Co-Offending and Offender Attributes Research Paper

This sample Co-Offending and Offender Attributes Research Paper is published foreducational and informational purposes only. If you need help writing yourassignment, please use our research paper writing service and buy a paper on any topic ataffordable price. Also check our tips on how to write a research paper, see the lists of criminal justice research paper topics, and browse research paper examples.

Co-offending, the actof committing crime alongside one or more accomplices, is a well-documented and widespread phenomenon (Carrington & van Mastrigt, 2013). However, it is farfrom uniform. Research has uncovered sizeable variations in the rates and formsof co-offending observed both between and within individuals. In this researchpaper, empirical evidence linking these variations to the offender attributesof sex, ethnicity, age, and criminal career is reviewed and discussed inrelation to its implications for theory and policy. It is concluded that theseoffender characteristics generate, and sometimes reflect, meaningfuldifferences in co-offending and that treating co-offending as a unitaryphenomenon risks masking some of its most interesting, and policy relevant,features. Open questions and directions for future research are identified,together with a need for greater attention to person-by-situation interactions.


Research dating back nearly a century has noted that many crimes are committed in groups, particularly during adolescence (Shaw and McKay 1931). This finding is so robust across time and geography that co-offending is now widely regarded as criminological “fact.” Even so, relatively little research has been dedicated to understanding co-offending patterns and their variations between and within individuals. This is likely due, in part, to the fact that many of the early investigations of co-offending were cross-sectional in nature and based on small samples of predominantly white, young males. Although these studies were quick to identify variations in rates of co-offending observed for different crime types (which tend to be high for property crimes like robbery, burglary, and theft of motor vehicles and very low for sex crimes), their restricted samples often precluded detailed analyses of the relationships between co-offending and key offender characteristics like sex, age, ethnicity, and criminal history.

It is only relatively recently that growing empirical interest in the topic of group crime, and the availability of larger and more inclusive datasets, has led to considerable advances in our understanding of attribute-based variations in co-offending. To date, most research has focused on identifying variations in the extent of co-offending; however, a growing body of literature suggests that there are also notable individual-level variations in the nature of co-offending with respect to the composition and character of one’s co-offending involvement. Understanding these variations is important not only for theory development but also because accurate estimates of crime rates and key criminal justice statistics, like the probability of conviction and the effects of incapacitation, must take account of varied rates of co-offending for different social groups. Because an offense committed by two offenders is counted twice in many criminal justice statistics (e.g., arrest and conviction data), the share of crime attributed to offender groups that have a greater tendency to co-offend will be artificially inflated, and estimates of interventions targeting these offenders will be subject to greater distortions (Andresen and Felson 2009; van Mastrigt and Farrington 2009; Zimring 1981). Thus, as McCord and Conway (2005: 11) have noted, “crime.. .cannot be accurately portrayed unless co-offending is accurately recorded” and disaggregated for different offender groups.

The importance of information on individual-level variations in co-offending is clear. Unfortunately, knowledge in this area is still underdeveloped (in particular that exploring within individual co-offending across the criminal career). Nevertheless, a number of consistent attribute-based variations in the prevalence and form of co-offending have been identified in the extant literature. These are outlined below.

Empirical Evidence


Compared to the vast literature dedicated to exploring other gender differences in crime, relatively little work has considered how patterns of group crime vary for males and females.

With respect to sex differences in the prevalence of co-offending, the bulk of the existing evidence suggests that females are slightly more likely than males to offend with accomplices, although a handful of studies have reported the opposite. Citing West German police statistics from 1982, for example, Reiss (1988: 134) reported that across age and crime type, males displayed higher aggregate rates of co-offending compared to females (32 % vs. 24 %), a finding that was later replicated (to a lesser extent) in a number of Swedish investigations of juvenile co-offending (Pettersson 2005; Sarnecki 1990, 2001). More commonly, however, females are found to have higher co-offending rates. In one of the earliest studies to explicitly compare male and female co-offending patterns, it was concluded that in the US context, “females in both urban and rural settings tend to have group violation rates that are higher than their male counterparts” (Erickson and Jensen 1977: 268). Similar conclusions have since followed from research conducted in other countries, including Holland (Hakkert 1998: 466), Scotland (Emler et al. 1987: 104), and England (Budd et al. 2005), although the magnitude of difference is typically small (often less than 10 %).

