Sources and References

This document draws on the following scholarly works and research contributions: Ronald L. Akers, Social Learning and Social Structure and related chapters on social learning theory; Jean Marie McGloin and Kyle J. Thomas on peer influence mechanisms and network dynamics; Bill McCarthy on differential association and its empirical measurement; Owen Gallupe and colleagues on deviant modeling and observational learning; Frank M. Weerman and Wilma H. Smeenk on peer similarity and the selection–socialization debate; Robert J. Sampson on techniques of neutralization and cognitive frameworks for moral disengagement; Chad Posick and Michael Rocque on the tension between control theory and cultural deviance perspectives; Peggy C. Giordano on life-course transitions and peer relationships; Timothy McCuddy on social media, digital peer networks, and online deviance; Evelien M. Hoeben and Kyle J. Thomas on decision-making processes in peer contexts; Dana L. Haynie on network analysis, social status, and delinquent peer influence; John H. Boman IV and Thomas J. Mowen on peer-based interventions and reentry programming; Faye S. Taxman on evidence-based correctional policy; Andrew V. Papachristos and David S. Kirk on street dynamics, social networks, and neighborhood violence. Additional theoretical foundations are drawn from Edwin H. Sutherland’s differential association theory, Mark Warr’s work on peer influence and delinquency, David Matza’s drift theory and techniques of neutralization, Albert Bandura’s social learning and social cognitive theory, Travis Hirschi’s social control theory, Robert Burgess and Ronald Akers’s differential association–reinforcement theory, Terrie Moffitt’s developmental taxonomy, Robert Sampson and John Laub’s age-graded theory of informal social control, Clifford Shaw and Henry McKay’s social disorganization theory, Terence Thornberry’s interactional theory, Frederic Thrasher’s early gang research, James F. Short Jr.’s group process studies, Scott Decker and David Pyrooz on modern gang dynamics, Warr’s work on peer fear and loyalty, Christopher Sullivan’s work on the peer delinquency paradox, Henneberger and colleagues on gender differences in peer influence, and cross-cultural research from multiple national contexts.

Chapter 1: Foundations of Peer Influence in Criminology

The Centrality of Peers in Understanding Crime

The study of crime has long recognized that delinquency (青少年犯罪) is overwhelmingly a group phenomenon. From the earliest sociological investigations of urban youth gangs in Chicago to contemporary network analyses of gun violence, scholars have returned again and again to the observation that criminal behavior is not typically a solitary enterprise. Adolescents and young adults who engage in law-violating behavior tend to do so in the company of others, and the people with whom one associates exert a powerful influence on whether, when, and how one offends. This insight forms the bedrock of an entire tradition in criminological thought, one that stretches from Edwin Sutherland’s foundational propositions through Ronald Akers’s elaboration of social learning mechanisms and into the sophisticated network analyses of the twenty-first century.

The relationship between peers and crime is one of the most robust empirical findings in all of criminology. Across decades of research, spanning self-report surveys, longitudinal panel studies, experimental designs, and ethnographic fieldwork, the association between having delinquent friends and engaging in delinquent behavior has been confirmed with remarkable consistency. Yet this apparent simplicity masks profound theoretical complexity. What, precisely, is the mechanism by which peers influence criminal behavior? Do individuals learn to offend from their associates, or do they simply gravitate toward others who share their pre-existing antisocial inclinations? How should researchers measure the peer–delinquency relationship without introducing systematic biases? And can interventions that target peer dynamics effectively reduce crime?

These questions animate the field of peers and crime, and answering them requires careful engagement with multiple theoretical traditions, methodological innovations, and interdisciplinary insights. This text examines the full breadth of this enterprise, from classical theory to cutting-edge empirical research, providing a comprehensive account of how peers shape criminal trajectories and how society might respond.

The Chicago School and Social Disorganization as Precursor

To understand Sutherland’s differential association theory, one must first appreciate the intellectual milieu from which it emerged. The Chicago School (芝加哥学派) of sociology, centered at the University of Chicago in the early twentieth century, fundamentally shaped how criminologists thought about the social origins of crime. Clifford Shaw and Henry McKay’s social disorganization theory (社会解组理论) demonstrated that crime rates remained persistently high in certain urban neighborhoods regardless of which ethnic or racial groups inhabited them at any given time. This finding was revolutionary because it directed attention away from individual pathology and toward the social-structural conditions of communities — poverty, residential mobility, and ethnic heterogeneity — that weakened the capacity of local institutions to exercise informal social control.

Shaw and McKay’s research on delinquency areas revealed that criminal traditions were transmitted across generations of residents within disorganized neighborhoods. Their concept of cultural transmission (文化传递) — the idea that criminal values, techniques, and opportunities were passed from older to younger residents through everyday social interaction — provided the immediate intellectual precursor to Sutherland’s differential association theory. Shaw and McKay documented how delinquent gangs served as vehicles for this transmission, socializing new members into patterns of criminal behavior through mentoring, modeling, and the enforcement of group norms. Their classic studies, including The Jack-Roller and The Natural History of a Delinquent Career, provided vivid ethnographic evidence of how individual criminal careers were shaped by the social environments in which they unfolded.

The ecological approach of the Chicago School also introduced the idea that crime was not randomly distributed across urban space but was concentrated in particular areas characterized by specific structural features. This spatial dimension of crime would later inform network approaches to understanding how criminal behavior diffuses through social ties, but in Shaw and McKay’s formulation, the emphasis was on the neighborhood as the unit of analysis and on the mechanisms — including peer association — through which structural disadvantage was translated into individual criminal behavior.

Sutherland’s Differential Association Theory

Sutherland’s differential association (差别交往理论) theory, first articulated in 1939 and refined in subsequent editions of his influential textbook, represents the single most important theoretical statement on the role of peers in crime. Sutherland proposed nine principles that together constitute a comprehensive learning theory of criminal behavior. At its core, the theory asserts that a person becomes criminal when definitions favorable to violation of law exceed definitions unfavorable to violation of law. These definitions are acquired through associations with others, and the impact of these associations varies according to their frequency, duration, priority, and intensity.

Sutherland’s Nine Propositions in Detail

Sutherland’s nine propositions merit individual examination, as each contributes a distinct element to the theoretical framework. The first proposition states that criminal behavior is learned, not inherited or invented anew by each offender. This seemingly simple assertion carried enormous theoretical weight in an era when biological and psychiatric explanations of crime dominated scholarly and public discourse. The second proposition specifies that criminal behavior is learned in interaction with other persons in a process of communication — primarily verbal, but also gestural. The third proposition narrows the locus of learning to intimate personal groups (亲密人际群体), explicitly rejecting the idea that impersonal agencies of communication such as newspapers or movies play a significant role. This proposition has been contested in the digital age, as online peer groups blur the distinction between intimate and impersonal communication.

The fourth proposition states that learning criminal behavior includes both the techniques of committing crime (which may be simple or complex) and the specific direction of motives, drives, rationalizations, and attitudes. The fifth proposition specifies that the direction of motives and drives is learned from definitions of legal codes as favorable or unfavorable. The sixth proposition — the principle of differential association itself — states that a person becomes delinquent because of an excess of definitions favorable to violation of law over definitions unfavorable to violation of law. This is the theoretical core of the entire framework, and it distinguishes differential association from simple “bad company” explanations by emphasizing the cognitive content of associations rather than merely the behavioral characteristics of associates.

The seventh proposition introduces the crucial modalities of association (交往的多种模式): differential associations vary in frequency, duration, priority, and intensity. Frequency refers to how often one interacts with particular associates; duration refers to how long associations last; priority refers to how early in life the associations are formed; and intensity refers to the emotional significance and prestige of the associational partners. The eighth proposition states that the process of learning criminal behavior by association involves all of the mechanisms involved in any other learning — a deliberate effort to place criminal learning on the same footing as conventional learning. The ninth proposition clarifies that while criminal behavior is an expression of general needs and values, it is not explained by those general needs and values, since conforming behavior is an expression of the same needs and values.

Several features of Sutherland’s theory deserve emphasis. First, the theory is explicitly anti-individualistic: it locates the causes of crime not in biological predisposition, psychological deficit, or rational calculation, but in the normative content of one’s social relationships. Second, it is a general theory, intended to explain all forms of criminal behavior, from juvenile delinquency to white-collar crime. Third, it treats criminal behavior as fundamentally no different from conforming behavior in its learning processes; the difference lies only in the content of what is learned. Fourth, and crucially for the study of peers, Sutherland specified that the most important learning occurs in intimate personal groups, not through mass media or impersonal contact.

McCarthy’s empirical work on differential association has been instrumental in demonstrating that the theory’s core propositions can be operationalized and tested with considerable precision. McCarthy showed that the specific dimensions Sutherland identified — frequency, duration, priority, and intensity of associations — each contribute independently to the prediction of criminal behavior. This research was important because it moved beyond the common practice of simply counting the number of delinquent friends a person has and instead measured the qualitative dimensions of associational patterns that Sutherland had theorized were crucial. McCarthy’s findings confirmed that not all associations are equal: close, frequent, long-standing relationships with criminal others exert far more influence than casual or superficial contacts.

