transitions turning points SAOMs

During adolescence, individuals face various turning points or life transitions that impact behavioral trajectories in meaningful ways, including their involvement in crime and delinquency (Laub & Sampson, 1993, 2009). One exceptionally common, yet understudied, transition that adolescents experience is the normative, structural change from lower- to higher-level schools, such as the move from elementary to middle school. These anticipated school transitions are crucial events for young people because they impact various academic, social, and health outcomes (Benner & Graham, 2009; Felmlee et al., 2018; Grigg, 2012; Schwerdt & West, 2013). A growing body of work argues that school transitions should be conceptualized as turning points that reduce youth involvement in deviant and criminal behaviors during both the immediate and long term (Carson et al., 2017; Freelin et al., 2023). However, we currently know little about why normative school moves discourage individuals from participating in delinquency. By focusing on the transition from elementary to middle school, the current project examines whether social network processes that relate to youth crime can help explain the link between normative school transitions and participation in delinquent behavior.

Young people do not experience significant life transitions in isolation from their peers. Instead, the potential consequences of various turning points are informed by one's position in their peer network, and these life changes can restructure relational patterns in ways that either protect or expose youth to sources of harmful influence (Sutherland, 1939; Warr, 1998). Given the well-documented link between adolescent delinquency and school peers (e.g., Haynie, 2001), there is particular promise in applying a social network approach to evaluate why changing from lower- to higher-level schools can impact delinquent behaviors. For example, school transitions may lead newly reformed adolescents to knife off connections to antisocial peers, discourage delinquent youth from maintaining amicable ties with prosocial classmates, or restructure patterns of network clustering on shared delinquency participation. Understanding how school transitions shape delinquency-related peer processes is crucial because these network phenomena impact the ways that harmful behaviors diffuse through populations, as well as the establishment of group norms.

The present study adopts a network perspective to reconsider why normative school changes carry important implications for individuals’ delinquency trajectories. Using data from the PROSPER study on 51 longitudinal, grade-cohort networks of adolescents, we evaluate whether there are variations in three delinquency-related friendship selection processes over time and across transition structures. More specifically, we consider whether delinquent youth that transition to higher-level schools between 6th and 7th grade are less popular, less social, and more likely to cluster together when compared to deviant peers who do not experience these school changes. Our results highlight the peer-related mechanisms that inform the link between school transitions and delinquency. We also discuss how the network changes that accompany these transitions can further disadvantage delinquent youth and impact their criminal trajectories over the life span.

School Transitions, Peers, and Delinquency

Perspectives from life course theory define life transitions as changes in one's state or role (Elder, 1998), such as moving from unemployed to employed or from gang-affiliated to not affiliated. Such transitions can provide opportunities for people to alter their individual behaviors in either beneficial or harmful ways (Elder et al., 2003; Laub & Sampson, 1993, 2009). Many life transitions occur during adolescence, and one change that is particularly common is the normative, structural move from lower- to higher-level schools embedded in the same geographic area (Benner & Graham, 2009; Langenkamp, 2010). These anticipated school changes are studied extensively outside of criminology, with research typically finding detrimental effects on various outcomes of interest. For example, experiencing a normative school shift tends to lower academic achievement after the transition occurs (Blyth et al., 1983; Grigg, 2012; Seidman et al., 1994). School transitions are also associated with unfavorable social consequences (Pribesh & Downey, 1999), such as changes in who youth perceive as their best friends (Aikins et al., 2005). Students who transition end up with smaller friendship networks (Hardy et al., 2002), particularly after multiple feeder schools merge into a single, higher-level school (Temkin et al., 2018).

When life transitions represent a substantial redirection in one's behavioral trajectory, these changes can be conceptualized as turning points (Elder et al., 2003). Although some turning points lead to harmful outcomes, others afford opportunities to start anew and adjust one's involvement in crime and delinquency (Laub & Sampson, 1993, 2009; Sampson & Laub, 2005; Sweeten et al., 2013). Criminological research is beginning to recognize that normative school transitions can be conceptualized as positive turning points that reduce delinquent behavior by restructuring youth routines and introducing them to new classes, extracurricular activities, and school personnel. For example, Freelin and colleagues (2023) find that students who transition to high school in nonmetropolitan areas are less likely to participate in violence and property crime when compared to same-age students who do not move schools. Similarly, Carson and colleagues (2017) uncover evidence that the contextual changes associated with school transitions lead to decreased gang involvement for at-risk youth in urban settings. However, other research finds that school transitions are not significantly associated with one's participation in property crime (Weiss & Bearman, 2007).

Conceptualizing normative school transitions as a turning point represents a promising line of inquiry in criminology because these school moves occur at a life stage when levels of delinquency change for most individuals (Hirschi & Gottfredson, 1983). Specifically, population-level delinquency increases in early adolescence, peaks around age 15, and then decreases considerably by the mid-to-late 20s (Farrington, 2003; Hirschi & Gottfredson, 1983). As such, transitions to middle and high school occur at key ages that may affect delinquency trajectories. Furthermore, normative school transitions may be less prone to issues of selection bias than other life transitions such as securing paid employment in high school (Staff et al., 2010) or residential changes (Gasper et al. 2010). This is because the timing of transitions from lower- to higher-level schools is compulsory and imposed upon all students in a district, regardless of their prior delinquency or other risk factors. Understanding whether normative school transitions affect criminal behavior can, therefore, provide unique insight into ways to disrupt criminal trajectories when levels of delinquent involvement change rapidly for many young people.

Although previous work considers the impact of normative school transitions on patterns of delinquency, it remains unclear how these changes shape the social network mechanisms that inform crime and other antisocial behaviors. The life course concept of linked lives embeds individuals in an interdependent network of relational ties that shapes their experiences throughout their life span (Elder, 1994). Linked lives can be conceptualized as a convoy that provides access to tangible and intangible resources and varies in its membership as individuals age (Kahn & Antonucci, 1980). In adolescence, contextual and developmental changes encourage young people to begin forming more connections outside of the family, particularly with same-age peers from school (Wrzus et al., 2013). Previous work finds that adolescents’ connections to school peers are consequential because they both shape and are shaped by their participation in problem behavior. Guided by differential association theory (Sutherland, 1939) and social learning perspectives (Akers, 1977; Burgess & Akers, 1966), for example, empirical work finds consistent evidence that adolescents are influenced by their friends’ delinquency (Light et al., 2013; Weerman, 2011). At the same time, problem behavior participation is known to impact where young people are positioned in their networks. Delinquent youth tend to select one another as friends (Knecht et al., 2010) and deviant behavior is often associated with popularity and sociability (Haynie, 2001; Young, 2014).

The present study reconsiders how school transitions shape delinquency by evaluating whether the change from elementary to middle school impacts three network processes of interest: (1) delinquent popularity, (2) delinquent sociability, and (3) friend selection on shared delinquency. Below, we outline how each delinquency-related network process is expected to change following normative school transitions and its implications for individuals’ criminal trajectories.

Delinquent popularity

Adolescents who engage in delinquent behaviors are frequently more popular and well-liked than their nondeviant peers (Allen et al., 2005; Dijkstra et al., 2010; Haynie, 2001). These variations in social standing reflect youth subcultures that attach value to delinquent acts (Eckert, 1989; Matza, 1964), as well as the fact that young people tend to associate risky behaviors with higher levels of maturity (Dijkstra et al., 2015; Moffit, 1993). However, the relationship between delinquency and popularity is not consistent. Previous work finds that the magnitude of the association varies according to the severity of the act (Allen et al., 2005; Kreager et al., 2011), as well as across social contexts (Henneberger, et al., 2013; Staff & Kreager, 2008) and developmental periods (Allen et al., 2014, Burk et al., 2012). For example, Allen and colleagues (2014) find that delinquency increases one's odds of being well-liked by their peers in early adolescence but this association diminishes as youth age to adulthood. By age 15, differences in the popularity levels of delinquents and nondelinquents are no longer statistically significant (Allen et al., 2014).

We anticipate that normative school transitions can help explain why the association between youth delinquency and popularity is defined by such variation. Perspectives from life course criminology argue that various types of positive turning points can encourage individuals to refrain from socializing with delinquent peers because of the heightened social control and daily routine changes that accompany these transitions (Laub & Sampson, 1993, 2009; Warr, 1998). Following a young person's disengagement from a gang, for example, they are expected to report fewer connections with antisocial associates than when they were affiliated with the group (Pyrooz et al., 2021; Sweeten et al., 2013). Even though friendships with delinquent peers are understood to be “sticky” (Warr, 1993), transitions can inspire structural changes that encourage young people to start fresh and report fewer deviant associates. These shifting expectations will reduce delinquent popularity, or the number of friendship nominations they receive, to lower levels than would be expected in absence of such a transition.