Most studies exploring gender differences in co-offending, including those cited above, have been based on juvenile samples and have been limited to bivariate analysis of the sex/co-offending relationship. Because the age distributions and types of offenses committed by male and female offenders is known to differ, however, it is important to consider whether the demonstrated sex/co-offending relationship could be spurious and, in fact, reflect variations in the ages or types of crimes committed by male and female co-offenders. The findings from two recent multivariate investigations of co-offending suggest that this is not likely to be the case. Based on a large-scale analysis of Canadian Uniform Crime Report (UCR) 2 incident-based data, Carrington (2009: 1306) reported that with the exception of sex offending, females’ rates of juvenile co-offending were higher across nearly all major categories of offense and that this pattern was maintained across age and criminal experience. His overall conclusion that “the


offense participations of females were more likely to involve accomplices than those of males, but the difference is not large (57 % vs. 51 %)” echoes that reported by van Mastrigt and Farrington (2009: 565) across a wider age range (10–74 years); in their study, females had approximately 20 % higher logged odds of co-offending, even when controlling for age and crime type. These multivariate findings provide the strongest evidence yet that females are (slightly) more likely than males to engage in group offending and that the sex/co-offending relationship cannot be attributed solely to the types of crimes that males and females commit, or to the age of particular study populations. What has yet to be determined is what can explain this sex difference. Whether females are more likely to co-offend because they are more socially oriented in general, more susceptible to peer pressure, or something else entirely remains an empirical question.

In addition to variations in the extent of male and female co-offending, a number of sex-linked variations in the composition of males’ and females’ co-offending groups have also been identified. One of the most consistent findings to emerge in this regard is that across age, males tend to offend in larger groups. Although numerous studies have shown that offending in groups of two is the overwhelming norm for both sexes, pair offending is “somewhat.. .more common among females and offending in groups of three or more is slightly more common among males” (Carrington 2002a: 289; Hakkert 1998; Sarnecki 2001).

Another interesting difference concerns the gender composition of males’ and females’ offending groups. In general, while both males and females appear to choose co-offenders of the same sex in the majority of cases (especially during youth), “females are [nonetheless] more often found in the company of males than vice versa” (Warr 2002: 79, also see Pettersson 2005; Reiss and Farrington 1991; Sarnecki 2001). In their analysis of 400 Philadelphia juveniles’ court records, for example, Conway and McCord (2002) reported that whereas 95.5 % of male co-offenders had a male accomplice in their first recorded co-offense, only 80.6 % of females had female co-offenders. In older samples, even more striking differences are evident. In Carrington’s (2002b) Canadian analysis, 87 % of co-offending males compared to only 50 % of co-offending females were involved in same-sex incidents. Often, these findings are interpreted, either explicitly or implicitly, to indicate that males have a much stronger preference for same-sex accomplices, perhaps because they are viewed as more competent or able crime partners (Steffensmeier and Terry 1986). However, in a population highly skewed in favor of males, as the offending population is, a preponderance of all-male groups and a scarcity of all-female groups would be expected even if offenders chose their accomplices at random (Carrington 2002b). Recent analyses have shown that when the observed sex compositions of offending groups are compared with those expected based on the gender distribution of the offending population, males do, in fact, exhibit a slightly greater (statistical) bias for same-sex accomplices overall, but this difference is relatively small, and for offending groups with two offenders (the vast majority), this bias is greater for females (van Mastrigt and Carrington 2013). These findings highlight the importance of considering chance expectations when comparing the compositions of males’ and females’ co-offending groups.

Such considerations will also be relevant for the interpretation of other recent findings reported in the co-offending literature, for example, the tendency for boys involved in violent offending to be tied to co-offenders from the same neighborhood to a greater extent than females (Pettersson 2005), the suggestion that when males are involved in mixed-sex offending it is usually with family members or romantic partners (Reiss and Farrington 1991; Warr 2002), and the finding that mixed-sex offending groups appear to be slightly more likely to reoffend than same-sex groups of either gender (van Mastrigt 2008:203). Going forward, it will be important to determine whether these patterns reflect meaningful sex differences in the preference for particular types of co-offenders, or differential opportunity or social network structures that render some types of accomplices more “available” to one sex or the other (Andresen and Felson 2009; Weerman 2003).