Akers and Social Learning Theory

If Sutherland provided the conceptual blueprint, it was Ronald Akers who constructed the theoretical architecture that would allow differential association to be tested, refined, and extended. Social learning theory (社会学习理论), as developed by Akers in collaboration with Robert Burgess and subsequently elaborated across several decades of scholarship, integrates Sutherland’s sociological insights with the behavioral and cognitive principles of modern learning theory, particularly the work of Albert Bandura and B. F. Skinner.

The initial Burgess-Akers formulation, published in 1966 as differential association-reinforcement theory (差别交往-强化理论), represented the first systematic effort to restate Sutherland’s propositions in the language of modern behavioral psychology. Burgess and Akers drew on operant conditioning principles to specify the mechanisms through which criminal definitions and behaviors are acquired, maintained, and extinguished. This reformulation addressed a key criticism of Sutherland’s original theory — that it was too vague about the actual learning processes involved — by grounding the theory in established principles of behavioral science. Over the following decades, Akers refined this initial statement into a comprehensive social learning theory that incorporated not only operant conditioning but also Bandura’s cognitive social learning theory, with its emphasis on observational learning, self-regulation, and reciprocal determinism.

Akers identifies four principal mechanisms through which criminal behavior is learned and maintained:

Differential Association

In Akers’s reformulation, differential association refers to the process of interacting with others who model criminal behavior, express attitudes favorable to crime, and provide the social context within which criminal learning occurs. This concept retains the core of Sutherland’s original formulation but gives it sharper operational definition. Differential association encompasses both direct interaction with criminal peers and exposure to normative definitions that either encourage or discourage law violation. Akers emphasized that differential association is the mechanism that provides the social context for the operation of the other three learning mechanisms; it is the process through which individuals are exposed to definitions, models, and reinforcement contingencies.

Definitions

Definitions (定义) are the attitudes, rationalizations, and orientations toward behavior that individuals acquire through social interaction. Akers distinguishes between general definitions — broad moral beliefs about right and wrong — and specific definitions — attitudes toward particular acts in particular situations. A person may hold a general belief that stealing is wrong while simultaneously maintaining a specific definition that shoplifting from large corporations is acceptable. These definitions serve as discriminative stimuli that signal whether a given behavior is likely to be rewarded or punished in a particular context.

Akers further distinguished between positive definitions (those that define behavior as morally desirable or wholly permissible), neutralizing definitions (those that justify or excuse behavior that would otherwise be considered wrong, paralleling Sykes and Matza’s techniques of neutralization), and negative definitions (those that define behavior as morally undesirable). The balance of these definitions — the ratio of favorable to unfavorable definitions to which one has been exposed — determines the probability that an individual will engage in the behavior in question.

Differential Reinforcement

The concept of differential reinforcement (差别强化) draws directly from operant conditioning principles. Behavior is shaped by its consequences: positive reinforcement (gaining rewards), negative reinforcement (removing aversive stimuli), positive punishment (receiving penalties), and negative punishment (losing rewards). In the context of peer influence, the most important forms of reinforcement are often social: approval, status, acceptance, and belonging. When a peer group rewards criminal behavior with admiration and respect, and when conformity is met with ridicule or exclusion, the reinforcement contingencies powerfully favor delinquency.

Akers emphasized that reinforcement operates through both direct and vicarious channels. Direct reinforcement involves the consequences that follow one’s own behavior, while vicarious reinforcement involves observing the consequences that follow others’ behavior. An adolescent who sees a peer gain social status and material rewards through drug dealing experiences vicarious reinforcement that increases the probability that the observer will consider similar behavior. Importantly, the anticipation of reinforcement — the expectation that certain consequences will follow a behavior — is itself a powerful determinant of action, meaning that reinforcement processes operate cognitively as well as behaviorally.

Imitation

Imitation (模仿), also termed modeling or observational learning, refers to the process of acquiring behavior by observing others. Drawing on Bandura’s extensive experimental work, Akers argues that individuals can learn criminal behavior by watching others engage in it, observing the consequences that follow, and forming expectations about the likely outcomes of similar behavior. Imitation is particularly important in the initial acquisition of novel forms of behavior — a young person may learn how to shoplift, use drugs, or commit a robbery by watching a more experienced peer — but its influence typically diminishes relative to reinforcement processes once the behavior has been established.

Bandura’s original research identified four component processes of observational learning: attention (the observer must attend to the model’s behavior), retention (the observer must remember the modeled behavior), motor reproduction (the observer must be capable of performing the behavior), and motivation (the observer must have reason to perform the behavior). Each of these processes can be influenced by the characteristics of the model, the observer, and the social context. Models who are perceived as having high status, competence, and social power are more likely to be attended to and imitated, a finding that has direct implications for understanding peer influence in hierarchically organized groups such as gangs.

Gallupe and colleagues’ research on deviant modeling (越轨行为模仿) provides important empirical support for the imitation component of social learning theory. Their work demonstrates that exposure to peers who model deviant behavior significantly increases the likelihood that an observer will engage in similar acts, even after controlling for the observer’s pre-existing attitudes and prior behavior. Critically, Gallupe et al. found that the effect of deviant modeling operates through both direct observation (seeing a peer commit a deviant act) and indirect communication (hearing a peer describe deviant acts they have committed). This distinction matters because it suggests that peer influence does not require physical co-presence; narratives and accounts of deviant behavior can themselves serve as powerful modeling stimuli.

The SSSL Model

Akers’s mature theoretical framework is encapsulated in the Social Structure–Social Learning (SSSL) model (社会结构-社会学习模型), which represents his effort to integrate macro-level structural variables with micro-level learning processes. The SSSL model proposes that social structural conditions — including differential social organization, differential location in the social structure (class, gender, race, age), theoretically defined structural variables (anomie, social disorganization, group conflict), and differential social location (family, peers, school, community) — affect individual behavior primarily through their influence on the social learning variables of differential association, definitions, differential reinforcement, and imitation.

The SSSL model addresses a persistent criticism of social learning theory — that it is purely a micro-level theory that ignores the structural and cultural context in which learning occurs. By specifying the pathways through which structural variables influence learning processes, Akers provided a framework for understanding why rates of crime vary across communities, social groups, and historical periods while maintaining that the proximate causes of individual criminal behavior are the social learning mechanisms he identified. Structural disadvantage, for example, increases crime not by directly motivating individuals to offend but by increasing their exposure to criminal models, criminal definitions, and reinforcement contingencies that favor criminal behavior.

Akers’s social learning theory has been subjected to extensive empirical testing and has consistently received strong support. Meta-analyses of the literature confirm that social learning variables — particularly differential association and definitions — are among the strongest predictors of criminal and deviant behavior across a wide range of offenses, populations, and research designs. Pratt and colleagues’ meta-analysis found that social learning variables explained more variance in criminal behavior than variables from any other major criminological theory, a finding that has been replicated across multiple reviews.

Chapter 2: Measuring Peer Effects and Methodological Challenges

The Measurement Problem

One of the most vexing challenges in the study of peers and crime concerns how to measure peer delinquency (同伴犯罪行为). The most common approach in criminological research has been to ask respondents to report on the delinquent behavior of their friends — a method known as perceptual measurement. Respondents are typically asked questions such as “How many of your close friends have stolen something worth more than $50 in the past year?” or “How often do your friends use illegal drugs?” The resulting measure of peer delinquency is then correlated with the respondent’s own self-reported delinquency.

This approach, while practical, introduces a serious methodological concern: projection bias (投射偏差). Respondents may systematically overestimate the similarity between their own behavior and that of their friends, either because they assume their friends behave as they do (false consensus) or because they use their own behavior as an anchor when estimating their friends’ behavior. If projection bias inflates the correlation between self-reported delinquency and perceived peer delinquency, then the apparent influence of peers on crime may be substantially overstated.

The Peer Delinquency Paradox

The measurement problem gives rise to what has been termed the peer delinquency paradox (同伴犯罪悖论). The paradox consists in the observation that the correlation between an individual’s own delinquency and their perception of their friends’ delinquency is consistently stronger than the correlation between an individual’s own delinquency and their friends’ actual self-reported delinquency. In other words, respondents perceive their friends as more similar to themselves than those friends actually are. This discrepancy raises troubling questions about the validity of the vast body of research that relies on perceptual measures of peer delinquency.

The paradox has multiple potential explanations. Projection bias, as described above, is one. Another is selective attention (选择性注意): individuals may disproportionately notice and remember instances of their friends’ behavior that are consistent with their own, while failing to attend to or recall inconsistent behavior. A third explanation involves social desirability (社会期望偏差): friends who are reporting their own behavior may underreport delinquency, making direct measures lower than they “should” be. The paradox thus may reflect not only inflated perceptual measures but also deflated direct measures, with the true level of peer similarity falling somewhere between the two.