There are two primary mechanisms through which school transitions could decrease the popularity of delinquent youth, particularly within same-age, school-based friendship networks. First, adolescents who participate in delinquency may lack the social skills and developmental maturity necessary to attract friendship nominations in higher-level schools. Cumulative continuity theory (Caspi et al., 1989) suggests that early adolescent delinquents learn how to gain peer status through participation in risky, pseudo-mature behaviors. These youth are less likely to cultivate the positive social skills needed to attract and maintain more meaningful friendships, leading them to encounter relational difficulties as they age through adolescence (Allen et al., 2014). Normative school transitions may operate as a catalyst in this process. Higher-level school contexts tend to reorganize the social hierarchy by introducing young people to new potential friendship partners, particularly when school transitions bring together youth from multiple, lower-level schools (Eckert, 1989; Temkin et al., 2018). With underdeveloped social skills and limited relational maturity, it may be challenging for delinquent youth to gain acceptance in these restructured social spheres.

Second, academic expectations are more rigorous in higher-level schools, and adolescents tend to see a decline in their grades following these normative changes (Felmlee et al., 2018). As youth attempt to get back on track, they may reorganize their schedules in- and outside-of-school around prosocial, academically oriented activities, limiting the time they can spend maintaining prior friendships with delinquent peers. From interviews with gang-affiliated adolescents, Carson and colleagues (2017) find that normative school transitions often functioned as a hook for change that led respondents to reorient their academic goals and sever previous ties to antisocial peers. The increasing demands of new courses and heightened emphasis on college preparation, for example, made it tactically difficult for these reformed youth to remain in contact with their delinquent friends from the past.

As a result of these potential mechanisms, we hypothesize that the association between delinquency and popularity will be significantly less pronounced in districts that transition between 6th and 7th grade than nontransition districts during the years that follow the contextual change (Hypothesis 1). After the transition, we expect that youth involved in problem behaviors will begin to drift from the central cores of their school networks toward the peripheries. When school districts do not undergo these changes, delinquent youth should remain central in their school-based networks.

Delinquent sociability

Delinquent and nondelinquent youth report friendships defined by similar levels of closeness, interaction, and trust (Giordano et al., 1986). However, there is less consensus as to whether adolescent delinquents identify the same number of friends as their nondelinquent classmates. Some research finds that delinquent youth tend to recognize fewer peers as friends since their involvement in risky and illicit behaviors leads them to become more discerning when forging relationships (Dijkstra et al., 2010; Osgood et al., 2015). These patterns do not appear to be consistent across all contexts or developmental stages, however. Early adolescent drinkers report more friendships than nondrinkers, for example, while alcohol use does not explain variations in older adolescents’ tie-sending patterns (Burk et al., 2012).

Normative school transitions may help explain these variations in delinquent youths’ sociability, or their tendency to make friendship nominations. Although life course perspectives emphasize the agency of newly reformed individuals (Elder, 1998; Laub & Sampson, 1993, 2009; Warr, 1998), relationships are not a one-way street. In addition to experiencing declines in their popularity, other aspects of normative school transitions may lead delinquent youth to initiate fewer amicable ties with their new classmates. For example, school transitions are known to reshuffle relational patterns, and this can encourage students to focus on expanding their personal networks to include new friends as a strategy to avoid social isolation in these unfamiliar settings (Eckert, 1989). Delinquent youth may have less interest in extending their networks in these new contexts, however, because of the potential overlap among friendship and co-offending (McGloin & Piquero, 2010). Both friendship and co-offending relationships rely on high degrees of trust (Schaefer et al., 2014), and as a result, deviant adolescents may be cautious when forging these social connections since not all peers will become accomplices, or even supporters, of delinquent activity. Instead, deviant youth are apt to be less concerned about “being known” in higher-level schools (Eder, 1985) and concentrate their efforts on fostering friendships with small, intimate peer groups.

At the same time, higher-level schools are defined by unique structural features that may impact the sociability of delinquents and nondelinquents in different ways. Compared to elementary schools, middle and high schools tend to offer a greater diversity of extracurricular activities, which represent key foci for adolescent friendship formation (Schaefer et al., 2011), particularly among same-school peers. Delinquent youth, however, participate in clubs, sports, and other after-school activities less frequently than their nondelinquent classmates (Osterman et al., 2016), which will limit their opportunities to form new same-age, within-school friendships in these contexts.

As a result, we suspect that delinquent youth will report fewer friendships following a school transition from elementary to middle school when compared to antisocial adolescents who do not make this change (Hypothesis 2). If school transitions encourage students involved in criminal behavior to be less social, they will begin to occupy more marginalized positions in their school networks than delinquent youth who do not experience this change in context.

Selection on similar delinquency

Young people tend to form friendships with peers who participate in similar levels of delinquency as their own, a phenomenon that is often referred to as homophily (McPherson et al., 2001). For instance, a recent meta-analysis (Gallupe et al., 2019) and systematic review (Sijtsema & Lindenberg, 2018) find evidence for homophily on delinquency net of the effects of peer influence. Not unlike the developmental trends that define the age-crime curve (Farrington, 2003; Hirschi & Gottfredson, 1983), the magnitude of problem behavior homophily varies across the life course, with levels intensifying and reaching their peak in early- to middle-adolescence (Burk et al., 2012; Ragan, 2020). Although universal changes in adolescent development and cognitive capacities can explain these patterns in part (Ragan, 2020), the normative school transitions that occur during this period may further elucidate these trends.

As mentioned previously, the structural organization of higher-level schools differs from that of their lower-level counterparts in distinct ways. We suspect that these changes will not only shape the tendencies for delinquent youth to send and receive ties but that they will also impact adolescents’ choices of who they befriend. For example, starting in middle school and continuing through high school, educational institutions become increasingly characterized by rigid systems of academic tracking (Frank et al., 2013). In schools that are tracked, course-taking patterns segregate students according to academic grades, test scores, and levels of school engagement, which amplify friendship homophily on correlated factors because of the way classrooms restrict opportunities for relationship formation (McFarland et al., 2014). Given that illicit behavior is associated with lower grades and less school involvement (Felson & Staff, 2006; Hoffmann et al., 2013), the structural changes that accompany transitions should enhance the tendency for youth to select friends with similar levels of delinquent involvement as their own. This link between course-taking patterns and clustering on delinquency is concerning since classes tend to offer uneven access to social capital (Frank et al., 2013), which may carry long-term implications for individuals’ criminal trajectories.

We suspect that the structural changes that accompany school transitions will lead to greater levels of problem behavior homophily than would be expected in absence of this transition (Hypothesis 3). As homophily on delinquency intensifies in transition districts, networks will become increasingly fragmented into pockets of prosocial versus deviant adolescents. Segregated into distinct clusters, delinquent youth in transition districts will have even fewer opportunities to befriend prosocial peers than delinquent youth who do not experience these school changes.

Types of school transitions

It is important to note that normative school transitions come in various forms that are expected to shape the social processes that guide friendship patterns in different ways. For instance, some school changes represent multifeeder transitions where students move from multiple lower-level schools to a single higher-level school. In other instances, these changes are defined as single-feeder transitions where a school district shifts students from a single lower-level school to a single higher-level school. Previous work finds that the impact of school transitions on youth outcomes tends to be the most pronounced in multifeeder arrangements (e.g., Felmlee et al., 2018; Freelin et al., 2023). This is because a school's social hierarchy remains relatively stable when the same cohort of youth moves from one physical building to another, while multifeeder transitions are characterized by greater disruptions (Langenkamp, 2010; Schiller, 1999). As a result, we suspect to uncover particularly sizable declines in delinquent popularity and sociability, as well as more clustering on delinquency, after individuals undergo multifeeder school changes when compared to their peers in nontransition districts (Hypothesis 4). We do not expect differences to be as pronounced when we compare delinquent youth who attend school in single-feeder versus nontransition districts.

Implications for network structure and delinquency

Following our hypotheses about the impact of school transitions on delinquency-related network processes, we anticipate that normative school changes carry implications for the structures of connectivity that define adolescents’ social worlds. If transitions cause delinquent youth to lose popularity, retreat socially, and form more homogenous subgroups, these phenomena will carry important implications for network structure (see hypothetical network in Figure 1). Prosocial youth will begin to occupy more central positions and report a larger number of ties to nondelinquent peers, as can be seen in the dense core located in the center of the post-transition network. At the same time, antisocial adolescents will lose friendships and become increasingly partitioned into isolated clusters with other delinquent peers, such as the group of youth in the bottom left of the post-transition network.

Figure 1.

Hypothetical example of a friendship network before and after a normative school transition. Lighter, green circles represent non-delinquent youth. Darker, blue circles represent delinquent youth. Ties between adolescents represent directed friendships.

These differences in network position are noteworthy because they will advantage prosocial youth at the expense of their delinquent peers. In the short-term, delinquent youth will have limited opportunities to influence their prosocial peers’ behaviors negatively. Adolescents in central positions carry more influence over their friends’ involvement in crime because they play a prominent role in establishing the delinquency-related norms of their networks (Dijkstra et al., 2010; Haynie, 2001). Thus, the network changes we expect to observe following a normative school transition should protect most youth from exposure to pro-delinquency influences and reduce average levels of criminal activity for the general population. Furthermore, occupying central network locations is associated with better mental health (Ueno, 2005) and higher academic grades (Vignery, 2022), two outcomes that are known to inspire long-term desistance from illicit activities (Hoffmann et al., 2013; Siennick et al., 2017). As delinquent youth become more socially marginalized after experiencing school transitions, this will jeopardize their ability to reap benefits from their network positions.