Additional attention is also warranted with respect to the offending roles adopted by males and females engaged in group crime. More than 25 years ago, Reiss (1988) stressed the importance of identifying potential co-offending recruiters or instigators, that is, individuals who introduce others into crime and/or play a primary role in initiating offenses. Although there is often a suggestion that males are more likely than females to adopt this role and that male instigation plays a particularly important part in female offending (Sarnecki 2001), very little empirical work has examined this question. Although an earlier analysis of National Youth Survey data showed that females were less likely than males to report being an instigator of crime (Warr 1996), two more recent multivariate analyses of recruitment and instigation, respectively, failed to identify a gender effect when controlling for other person and situational-level variables (McGloin and Nguyen 2012; van Mastrigt and Farrington 2011).

Taken together, the findings outlined above suggest that there are a number of consistent gender differences in the prevalence and character of co-offending and that sex-linked variations deserve serious attention from both researchers and theorists. At the same time, many of the identified differences are small, and a number of the group crime patterns observed across sex are more similar than different (e.g., with respect to the overall tendency for offending groups to be small, homogeneous, and unstable). Given that many features of crime are very strongly differentiated by gender (including the basic tendency to offend), it is perhaps surprising that the sex variations noted here are not of a greater magnitude. This in itself raises interesting questions for future work. Clearly, there is much more to be learned about how gender shapes the dynamics of co-offending.


One of the attributes that has received the most attention in the general criminological literature and the least in the co-offending field is ethnicity. Very little systematic research has explored ethnic or racial variations in the extent and form of co-offending. Citing US victimization statistics from 1982, Reiss (1988:134) was the first to identify the potential importance of this attribute, noting that for violent offenses, black offenders were “somewhat less likely than whites to be solo offenders” (60 % vs. 72 %). More recently, Felson (2003: 154) reported a similar finding based on his analysis of the same data from the year 2000, noting that “although nonwhites constitute 20 % of the total U.S. population, violent victims of lone offenders report their attackers to be nonwhite 34 % of the time. By comparison, victims of group violence describe their attackers as nonwhite or mixed in 56 % of reports.” Unfortunately, comparative prevalence analyses for non-US populations, nonviolent crimes, and other ethnic and racial groups are lacking. Thus, while there is preliminary evidence that race may play a role in whether American offenders commit their violent crimes alone or in groups, the extent to which this is the case more generally is unclear. Additional data and multivariate analyses which can disentangle the effects of ethnicity and related variables (e.g., SES, crime type) will be needed before definitive statements can be made regarding ethnic and racial differences in the tendency to co-offend; however, it appears likely that some variation exists.

In addition to impacting whether an offense involves co-offenders, a handful of studies suggest that ethnicity also plays a role in the selection of those co-offenders. In the American context, individuals from different racial backgrounds are rarely suspected of committing offenses together. Based on the same US victimization data noted above, Reiss (1988: 135) reported that less than 6 % of violent group victimizations were perpetrated by mixed-race groups, leading him to conclude that offending with an accomplice of a different race is even less likely than offending with a co-offender of the opposite sex. In a more recent study of violent crime committed by juveniles in Philadelphia, co-offending was similarly characterized by marked racial homogeneity such that 96 % of the accomplices linked to African-American suspects and 83 % of those linked to Whites and Latinos were of the same ethnic background (Conway and McCord 2002). Outside of North America, the results are more mixed. In the UK-based Cambridge Study in Delinquent Development, high levels of homogeneity were reported for whites, but not blacks, a finding that the authors attributed in part to “the relative distribution of whites and blacks in the London population” (Reiss and Farrington 1991: 391).

