Over the last decades, no area of inquiry within the field of juvenile offending has generated as much enduring interest as the study of peer influence. One of the most consistent and robust findings in the research on juvenile offending is that affiliation with deviant peers is an important contributor to adolescent delinquency. However, the literature has been inconsistent and confusing in developing the causal mechanisms of peer influence (Giordano, 2003; Warr, 2002). It is still open to speculation as to whether juvenile offending is the result of having delinquent friends or, alternatively, is mainly due to the fact that individuals with delinquent inclinations select deviant peers to befriend (Monahan, Steinberg & Cauffman, 2009). The principle aim of this essay is to examine the extent to which juvenile offending is the result of having friends who are involved in crime by integrating relevant theories and research evidence.
Before going any further it would seem sensible to first define what is meant by juvenile offending. A brief and working definition is used to denote those actions, committed by young people between the ages of 12 to 18, that violate the law (Burfeind & Bartusch, 2006). More specifically, a juvenile offender is "a youth who has committed an offence that would be considered a felony offence if committed by an adult and that is an offence against a person, an offence against a property or an offence involving dangerous drugs" (Bufreind & Bartusch, 2006, p.51). A broader conceptualization is offered by sociologists who, under the umbrella of juvenile delinquency, include an extensive amount of different kinds of acts against the legal and social norms, from minor offences to serious crimes, committed by juveniles (World Youth Report, 2003).
At this point, a question that needs to be addressed is why this strong link between deviant peers and adolescent delinquency exists. Adolescence is, without a doubt, a particularly vulnerable phase of development characterized by immense change in school environment, social relationships, and physiological development (Abela & Sullivan, 2003). During this critical transition phase, delinquent behaviour features begin to develop (Elliott & Menard, 1996; Snyder et al., 1986) and friends gain a salient role in the developmental adjustment of adolescents (Warr, 1993). As youths start to distance themselves from their families (Coleman, 1961), a great deal of time is spent with same-age peers, extending the importance that is placed on peer affiliations (Corsaro & Eder, 1990; Schreck et al., 2004). Moreover, research has demonstrated that among variables such as family, school and community characteristics, affiliation with deviant peers is the strongest predictor of delinquency (Agnew, 2001; Elliott & Menard, 1996; Warr, 2002). It is therefore not surprising that the peer context has been a central tenet of most delinquency theories (e.g. Akers, 1985; Hirschi, 1969; Sutherland, 1947; Thornberry, 1987). According to McGloid (2009), a vast amount of research has attempted to operationalise the exposure to deviant peers by measuring peer influence in one of the following ways: 1) the number of friends who are involved in crime (e.g. Matsueda & Anderson, 1998); 2) the amount of criminal acts committed by friends (e.g. Bauman & Fisher, 1986; Reitz et al., 2006); and 3) the proportion of delinquent friends an individual has (e.g. Haynie, 2002).
An impressive number of inquiries across an array of methodological specifications have demonstrated that youths who have delinquent friends are themselves more likely to be delinquent. The Gluecks (1950) were among the first to report this relationship. Their findings indicated that 98% of 500 delinquent adolescents had delinquent friends, whereas only 8% of 500 non-delinquents had friends that were involved in crime (as cited in Matsueda & Anderson, 1998). An extensive amount of research revealed that exposure to deviant peers has been linked to norm-breaking behaviour (Brendgen et al., 2000), substance use (Aseltine, 1995, Dishion & Owen, 2002), school problems (Berndt & Keefe, 1995), violent offences (Elliott & Menard, 1996), and early and high-risk sexual behaviour (Dishion, 2000). However, an issue that has still to be investigated is the extent to which juvenile offending is the result of having peers who are involved in crime. Do adolescents become delinquent because they associate with deviant friends, as social learning theories contend, or do delinquents merely seek out friends like themselves, as social control theorists propose?