Methods

Sample

We analyzed data on roughly 14,000 students who participated in the Promoting School-Community Partnerships to Enhance Resilience (PROSPER) study. All adolescents who attended school in one of 28 small public school districts during their 6th- through 12th-grade years were invited to participate starting when they were roughly 11–12 years old. The various school districts were located in rural or suburban communities in either Pennsylvania or Iowa. Self-administered surveys were distributed across eight waves for two cohorts of students: one began 6th grade in 2002 and the other started in 2003. In the current study, we focus on five waves of panel data that were collected annually during students’ 6th- through 10th-grade years. Response rates remained high throughout this period, ranging from 86–90% across each wave. For the purpose of our analysis, it was necessary to omit students who attended school in districts that experienced abnormal transition patterns (e.g., one district experienced a fire that caused the reassignment of many students to different schools). After these omissions, our sample includes 13,752 students, or an average of 9,223 per wave, with all respondents included in one of 51 school networks.

Individual-Level Measures

To calculate our measure of delinquency, we combined student responses from 12 survey questions related to different types of antisocial and criminal activities (following Freelin et al., 2023; Kreager et al., 2011; Osgood et al. 2015). Students were asked to report how many times they participated in various delinquent behaviors during the past year, such as getting into physical fights, stealing property, and vandalizing (see Supplemental Materials, Part A for a complete list of survey items). Using students’ responses, we constructed a variety score of past-year delinquency by summing the number of deviant behaviors that each adolescent participated in at least one time during the past 12 months (following Sweeten, 2012). The resulting variety score ranges from 0 to 12, with 0 suggesting no delinquent activity and 12 indicating maximal involvement.¹

We constructed complete, sociocentric friendship networks at each wave of the survey by considering students’ nominations of their “best and closest friends” in their grade. Respondents were permitted to nominate a maximum of seven friends. To match each nominated peer to their own individual-level data reported at each survey wave, we could only consider within-grade friendship nominations to peers who attended the same school as the respondent. No information was collected from out-of-grade or out-of-school peers about their individual-level characteristics or relational patterns, which precluded us from incorporating these youth in our sociocentric, friendship networks.²

Additionally, we included several individual-level control variables that are known to shape adolescent delinquency and friendship network structure. All models include binary measures for race (0 = racial minority, 1 = white), gender (0 = girl, 1 = boy), and family structure (1 = lives with both biological parents) (following McMillan & Schaefer, 2021; Osgood et al., 2015). We also control for individuals’ family relations, school bonding/adjustment, and sensation-seeking patterns by averaging students’ responses to various, related survey questions (following Simons et al., 1991; Zuckerman, 1994). This resulted in three quantitative measures that are coded such that higher values suggest better family relations, greater levels of school bonding/adjustment, and a higher tendency toward risk-taking (see Supplemental Materials, Part A for more details on the survey items).

Plan of Analysis

Stochastic actor-oriented models

To evaluate our hypotheses, we applied stochastic actor-oriented models (SAOMs) (Snijders et al., 2010). SAOMs represent a simulation-based method that is ideal for evaluating the connection between network processes and adolescent delinquency. Using data from an initial wave as a starting point, the SAOM simulates the coevolution of network ties and individual-level behavior by allowing randomly selected participants to make forward-looking changes. The SAOM consists of two components that operate simultaneously: the network evolution model and the behavioral change model. Within the network evolution model, randomly selected respondents, or actors, are permitted to form new network ties, dissolve existing connections, or make no changes regarding their position. In the behavior change model, actors can increase their participation in the dependent variable by one unit, decrease it by one unit, or keep it constant.

Actors’ decisions to change their network ties and behaviors are probabilistically governed by a set of user-specified terms or effects. The SAOM simulates a population of networks according to the set of effects included by the user and then employs these networks to calculate the distribution of predicted changes across network ties and individual behaviors. Once the researcher identifies a model that best fits the observed data, the coefficients for the resulting model's network and behavioral effects can be interpreted as log odds ratios.

Effects of interest

We include three effects in the network component of each SAOM to test our hypotheses about the connection between individual delinquency, friendship patterns, and school transitions. To evaluate whether delinquent popularity declines after transitions (Hypothesis 1), we include delinquency alter effects that test whether individuals who report greater involvement in deviant behavior receive more friendship nominations. We assess the impact of transitions on delinquent sociability (Hypothesis 2) by incorporating delinquency ego effects to measure whether individuals who report high levels of delinquency send ties at greater rates than expected. Finally, the delinquency similarity effect is included to test whether respondents are more likely to select friends who participate in delinquency at similar levels to their own (Hypothesis 3).

When estimating a SAOM, it is standard to assume that processes of network and behavior change occur at constant rates throughout the period of interest. This results in effect coefficients that represent the average level of change across all waves. Recently developed techniques enable one to collapse these standard SAOM effects into time-varying estimates (Lospinoso & Satchell, 2011; Lospinoso et al., 2011), allowing researchers to answer novel questions about variations in network-based phenomena over time. This is accomplished by first including a set of dummy variables for each time period between the waves of the survey. In our case, these dummy variables represent the periods between 7th through 8th grade, 8th through 9th grade, and 9th through 10th grade, with 6th through 7th grade serving as the reference. Then, we included sets of interactions between the time period dummy variables and each network effect of interest. This resulted in three separate models with interactions between the time period dummy variables and the delinquency alter, delinquency ego, and delinquency similarity effects, respectively.

By including these interactions in our models, the coefficient for the original SAOM effect of interest represents the strength and direction of the network process between the first two panels of data, or between 6th and 7th grade in our sample. The coefficients for each interaction term indicate the difference between the first period and each subsequent period of time. Thus, significant coefficients demonstrate that there is a statistically significant difference in the strength of the network processes during the initial time point versus the later period of interest. Coefficients for the time period dummy variables are fixed at zero and not interpreted substantively (Lospinoso & Satchell, 2011; Ragan, 2020).

Control effects

All SAOMs include a variety of effects that are known to shape adolescent friendship patterns and participation in delinquency. We began by controlling for several effects in the network evolution model that shape the structural patterns of the friendship ties (Jacobsen et al., 2022; Ragan, 2020). The density effect reflects the general tendency to form relationships. By including the density effect, we can interpret the coefficients of other included effects net of the number of students included in the observed networks and their general tendency to make friendship nominations (Steglich et al., 2010). The reciprocity effect accounts for the tendency for friendships to be mutual, while the transitive triplets effect considers whether adolescents are more likely to be friends with their friends’ friends. Following Block (2015), we also include the reciprocated transitive triplets effect, or an interaction between the two aforementioned terms, to reduce bias in the network-level estimates of reciprocity and transitivity. We incorporate the square root version of the indegree popularity measure to account for the fact that connected adolescents tend to attract more friendship nominations over time. Relatedly, the in-in degree assortative term reflects the tendency for adolescents to forge friendships with peers who report similar popularity levels as their own. We also include the truncated outdegree effect, which captures the tendency for actors to be outdegree isolates, or to make no friendship nominations. Finally, the network component of each SAOM includes several attribute-based effects to account for friendship patterns that are informed by the race and gender of respondents. Similar to our key independent variables, we include ego, alter, and same attribute effects to account for the ways race and gender shape network ties.³

We also include several controls in the behavioral change model to account for individual-level factors that are associated with delinquency. Importantly, we include the average delinquency similarity effect to measure whether individuals are influenced by the deviant behaviors of their friends. This effect evaluates whether actors’ levels of delinquency are more likely to resemble that of their friends over time, with positive coefficients providing evidence for peer influence. We also incorporate several effect from attribute parameters to measure how race, gender, living with both biological parents, family relations, school bonding/adjustment, and sensation-seeking impact each respondent's delinquency.

Meta-analysis and variations by transition structure

To address our hypotheses, we estimated SAOMs on each of the 51 networks in our sample. We applied three sets of SAOMs to each of our networks. Each set considered time variations in the delinquency alter, delinquency ego, or delinquency similarity SAOM effects. Only models that converged and produced reasonable goodness-of-fit statistics were included in our final analyses.⁴ To aggregate our findings across the SAOMs, we performed several multilevel random effects meta-analyses, where the first level represents the network coefficients, and the second represents the network's community-cohort. Our meta-analysis technique averaged each SAOM coefficient across all models, giving greater weights to those estimates that were more precise (following Jacobsen et al., 2022).⁵

Most crucially, we include time-invariant, dichotomized indicators of transition patterns in our meta-analyses to test whether network processes differ across our sample of school districts. In the current project, we focus on normative, district-wide transitions that occur between 6th and 7th grade (1 = 6/7th transition district) because this matriculation occurs when delinquency and peers are highly salient in day-to-day life (Moffitt, 1993; Veenstra & Laninga-Wijnen, 2023). In our sample, 35% of school-cohorts experience a transition between 6th and 7th grade, while the remaining 65% do not transition between these two grade levels.⁶ Among the school districts that transition students between these grades, there are two distinct transition patterns that further define the communities. Some districts merge 6th-grade students from multiple elementary schools into a single middle school (44% of transitioning districts), while others shift 6th graders from a single elementary school to a single middle school (56%). As a result, subsequent models use two binary indicators to distinguish between multifeeder (1 = 6/7th multifeeder transition district) and single-feeder transition districts (1 = 6/7th single-transition district), comparing each to the nontransition districts.