Weerman and Smeenk addressed this problem directly by comparing perceptual measures of peer delinquency (respondent reports about friends) with direct measures (friends’ own self-reports). Their findings revealed that while perceptual measures do tend to inflate the peer–delinquency correlation, the association remains significant and substantively important even when direct measures are used. This research provided crucial evidence that peer influence is not merely a methodological artifact but a genuine causal process, albeit one whose magnitude may be somewhat exaggerated by conventional measurement approaches.

Warr on Peer Fear and Loyalty

Mark Warr’s contributions to understanding peer influence extend beyond the measurement issue to encompass the emotional and relational dynamics that drive conformity within peer groups. Warr’s research on peer fear (同伴恐惧) and peer loyalty (同伴忠诚) revealed that adolescents often participate in delinquent acts not because they have internalized deviant definitions or because they find the behavior intrinsically rewarding, but because they fear the social consequences of refusal. The fear of ridicule, exclusion, or loss of status within the peer group can be a more powerful motivator than any positive attraction to the delinquent behavior itself.

Warr found that loyalty to friends — the sense of obligation and emotional commitment that characterizes close relationships — also plays a crucial role. Adolescents may engage in delinquency to demonstrate loyalty to their friends, to avoid appearing disloyal or cowardly, or to fulfill perceived obligations within the friendship group. This finding complicates the social learning framework by suggesting that peer influence operates not only through the cognitive mechanisms of definitions and reinforcement but also through affective mechanisms rooted in the emotional dynamics of friendship. An adolescent who participates in a robbery not because he believes it is acceptable but because he fears losing his friends if he refuses is being influenced by a qualitatively different process than one who has internalized definitions favorable to robbery.

Warr also demonstrated that the structure of peer groups matters for understanding these dynamics. In groups with a clear leader or dominant individual, fear of that specific person may drive conformity. In more egalitarian groups, the diffuse pressure to maintain group cohesion and avoid being seen as an outsider may be more important. These relational dynamics cannot be captured by standard measures of peer delinquency, which typically focus on the behavioral characteristics of friends rather than the emotional quality of the relationships.

Perceptual versus Behavioral Measures

The distinction between perceptual measures (感知测量) and behavioral measures (行为测量) of peer influence has generated a substantial methodological literature. Perceptual measures ask respondents about their friends’ behavior; behavioral measures obtain information directly from the friends themselves. While behavioral measures are generally considered more valid, they are also more costly and difficult to obtain, requiring data collection from complete friendship networks rather than from individual respondents.

Research comparing the two types of measures has consistently found that perceptual measures produce stronger associations with respondent delinquency than behavioral measures do. However, this does not mean that perceptual measures are simply “wrong.” Some scholars have argued that perceptions of peer behavior may be more causally relevant than actual peer behavior, because it is the perceived normative environment — not the objective normative environment — that shapes individual behavior. If an adolescent believes that all of her friends use drugs, this perception may influence her behavior regardless of whether it is accurate. From this perspective, the “bias” in perceptual measures may actually capture an important aspect of the peer influence process.

Selection versus Socialization

The selection–socialization debate (选择-社会化辩论) represents one of the most enduring controversies in the study of peers and crime. The core question is deceptively simple: does associating with delinquent peers cause an individual to become more delinquent (socialization), or do individuals who are already inclined toward delinquency seek out like-minded peers (selection)?

The socialization perspective aligns naturally with differential association and social learning theories, which posit that criminal behavior is learned through interaction with others. From this standpoint, having delinquent friends is a cause of delinquency: exposure to criminal definitions, modeling, and reinforcement processes transforms the individual over time. The selection perspective, by contrast, is more consistent with control theory and propensity-based approaches, which argue that the peer–delinquency correlation arises because individuals with weak social bonds or high criminal propensity gravitate toward each other. On this view, delinquent peer association is a consequence, not a cause, of individual-level risk factors.

Thornberry’s Interactional Theory

Terence Thornberry’s interactional theory (互动理论) offered a sophisticated resolution to the selection–socialization impasse by proposing that the relationship between delinquent peers and delinquent behavior is reciprocal and dynamic. Rather than treating selection and socialization as competing explanations, Thornberry argued that they are complementary processes that operate simultaneously, each reinforcing the other over time. Individuals with weakened bonds to conventional institutions (family, school) are more likely to associate with delinquent peers (selection), and association with delinquent peers further weakens conventional bonds while strengthening delinquent behavior (socialization). This delinquent behavior, in turn, deepens involvement with delinquent peers, creating a self-reinforcing cycle that Thornberry termed bi-directional causation (双向因果关系).

Thornberry’s interactional theory also introduced an explicitly developmental dimension, arguing that the relative importance of different causal factors changes across the life course. In early adolescence, weakened attachment to parents is the primary factor driving association with delinquent peers. In middle adolescence, commitment to school and conventional activities becomes more salient. In late adolescence and early adulthood, commitment to conventional adult roles (employment, intimate partnerships) increasingly supplants peer relationships as the primary source of social influence. This developmental sequencing means that effective intervention strategies must be calibrated to the specific developmental stage of the target population.

Longitudinal Evidence

Decades of research have converged on a middle-ground position: both selection and socialization processes operate, often simultaneously and reciprocally. Longitudinal studies that track individuals over time have consistently found evidence for both effects, though their relative magnitude varies across studies, developmental periods, and types of offending.

Three major longitudinal studies have been particularly influential in disentangling selection and socialization effects. The Rochester Youth Development Study (罗切斯特青年发展研究), directed by Thornberry, followed a sample of urban youth from early adolescence through young adulthood and provided strong evidence for the reciprocal relationship between delinquent peers and delinquent behavior that interactional theory predicted. The study found that association with delinquent peers predicted subsequent increases in delinquency (socialization), while prior delinquency predicted subsequent increases in association with delinquent peers (selection), with both processes operating simultaneously across developmental periods.

The Pittsburgh Youth Study (匹兹堡青年研究), directed by Rolf Loeber, followed three cohorts of boys from childhood through adulthood and provided detailed information on the developmental sequencing of peer influence and delinquent behavior. The Pittsburgh data revealed that association with deviant peers typically preceded the escalation of delinquent behavior, supporting the socialization hypothesis, but also that early behavioral problems predicted later association with deviant peers, supporting the selection hypothesis. Notably, the Pittsburgh study found that the socialization effect was stronger for some types of offending (particularly drug use) than for others (particularly violence).

The Denver Youth Survey (丹佛青年调查), directed by David Huizinga, provided additional longitudinal evidence from a different urban context and confirmed the bidirectional relationship between peers and delinquency. The Denver data also highlighted the importance of peer group stability (同伴群体稳定性): individuals whose peer groups were more stable over time showed stronger socialization effects, suggesting that the duration and consistency of peer associations moderates their influence on behavior.

McGloin and Thomas made important contributions to this debate by examining the mechanisms through which peer influence (同伴影响) operates, arguing that the field needed to move beyond the simple question of “selection or socialization” and instead investigate the specific processes by which peers shape behavior in real-time interactional contexts.

Innovative Measurement Approaches

McCarthy’s work on operationalizing the dimensions of differential association — frequency, duration, priority, and intensity — represented one important methodological advance. By moving beyond simple counts of delinquent friends and instead measuring the qualitative characteristics of peer relationships, McCarthy demonstrated that the strength of peer influence depends not just on how many delinquent friends one has but on how close, how frequent, and how long-standing those relationships are.

Gallupe and colleagues contributed additional methodological innovation by developing measures of deviant modeling (越轨行为模仿) that distinguish between direct observation of peer behavior and indirect exposure through narratives and accounts. This distinction is theoretically important because it maps onto different learning mechanisms (observational learning versus symbolic communication) and has practical implications for understanding how peer influence operates in an era of digital communication, where much exposure to peer behavior occurs through mediated channels rather than face-to-face interaction.

Network-based measurement represents perhaps the most significant methodological advance in recent decades. Rather than relying on respondent perceptions of their friends’ behavior, network approaches collect data from all members of a bounded social unit (such as a school or neighborhood) and use nomination data to reconstruct the structure of social relationships. This approach allows researchers to measure peer influence and peer selection simultaneously, to identify the role of indirect ties and network position, and to test hypotheses about how the structure of social relationships — not just their content — shapes criminal behavior.

Chapter 3: Challenges to the Peer Influence Paradigm

Control Theory and the Critique of Cultural Deviance

Not all criminological theories assign peers a causal role in producing criminal behavior. Social control theory (社会控制理论), most fully articulated by Travis Hirschi, begins from the premise that the motivation to commit crime is universal — or at least sufficiently widespread as to require no special explanation — and asks instead why most people conform. The answer, according to control theory, lies in the social bonds that tie individuals to conventional society: attachment to significant others, commitment to conventional pursuits, involvement in conventional activities, and belief in the moral validity of social rules.