The causal efficacy of delinquent peers in transmitting delinquency to adolescents has been emphasized by the advocates of socialization tradition. Sutherland's (1947) differential association theory and Burgess and Akers' (1966) social learning perspective played a key role in the criminological literature in bringing the phenomenon of peer effect to the forefront of theoretical and research attention. The primary mechanism for influence is regarded to be the transmission of behaviours between adolescents who are closely associated with one another (Payne & Cornwell, 2007). Specifically, the theories' foremost proposition is that juvenile delinquency occurs in the course of social interaction, through learning. Consequently, delinquent behaviour occurs as a result of peer influence, as, for example, when the direct pressure or example of peers leads to smoking behaviour (Ennett & Bauman, 1994).
In more detail, differential association theory contends that an individual becomes delinquent due to an exposure to definitions favourable to violating the law compared to unfavourable definitions (Sutherland, 1947). As such, delinquent definitions are learned through social interaction in intimate peer groups who have already incorporated criminal values. This learning involves not only the skills for committing delinquent acts but also the specific motivation and justifications that encourage violation of the law (Bufreind & Bartusch, 2006; Matsueda, 1982). The theory proposes that the association with different groups ("differential association") is determined by several aspects of peer relationships, specifically the frequency, duration, priority and intensity (Agnew, 1991). To illustrate, an adolescent is more likely to engage in criminal behaviour if the interaction with his deviant friends is strong (intensity), emerges early in life (priority), occurs frequently (frequent), and lasts for long periods (duration) (Bufreind & Bartusch, 2006). In brief, Sutherland's (1947) differential association attributes the cause of delinquent behaviour to the association with delinquent friends since the roots of crime are placed in the close social network of adolescents (Eliesen, 1996).
Similarly, Burgess and Akers' (1966) corroborated and extended the above ideas, giving more weight to the direct social mechanisms involved in the learning processes by postulating that crime is learned initially through direct imitation and reinforcement (Eliesen, 1996; Payne & Cronwell, 2007). During adolescence the peer group is the most important reinforcer for young people. As such, its effect on adolescents' behaviour is crucial. To be more specific, taking into account the actual or expected consequences of a given behaviour, juveniles learn through differential reinforcement how to gain rewards and/or avoid punishments. In fact, an adolescent can be excluded from the peer group or lose his/her status as a punishment for not obeying to the norms of deviance. In contrast, approval and recognition can be gained through conformity to the rules of law-breaking behaviour (Akers, 1985; Coleman, 1961). As social learning theory proposes, the source of juvenile offending is located in the learning of beliefs and attitudes that are favourable to crime (definitions), exposure to criminal models (imitation) and reinforcement of criminal behaviour (differential reinforcement) through the association with delinquent friends (differential association). As a consequence, crime comes to be an acceptable and justifiable action that adolescents have to follow. While Sutherland's (1947) theory emphasizes the attitudes of peers in the transmission of delinquency, Burgess and Akers' (1966) revision of differential association theory gives priority to the role of peers' behaviour. To sum up, the whole process of social learning is complex and implies causality (Payne & Cornwell, 2007). As such, juvenile delinquency emerges as a result of the affiliation with deviant peers (Akers, 2000).
Consistent with the socialization perspective is a vast amount of evidence indicating that the association with deviant peers strongly predicts juvenile offending. To start with, Elliott and Menard's (1996) analysis of the National Youth Survey (NYS) data confirmed the causal ordering offered by social learning theories. The structural equation model revealed that exposure to delinquent friends typically precedes the onset of one's own law-breaking behaviour, since the influence of delinquent peer group on deviant behaviour was consistently stronger than the influence of delinquent behaviour on the affiliation with delinquent peers. The pattern of delinquent behaviour in the NYS sample followed a similar developmental transition as the deviant's group behaviour has followed, moving from minor to more serious delinquency through the years and decreasing in adulthood (Grifford-Smith et al., 2005). In the same line, Akers (1998) focused on the attitudes of juveniles' alcohol and marijuana use. The findings revealed that 59% and 61% of his adolescent sample used alcohol and marijuana, respectively, as a result of peer influence. McCarthy and Hagan (1995) came to similar conclusions, showing that getting involved with crime is a social phenomenon. Their findings confirmed the importance of criminal contacts in peer networks by indicating that adolescents living on the street were more likely to become involved in criminal activities if the people whose company they chose were already involved in criminal behaviour and who could teach them criminal skills.