It is important to note that we uncover minimal, statistically significant differences in community-level characteristics across our sample of school districts (see Supplemental Materials, Part C). For example, the nontransition, single-feeder, and multifeeder districts are nested in counties with similar population sizes, levels or rurality, and measures of socioeconomic status. Additionally, all districts report comparable dropout rates and employ roughly the same number of full-time staff. Although we are unable to assess whether the districts in our sample are defined by different cultural norms or varying levels of parental involvement, we believe these findings reduce the likelihood that other community- or school-level factors resulted in preexisting differences across the groups of respondents.⁷

We account for the impact of various transition patterns by interacting the binary indicators for transition patterns with the time-varying estimates of the network processes of interest in our multilevel random effects meta-analyses. By including these interactions, we can evaluate whether there are significant differences in popularity, sociability, and homophily by a district's transition structure, as well as over time. More specifically, the coefficient for each interaction indicates whether a network process operates differently in a transition district versus a nontransition district during a specific period of interest. We expect these differences to be significant in the time periods following the transition, but not in those that capture periods of time before the transition occurred.⁸

Results

Descriptive Results

Regardless of the transition patterns that characterize a student's district, average delinquency participation tends to increase steadily from 6th grade until reaching its peak in 10th grade (see Table 1). However, the magnitude of this increase varies across the district types in our sample. In 6th grade, or before the transition of interest occurs, delinquency participation is relatively similar across school types. During the years immediately following the transition, the average delinquency of students in nontransition districts is significantly higher than it is for those students who made these structural changes. In 7th grade, for example, students attending school in transition districts report an average of only 1.05 delinquent acts in the past year compared to an average of 1.31 among students in the nontransition districts. This difference does not persist until the end of the sample, however. In the later waves of the study, students in the transition districts report average levels of delinquency that are similar, if not greater, than their peers who attended school in districts that do not transition students after 6th grade.

Table 1.

Average Delinquency Over Time and Across District Types.

	Nontranstion		Transition Districts
	Mean	SD	Mean	SD
6th Grade	0.965	1.858	0.934	1.804
7th Grade	1.309	2.256	1.185	2.127*
8th Grade	1.616	2.495	1.595	2.445
9th Grade	1.788	2.669	1.841	2.727
10th Grade	1.787	2.728	2.005	2.836*

Notes: *A significant difference from the nontranstion districts according to a t-test (p < .05).

Changes Over Time Across all Districts

Across our full sample, adolescents’ experiences with delinquency and the three network processes of interest complement findings from previous work. Participating in an additional delinquent act tends to increase one's odds of receiving a friendship nomination (see Table 2, Model 1: delinquency alter (baseline): b = .053, p < .001), though this association significantly weakens starting in 9th grade (delinquency alter (9th): b = −.028, p < .05). Involvement in delinquency also increases one's odds of sending friendship nominations (Model 2: delinquency ego (baseline): b = .016, p < .001), and this continues throughout all waves. Finally, adolescents tend to befriend peers who participate in similar levels of delinquency as their own (Model 3: delinquency similarity (baseline): b = .629, p < .001), and this persists throughout the study. SAOMs include all effects mentioned previously, such as controls for being influenced by the delinquent behavior of one's friends and controls for how race and gender impact network patterns. For a substantive interpretation of the control variables’ coefficients, we point the reader to previous work that also applies SAOMs to the PROSPER data (e.g., Osgood et al., 2015).

Table 2.

Meta-Analysis Results for SAOMs Considering Changes in Delinquency-Related Peer Processes Over Time.

	Model 1: Alter (Popularity)		Model 2: Ego (Sociability)		Model 3: Similarity (Selection)
	b	SE	b	SE	b	SE
Key Effects of Interest:
Network process (baseline)	0.053	(0.005)***	0.016	(0.004)***	0.629	(0.065)***
Network process (7th)	0.001	(0.011)	0.010	(0.009)	0.010	(0.044)
Network process (8th)	−0.015	(0.011)	0.015	(0.008)	−0.049	(0.125)
Network process (9th)	−0.028	(0.011)*	0.004	(0.009)	0.150	(0.105)
Network Evolution Controls:
Constant friends rate (baseline)	20.248	(0.914)***	19.977	(0.911)***	20.087	(0.934)***
Constant friends rate (7th)	17.677	(0.725)***	17.515	(0.743)***	17.417	(0.718)***
Constant friends rate (8th)	14.244	(0.512)***	13.960	(0.503)***	14.100	(0.511)***
Constant friends rate (9th)	11.877	(0.368)***	11.643	(0.346)***	11.758	(0.358)***
Density	−3.533	(0.056)***	−3.556	(0.053)***	−3.532	(0.055)***
Reciprocity	2.844	(0.035)***	2.843	(0.034)***	2.841	(0.035)***
Transitive triplets	0.733	(0.015)***	0.729	(0.015)***	0.737	(0.014)***
Transitive reciprocated triplets	−0.513	(0.015)***	−0.502	(0.016)***	−0.509	(0.016)***
Indegree popularity	0.730	(0.016)***	0.755	(0.015)***	0.740	(0.015)***
Outdegree truncated	−2.570	(0.083)***	−2.519	(0.081)***	−2.557	(0.083)***
In-in degree assortativity	−0.401	(0.011)***	−0.413	(0.011)***	−0.407	(0.011)***
Male alter	0.145	(0.014)***	0.147	(0.013)***	0.147	(0.013)***
Male ego	−0.163	(0.015)***	−0.169	(0.015)***	−0.167	(0.014)***
Same gender	0.774	(0.024)***	0.785	(0.024)***	0.773	(0.023)***
White alter	−0.179	(0.017)***	−0.175	(0.018)***	−0.173	(0.019)***
White ego	−0.040	(0.017)*	−0.032	(0.018)	−0.041	(0.017)*
Same race	0.243	(0.028)***	0.232	(0.028)***	0.246	(0.028)***
Delinquency alter			0.051	(0.004)***	0.048	(0.004)***
Delinquency ego	0.017	(0.003)***			0.018	(0.003)***
Delinquency similarity	0.673	(0.070)***	0.613	(0.068)***
Behavior Change Controls:
Rate of delinquency change (baseline)	7.309	(0.552)***	7.073	(0.542)***	7.277	(0.525)***
Rate of delinquency change (7th)	10.656	(0.790)***	11.498	(0.875)***	9.950	(0.680)***
Rate of delinquency change (8th)	12.201	(0.914)***	12.131	(0.895)***	11.528	(0.838)***
Rate of delinquency change (9th)	11.507	(0.771)***	12.381	(0.839)***	11.618	(0.769)***
Behavior delinquency linear shape	−0.462	(0.014)***	−0.470	(0.014)***	−0.466	(0.014)***
Behavior delinquency quadratic shape	0.030	(0.002)***	0.031	(0.002)***	0.030	(0.002)***
Friends’ average delinquency	0.237	(0.301)	0.271	(0.276)	0.323	(0.282)
Effect from male	0.084	(0.011)***	0.079	(0.010)***	0.082	(0.011)***
Effect from white	−0.044	(0.014)**	−0.046	(0.014)**	−0.043	(0.014)**
Effect from living with both bio. parents	−0.070	(0.008)***	−0.072	(0.009)***	−0.075	(0.009)***
Effect from family relations	−0.098	(0.012)***	−0.097	(0.011)***	−0.098	(0.012)***
Effect from school bonding	−0.071	(0.006)***	−0.072	(0.006)***	−0.071	(0.006)***
Effect from sensation seeking	0.053	(0.004)***	0.052	(0.005)***	0.055	(0.005)***

Notes: *p < .05, **p < .01, ***p < .001. Model 1 includes 49 networks, Model 2 includes 51, and Model 3 includes 49.

SAOM: stochastic actor-oriented model.

Changes Over Time by Transition Patterns

All 6/7th transitions

Important variations begin to emerge when we consider how rates of change for the three delinquency-related network processes differ across districts that transition between 6th and 7th grade versus those that do not. In support of our second hypothesis, we find that delinquent youth who attend school in transition districts make significantly fewer friendship nominations than nontransitioning delinquents during all periods that occur after the school change (see Table 3, Model 2). For instance, 7th graders who report past-year delinquency are significantly less likely to send friendship nominations if they attend school in transition versus nontransition districts (delinquency ego (7th grade) × transition: b = −.052, p < .01). Importantly, the association between individual delinquency and sociability does not vary by district type before the transition occurs (delinquency ego (baseline) × transition: b = .016, p > .05). Measures of the association between individual delinquency and popularity, as well as the tendency to select friends with similar levels of delinquency participation as one's own, are not significantly different across the transition versus nontransition networks during any period of the study.

Table 3.