In the most comprehensive analyses of ethnicity and group crime published to date, the co-offending networks of all youths (under the age of 21) suspected of committing a violent offense in Stockholm, Sweden, in 1995 were constructed (Pettersson 2003; Sarnecki 2001). Using country of birth and citizenship status to code offenders into one of eight ethnic groups, the authors found that the probability of choosing a person from one’s own ethnic group as a co-offender was about three times as high as would have been expected if the choices had been made at random. Further, some differences between ethnic groups and crime types were observed. Individuals with an African or a South/Central American background were particularly likely to choose co-offenders of the same ethnicity, and assaults and threatening behavior were more likely than muggings to involve ethnically homogeneous groups.

These findings suggest that complex relationships between ethnicity, crime type, and co-offending are likely to exist. However, the varied ethnic profiles of previous samples together with a nearly exclusive focus on violent crime make detailed conclusions about the ethnicity/co-offending relationship, and hypotheses regarding its roots, impossible to draw at this stage. What does seem clear is that overreliance on simple racial dichotomies like “black” and “white” may fail to detect important co-offending variations like those reported in the Swedish data. As research in this area develops, it will be important to consider the most appropriate ethnic categorizations for different geographical areas, and how to best compare them. Another important issue for future investigations of the ethnicity/ co-offending relationship and, in particular, the ethnic composition of offending groups will be to control for the effects of potential confounds like geographic racial segregation. As Sarnecki (2001: 136) has warned, given that many co-offending relationships are formed within the confines of one’s local neighborhood and activity spaces (see also, Reiss and Farrington 1991), it could be the case that the “the ethnic dimension of co-offending is actually governed in the first instance not by the ethnic background of the actors.. .but rather by the type of residential segregation to be found in a society.” Investigations which assess the effect of ethnicity above and beyond that of propinquity (and other variables) will be needed to clarify their respective roles in the selection of co-offenders.


One of the most consistent findings in the co-offending literature is that youths are significantly more likely than adults to commit crimes in groups. Numerous studies conducted across historical time periods and geographical areas have shown that in the aggregate, the age/ co-offending curve rises sharply in early adolescence, peaks in the mid to late teens, and decreases steadily thereafter, mirroring the shape of the general age/crime curve (Andresen and Felson 2012; Carrington 2002a; Piquero et al. 2007; Stolzenberg and D’Alessio 2008; van Mastrigt and Farrington 2009). Although this aggregate decrease in co-offending with age is observed across all major studies, there are considerable variations in the specific prevalence rates reported by year of age in different investigations. On the one hand, many studies suggest that “solo offending.. .does not become the modal form of offending until the late teens or early twenties” (Reiss 1988: 151, see also Andresen and Felson 2012). In the Cambridge Study in Delinquent Development, for example, 75 % of offense participations committed by boys at ages 10–13 involved co-offenders, solo offending became the norm between ages 21 and 24 and accounted for 72 % of offense participations at ages 37–40 (Piquero et al. 2007:104). In contrast, some studies suggest that co-offending accounts for less than 50 % of offending at all ages. In an influential Canadian study, lone offending exceeded co-offending from the age of 11 onwards (Carrington 2002a: 289), and two recent large-scale analyses of police data from the USA (Stolzenberg and D’Alessio 2008) and the UK (van Mastrigt and Farrington 2009) revealed that, in the aggregate, the majority of even the youngest offenders committed their offenses alone (see also, Carrington & van Mastrigt 2013). The reasons for these varied findings are somewhat unclear, but as Andresen and Felson (2012) have stated, “whether or not co-offending is reported to be the norm at any given age, all existing research agrees that on the whole it is nonetheless more common amongst younger compared to older offenders.” Further, this holds even when controlling for potential confounds like gender and crime type (Carrington 2009; van Mastrigt and Farrington 2009). The decrease in co-offending with age thus appears to occur independently of variations in other offender and offense attributes.