A study conducted by Loeber and colleagues (1991) adds to the extant body of research favouring a social learning explanation of deviant behaviour. The significance of peer variables as reported by the caretakers and the adolescents confirmed the role that peers play in inducing youth to commit various types of offences. Agnew (1991) employed peer measures to explore the influence of delinquent friends on adolescents' behaviour. The analysis revealed that when attachment to peers, time spent with peers, and the extent to which peers display criminal patterns of behaviour are above average they strongly predict delinquency. Relevant evidence is provided by Loeber and Dishion (1987) who showed that the best predictor of the onset of crime and delinquency is differential association with law-violating peers. Similarly, Warr and Stafford (1991) provided evidence in favour of Akers' (1985, 2000) differential association reinforcement theory by concluding that imitation and reinforcement are the best ways to interpret the strength of peer effect.
This long line of research supporting social learning theories has extended to the present time. In a study conducted by Brendgen et al (2000), adolescents having deviant friends were compared to adolescents with non-deviant friends and to friendless adolescents in order to examine whether mutual friendship with delinquent peers would cause delinquent behaviour. The results confirmed the hypothesis that youth with friends who are involved in crime exhibit more norm-breaking behaviour than adolescents with conventional friends (Brendgen et al., 2000). By the same token, Ardelt and Day (2002) examined the relations between parents, siblings, peers, adolescents' individual characteristics and adolescents' deviant attitudes and behaviours, on a sample of 121 families, using a social learning perspective. As predicted, the variable of deviant peers appeared to be positively related to adolescents' delinquent attitudes and behaviours, even after controlling for the other variables in the model. As indicated by the results, the more deviant friends adolescents tended to have the more likely they were to report engaging in deviant behaviour themselves (Ardelt & Day, 2002).
In a similar vein, Kim and Goto (2000) examined the factors that predict Asian American adolescent delinquency in a sample of 101 youth ranging in age from 14 to 18 indicating that the stronger predictor of adolescent delinquency was high peer delinquency. Recently, Haynie's (2002) study revealed that the proportion of delinquent peers exerted a strong positive effect on adolescents' involvement in delinquency. As Haynie (2002) notes, the results were concluded after "carefully specifying the peer network based on sociometric data, accounting for prior delinquency, omitting self-projection bias, and controlling for other potentially important characteristics of friendship networks (including involvement with friends, age of friends, and the cohesion in the network" (p. 124).
Notwithstanding the large number of studies that have examined the effects of delinquent friends on delinquency, the extent to which juvenile offending is the result of having friends who are involved in crime cannot be clearly determined, since a number of issues remain inherently questionable. First and foremost, nearly all previous studies have relied on cross-sectional research designs making it impossible to disentangle the bidirectional relation between affiliation with deviant peers and juvenile offending (Monahan et al., 2009). Consequently, the findings presented so far can only be subjective of causes. Any interpretation must be treated with caution as a reciprocal causation can be derived from correlational findings. To illustrate, social learning theory fails to acknowledge the reciprocal effect that deviant behaviour may have on social learning variables, emphasizing solely the uni-directional causal mechanisms via which social learning affects juvenile offending through delinquent peers (Thornberry, 1987). According to Vitaro et al., (2000) it may be that affiliation with deviant peers increases adolescents' engagement in delinquent behaviour (peer socialization), but it is just as possible that juvenile offenders seek peers who show similar patterns of behaviour (peer selection).