Meta-analysis results for SAOMs considering changes in delinquency-related peer processes over time for transition versus nontransition districts

	Model 1: Alter (Popularity)			Model 2: Ego (Sociability)			Model 3: Similarity (Selection)
	b	SE		b	SE		b	SE
Network process (baseline)	0.050	(0.006)	***	0.011	(0.005)	*	0.634	(0.081)	***
Network process (baseline) × transition district	0.009	(0.011)		0.016	(0.009)		–0.014	(0.145)
Network process (7th)	0.010	(0.013)		0.028	(0.010)	**	–0.059	(0.121)
Network process (7th) × transition district	–0.029	(0.024)		–0.052	(0.017)	**	0.258	(0.239)
Network process (8th)	–0.005	(0.014)		0.031	(0.010)	**	–0.174	(0.147)
Network process (8th) × transition district	–0.032	(0.025)		–0.046	(0.016)	**	0.450	(0.278)
Network process (9th)	–0.016	(0.013)		0.021	(0.011)		0.107	(0.127)
Network process (9th) × transition district	–0.037	(0.024)		–0.048	(0.018)	**	0.159	(0.238)

Notes: *p < .05, **p < .01, ***p < .001. All models include the structural and behavioral controls presented in Table 2. Model 1 includes 49 networks, Model 2 includes 51, and Model 3 includes 49.

Single- and multifeeder 6/7th transitions

When we consider single- and multifeeder transition patterns separately, we find that the coevolution of adolescent friendship and delinquency is defined by further variations. During the years that follow the normative transition, all three delinquency-related network processes vary notably for adolescents attending school in the multifeeder versus the nontransition districts. When differences are significant, they are in the expected direction of our hypotheses. In support of Hypothesis 1, we find that levels of delinquent popularity are significantly lower in multifeeder versus nontransition networks across all periods that follow the change in schools (see Table 4, Model 1, e.g., delinquency alter (7th) × multifeeder: b = −.108, p < .01). These differences only emerge after the school change occurs, as the association between popularity and delinquency is indistinguishable across the districts during 6th grade (delinquency alter (baseline) × multifeeder: b = .008, p > .05). We visualize these trends in Figure 2. Across all post-transition waves, the association between delinquency and receiving friendship nominations is significantly less pronounced in multifeeder districts than nontransition districts. In fact, in the years immediately following the transition, there is a slight negative correlation between delinquency and adolescent popularity in the multifeeder districts.

Figure 2.

Expected value of the delinquency alters coefficient (i.e., popularity) by transition type. *p < .05, **p < .01 indicates a significantly different from the nontranstion districts.

Table 4.

Meta-analysis results for SAOMs considering changes in delinquency-related peer processes over time for multi- and single-feeder transition districts

	Model 1: Alter (Popularity)			Model 2: Ego (Sociability)			Model 3: Similarity (Selection)
	b	SE		b	SE		b	SE
Network process (baseline)	0.050	(0.006)	***	0.011	(0.005)	*	0.635	(0.082)	***
Network process (baseline) × multifeeder	0.008	(0.015)		0.034	(0.013)	**	0.013	(0.236)
Network process (baseline) × single-feeder	0.010	(0.013)		0.003	(0.011)		−0.026	(0.167)
Network process (7th)	0.010	(0.012)		0.028	(0.009)	**	−0.053	(0.111)
Network process (7th) × multifeeder	−0.108	(0.040)	**	−0.102	(0.024)	***	1.123	(0.403)	**
Network process (7th) × single-feeder	0.001	(0.026)		−0.025	(0.018)		−0.072	(0.255)
Network process (8th)	−0.005	(0.013)		0.031	(0.009)	**	−0.173	(0.145)
Network process (8th) × multifeeder	−0.094	(0.040)	*	−0.088	(0.024)	***	1.313	(0.468)	**
Network process (8th) × single-feeder	−0.007	(0.027)		−0.025	(0.018)		0.120	(0.309)
Network process (9th)	−0.016	(0.013)	***	0.021	(0.011)		0.100	(0.117)
Network process (9th) × multifeeder	−0.076	(0.039)	*	−0.098	(0.026)	***	1.043	(0.395)	**
Network process (9th) × single-feeder	−0.021	(0.027)		−0.020	(0.020)		−0.139	(0.247)

Notes: *p < .05, **p < .01, ***p < .001. All models include the structural and behavioral controls presented in Table 2. Model 1 includes 49 networks, Model 2 includes 51, and Model 3 includes 49.

Regarding the association between delinquency and sending friendship nominations, we find that before the transition occurs, delinquent students in multifeeder districts send significantly more ties than deviant youth attending nontransition schools (see Table 4, Model 2, delinquency ego (baseline) × multifeeder: b = .034, p < .01). However, in all time periods after the transition, this relationship reverses. Starting in 7th grade, greater involvement in delinquency is associated with lower levels of sociability in the multifeeder districts when compared to districts that do not transition students after 6th grade (Hypothesis 2). This significant difference persists through the later periods of the study as illustrated in Figure 3.

Figure 3.

Expected value of the delinquency ego coefficient (i.e., sociability) by transition type. **p < .01, ***p < .001 indicates a significantly different from the nontranstion districts.

Finally, there are no differences in selection on similar delinquent behaviors between the multifeeder and nontransition districts before the school change (see Table 4, Model 3, delinquency similarity (baseline) × multifeeder: b = .013, p > .05), but significant variations emerge at all other time points that support our third hypothesis. After the transition, students attending schools in multifeeder districts are even more likely to select friends who participate in similar levels of delinquency as their own than students who attend school in nontransition districts (see Figure 4). In the multifeeder districts, for example, 7th graders who participate in one type of delinquency are 2.48 times more likely to befriend a peer who reports this same level of involvement than a nondelinquent (exp(.26)/exp(−.65); see Supplemental Materials, Part D for ego-alter selection tables). In nontransition districts, the same 7th grader would only be 1.07 times more likely to befriend an equally delinquent peer than a nondelinquent (exp(.07)/exp(−.27)). Importantly, this difference did not exist before the transition. In nontransition and multifeeder districts, 6th graders who participate in one type of delinquent behavior are, respectively, 1.44 and 1.45 times more likely to select same-delinquency peers as friends than nondelinquents.

Figure 4.

Expected value of the delinquency similarity coefficient (i.e., selection) by transition type. **p < .01 indicates a significantly different from the nontranstion districts.

Although adolescents have significantly different experiences with key network processes in multifeeder versus nontransition districts, the experiences of students in single-feeder districts closely resemble those of peers who do not change schools after 6th grade (see Table 4 and Figures 2–4). No significant differences in delinquent popularity, delinquent sociability, or selection on shared delinquency emerge between the single-feeder and nontransition districts. This is the case for all time periods considered, including those that precede and follow the transition's occurrence. Furthermore, supplemental analyses suggest that adolescents’ experiences with delinquency-related network processes significantly differ in single- versus multifeeder districts (see Supplemental Materials, Part E). Although there are some exceptions, delinquent adolescents attending school in multifeeder districts tend to be less popular, less social, and more clustered than delinquent youth in single-feeder communities.

Supplemental analyses

Although all SAOMs include controls to account for various individual-level and structural factors that shape network patterns and delinquency, we also estimated supplemental meta-analyses to ensure that other community-level factors do not account for the variations across district types (see Supplemental Materials, Part F). After controlling for each community's racial diversity, urbanization, economic resources, size, and other contextual differences, we continue to observe the same patterns of significant differences in delinquency-related network processes across the different types of districts. Although patterns largely mimic those presented here, there is one exception. In the last period of the study, delinquent students in multifeeder districts are no longer significantly less popular than delinquent students in nontransition districts.

Since previous work finds that the association between delinquency and popularity varies according to the severity of the deviant acts (Allen et al., 2005; Kreager et al., 2011), we also estimated supplemental SAOMs that model the evolution of friendship and involvement in violent delinquency, specifically (see Supplemental Materials, Part G). We focused on violent delinquency because most of the delinquent acts in our variety scale represent minor property crimes and we wanted to ensure that our results were not driven by these offenses. Significant variations continue to emerge across the districts that complement our main findings.

Discussion

For many adolescents, normative transitions to higher-level schools represent important life changes that offer opportunities to scale back on prior involvement with delinquent activities and start anew. The current project emphasizes how various network processes facilitate this role by comparing changes in delinquent popularity, sociability, and clustering across a large, longitudinal sample of students enrolled in school districts defined by various transition patterns. After a student body experiences a normative transition between 6th and 7th grade, the social dynamics that guide friendship patterns change in ways that would not be anticipated in the absence of this contextual shift. For instance, after the grade cohorts in our sample experience multifeeder transitions (i.e., when multiple elementary schools merge students into a single middle school), delinquent youth see a reduction in their popularity and sociability, and they are more likely to report friendships with same-grade peers who also engage in problematic behaviors. The delinquency-related peer processes that define networks of youth who do not transition to middle school after 6th grade, as well as those who experience single-feeder school changes, are not characterized by these same trends. Our findings suggest that normative school transitions can restructure the social landscape in ways that advantage youth who refrain from antisocial behaviors, while further jeopardizing the trajectories of peers who remain involved in crime and delinquency.