As Reiss and Farrington (1991) discuss, aggregate declines in co-offending with age could either reflect changes in the population of offenders (e.g., if co-offenders tend to desist at early ages while solo offenders persist) or behavioral changes within offenders (e.g., if offenders become more likely to commit crimes alone with age). In the first empirical test of this question, Reiss and Farrington analyzed prospective longitudinal data from the Cambridge Study. They found a significant negative correlation between age and average number of co-offenders for even the most persistent offenders in their sample, a finding that led them to conclude that the decrease in co-offending with age was a developmental process, which was “not caused by the persistence of solo offenders and/or the dropping out of co-offenders, but instead reflects changes within individual criminal careers” (1991:382; also see Piquero et al. 2007).

More recently, however, evidence for both selective desistance and shifts in individual careers has emerged (McGloin et al. 2008: 174). Trajectory analyses of juvenile co-offending careers in Philadelphia revealed that approximately 50 % of juveniles who demonstrated a pattern of decreasing co-offending in late adolescence desisted, while the other half appeared to shift their criminal careers towards solo offending. Still others exhibited a pattern of increased co-offending behavior over time. The varied findings that have emerged from this and other (e.g., Carrington 2009) studies highlight the risk that “aggregate indicators of co-offending over time can be analytically insensitive and misleading since they mask unique and varying developmental trajectories” at the individual level (McGloin et al. 2008). The need for more disaggregated longitudinal analyses of the age/ co-offending relationship, and further research on the impact of early co-offending trajectories for the subsequent criminal career, is paramount (see next section).

In cases where within-individual changes from joint to lone offending do occur, the obvious question is: what underlies the shift? One possibility is that the age-related change in co-offending reflects variations in the normative influence and salience of peers across the life course. Some argue that youths are simply more gregarious and peer oriented than adults; thus, offenders tend to offend alone more often as they grow older because they tend to do everything less often with peers as they age (Stolzenberg and D’Alessio 2008; Warr 2002). As Carrington (2009: 1299) states, “children and teenagers like to do things with their peers and in groups, and committing crimes is just another of their group activities. With maturity comes greater autonomy and, therefore, more solo offending.”

An alternative explanation is that the shift from joint to lone offending is related to the accumulation of criminal experience (Reiss and Farrington 1991). According to this view, as offenders develop criminal expertise with age, the need for and/or attractiveness of accomplices decreases, particularly in cases where crime proceeds would need to be shared. While this may, indeed, be the case, the current evidence suggests that criminal experience alone does not explain the age/co-offending relationship. In the most recent analysis of the Cambridge Study, for example, Piquero and colleagues (2007) concluded that “the rate of co-offending remains relatively constant for up to eight offenses before declining.. .[thus].. .age may be more important than experience in explaining the decline in co-offending.” Recent multivariate evidence from a Canadian juvenile criminal career study also showed that age and experience had independent effects in the prediction of co-offending, although these variables were also related in complex ways (Carrington 2009).

To recap, the current evidence suggests that the offender attribute of age is one of the strongest aggregate-level predictors of co-offending (typically, second only to crime type). The shape of the aggregate age/co-offending curve is robust, observed across gender and criminal experience, and for most major crime types. When the age/co-offending relationship is examined within individuals, however, it is considerably more varied. Many individuals’ co-offending trajectories mirror the aggregate curve while others diverge, for reasons that have yet to be identified. Thus, exclusive reliance on aggregate analysis of the age/co-offending relationship may neglect important offender groupings (e.g., individuals with increasing rates of co-offending with age). These “atypical” offenders warrant further attention.

In addition to consistent evidence that co-offending prevalence changes with age, a growing body of research also shows that co-offending dynamics change with age such that, overall, the offending groups formed at older ages tend to be smaller and more heterogeneous. With respect to group size, for example, whereas the majority of co-offenses involve two offenders at all ages, incidents with three or more offenders are fairly common in adolescence, but become very unlikely after age 20 (Reiss and Farrington 1991). In addition to changes in the number of accomplices selected by offenders of different ages, there is also strong evidence that the types of accomplices selected vary with age. Research from the USA (Warr 2002), Sweden (Sarnecki 2001), and England (van Mastrigt 2008) all reveal a “slight tendency for the age difference between actors and their co-offenders.. .to increase as actors became older.” This finding was explored in more detail in a recent analysis of age homophily (van Mastrigt and Carrington 2013) in which youths were found to exhibit significantly greater clustering as compared to adults, even when controlling for chance expectations based on the age distribution of the offending population.