Beside the concerns of causality, certain criticisms have been applied to those studies in connection with the use of self-report measures. In fact, by utilizing youths' perceptions of their friends' delinquent behaviour, there is always a possible risk of overestimating the effect of peer delinquency. Consequently, the aforementioned findings may generate spuriously strong correlations between respondents' and friends' deviant behaviour (Asetline, 1995). According to the "influence of assumed similarity" (Jussim and Osgood, 1989), respondents usually tend to project their own attributes and behaviours on to others. To illustrate, self-report measures usually ask the respondent about the number of delinquent acts his or her friends have committed. In this manner, there is always the possibility for the adolescent to report his or her own criminal acts, rather than that of his or her friends. Therefore, the correlation between juvenile offending and delinquent peers may be attributed to exaggerated similarity due to the methodological invalidity of the self-report measure of delinquent peers (Matsueda & Anderson, 1998).
In support of this assumption, a study by Aseltine (1995) revealed that the relationship between adolescent's drug use and best friends' drug use was overestimated as the peer groups were not as homogeneous as the adolescents' reports showed. In order to reach valid conclusions, it is important to include friends' own reports of their behaviours into studies, rather than solely relying on respondents' subjective perceptions (Haynie, 2002). It would be desirable in future research to obtain additional data from other sources (e.g. parents and peers or criminal record data) and to employ other methods of data collection (Dekovic et al., 2004).
In addition, a number of impurities arise from the theoretical underpinnings of this causal relation, adding to the concerns mentioned so far. To start with, social learning theories have been challenged for not being explicit about the way an adolescent comes into contact with peers who exhibit criminal behaviour (Jeffery, 1990). Therefore, the role of opportunity to engage in delinquency is disregarded by the theory. Moreover, certain criticisms have been implied regarding Sutherland's failure to operationalize differential association theory. As Matsueda (1988) claims, without this ability the theory cannot be empirically verified. In fact, each individual perceives and judge criminal acts in a different way; hence, it is not possible to reliably measure and reproduce each definition that an adolescent is exposed to. Finally, the potential of the socialization perspective to provide implications for prevention of juvenile offending is limited.
Based on the above inconsistencies, the advocates of social control theory challenge the argument that delinquent peers cause delinquency and propose that individuals are inherently antisocial and naturally able to commit criminal acts (Hirschi, 1969; Gottfredson & Hirschi 1990). Contrary to social learning theories, deviant peers are almost always a consequence rather than a cause of one's own deviant behaviour. It is possible for an adolescent's delinquent behaviour to precede the selection of deviant friends. Therefore, the causal relationship proposed by social learning theories could be spurious and could be explained in terms of social selection (reverse causality), and response effects in measurement (Matsueda & Anderson, 1998).
Specifically, Hirschi (1969) supported the claims for a spurious correlation by suggesting that both delinquent behaviour and delinquent peer associations can be determined as a result of a third causal variable (Matsueda & Anderson, 1998). According to social control theory, the concept of weak bonds to conventional society can explain the association between adolescents' delinquency and affiliation with deviant peers. As Matsueda and Anderson (1998) note: "this concept consists of attachments to others, commitments to conventional lines of action, involvement in conventional activities, and belief in morality" (p.272). These four conditions seem to account for delinquent behaviour. If those bonds to others are strong, an involvement in deviant activities is prevented (Hayine, 1999). Individuals with strong bonds to society are less likely to violate the law, act on their natural deviant disposition and seek association with delinquent companion. Therefore, weak social bonds lead to juvenile offending which subsequently increases the affiliation with delinquent peers (Eliesen, 1996). For control theory, delinquency does not occur by learning from others in the course of social interaction, but instead each member of the society is equally predisposed to deviance (Matsueda & Anderson, 1998).
A dominant proposition of social control theory refers to the process of social selection. Maintaining that "birds of a feather flock together" the theory asserts that juvenile offenders seek each other out for friendship (Eliensen, 1996). Therefore, the attraction of similar youths is as strong, even stronger sometimes, than peer influence in fostering similar patterns of delinquent behaviour among adolescent peers. In fact, juvenile offenders engage in delinquent behaviour and then seek out friends who are also engaged in delinquency. Sheldon and Glueck (as cited in Matsueda & Anderson, 1998) were the first to report selection effects, providing evidence that delinquent patterns of behaviour were present in gang members before even getting into the selected gangs. Consequently, according to the social control perception, social selection of delinquent friends seems to be a more salient factor in the analysis of the causes of juvenile crime than socialization with deviant peers.