From the control theory perspective, the correlation between delinquent peers and delinquency is largely spurious. Individuals with weak social bonds are simultaneously more likely to engage in delinquency (because they lack the stakes in conformity that restrain most people) and more likely to associate with delinquent peers (because they are freed from conventional relationships and attracted to others in similar circumstances). Peer association, on this view, is an indicator of low social control rather than a cause of crime.

Posick and Rocque provide a careful analysis of the tension between control theory (控制理论) and what they term the cultural deviance (文化越轨) perspective. The cultural deviance label, which Posick and Rocque use to characterize theories like differential association and social learning, implies that crime results from conformity to deviant cultural norms learned through association with others. Posick and Rocque argue that the empirical evidence does not clearly favor one perspective over the other, and that both control and learning processes likely contribute to criminal behavior. Their work highlights the importance of theoretical integration, suggesting that a complete account of peers and crime must incorporate both the motivational processes emphasized by learning theories and the restraining processes emphasized by control theories.

Techniques of Neutralization and Cognitive Frameworks

A related challenge to straightforward peer influence accounts comes from research on techniques of neutralization (中和技术). Originally formulated by Gresham Sykes and David Matza, neutralization theory argues that most delinquents hold conventional moral beliefs but employ specific cognitive techniques to temporarily suspend those beliefs, thereby freeing themselves to engage in law-violating behavior without experiencing guilt or self-censure.

Sampson’s work on techniques of neutralization extends this theoretical tradition by examining how neutralizing definitions are transmitted and reinforced within peer groups. Sampson demonstrates that peer groups do not merely provide opportunities for crime or model criminal behavior; they also provide the cognitive frameworks that individuals use to justify their actions. When a peer group collectively endorses neutralizing definitions — “everyone steals from that store,” “the police don’t care about this neighborhood,” “we’re just having fun” — these definitions become part of the shared normative vocabulary of the group, making it easier for individual members to overcome moral inhibitions.

This insight is important because it suggests that peer influence operates at the level of cognition and meaning, not just behavior. Peers shape not only what individuals do but how they think about what they do. The social construction of definitions favorable to law violation is a collective process, one that unfolds through everyday conversation, storytelling, and the gradual negotiation of group norms. Sampson’s research thus bridges the gap between Sutherland’s emphasis on definitions and Matza’s emphasis on neutralization, showing that the two concepts describe overlapping aspects of the same social learning process.

Drift Theory and Moral Ambiguity

David Matza’s broader theoretical framework, often referred to as drift theory (漂移理论), provides additional nuance to the understanding of peer influence. Matza argued that most delinquents are not deeply committed to a criminal lifestyle but rather “drift” between conformity and delinquency, occupying a liminal space in which they are neither fully constrained by conventional norms nor fully liberated from them. In this state of drift, situational factors — including the immediate presence and influence of peers — can tip the balance toward or away from criminal behavior.

Drift theory challenges both the deterministic implications of differential association (which might seem to suggest that exposure to criminal definitions inevitably produces criminal behavior) and the static implications of control theory (which might seem to suggest that individuals with weak bonds are uniformly predisposed to crime). Instead, Matza’s framework emphasizes contingency, context, and the fluid nature of moral commitment, providing a more realistic account of the intermittent and episodic character of most delinquent careers.

Chapter 4: Social Networks, Status, and Peer Dynamics

Network Analysis in Criminology

The application of social network analysis (社会网络分析) to the study of crime represents one of the most significant methodological and theoretical advances in contemporary criminology. Traditional approaches to studying peers and crime typically measured the characteristics of an individual’s immediate friends — how many are delinquent, how close the relationships are, and so on. Network analysis goes further by examining the structure of relationships within a population, mapping patterns of connection and disconnection, and identifying how an individual’s position within a larger social structure shapes their behavior.

Dana Haynie’s pioneering research on social networks (社会网络) and delinquency demonstrated that network position — particularly measures of centrality, popularity, and prestige — significantly predicts involvement in delinquent behavior. Haynie found that adolescents who occupy central positions in their peer networks are more susceptible to the influence of their network contacts. Moreover, the effect of peer delinquency on individual behavior is conditioned by network structure: having delinquent friends matters more when those friends are themselves well-connected and when the individual’s network is dense (meaning that friends are connected to each other).

Haynie’s work also explored the concept of social status (社会地位) within peer networks. High-status individuals — those who are popular, admired, or otherwise influential — disproportionately shape group norms and behavioral expectations. When high-status peers engage in delinquency, the normalizing effect on others in the network is greater than when low-status peers do so. This finding aligns with social learning theory’s emphasis on the characteristics of models: Akers and Bandura both predict that modeling effects are strongest when the model is perceived as having high status, competence, or social power.

Network Exposure Models

Building on Haynie’s foundational work, scholars have developed increasingly sophisticated network exposure models (网络暴露模型) that quantify an individual’s exposure to delinquent behavior through their position in a social network. Rather than simply counting the number of delinquent friends, network exposure models calculate a weighted measure of exposure that takes into account the delinquency levels of all network contacts, the strength of ties to those contacts, and the position of both the individual and their contacts within the broader network structure.

These models have revealed several important findings. First, indirect exposure to delinquency — contact with the friends of one’s friends, even when those individuals are not personally known — significantly predicts individual delinquent behavior. This suggests that peer influence extends beyond the immediate friendship circle to encompass broader network neighborhoods. Second, the effect of network exposure varies depending on the individual’s own position in the network: centrally located individuals are more strongly affected by their network exposure than peripheral individuals. Third, the composition of one’s network matters in complex ways: having a few highly delinquent friends may exert more influence than having many moderately delinquent friends, depending on the strength and structure of those relationships.

The Embeddedness of Crime in Social Structure

Papachristos and Kirk extended the network perspective to examine how violence is embedded in the social structure (社会结构) of urban neighborhoods. Their research on street dynamics demonstrated that gun violence spreads through social networks in patterns that resemble the diffusion of a contagious disease. Individuals who are socially connected to victims of gun violence face substantially elevated risks of victimization themselves, and this risk propagates through network ties in predictable ways.

Social Contagion of Gun Violence

Papachristos’s research on the social contagion (社会传染) of gun violence represents one of the most important applications of network analysis to criminology. Working with data from several American cities, Papachristos demonstrated that gun violence clusters not randomly across urban space but within specific network communities — densely connected subgroups within the broader social network. Individuals whose network contacts have been involved in gun violence (as either perpetrators or victims) face dramatically elevated risks of gun violence involvement themselves, even after controlling for individual-level characteristics such as age, race, gang membership, and prior criminal history.

The contagion metaphor is not merely rhetorical. Papachristos and colleagues demonstrated that the network diffusion of gun violence follows epidemiological patterns, with identifiable risk factors (network proximity to prior violence), incubation periods (the temporal lag between exposure and involvement), and dose-response relationships (greater exposure to violence in one’s network predicts greater risk). This framing has profound implications for intervention, suggesting that violence prevention might benefit from the same strategies used in disease control: identifying high-risk individuals through network mapping, deploying “treatment” (conflict mediation, case management) to those at greatest risk, and interrupting transmission pathways before violence spreads further.

This network embeddedness perspective has profound implications for understanding peer influence. It suggests that the relevant “peers” for crime are not limited to close friends but extend to the broader network of social connections within which an individual is situated. Exposure to violence, criminal opportunity, and normative definitions may flow through weak ties as well as strong ones, through acquaintances as well as intimates, and through indirect connections as well as direct ones.

Papachristos and Kirk also demonstrated that network dynamics play a crucial role in the escalation and perpetuation of violence. Retaliatory violence follows network pathways, with acts of violence targeting not just the original offender but members of the offender’s broader network. This pattern creates cycles of reciprocal violence that are sustained by the structure of social relationships, making it extremely difficult for any individual to exit the cycle without severing network ties.

Co-Offending Networks

Research on co-offending networks (共同犯罪网络) has revealed additional dimensions of the relationship between social structure and criminal behavior. Co-offending refers to the commission of crimes by two or more individuals acting together, and the study of co-offending networks maps the patterns of criminal collaboration within and across offenses. Research consistently shows that a substantial proportion of crimes — particularly among younger offenders — involve multiple participants, and that co-offending relationships tend to be fluid, with individuals cycling through different partnerships across different offenses.

Analysis of co-offending networks has revealed that a relatively small number of highly connected individuals — sometimes termed key players (关键行为者) or “super-connectors” — link otherwise disconnected segments of criminal networks. These individuals may not necessarily be the most prolific or serious offenders themselves, but they serve as bridges through which criminal opportunities, techniques, and normative definitions flow. The identification and targeted intervention with these key players has emerged as a promising strategy for disrupting criminal networks, based on the theoretical insight that removing bridging individuals can fragment networks and reduce the flow of criminal influence.