Patterns of gender homogeneity also appear to change with age. Among Swedish juveniles, for example, most co-offending dyads up to age 20 were unisexual (and in particular, all male); however, the likelihood of committing crimes with accomplices of the opposite sex was significantly increased for the oldest offenders (Sarnecki 2001: 54). Similar evidence has also emerged in England, where adults ( 18) in one study were found to have more than twice the odds of offending in mixed-sex groups compared to youths (van Mastrigt and Farrington 2009).

Research further suggests that the locations from which co-offenders are selected varies with age, such that geographic proximity plays a more fundamental role in the selection of co-offenders in adolescence, and becomes less important later in life. Whereas 100 % of co-offenders aged 10–13 in the Cambridge Study lived in the same postal district in London, only 53 % of those aged 21–32 lived in such close proximity. Thus, “although it is clear.. .that propinquity in the selection of co-offenders declines with age, at all ages a sizable proportion of all co-offenders in any offense reside in fairly close proximity to each other” (Reiss and Farrington 1991:389). This finding raises interesting questions about the shared (and divergent) “offender convergence settings” and accomplice pools that facilitate co-offender selection at different ages (see Andresen and Felson 2012).

The aggregate tendency for individuals to select co-offenders less similar to themselves as they age likely reflects the well-known sociological fact that friendships and relationships tend to diversify with age. In childhood, individuals tend to form large, predominately same-sex friendship groups with other kids from the same neighborhoods or schools. As children become teenagers and then adults, they gain increased mobility, enter the workforce, and start to form more lasting friendships and relationships with members of the opposite sex. All of these milestones act to expand the boundaries of individuals’ social networks which, in turn, increases their exposure to less similar others, including potential accomplices. More research is needed to explore whether the age-related changes that have been observed in offending groups simply reflect structural constraints related to the changing constellations of individuals’ general social networks (from which accomplices are often drawn) or whether these patterns reflect different, crime-specific processes related to aging and co-offending. Teasing these issues apart will require longitudinal comparative analyses of co-offenders’ conventional and illicit network ties across the life course. Unfortunately, such data is extremely rare.

In sum, there is clear evidence that in the aggregate, joint participation in crime peaks in adolescence, drops sharply in early adulthood, and settles to a low plateau in later life. However, disaggregated analyses show that the age/co-offending relationship varies somewhat more when examined from the perspective of individual criminal careers. With respect to changes in the composition of offending groups across age, there is evidence for both similarities and differences between youths’ and adults’ co-offending groups. While the vast majority of all co-offending groups are small, unstable, and relatively homogeneous at all ages, adults nonetheless tend to offend in slightly smaller, more heterogeneous groups than youths.

Criminal Careers

The notion that co-offending involvement generates and/or reflects differential criminal career trajectories is long standing but rarely investigated empirically (not least due to a shortage of the requisite data). In their infamous description of Sidney’s criminal career, Shaw and McKay (1931) noted clear changes in both the level and form of Sidney’s criminal behavior that were closely tied to his changing accomplice associations. Despite frequent references to this finding across the criminological literature, co-offending remains “one of the most ill studied of all criminal career dimensions” (Piquero et al. 2007: 120). In this section, various dimensions of the criminal career are framed as offender “attributes” and considered as either causes or effects of co-offending. Although research in this area is still in its infancy, a number of key criminal career features including offending style, persistence, and versatility appear to be important.

Co-offending careers may take one of three forms: (1) an exclusively solo career, where the individual always engages in crime alone; (2) an exclusively co-offending career, where the individual never offends alone; and (3) a mixed co-offending career, where the individual commits a combination of joint and lone offenses. For offenders with numerous offenses, “the typical criminal career is a mix of offenses committed alone and with others” (Reiss 1988: 117; Reiss and Farrington 1991; Sarnecki 2001). At the same time, it appears that active offenders also exhibit “a style of offending” (i.e., predominantly group or loner) over time (Warr 1996: 22; Reiss and Farrington 1991; Weerman 2003). In a recent examination of juvenile co-offending careers in Philadelphia, for example, considerable consistency was observed in many individuals’ co-offending behavior, such that “some classes [of offenders] contain relatively high-rate co-offenders.. .whereas others comprise relatively low-rate co-offenders,” although a number of individuals also exhibited a high degree of versatility in their co-offending behavior with some increasing and others decreasing their co-offending activity over time (McGloin et al. 2008: 174).