Gottfredson and Hirschi's (1990) extension of control theory, highlights the concept of low-self control as an explanation of the causal pattern underlying juvenile offending and deviant peers. As such, people with limited self-control tend to be impulsive, insensitive and short-sighted. According to this point of view, delinquent behaviour occurs as a result of an adolescent's low self-control. In brief, as Gottfredson and Hirschi (1990) state, "the affiliation with deviant peers has no independent or causal effect on the adolescent's behavioral patterns, other than to ease or facilitate acts that would be too difficult or dangerous to do alone â€¦" (p. 159).
A noteworthy study by Matsueda and Anderson (1998) gives weight to Gottfredson and Hirschi's (1987) arguments regarding the ambiguous interpretation of the correlation between delinquent behaviour and delinquent peer affiliation from the socialization perspective. Consistent with control theories' assumption that juvenile offenders are likely to be attracted to delinquent friends, the authors reported a relatively larger effect (.26) of delinquent behaviour on delinquent peer associations than the reverse. In addition, by attributing the stability in delinquency to low-self-control, the authors confirmed Gottftredson and Hirschi's (1969) thesis. However, contrasting with the prior findings, the analysis revealed also a significant effect of delinquent behaviour on the association with friends who are involved in crime, suggesting a reciprocal relation between the two concepts. In a similar vein, Ennett and Bauman (1994) examined the contribution of peer influence and selection to adolescent peer group homogeneity regarding smoking behaviour. Selection process was found to be the most important determinant of friend similarity regarding a variety of attitudes and behaviours, such as drug use, sexual behaviour and deviant values.
Finally, Farell and Danish (1993) assessed a large sample of middle-school students at three time-points over a period of 18 months in order to investigate the association between peer influences and drug use. Adolescent drug use was found to be a strong predictor of later peer drug use, whereas neither peer pressure nor peers' use of drugs predicted later adolescent drug use. As Farell and Danish (1993) state: "it appears that changes in the frequency of drug use precede changes in peer variables rather than vice-versa" (p. 332). Therefore, attributing the association between friends and adolescents' behaviour solely to peer influence may be misleading. However, strong evidence in favour of the assumptions proposed by control theory does not seem to exist.
Obviously, social control theory seems to imply some important shortcomings. First and foremost, the theory's propositions are not empirically supported through systematic research. Additionally, the context in which social bonds occur is neglected. As Erikson et al., (2000) mention, social control theory displays a general failure to acknowledge sufficiently the well-documented role of peer relations in youth offending. A final criticism has been related to the theory's omission to take into account the possibility of a reciprocal relationship between delinquency and weak social bonds (Conklin, 2010).
A balance between the two competing traditions is proposed by interactional theorists who have explicitly emphasized that both selection and socialization unfold reciprocally in a developmental process (Thornberry, 1987; Thornberry et al., 1994). The theory's basic premise is that both delinquent behaviour and exposure to deviant peers contribute to juvenile offending. In particular, youths associated with delinquent peers are prone to violate the law, and those who commit such acts tend to select or maintain friends who have common experiences and display similar patterns of behaviour (Simons-Morton & Chen, 2006). A different perspective is adopted by the supporters of this tradition as they put emphasis on the interrelations between peers who are involved in crime, delinquent beliefs and criminal behaviour (Eliesen, 1996). Instead of giving causal priority to affiliation with friends as social learning theories do or interpreting delinquent behaviour in terms of social selection as control theory contends, the interactional perspective grants a significant causal role to each concept (Thornberry et al., 1994).