Peer Hierarchies and Influence Mechanisms

The recognition that peer groups are internally differentiated — that not all members are equally influential — has generated important research on the mechanisms through which influence flows within groups. McGloin and Thomas contributed to this line of inquiry by examining how the structure of peer relationships moderates the effect of peer influence on criminal behavior. Their research suggests that influence is not a simple, uniform process but rather one that varies in intensity and direction depending on the characteristics of the relationship, the characteristics of the individuals involved, and the broader social context in which the interaction takes place.

Chapter 5: Gangs and Group Processes

Early Gang Research: Thrasher and the Chicago Tradition

The study of gangs has been intertwined with the study of peers and crime since the earliest days of American criminology. Frederic Thrasher’s landmark 1927 study, The Gang, documented over 1,300 gangs in Chicago and provided the first systematic analysis of gang formation, organization, and behavior. Thrasher defined the gang (帮派) as an interstitial group — one that forms in the gaps and interstices of the institutional structure of society — and argued that gangs emerge naturally from the play groups of childhood when those groups come into conflict with their social environment.

Thrasher’s work established several foundational insights that continue to inform the study of peers and crime. First, he demonstrated that gangs are products of social disorganization: they form most readily in neighborhoods where conventional institutions (family, school, church) are weak and where poverty, crowding, and residential instability create conditions conducive to group formation outside institutional channels. Second, he showed that gangs provide their members with functions that conventional institutions have failed to provide: a sense of belonging, identity, excitement, protection, and status. Third, he documented the internal dynamics of gangs — their hierarchies, norms, rituals, and conflict processes — in a way that revealed them to be complex social organizations, not mere collections of antisocial individuals.

Short and Group Process Perspectives

James F. Short Jr. extended the study of gang dynamics through a group process (群体过程) perspective that emphasized the situational and interactional mechanisms through which gangs produce criminal behavior. Short argued that much gang violence is not the product of deep-seated criminal values or calculated criminal enterprise but rather emerges from the group dynamics that unfold in everyday gang life — status threats, challenges to group solidarity, inter-group conflicts, and the pressures of maintaining one’s position within the gang hierarchy.

Short’s concept of status threat (地位威胁) was particularly influential. He argued that gang leaders face a constant challenge to their status from both within and outside the group. When a leader’s status is threatened — by a challenge from a rival, a perceived insult, or a loss of face — the leader may resort to violence not because violence is valued for its own sake but because it serves to restore status and demonstrate the strength necessary to maintain leadership. This situational dynamic means that much gang violence is reactive and expressive rather than instrumental, driven by the immediate social pressures of the group context rather than by long-term criminal planning.

Modern Gang Research

Contemporary gang research has moved beyond the classic Chicago tradition to examine gangs in diverse settings and through multiple theoretical lenses. Scholars such as Scott Decker and David Pyrooz have investigated the processes of gang joining (加入帮派) and gang leaving (退出帮派), finding that both are complex social processes shaped by peer dynamics. The decision to join a gang is influenced by exposure to gang members in one’s immediate social network, the perceived benefits of gang membership (protection, status, income, belonging), and the absence of attractive conventional alternatives. The decision to leave a gang is complicated by the social embeddedness of gang ties: members who wish to desist must navigate the expectations and potential sanctions of their gang peers while simultaneously establishing new prosocial relationships.

Research on the gang enhancement effect (帮派增强效应) has demonstrated that gang membership amplifies the criminogenic influence of delinquent peers beyond what would be expected from the simple aggregation of delinquent individuals. Gang members commit more crime than equally delinquent non-gang youth, suggesting that the organizational features of gangs — their norms, hierarchies, group processes, and collective identity — intensify the social learning processes that drive criminal behavior. This enhancement effect supports the socialization perspective: it is not simply that delinquent individuals join gangs (selection) but that the gang context itself produces additional criminal behavior beyond what members would engage in on their own.

Chapter 6: Life-Course Perspectives on Peers and Crime

Developmental Transitions and Peer Relationships

The life-course perspective (生命历程视角) in criminology examines how criminal behavior unfolds over the lifespan, attending to the ways in which developmental transitions, turning points, and age-graded social roles shape trajectories of offending and desistance. Within this framework, peer relationships are understood not as static influences but as dynamic forces whose significance, character, and consequences change as individuals move through different stages of life.

Giordano’s research on life-course transitions and peer relationships has been particularly influential in illuminating the complex interplay between peers, identity, and behavioral change. Giordano developed a theory of cognitive transformation (认知转变) that emphasizes the role of subjective processes — shifts in how individuals think about themselves, their relationships, and their futures — in facilitating desistance from crime. Within this framework, peers can serve as both obstacles to and catalysts for change.

Giordano argued that the traditional life-course emphasis on structural turning points — marriage, employment, military service — needed to be supplemented by attention to the cognitive and emotional processes through which individuals make sense of these transitions. Peers are central to this process in multiple ways. During adolescence, peer groups provide the primary context for identity formation, and identification with a delinquent peer group can solidify a criminal self-concept. In young adulthood, the weakening of delinquent peer ties and the formation of new prosocial relationships can provide the relational hooks that facilitate cognitive transformation and desistance.

Importantly, Giordano’s research found that cognitive transformation is not simply a passive response to changed circumstances but an active process of meaning-making. Individuals who desist from crime do not merely find themselves in better social environments; they actively reinterpret their past behavior, reconstruct their identities, and develop new frameworks for evaluating opportunities and relationships. Prosocial peers play a crucial role in this process by providing alternative models of identity, validating the emergent conventional self-concept, and reinforcing behavioral changes through positive feedback and social support.

Sampson and Laub’s Age-Graded Theory

Robert Sampson and John Laub’s age-graded theory of informal social control (年龄分级非正式社会控制理论) provides the most comprehensive structural framework for understanding how peer influence changes across the life course. Drawing on a remarkable longitudinal data set — the Glueck study, which followed 500 delinquent and 500 non-delinquent boys from childhood into old age — Sampson and Laub demonstrated that criminal trajectories are shaped by the accumulation and disruption of social bonds across the lifespan.

Sampson and Laub identified several key turning points (转折点) that redirect criminal trajectories: marriage to a prosocial partner, stable employment, and military service. Each of these turning points operates by creating new social bonds that invest the individual in conventional society, provide routine and structure that reduce opportunities for criminal behavior, and — critically — restructure the individual’s peer environment. Marriage, for example, typically involves a physical and social separation from the peer groups of adolescence and young adulthood, replacing the unstructured socializing with delinquent friends that characterizes adolescence with domestic routines and obligations that leave less time and opportunity for criminal activity.

The mechanism through which turning points operate is what Sampson and Laub term social capital (社会资本) — the investment in social relationships that creates stakes in conformity. As individuals accumulate social capital through stable marriages, employment, and community ties, the costs of criminal behavior increase (because one has more to lose) and the benefits decrease (because conventional relationships provide alternative sources of satisfaction, identity, and belonging). Peer influence thus becomes less salient as individuals age not because the psychological capacity for social influence diminishes but because the structural conditions that facilitate peer influence — unstructured time with peers, weak conventional bonds, limited stakes in conformity — are progressively replaced by conditions that favor conventional behavior.

Marriage, Military, and Employment as Hooks for Change

Each of the turning points identified by Sampson and Laub has a distinctive relationship to the peer environment. Marriage (婚姻) to a prosocial partner operates partly through what Warr termed “the sticky friends” phenomenon — the observation that marriage dramatically reduces time spent with peers, particularly peers who engage in delinquent behavior. Warr found that the decline in delinquent behavior associated with marriage could be largely explained by the reduction in exposure to delinquent friends that marriage produces. This finding suggests that one of the primary mechanisms through which marriage promotes desistance is the restructuring of the individual’s social network away from criminogenic peers and toward conventional domestic and family relationships.

Military service (服兵役) operates through a similar mechanism of social restructuring, physically removing individuals from their pre-existing peer networks and placing them in a highly structured environment with strong norms of discipline, obedience, and conventional behavior. The military experience also provides opportunities for skill development, credential acquisition, and the formation of new social ties that may facilitate successful reintegration into civilian life. However, the desistance-promoting effects of military service are contingent on the quality of the military experience; those who serve in combat zones or who experience military-related trauma may not benefit from the same protective effects.

Employment (就业) operates through multiple mechanisms: it occupies time that might otherwise be spent in unstructured socializing with deviant peers; it provides a source of legitimate income that reduces the material incentives for criminal behavior; it creates social bonds with coworkers and supervisors who typically endorse conventional norms; and it provides a basis for a conventional social identity that is incompatible with a criminal self-concept. Research consistently shows that the quality and stability of employment matter more than the mere fact of being employed: steady, satisfying work with advancement opportunities is associated with greater desistance than unstable, low-wage work without prospects.