Individuals who experience positive turning points are hypothesized to sever ties with their previous delinquent contacts as they attempt to start anew (Laub & Sampson, 1993, 2009; Warr, 1998), a process that should reduce delinquent popularity. We find evidence for these expectations following multifeeder, normative school transitions. Delinquent youth who change schools after 6th grade receive significantly fewer friendship nominations than their deviant peers in nontransition districts, which gives support to our first hypothesis. In the years that directly follow a multifeeder transition, we even uncover a slight negative association between delinquency and popularity, though this relationship does not persist throughout the study. These findings suggest that the shock of a normative structural transition can resonate through a grade-cohort's social hierarchy, particularly when students from multiple lower-level schools are merged into a single higher-level school (Langenkamp, 2010; Schiller, 1999). Although navigating these social dynamics is apt to be challenging for many young people, it may be particularly difficult for early adolescent delinquents who lack the social skills necessary to gain peer status outside of risky behavior participation (Caspi et al., 1989). In fact, our findings suggest that normative transitions may help explain why the association between minor delinquent involvement and popularity dissipates as youth age through adolescence (Allen et al., 2014).

We also find that prosocial adolescents are not the only individuals who play an active role in reshaping peer networks after school transitions occur. Instead, the tie-sending patterns of youth who remain delinquent undergo significant changes that complement these broader trends. In line with our second hypothesis, we find that after the transition to 7th grade, delinquents are expected to send fewer friendship nominations to their in-school peers when compared to deviant youth who did not make this structural change. This decline in delinquent sociability is particularly pronounced following multifeeder transitions. We suspect that this reduction in delinquent sociability is due to growing divides in school engagement between anti- and prosocial youth.⁹ With higher levels of school involvement and lower rates of truancy (Osterman et al., 2016), nondelinquent youth can use the various foci of interaction that define higher-level schools as opportunities to forge new friendships. Delinquent adolescents are less likely to partake in new extracurriculars and after-school activities, and instead appear to focus on maintaining small groups of trusted, within-school friends (Schaefer et al., 2014).

In addition to changes in delinquent popularity and sociability, we also find that adolescents are more likely to select friends who participate in similar levels of delinquency as their own after experiencing the transition to middle school, a finding that supports our third hypothesis. This problem behavior homophily exists across all district types in our sample. However, friendship networks in multifeeder communities are defined by significantly more clustering on shared delinquency when compared to the networks of nontransition districts, which supports our fourth hypothesis. Since multifeeder transitions introduce new groups of students to one another, they may provide young people more opportunities to select friends with whom they share common characteristics, including problem behavior participation (McPherson et al., 2001). Additionally, higher-level schools are more likely to be defined by systems of academic tracking (Frank et al., 2013). Given the overlap between academic performance, socioeconomic status, and delinquency participation (Felson & Staff, 2006; Hoffmann et al., 2013), the consolidation introduced by these structural forces appears to further segregate pro- and antisocial youth (Blau, 1977).

Our findings carry implications for our understanding of how normative school transitions shape individuals’ delinquency in the short-term, as well as across the life span. The changing peer processes documented above will restructure youth networks such that individuals who abstain from delinquency will become more central, while those who participate in problematic behaviors are increasingly segregated into homogenous clusters on the peripheries. On the one hand, these relational modifications should inspire beneficial outcomes for nondelinquent youth. As their delinquent peers become more clustered into marginalized pockets, prosocial youth will spend less time with classmates who participate in criminal activities, which will limit the number of avenues through which negative influence can diffuse across the network. In fact, this reduction in exposure to delinquent peers should help explain the lower population levels of delinquency that follow school transitions (Freelin et al., 2023). There are also long-term benefits that accompany the central network positions nondelinquent youth occupy after the school change. Being centrally located in one's school network is not only associated with healthier indicators of well-being and stronger academic performance during adolescence (Ueno, 2005; Vignery, 2022) but also lower rates of depression and higher incomes in adulthood (Copeland, 2021; Shi & Moody, 2017). As a result, the relational changes that follow school transitions should equip prosocial students with advantageous resources to facilitate desistance from criminal activities throughout the life course (Hoffmann et al., 2013; Siennick et al., 2017). The network positions of delinquent youth, on the other hand, are not associated with these same benefits. Instead, antisocial youth will accrue more disadvantages as they drift to their network's peripheries and lose ties to nondelinquent peers.

Our findings suggest that part of the reason normative school transitions impact individuals’ criminal trajectories is because of how they alter the delinquency-related network process that guides our social worlds. The effect school transitions have on these network phenomena should be understood as a type of cumulative (dis)advantage that individuals acquire as they age (Elder, 1998; Sampson & Laub, 1997). Normative school transitions are pivotal moments for those who are able to start anew, while they further disadvantage those individuals who remain involved in delinquent behaviors. This is because school transitions limit delinquent adolescents’ access to the positive influences from prosocial peers and other advantages associated with central network positions.

Limitations

Our study not only makes several important contributions but also raises questions for future work. First, our sample does not include schools in large cities or with sizeable numbers of adolescents who identify as racial/ethnic minorities. Instead, we consider a sample of young people from numerous school districts in rural areas and small towns. Although our study considers an important population that has been understudied in social science research (White & Corbett, 2014), the associations documented here may be distinct to these communities and future research should assess the generalizability of our results to other, more diverse contexts. Second, structural school transitions may shape additional types of antisocial behavior that are not included in our delinquency measure, such as drinking, smoking, and other types of substance use. Since peer processes can impact substance use in different ways than involvement in property and violent crimes (Kreager et al., 2011), it would be beneficial to examine substance use outcomes in future work on normative school transitions.

Third, the current project only considers patterns of same-school, within-grade friendships. Although these friendships have particularly salient impacts on youth behavior (Veenstra & Laninga-Wijnen, 2023), future work should also evaluate whether school transitions impact the delinquency-related network dynamics that shape friendships formed in other contexts. Fourth, it remains unknown whether transitioning from middle to high school also impacts delinquency-related peer network processes. We suspect that matriculating to high school will carry similar consequences as the transition studied here. In fact, the network-related disadvantages delinquent youth accumulate from earlier school transitions should compound over time, and such a finding would provide additional insight into how peer processes affect long-term criminal trajectories. More generally, future research should consider how various turning points that occur across the life span (e.g., experiencing parental divorce, securing full-time employment) impact peer network processes that inform criminal behaviors.

Conclusion

Overall, the present study argues that delinquency-related network processes can help explain why normative school transitions carry lasting implications for individuals’ criminal trajectories. For youth who abstain from delinquency, moving to higher-level schools can reorganize young people's social networks in ways that promote abstinence from crime, as well as other long-term academic, social, and behavioral benefits. For youth who continue to commit deviant acts, these network changes are expected to jeopardize their delinquency trajectories in both the immediate and long term. By understanding the interaction between patterns of friendship and delinquency, school officials can better prepare students for success across districts defined by numerous transition structures. For instance, school districts should invest in programing that gives at-risk youth a chance to start fresh, sever delinquent connections, and decrease criminal behavior in the years that follow these structural school changes.

Supplemental Material

sj-docx-1-jrc-10.1177_00224278231167841 - Supplemental material for School Transitions, Peer Processes, and Delinquency: A Social Network Approach to Turning Points in Adolescence

Supplemental material, sj-docx-1-jrc-10.1177_00224278231167841 for School Transitions, Peer Processes, and Delinquency: A Social Network Approach to Turning Points in Adolescence by Cassie McMillan and Brittany N. Freelin in Journal of Research in Crime and Delinquency

Footnotes

Acknowledgments

The authors thank Mark Feinberg, Gregory Zimmerman, Diane Felmlee, Wayne Osgood, and Scott Duxbury for feedback on earlier versions of the paper. The authors also gratefully acknowledge research assistance from Alana Colindres and support from the School of Criminology & Criminal Justice at Northeastern University.

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

This research was supported in part by the W.T. Grant Foundation (8316) and National Institute on Drug Abuse (RO1-DA08225; T32-DA-017629; F31-DA-024497) and uses data from PROSPER, a project directed by R. L. Spoth and funded by grant RO1-DA013709 from the National Institute on Drug Abuse.

ORCID iDs

Cassie McMillan

Brittany N. Freelin

Supplemental Material

Supplemental material for this article is available online.

Notes

Author Biographies

Cassie McMillan is an assistant professor of sociology and criminology at Northeastern University. Her research focuses on social networks, youth crime, substance use, and social inequality. In particular, she is interested in developing and applying network methodologies to better understand patterns of adolescent delinquency.

Brittany N. Freelin is an assistant professor of criminology at Randolph–Macon College. Her research examines how educational policies influence life–course trajectories of social control, crime, and delinquency, with a focus on racial disparities in outcomes.

References

Aaron

Dallaire

D. H.

. 2010. “Parental Incarceration and Multiple Risk Experiences: Effects on Family Dynamics and Children’s Delinquency.” Journal of Youth and Adolescence 39(12):1471-84.

Aikins

J. W.

Bierman

K. L.

Parker

J. G.

. 2005. “Navigating the Transition to Junior High School: The Influence of pre-Transition Friendship and Self-System Characteristics.” Social Development 14(1):42-60.

Akers

R. L.

1977. Deviant Behavior: A Social Learning Approach. Belmont, CA: Wadsworth.

Allen

J. P.

Porter

M. R.

McFarland

F. C.

Marsh

McElhaney

K. B.

. 2005. “The two Faces of Adolescents’ Success with Peers: Adolescent Popularity, Social Adaptation, and Deviant Behavior.” Child Development 76(3):747-60.