According to Thornberry et al., (1994) the underlying cause of juvenile delinquency may be the weakening of social bonds, but it will not emerge unless it is learned through the association with delinquent peers (Haynie, 1999). Moreover, interactional theory proposes that the reciprocal effects between delinquent behaviour and delinquent peers remain invariant, despite the fact that developmental processes may affect the relationship between those variables and other concepts of adolescent development (Matsueda & Anderson, 1998). Based on the above assumptions, interactional theorists conclude that juvenile offending is a dynamic social process not the result of that process (Eliesen, 1996).
The reciprocal association between delinquent behaviour and deviant peers has been established firmly in numerous studies. Kandel (1978) was one of the first researchers to report the recicprocal nature of this relationship using longitudinal data on adolescent friendship pairs. The results indicated that adolescents who use marijuana are not only likely to be influenced by friends using this drug, but are also likely to select friends with similar behaviour patterns once they start using marijuana (Kandel, 1978). Dishion and Owen (2002) reached similar conclusions in an examination of the reciprocal relation between deviant friendships and substance use in a sample of 206 boys from early adolescence (age 13-14) to young adulthood (age 22-23). Using a multivariate model, the authors indicated that deviant friendship process predicts adolescent marijuana use, and in turn, early marijuana use predicts delinquent behaviour in late adolescence. Finally, a large longitudinal study conducted by Monahan et al (2009) with 1354 antisocial youth, revealed that both selection and socialization processes determine juvenile delinquency during middle adolescence (age 14-15). However, socialization effects become stronger as adolescents grow older (age 16-20). In addition, a resistance to peer influence was found after the age of 20 years due to the individual's social and emotional maturity.
Despite the long and robust tradition in the field of juvenile offending, a review of the literature reveals some general methodological inconsistencies that need to be highlighted. To begin with, research has been proved to be weak in exploring peer relationships. The nature and quality of relationships that youth have with their delinquent peers have been inadequately investigated (Haynie, 1999). The typical pattern used in the majority of the studies presented so far simply measures the frequency with which peers engage in criminal activities or the number of delinquent friends an adolescent has. There is always the likelihood that the link between deviant peers and juvenile offending exists due to some unmeasured confounding variable. For instance, by measuring only the number of friends who are involved in crime, the aforementioned findings seem to assume that each individual is affected in the same way by his or her delinquent friends, no matter his/her position within the peer group or his/her ability to resist peer influence (Matsueda & Anderson, 1998). In addition, differences in the peer group composition, such as gender, race, and socioeconomic status or age may attribute to the strong correlation between delinquent behaviour and affiliation with deviant peers (Ennett & Bauman, 1994). Obviously, no attempt has been made to investigate the types and attributes of peer relations and its possible links to adolescents' delinquent behaviour (Pabon et al., 1992).
A further criticism is related to the fact that some of these studies generalize delinquency while focusing only on a few minor offences (e.g. using marijuana) that do not represent the range of criminal acts committed by juveniles (Haynie, 1999). Finally, it has to be remembered that, the evidence presented does not provide the opportunity to explore issues of causality. Hence, any interpretation deriving from the above can be misleading.
To summarise, the extensive body of literature considering the role of delinquent peer influence on juvenile offending overwhelmingly lends support to the assumption that keeping company with friends who are involved in crime significantly increases the likelihood of adolescents' norm-breaking behaviour. However, depending on different theoretical perspectives, various explanations have been developed for the causal nature of these particular variables. Consequently, the extent to which juvenile offending is the result of having friends who are involved in crime remains controversial as it is not entirely clear whether the association reflects peer influence, self-selection or the combination of these social processes.
To date, most evidence is correlational, thus any implications for causality can lead to spurious results. In brief, the issue merits further investigation; certainly more longitudinal studies are needed in order to be able to provide more valuable insights into the theoretical underpinnings of this association, unscramble the causal pattern and be able to support with confidence that juvenile offending is the result of the exposure to friends who are involved in crime. As Agnew (1991, p. 48) notes: "examining the impact of delinquent peers on delinquency may greatly improve our ability to explain and control delinquency".
Word Count: 4468