Age-Graded Patterns of Peer Influence

Research consistently shows that the influence of peers on criminal behavior follows a distinctive age-graded pattern. Peer influence peaks during adolescence — roughly between the ages of 14 and 18 — and declines thereafter. This pattern corresponds to the well-documented age-crime curve (年龄-犯罪曲线), which shows that criminal behavior rises sharply in early adolescence, peaks in the late teenage years, and declines steadily through adulthood.

Several factors account for the heightened susceptibility of adolescents to peer influence. Developmental neuroscience has demonstrated that the brain’s reward-processing systems mature before the prefrontal cortical regions responsible for impulse control, risk assessment, and long-term planning. This maturational imbalance means that adolescents are more responsive to the social rewards associated with peer approval and more susceptible to the emotional arousal generated by peer presence than adults are. Moreover, the adolescent period is characterized by intensified orientation toward peers and away from family, as young people seek to establish autonomous identities and social status within the peer group.

Samson and Laub’s age-graded theory of informal social control provides a complementary structural explanation. As individuals age, they accumulate “social capital” in the form of stable employment, intimate partnerships, and other social bonds that create stakes in conformity. These bonds increasingly substitute for peer relationships as the primary source of social influence, making adult behavior less susceptible to peer pressure and more responsive to conventional social controls.

Desistance and the Peer Context

Desistance (犯罪中止) — the process by which individuals cease criminal behavior — has become a major focus of life-course criminology, and the role of peers in this process is complex and contested. On one hand, maintaining ties to delinquent peers is one of the strongest predictors of continued offending, and the severing of such ties is frequently cited as a precondition for successful desistance. On the other hand, as Giordano’s work emphasizes, the mere absence of delinquent peers is insufficient for desistance; what matters is the presence of positive relational alternatives and the cognitive transformations that accompany them.

Boman and Mowen’s research on peer-based interventions and reentry programming speaks directly to the practical challenges of managing peer transitions during the desistance process. Their work examines how formerly incarcerated individuals navigate the tension between the need to distance themselves from criminogenic peer networks and the equally pressing need for social support and belonging. Boman and Mowen demonstrate that successful reentry often depends on the ability to form new prosocial peer relationships that provide alternative sources of identity, reinforcement, and social integration.

The challenge of desistance through peer change is compounded by what scholars have termed the knifing off (切断联系) problem. Sampson and Laub used this metaphor to describe the process by which individuals sever ties to their criminal past, including criminal peer networks. While knifing off is often necessary for desistance, it carries significant costs: the loss of social support, the disruption of identity, and the loneliness and isolation that can accompany the severing of long-standing relationships. Individuals who knife off from criminal peers without simultaneously forming new prosocial connections may find themselves socially isolated, which can itself be a risk factor for relapse into criminal behavior.

Chapter 7: Social Media, Digital Peers, and Contemporary Challenges

The Digital Transformation of Peer Influence

The rapid proliferation of digital communication technologies has fundamentally altered the landscape of peer interaction, creating new channels through which peer influence can operate and new challenges for researchers seeking to understand its effects. Social media (社交媒体) platforms, messaging applications, and online communities have expanded the scope of peer influence beyond the geographic and temporal boundaries that previously constrained face-to-face interaction, enabling continuous, asynchronous, and often public communication among peers.

McCuddy’s research on social media and deviance examines how digital peer networks (数字同伴网络) shape criminal and deviant behavior. McCuddy argues that social media platforms function as contexts for the same social learning processes that Akers identified in face-to-face settings. Users are exposed to deviant modeling through posts, stories, and videos that depict illegal or antisocial behavior; they encounter definitions favorable to deviance in comments, captions, and shared narratives; and they receive differential reinforcement in the form of likes, shares, and positive responses when they post deviant content.

McCuddy’s work highlights several features of social media that may intensify peer influence relative to traditional face-to-face contexts. First, the permanence and reproducibility of digital content means that acts of deviance can be revisited, shared, and amplified long after they occur. Second, the public nature of much social media interaction means that peer influence operates through broadcast rather than dyadic communication, allowing single individuals to function as models for large audiences. Third, the algorithmic curation of content on major platforms may create filter bubbles that concentrate exposure to deviant content among users who have already expressed interest in it, reinforcing and intensifying existing behavioral tendencies.

Cyberbullying and Digital Aggression

The rise of digital communication has created new forms of peer aggression that extend the traditional concerns of peers and crime research into the online realm. Cyberbullying (网络欺凌) — the use of digital technologies to repeatedly harass, threaten, or humiliate others — has emerged as a significant public concern, with prevalence rates suggesting that a substantial proportion of adolescents experience cyberbullying as either victims or perpetrators.

Cyberbullying differs from traditional bullying in several ways that have implications for understanding peer influence. First, the anonymity or pseudonymity afforded by many digital platforms may reduce the social costs of aggression, making it easier for individuals to engage in behavior they would avoid in face-to-face settings. Second, the permanence and publicness (公开性) of digital content means that cyberbullying can reach larger audiences and persist longer than traditional bullying, amplifying its harmful effects on victims. Third, the boundary between perpetrators and bystanders is blurred in digital contexts: sharing, liking, or commenting on bullying content implicates a wider circle of peers in the aggressive behavior and contributes to the normalization of digital aggression within the peer group.

Research has demonstrated that exposure to cyberbullying by peers is associated with increased risk of perpetration, consistent with social learning theory’s prediction that observed and reinforced behavior will be modeled. The peer dynamics of cyberbullying also illustrate the role of bystander effects (旁观者效应) in digital contexts: the presence of a large, often silent audience may embolden aggressors (through perceived social support) while discouraging intervention (through diffusion of responsibility).

Online Radicalization

The peer influence framework has been extended to understand online radicalization (网络激进化) — the process by which individuals adopt extremist ideologies and become willing to engage in political violence through their participation in online communities. Online forums, social media groups, and encrypted messaging channels function as digital peer groups in which radical ideologies are transmitted through the same mechanisms identified by social learning theory: differential association with like-minded extremists, definitions favorable to political violence, reinforcement through group approval and status, and imitation of idealized models of violent action.

Research on online radicalization has emphasized the role of echo chambers (回音室效应) — digital environments in which users are predominantly exposed to information and opinions that reinforce their existing beliefs — in accelerating the radicalization process. Within these echo chambers, extremist definitions are progressively normalized, dissenting voices are marginalized or excluded, and the perceived consensus in favor of radical action grows stronger over time. The social learning processes that Akers described operate with particular intensity in these environments because the diversity of definitions that might counterbalance extremist messaging is systematically eliminated by the structure of the online community.

Sexting and Digital Sexual Behavior

The intersection of peer influence and digital technology has also generated concern about sexting (性短信) — the exchange of sexually explicit images or messages through digital devices. Research has consistently found that perceived peer norms regarding sexting are among the strongest predictors of individual sexting behavior, consistent with social learning theory’s emphasis on the role of definitions and differential association. Adolescents who believe that their peers engage in sexting are more likely to engage in it themselves, regardless of whether those perceptions are accurate — illustrating once again the power of perceptual measures and the potential for projection bias in shaping behavior.

The peer dynamics of sexting involve complex negotiations of status, intimacy, and gender, with research suggesting that the motivations for sexting and the social consequences of participation differ significantly between boys and girls. These gendered patterns intersect with broader questions about gender and peer influence that are examined in a subsequent chapter.

Digital Peer Pressure

The concept of digital peer pressure (数字同伴压力) captures the distinctive forms of social influence that operate through online communication. Unlike face-to-face peer pressure, which typically involves direct verbal or physical coercion, digital peer pressure often operates through more subtle mechanisms: the visibility of peers’ behavior on social media feeds, the quantification of social approval through metrics such as likes and followers, and the fear of missing out (FOMO) that arises from constant exposure to curated representations of peers’ activities.

Research has found that adolescents report feeling significant pressure to present themselves in particular ways on social media, to respond to messages and posts promptly, and to participate in online trends and challenges — some of which involve risky or illegal behavior. The gamification of social interaction through metrics and algorithms creates a reinforcement environment in which deviant behavior may be rewarded more consistently and more immediately than in face-to-face settings, potentially intensifying the social learning processes that drive peer influence.

Cybercrime and Online Peer Communities

The digital revolution has also created entirely new forms of criminal behavior that are inherently social in character. Cybercrime (网络犯罪) — including hacking, online fraud, identity theft, and the distribution of illegal content — frequently occurs within online communities that function as peer groups, complete with hierarchies, norms, mentoring relationships, and systems of status and reward. These communities provide the same social learning mechanisms identified by Akers: members learn criminal techniques through observation and instruction, acquire definitions favorable to cybercrime through group discourse, receive reinforcement for successful criminal activity, and model their behavior on high-status community members.

The study of online criminal communities has expanded the boundaries of the peers and crime field by demonstrating that peer influence can operate across geographic distances, cultural boundaries, and even language barriers. It has also raised new questions about the nature of peer relationships in anonymous or pseudonymous contexts, where the traditional dimensions of association — frequency, duration, priority, and intensity — take on new meanings.