Allen

J. P.

Schad

M. M.

Oudekerk

Chango

. 2014. “What Ever Happened to the “Cool” Kids? Long-Term Sequelae of Early Adolescent Pseudomature Behavior.” Child Development 85(5):1866-80.

2015. “Multilevel Meta Network Analysis with Application to Studying Network Dynamics of Network Interventions.” Social Networks 43:48-56.

Benner

A. D.

Graham

. 2009. “The Transition to High School as a Developmental Process among Multiethnic Urban Youth.” Child Development 80(2):356-76.

Blau

P. M.

1977. Inequality and Heterogeneity. New York: Free Press.

Block

2015. “Reciprocity, Transitivity, and the Mysterious Three-Cycle.” Social Networks 40:163-73.

10.

Blyth

Simmons

Carlton-Ford

. 1983. “The Adjustment of Early Adolescents to School Transitions.” The Journal of Early Adolescence 3(1–2):105-20.

11.

Burgess

Robert L.

Akers

Ronald L.

. 1966. “A Differential Association-Reinforcement Theory of Criminal Behavior.” Social Problems 14(2):128-47.

12.

Burk

W. J.

Van Der Vorst

Kerr

Stattin

. 2012. “Alcohol use and Friendship Dynamics: Selection and Socialization in Early-, Middle-, and Late-Adolescent Peer Networks.” Journal of Studies on Alcohol and Drugs 73(1):89-98.

13.

Carson

D. C.

Melde

Wiley

S. A.

Esbensen

F. A.

. 2017. “School Transitions as a Turning Point for Gang status.” Journal of Crime and Justice 40(4):396-416.

14.

Caspi

Bem

D. J.

Elder

G. H.

. 1989. “Continuities and Consequences of Interactional Styles across the Life Course”. Journal of Personality 57(2):375-406.

15.

Copeland

2021. “The Long Shadow of Peers: Adolescent Networks and Young Adult Mental Health.” Social Sciences 10(6):231.

16.

Dijkstra

J. K.

Kretschmer

Pattiselanno

Franken

Harakeh

Vollebergh

Veenstra

. 2015. “Explaining Adolescents’ Delinquency and Substance Use: A Test of the Maturity Gap: The SNARE Study.” Journal of Research in Crime and Delinquency 52(5):747-67.

17.

Dijkstra

J. K.

Lindenberg

Veenstra

Steglich

Isaacs

Card

N. A.

Hodges

E. V.

. 2010. “Influence and Selection Processes in Weapon Carrying During Adolescence: The Roles of Status, Aggression, and Vulnerability.” Criminology 48(1):187-220.

18.

Duxbury

Scott W.

Wertsching

. 2023. “Scaling Bias in Pooled Exponential Random Graph Models.” Social Networks 74:19-30.

19.

Eckert

1989. Jocks and Burnouts: Social Categories and Identity in the High School. New York: Teachers College Press.

20.

Eder

1985. “The Cycle of Popularity: Interpersonal Relations among Female Adolescents.” Sociology of Education 58(3): 154-65.

21.

Edin

Kefalas

. 2005. “Unmarried with Children.” Contexts 4(2):16-22.

22.

Elder

G. H.

1994. “Time, Human Agency, and Social Change: Perspectives on the Life Course.” Social Psychology Quarterly 57(1):4-15.

23.

Elder

G. H.

1998. “The Life Course as Developmental Theory.” Child Development 69(1):1-12.

24.

Elder

G. H.

Johnson

M. K.

Crosnoe

. 2003. “The Emergence and Development of Life Course Theory.” Pp. 3-19 in Handbook of the Life Course edited by Mortimer

J. T.

Shanahan

M. J.

. New York: Springer.

25.

Evans

S. Z.

Simons

L. G.

Simons

R. L.

. 2016. “Factors That Influence Trajectories of Delinquency Throughout Adolescence.” Journal of Youth and Adolescence 45(1):156-71.

26.

Farrington

D. P.

2003. “Key Results from the First Forty Years of the Cambridge Study in Delinquent Development.” Pp. 137-83 in Taking Stock of Delinquency. New York: Springer.

27.

Felmlee

McMillan

Inara Rodis

Osgood

D. W.

. 2018. “Falling Behind: Lingering Costs of the High School Transition for Youth Friendships and Grades.” Sociology of Education 91(2):159-82.

28.

Felson

R. B.

Staff

. 2006. “Explaining the Academic Performance-Delinquency Relationship.” Criminology 44(2):299-320.

29.

Frank

K. A.

Muller

Mueller

A. S.

. 2013. “The Embeddedness of Adolescent Friendship Nominations: The Formation of Social Capital in Emergent Network Structures.” American Journal of Sociology 119(1):216-53.

30.

Freelin

B. N.

McMillan

Felmlee

Osgood

D. W.

. 2023. “Changing Contexts: A Quasi-Experiment Examining Adolescent Delinquency and the Transition to High School.” Criminology 61(1):40-73.

31.

Gallupe

McLevey

Brown

. 2019. “Selection and Influence: A Meta-Analysis of the Association Between Peer and Personal Offending.” Journal of Quantitative Criminology 35(2):313-35.

32.

Gasper

DeLuca

Estacion

. 2010. “Coming and Going: Explaining the Effects of Residential and School Mobility on Adolescent Delinquency.” Social Science Research 39(3):459-76.

33.

Giordano

P. C.

Cernkovich

S. A.

Pugh

M. D.

. 1986. “Friendships and Delinquency.” American Journal of Sociology 91(5):1170-202.

34.

Grigg

2012. “School Enrollment Changes and Student Achievement Growth: A Case Study in Educational Disruption and Continuity.” Sociology of Education 85(4):388-404.

35.

Hardy

C. L.

Bukowski

W. M.

Sippola

L. K.

. 2002. “Stability and Change in Peer Relationships During the Transition to Middle-Level School.” The Journal of Early Adolescence 22(2):117-42.

36.

Haynie

D. L.

2001. “Delinquent Peers Revisited: Does Network Structure Matter?” American Journal of Sociology 106(4):1013-57.

37.

Henneberger

A. K.

Durkee

M. I.

Truong

Atkins

Tolan

P. H.

. 2013. “The Longitudinal Relationship Between Peer Violence and Popularity and Delinquency in Adolescent Boys: Examining Effects by Family Functioning.” Journal of Youth and Adolescence 42(11):1651-60.

38.

Herbers

J. E.

Reynolds

A. J.

Chen

C. C.

. 2013. “School Mobility and Developmental Outcomes in Young Adulthood.” Development and Psychopathology 25(2):501-15.

39.

Hirschi

Gottfredson

. 1983. “Age and the Explanation of Crime.” American Journal of Sociology 89(3):552-84.

40.

Hoffmann

J. P.

Erickson

L. D.

Spence

K. R.

. 2013. “Modeling the Association Between Academic Achievement and Delinquency: An Application of Interactional Theory.” Criminology 51(3):629-60.

41.

Hope

T. L.

Wilder

E. I.

Watt

T. T.

. 2003. “The Relationships among Adolescent Pregnancy, Pregnancy Resolution, and Juvenile Delinquency.” The Sociological Quarterly 44(4):555-76.

42.

Jacobsen

W. C.

Ragan

D. T.

Yang

Nadel

E. L.

Feinberg

M. E.

. 2022. “Arrested Friendships? Justice Involvement and Interpersonal Exclusion among Rural Youth.” Journal of Research in Crime and Delinquency 59(3):365-409.

43.

Kahn

R. L.

Antonucci

T. C.

. 1980. “Convoys Over the Life Course: Attachment, Roles, and Social Support.” Life-span Development and Behavior 3:253-67.

44.

Knecht

Snijders

T. A.

Baerveldt

Steglich

C. E.

Raub

. 2010. “Friendship and Delinquency: Selection and Influence Processes in Early Adolescence.” Social Development 19(3):494-514.

45.

Kreager

D. A.

Rulison

Moody

. 2011. “Delinquency and the Structure of Adolescent Peer Groups.” Criminology 49(1):95-127.

46.

Langenkamp

A. G.

2010. “Academic Vulnerability and Resilience During the Transition to High School: The Role of Social Relationships and District Context.” Sociology of Education 83(1):1-19.

47.

Laub

J. H.

Sampson

R. J.

. 1993. “Turning Points in the Life Course: Why Change Matters to the Study of Crime.” Criminology 31(3):301-25.

48.

Laub

J. H.

Sampson

R. J.

. 2009. Shared Beginnings, Divergent Lives. Cambridge, MA: Harvard University Press.

49.

Light

J. M.

Greenan

C. C.

Rusby

J. C.

Nies

K. M.

Snijders

T. A.

. 2013. “Onset to First Alcohol Use in Early Adolescence: A Network Diffusion Model.” Journal of Research on Adolescence 23(3):487-99.

50.

Lospinoso

J. A.

Satchell

D. J.

. 2011. “Smoking Behavior and Friendship Formation: The Importance of Time Heterogeneity in Studying Social Network Dynamics.” Proceedings of the 44th Hawaii International Conference on System Sciences: 1-11.

51.