Chapter 8: Decision-Making, Situations, and Peer Presence

The Situational Dynamics of Peer Influence

While much research on peers and crime focuses on the enduring characteristics of peer relationships — who one’s friends are, how delinquent they are, how long one has known them — an equally important line of inquiry examines the situational dynamics of peer influence: how the immediate presence and behavior of others shapes decision-making in real time.

Hoeben and Thomas’s research on decision-making (决策过程) in peer contexts represents an important contribution to this situational perspective. Their work examines how the presence of peers alters the cognitive processes through which individuals evaluate criminal opportunities, weigh risks and rewards, and arrive at behavioral choices. Hoeben and Thomas demonstrate that peer presence affects decision-making through multiple pathways: it increases arousal and excitement, shifts attention toward potential rewards and away from potential costs, alters perceptions of risk, and creates social pressure to conform to group expectations.

Hoeben and Thomas’s findings are consistent with a broader body of research in developmental psychology and neuroscience showing that the presence of peers activates reward-related neural circuits, increases risk-taking behavior, and heightens sensitivity to social evaluation. Adolescents, in particular, show marked increases in risk-taking when peers are present compared to when they are alone, a pattern that has been demonstrated in both laboratory experiments (using driving simulation tasks) and field studies of real-world behavior.

Routine Activities and Unstructured Socializing

The situational perspective on peer influence connects naturally to routine activities theory (日常活动理论), which argues that crime occurs when motivated offenders, suitable targets, and the absence of capable guardians converge in time and space. Unstructured socializing (非结构化社交) with peers — spending time with friends in the absence of adult supervision and without structured activities — has been identified as a particularly criminogenic routine activity, because it simultaneously increases exposure to peer influence, provides opportunities for deviance, and removes the constraints of formal supervision.

Mark Warr’s research on peers and delinquency emphasized the importance of the situational context of peer interaction, arguing that much delinquency is essentially a product of group dynamics that unfold in the course of everyday social interaction. The “dare” phenomenon — in which an individual commits an act they would not otherwise consider in order to maintain face or demonstrate courage in front of peers — exemplifies the power of situational peer influence. In such moments, the immediate social rewards of compliance (acceptance, admiration, status) overwhelm the individual’s assessment of the longer-term risks and moral costs of the behavior.

Chapter 9: Gender and Peer Influence

Gendered Patterns of Peer Association

The relationship between peers and crime is significantly shaped by gender (性别), yet much of the foundational research in this field was conducted primarily or exclusively with male samples. As scholars have increasingly incorporated female experiences into the study of peers and crime, important differences in the nature, structure, and consequences of peer influence across genders have emerged.

Research consistently shows that girls’ peer relationships tend to be characterized by greater intimacy, emotional depth, and dyadic structure compared to the larger, more loosely organized peer groups typical of boys. These structural differences have implications for the mechanisms of peer influence: the intensity dimension of Sutherland’s modalities of association may be particularly salient for girls, whose close, emotionally intense friendships may transmit definitions and reinforcement patterns with particular force, even in the absence of the large peer group contexts that characterize much male delinquency.

Gender Differences in Susceptibility and Mechanisms

Research by Henneberger and colleagues has examined whether girls and boys are differentially susceptible to peer influence on delinquent behavior. The findings are complex and do not support a simple “boys are more influenced by peers” narrative. While boys are more likely than girls to engage in delinquent behavior in general, the proportional contribution of peer influence to that behavior may be similar or even greater for girls. Some studies have found that the correlation between peer delinquency and individual delinquency is stronger for girls than for boys, suggesting that peer relationships may be a more potent risk factor for female than for male offending.

The mechanisms of peer influence also appear to differ across genders. For boys, peer influence on criminal behavior may operate primarily through modeling, reinforcement of aggression, and group processes such as status competition and the “dare” dynamics described in the previous chapter. For girls, peer influence may operate more through relational mechanisms — the fear of losing valued friendships, the desire to maintain intimacy and connection, and the use of relational aggression (gossip, social exclusion, manipulation of friendship networks) as both a means of social control within the peer group and a form of deviant behavior in its own right.

Romantic Partners as Peers

The role of romantic partners (恋人) in shaping criminal behavior represents an important intersection of gender and peer influence. Research consistently shows that having a romantically involved partner who engages in criminal behavior is a significant risk factor for one’s own offending, and this effect is particularly pronounced for women. The intimate, emotionally intense nature of romantic relationships may amplify the social learning processes of modeling, reinforcement, and the transmission of definitions, while the power dynamics that characterize some romantic relationships — particularly those involving coercive control — may create additional pathways of influence that go beyond the mechanisms typically studied in the peer influence literature.

Giordano’s research found that romantic relationships can serve as both risk factors and protective factors depending on the characteristics of the partner. A prosocial romantic partner can provide a powerful hook for change, motivating cognitive transformation and supporting desistance from crime. A criminal romantic partner can entrench criminal behavior and obstruct desistance, particularly when the relationship involves coercive control or economic dependence that limits the individual’s capacity to exercise autonomous choice.

Chapter 10: Cross-Cultural Perspectives on Peers and Crime

Universality and Cultural Specificity

The vast majority of research on peers and crime has been conducted in North American and, to a lesser extent, Western European contexts. This geographic concentration raises important questions about the cross-cultural generalizability (跨文化推广性) of the theories and findings reviewed in this text. Is the relationship between peers and crime a universal feature of human social life, or is it shaped by cultural factors that vary across societies?

Research from diverse cultural contexts has generally confirmed the basic finding that association with delinquent peers is correlated with individual delinquency across a wide range of societies. Studies from East Asia, Latin America, Sub-Saharan Africa, and the Middle East have all documented significant associations between peer delinquency and individual offending, suggesting that the fundamental social learning processes identified by Sutherland and Akers operate across cultural boundaries.

However, the strength and specific mechanisms of peer influence appear to vary across cultural contexts in ways that reflect broader cultural values and social structures. In collectivist cultures (集体主义文化) — societies that emphasize group harmony, social obligation, and respect for authority — peer influence may operate through different channels than in the individualist cultures of North America and Western Europe. The pressure to conform to group expectations may be stronger in collectivist societies, but this pressure may also be directed toward conventional behavior when the peer group endorses prosocial norms. The family, rather than the peer group, may serve as the primary source of social influence in societies where intergenerational bonds are particularly strong, potentially moderating the effect of peer delinquency on individual behavior.

East Asian Contexts

Research from East Asian societies (东亚社会) has revealed interesting patterns that both confirm and complicate Western theories of peer influence. Studies from China, Japan, South Korea, and Taiwan have documented significant effects of peer delinquency on individual offending, but the magnitude of these effects tends to be smaller than in North American samples. This difference may reflect the stronger role of family bonds in East Asian societies, where parental monitoring and family-based social control may buffer adolescents against peer influence to a greater degree than in Western societies where adolescent autonomy is more highly valued.

Latin American and Global South Contexts

Research from Latin American and Global South contexts has highlighted the role of structural inequality in shaping peer influence processes. In societies characterized by extreme economic inequality, limited access to education and employment, and weak state institutions, peer groups — including gangs — may provide the only available pathway to status, income, and social belonging. The social learning processes that drive peer influence operate within these structural constraints, and the relative absence of conventional alternatives amplifies the attractiveness of criminal peer networks. Research on maras (中美洲帮派) in Central America, for example, has documented how extreme poverty, social exclusion, and the legacy of political violence create conditions in which gang membership offers a rational, if tragic, response to the absence of legitimate opportunities.

Chapter 11: Interventions, Policy, and the Future of Research

Peer-Focused Interventions

The recognition that peers play a central role in criminal behavior has generated sustained interest in interventions that target peer dynamics as a strategy for crime reduction. These interventions take various forms, ranging from programs that seek to restructure peer relationships directly to those that aim to modify the cognitive and behavioral processes through which peer influence operates.

Boman and Mowen’s research on peer-based interventions examines both the promise and the pitfalls of programs that attempt to leverage peer relationships for prosocial purposes. They note that some well-intentioned programs have inadvertently produced iatrogenic effects (医源性效应) — harmful outcomes resulting from the intervention itself. The most well-known example is the aggregation of high-risk youth in group treatment settings, which can create opportunities for delinquent peer influence, reinforce deviant identities, and facilitate the formation of new criminal relationships. Programs such as Scared Straight, which exposed at-risk youth to incarcerated offenders, and certain group-based cognitive-behavioral programs have been found in rigorous evaluations to increase rather than decrease delinquency among participants.

Boman and Mowen argue that effective peer-focused interventions must be carefully designed to avoid these pitfalls. They identify several principles for successful programming: avoiding the aggregation of high-risk individuals without adequate structure and supervision; incorporating prosocial peer mentors who can model and reinforce conventional behavior; addressing the cognitive processes (definitions, neutralizations) through which peer influence operates, not just the behavioral outcomes; and recognizing that peer relationships are embedded in broader social contexts that may support or undermine programmatic goals.

Multisystemic Therapy (MST)

Multisystemic Therapy (MST) (多系统治疗) represents one of the most rigorously evaluated and empirically supported intervention approaches for addressing peer-related risk factors in juvenile offending. Developed by Scott Henggeler and colleagues, MST is an intensive, family-based treatment that addresses the multiple systems — family, peers, school, and community — in which youth behavior is embedded. Rather than targeting the individual adolescent in isolation, MST works with the entire family to restructure the youth’s social environment, strengthen parental monitoring and discipline, reduce association with delinquent peers, and increase involvement in prosocial activities and relationships.

MST is grounded explicitly in social learning theory and systems theory. The treatment targets the specific risk and protective factors that have been identified as causally related to criminal behavior, with particular attention to the peer context. MST therapists work with families to develop practical strategies for monitoring their children’s peer associations, setting and enforcing rules about acceptable and unacceptable friendships, and facilitating the youth’s involvement in structured prosocial activities (sports, clubs, employment) that provide access to conventional peer groups.

Randomized controlled trials of MST have consistently found that it produces significant reductions in criminal behavior, substance use, and out-of-home placements relative to usual services. Follow-up studies have demonstrated that these effects are sustained over periods of several years. Importantly for the peers and crime literature, research on MST has confirmed that reductions in association with delinquent peers partially mediate the treatment’s effects on criminal behavior, providing experimental evidence that changing the peer environment causes changes in offending.

Functional Family Therapy (FFT)

Functional Family Therapy (FFT) (功能性家庭治疗) is another evidence-based intervention that targets peer-related risk factors through a family-focused approach. Like MST, FFT recognizes that adolescent behavior is shaped by the family system in which it is embedded and that effective intervention requires changing family dynamics — including the family’s approach to monitoring and managing peer relationships.

FFT operates through three phases: engagement and motivation (building therapeutic alliance and reducing resistance), behavior change (developing specific communication and parenting skills), and generalization (maintaining gains and preventing relapse). Throughout these phases, attention to the peer context is woven into the treatment: therapists help parents develop skills for monitoring their children’s peer associations, negotiating rules about friendship choices, and responding constructively when children are exposed to delinquent peer influence.

The Cure Violence Model

The Cure Violence model (治愈暴力模型), formerly known as CeaseFire, represents a fundamentally different approach to peer-related crime reduction. Rather than targeting individual youth or families, Cure Violence treats violence as a public health problem that spreads through social networks and deploys community-based workers to interrupt the transmission of violent behavior.

The Cure Violence model employs three primary strategies. First, violence interrupters (暴力调解员) — credible community members with personal experience of violence, often former gang members or ex-offenders — identify and mediate conflicts before they escalate to violence. These interrupters use their knowledge of local social networks to anticipate retaliatory violence and intervene at critical moments. Second, outreach workers (外展工作者) provide intensive case management to high-risk individuals, connecting them with services (employment, housing, education, mental health) and working to change their attitudes and behaviors regarding violence. Third, the program engages in community mobilization activities designed to change neighborhood norms about the acceptability of violence.

The Cure Violence model is explicitly informed by the network epidemiology research of Papachristos and others. By targeting the network pathways through which violence spreads, the model aims to interrupt the contagion process that sustains cycles of retaliatory violence. Evaluations of Cure Violence programs in several cities have found significant reductions in shootings and killings in treatment areas relative to comparison areas, though the evidence base is still developing and some evaluations have produced mixed results.

Peer Mentoring Programs

Peer mentoring programs (同伴导师计划) represent another approach to leveraging peer dynamics for prosocial purposes. These programs pair at-risk youth with trained peer mentors — typically older adolescents or young adults who have demonstrated prosocial behavior — who provide guidance, support, and modeling of conventional behavior. The theoretical basis for peer mentoring draws directly from social learning theory: by providing exposure to prosocial models who are similar in age and social background to the mentee, these programs aim to shift the balance of definitions, modeling, and reinforcement in a prosocial direction.

The evidence on peer mentoring is mixed. While some programs have produced modest positive effects on behavior, school performance, and attitudes, others have found no significant effects or even negative effects — particularly when mentors are poorly trained, when the mentoring relationship is unstable, or when the matching of mentors and mentees is inappropriate. The potential for iatrogenic effects is a constant concern, as peer mentoring programs must navigate the risk that mentors may themselves model or reinforce deviant behavior, either intentionally or inadvertently.

Canadian Intervention Models

Canada has developed several notable intervention programs that address peer-related risk factors for criminal behavior. The Stop Now and Plan (SNAP) program (停止并计划项目), developed by the Child Development Institute in Toronto, is an evidence-based intervention for children aged 6-11 who are exhibiting early signs of antisocial behavior. SNAP uses cognitive-behavioral techniques to teach children self-regulation skills that enable them to resist peer pressure and make prosocial choices. The program includes a parent component that addresses family-level risk factors and a peer component that provides structured prosocial group activities as an alternative to unsupervised peer interaction.

SNAP has been recognized as a model program by multiple evidence-based clearinghouses and has been found in randomized trials to produce significant reductions in conduct problems and aggressive behavior. Its early-intervention approach is informed by developmental criminology, which emphasizes that intervening before antisocial patterns become entrenched and before association with delinquent peers intensifies is more effective than later intervention.

Youth Advocate Programs (YAP) (青年倡导计划), which operate in several Canadian jurisdictions, provide intensive community-based support to high-risk youth as an alternative to incarceration or residential placement. YAP assigns trained advocates who work with youth and their families for up to 30 hours per week, providing mentoring, connecting families to services, and actively managing the youth’s peer environment. The program explicitly targets peer-related risk factors by helping youth develop prosocial peer networks, providing structured activities as alternatives to unstructured socializing, and working with families to monitor and manage peer associations.

Evidence-Based Correctional Policy

Taxman’s work on evidence-based correctional policy provides a broader framework for thinking about how the criminal justice system can address the peer dimensions of criminal behavior. Taxman advocates for a systematic, research-driven approach to correctional programming that integrates knowledge about peer influence into the design and delivery of interventions at every stage of the criminal justice process, from pretrial diversion through incarceration and community reentry.

Taxman’s policy recommendations include the development of risk-responsive programming that tailors interventions to the specific peer-related risk factors of individual offenders; the use of network analysis to identify key influencers within criminal networks who might be targeted for intervention; the creation of structured prosocial environments that provide alternatives to criminogenic peer groups; and the implementation of reentry programs that actively support the formation of new prosocial relationships.

Street Dynamics and Network Interventions

Papachristos and Kirk’s research on street dynamics and social networks has generated important insights for violence intervention strategies. Their finding that violence spreads through social networks like a contagion has inspired the development of network-based intervention approaches, such as the Cure Violence model (formerly CeaseFire), which treats violence as a public health problem and deploys “violence interrupters” — credible community members with personal experience of violence — to disrupt the transmission of violent behavior through social networks.

These network interventions represent a fundamentally different approach to crime reduction than traditional law enforcement strategies. Rather than targeting individuals for arrest and prosecution, they target the social relationships and normative environments that sustain violence. By intervening at the relational level — interrupting conflicts before they escalate, mediating disputes, and changing community norms about the acceptability of violence — these programs aim to break the cycles of retaliatory violence that Papachristos and Kirk documented.

Toward an Integrated Understanding

The study of peers and crime has matured considerably since Sutherland first proposed that criminal behavior is learned through association with others. Contemporary research draws on multiple theoretical traditions — social learning theory, control theory, network analysis, developmental criminology, situational approaches, and public health frameworks — to construct a more nuanced and comprehensive understanding of how peers shape criminal trajectories.

Several key insights have emerged from this body of work. First, peer influence is real and consequential, not merely a methodological artifact or a reflection of selection processes. Both socialization and selection contribute to the peer–delinquency correlation, and the two processes operate reciprocally over time. Second, peer influence is not a monolithic process but operates through multiple mechanisms — modeling, reinforcement, the transmission of definitions, cognitive and emotional arousal, network contagion — that may vary in their importance across different contexts, developmental periods, and types of offending. Third, the structure of peer relationships matters as much as their content; network position, status, and the broader architecture of social connections all moderate the effects of peer influence. Fourth, effective interventions must be informed by an understanding of peer dynamics and must carefully avoid inadvertently intensifying the very processes they seek to counteract. Fifth, gender, culture, and digital technology introduce important variations in the operation of peer influence that require ongoing theoretical and empirical attention.

The digital transformation of social life, the growing sophistication of network analytical methods, the insights of developmental neuroscience, and the continuing refinement of social learning theory all promise to deepen understanding of peers and crime in the decades ahead. As Akers, McGloin and Thomas, Gallupe and colleagues, Haynie, Papachristos and Kirk, and others have demonstrated, the study of how people influence each other toward or away from crime remains one of the most vibrant and consequential enterprises in the social sciences.