Lospinoso

J. A.

Schweinberger

Snijders

T. A. B.

Ripley

R. M.

. 2011. “Advancing and Accounting for Time Heterogeneity in Stochastic Actor Oriented Modes.” Advances in Data Analysis and Classification 5(2):147-76.

52.

Matza

1964. Delinquency and Drift. Piscataway, NJ: Transaction Publishers.

53.

McFarland

D. A.

Moody

Diehl

Smith

J. A.

Thomas

R. J.

. 2014. “Network Ecology and Adolescent Social Structure.” American Sociological Review 79(6):1088-121.

54.

McGloin

J. M.

Piquero

A. R.

. 2010. “On the Relationship Between co-Offending Network Redundancy and Offending Versatility.” Journal of Research in Crime and Delinquency 47(1):63-90.

55.

McMillan

Schaefer

D. R.

. 2021. “Comparing Targeting Strategies for Network-Based Adolescent Drinking Interventions: A Simulation Approach.” Social Science & Medicine 282:114136.

56.

McPherson

Smith-Lovin

Cook

J. M.

. 2001. “Birds of a Feather: Homophily in Social Networks.” Annual Review of Sociology 27(1):415-44.

57.

Moffitt

T. E.

1993. “Adolescence-limited and Life-Course-Persistent Antisocial Behavior: A Developmental Taxonomy.” Psychological Review 100:674-701.

58.

Murray

Loeber

Pardini

. 2012. “Parental Involvement in the Criminal Justice System and the Development of Youth Theft, Marijuana use, Depression, and Poor Academic Performance.” Criminology 50(1):255-302.

59.

Osgood

D. W.

Feinberg

M. E.

Ragan

D. T.

. 2015. “Social Networks and the Diffusion of Adolescent Problem Behavior: Reliable Estimates of Selection and Influence from Sixth Through Ninth Grades.” Prevention Science 16(6):832-43.

60.

Osterman

K. F.

Bacigalupo

D. C.

Garrett

. 2016. “Middle School Students’ Understanding of Peer Alcohol Use and Implications for Secondary School Leaders.” Journal of School Leadership 26(4):661-89.

61.

Petts

R. J.

2009. “Family and Religious Characteristics’ Influence on Delinquency Trajectories from Adolescence to Young Adulthood.” American Sociological Review 74(3):465-83.

62.

Pribesh

Downey

. 1999. “Why are Residential and School Moves Associated with Poor School Performance?” Demography 36(4):521-34.

63.

Pyrooz

D. C.

Melde

Coffman

D. L.

Meldrum

R. C.

. 2021. “Selection, Stability, and Spuriousness: Testing Gottfredson and Hirschi’s Propositions to Reinterpret Street Gangs in Self-Control Perspective.” Criminology 59(2):224-53.

64.

Ragan

D. T.

2020. “Similarity Between Deviant Peers: Developmental Trends in Influence and Selection.” Criminology 58(2):336-69.

65.

Ripley

R. M.

Snijders

T. A. B.

Boda

Vörös

Preciado

. 2022. Manual for SIENA Version 4.0 (Version March 18, 2022). Oxford: University of Oxford, Department of Statistics; Nuffield College.

66.

Sampson

R. J.

Laub

J. H.

. 1997 “A Life-Course Theory of Cumulative Disadvantage and the Stability of Delinquency.” Pp. 133-61 in Developmental Theories of Crime and Delinquency edited by Thornberry

T. P.

. New York: Routledge.

67.

Sampson

R. J.

Laub

J. H.

. 2005. “A Life-Course View of the Development of Crime.” Annals of the American Academy of Political and Social Science 602(1):12-45.

68.

Schaefer

D. R.

Rodriguez

Decker

S. H.

. 2014. “The Role of Neighborhood Context in Youth co-Offending.” Criminology 52(1):117-39.

69.

Schaefer

D. R.

Simpkins

S. D.

Vest

A. E.

Price

C. D.

. 2011. “The Contribution of Extracurricular Activities to Adolescent Friendships: New Insights Through Social Network Analysis.” Developmental Psychology 47(4):1141.

70.

Schiller

K. S.

1999. “Effects of Feeder Patterns on Students’ Transition to High School.” Sociology of Education 72(4):216-33.

71.

Schwerdt

West

M. R.

. 2013. “The Impact of Alternative Grade Configurations on Student Outcomes Through Middle and High School.” Journal of Public Economics 97:308-26.

72.

Seidman

Allen

Aber

J. L.

Mitchell

Feinman

. 1994. “The Impact of School Transitions in Early Adolescence on the Self-System and Perceived Social Context of Poor Urban Youth.” Child Development 65(2):507-22.

73.

Shi

Moody

. 2017. “Most Likely to Succeed: Long-run Returns to Adolescent Popularity.” Social Currents 4(1):13-33.

74.

Siennick

S. E.

Widdowson

A. O.

Ragan

D. T.

. 2017. “New Students’ Peer Integration and Exposure to Deviant Peers: Spurious Effects of School Moves?” The Journal of Early Adolescence 37(9):1254-79.

75.

Sijtsema

J. J.

Lindenberg

S. M.

. 2018. “Peer Influence in the Development of Adolescent Antisocial Behavior: Advances from Dynamic Social Network Studies.” Developmental Review 50:140-54.

76.

Simons

R. L.

Whitbeck

L. B.

Conger

R. D.

Conger

K. J.

. 1991. “Parenting Factors, Social Skills, and Value Commitments as Precursors to School Failure, Involvement with Deviant Peers, and Delinquent Behavior.” Journal of Youth and Adolescence 20(6):645-64.

77.

Snijders

T. A.

Van de Bunt

G. G.

Steglich

C. E.

. 2010. “Introduction to Stochastic Actor-Based Models for Network Dynamics.” Social Networks 32(1):44-60.

78.

Staff

Kreager

D. A.

. 2008. “Too Cool for School? Violence, Peer status and High School Dropout.” Social Forces 87(1):445-71.

79.

Staff

Osgood

D. W.

Schulenberg

J. E.

Bachman

J. G.

Messersmith

E. E.

. 2010. “Explaining the Relationship Between Employment and Juvenile Delinquency.” Criminology 48(4):1101-31.

80.

Steglich

Snijders

T. A.

Pearson

. 2010. “Dynamic Networks and Behavior: Separating Selection from Influence”. Sociological Methodology 40(1):329-93.

81.

Sutherland

1939 “Principles of Differential Association.” Pp. 4-8 in Principles of Criminology. Chicago: J.B. Lippincott Company.

82.

Sweeten

2012. “Scaling Criminal Offending.” Journal of Quantitative Criminology 28(3):533-57.

83.

Sweeten

Pyrooz

D. C.

Piquero

A. R.

. 2013. “Disengaging from Gangs and Desistance from Crime.” Justice Quarterly 30(3):469-500.

84.

Temkin

D. A.

Gest

S. D.

Osgood

D. W.

Feinberg

Moody

. 2018. “Social Network Implications of Normative School Transitions in non-Urban School Districts.” Youth & Society 50(4):462-84.

85.

Ueno

2005. “The Effects of Friendship Networks on Adolescent Depressive Symptoms.” Social Science Research 34(3):484-510.

86.

Veenstra

Laninga-Wijnen

. 2023 “The Prominence of Peer Interactions, Relationships, and Networks in Adolescence and Early Adulthood.” Pp. 225-41 in APA Handbook of Adolescent and Young Adult Development edited by Crockett

L. J.

Carlo

Schulenberg

J. E.

. Washington, DC: American Psychological Association.

87.

Vignery

2022. “From Networked Students’ Centrality to Student Networks’ Density: What Really Matters for Student Performance?” Social Networks 70:166-86.

88.

Warr

1993. “Age, Peers, and Delinquency.” Criminology 31(1):17-40.

89.

Warr

1998. “Life-Course Transitions and Desistance from Crime.” Criminology 36(2):183-216.

90.

Weerman

F. M.

2011. “Delinquent Peers in Context: A Longitudinal Network Analysis of Selection and Influence Effects.” Criminology 49(1):253-86.

91.

Weiss

C. C.

Bearman

P. S.

. 2007. “Fresh Starts: Reinvestigating the Effects of the Transition to High School on Student Outcomes.” American Journal of Education 113(3):395-421.

92.

White

Corbett

. 2014. Doing Educational Research in Rural Settings: Methodological Issues, International Perspectives and Practical Solutions. New York: Routledge.

93.

Wrzus

Hänel

Wagner

Neyer

F. J.

. 2013. “Social Network Changes and Life Events Across the Life Span: A Meta-Analysis.” Psychological Bulletin 139(1):53-80.

94.

Yohros

Zimmerman

G. M.

. 2020. “Does the Residential Landscape Contextualize Friendships? Examining the Causes and Consequences of Affiliating with Older Friends.” Journal of Research in Crime and Delinquency 57(5):571-611.

95.

Young

J. T. N.

2014. “Role Magnets”? An Empirical Investigation of Popularity Trajectories for Life-Course Persistent Individuals During Adolescence.” Journal of Youth and Adolescence 43(1):104-15.

96.

Zuckerman

1994. Behavioral Expressions and Biosocial Bases of Sensation Seeking. New York: Cambridge University Press.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.09 MB