The observed between- and within-individual variability in co-offending trajectories begs two questions: what might explain these differences? And what do they matter for the continued criminal career? With respect to the first question, one possibility is that individual differences render particular “types” of offenders more or less likely to engage in co-offending across the criminal career. According to Moffitt’s (1993) model of adolescence limited (AL) and life course persistent (LCP) offenders, for example, LCP offenders might be expected to offend alone to a greater extent than AL offenders because they are more internally motivated to offend and less likely to require the social support or practical assistance of accomplices. According to this population heterogeneity view, patterns of co-offending are assumed to reflect, rather than shape, varied offender trajectories. Thus, in the AL vs. LCP example above, offenders with low co-offending involvement might be expected to exhibit more life course persistent criminal careers. Alternatively, involvement in co-offending might lead offenders down the road to persistent offending because it increases criminal opportunities, skills, and/or awareness (Andresen and Felson 2009; Carrington 2009; McGloin and Nguyen 2012). In this case, high levels of co-offending involvement might be expected to predict more persistent and chronic offending careers.

A number of studies have investigated the co-offending/persistence relationship, with mixed results. In support of the hypothesis that co-offending should be negatively correlated with offending frequency and persistence, Knight and West (1975) found that in their long-term follow up of London boys, offenders with a history of lone offending were, in fact, less likely than joint offenders to be “temporary” delinquents. Similarly, based on her analysis of UK police data, van Mastrigt (2008: 172) found that over a 3-year period, there was a negative relationship between the number of co-offenders per offense and offending frequency. In contrast, co-offending increased the likelihood of persistent criminality in the Philadelphia study, particularly when combined with an early age of onset (Conway and McCord 2002). Age was also found to play an important role in conditioning the relationship between co-offending and offending frequency in Carrington’s (2009: 1317) Canadian study. His analysis revealed two distinct delinquent careers (high activity and low activity), which were characterized by different relationships between co-offending, criminal experience, and age; whereas co-offending increased slightly with overall criminal activity for high-activity offenders, it decreased with experience for low-activity offenders. Further, for nearly three-quarters of offense participations – those involving low-activity offenders committed during the teenage years – co-offending decreased with the offender’s age, criminal experience, and overall career activity. In contrast, for offense participations of high-activity offenders as teenagers, co-offending decreased with age but varied little with experience and career activity. Finally, for offense participations committed during childhood, co-offending decreased with increasing career activity but varied only slightly, if at all, with the offender’s age and criminal experience. These complex differences highlight the need to consider potential interactions between multiple criminal career attributes and co-offending.

In addition to studies identifying positive and negative relationships between co-offending and career activity, a handful of studies have demonstrated null effects. For example, Reiss and Farrington (1991: 376) and Piquero and colleagues (2007: 110) found no relationship between average number of co-offenders per offense and total number of convictions in the Cambridge Study. The conclusion that “the proportion of an offender’s offenses that are committed with others does not differentiate chronic from non-chronic offenders” was further supported in a recent analysis of the Racine cohort data (McGloin and Stickle 2011: 14). Interestingly, although chronic and non-chronic offenders had comparable levels of co-offending with peers in the Racine data, less active offenders were significantly more likely to report offending “because of peers.” Thus, whether or not high- and low-activity offenders have similar co-offending participation rates, it seems likely that the role of co-offending differs in the criminal careers of chronic vs. non-chronic offenders.

Whereas co-offender influence may directly motivate offending for less serious offenders, accomplices likely play a less causal, but more practical, role in the criminality of chronic offenders, for example, by making some crimes easier to commit.

Irrespective of the relationship between co-offending and the persistence of the criminal career, it seems clear that involvement in co-offending with particular types of accomplices has the potential to shape the character of one’s subsequent criminal career in important ways. In the Philadelphia co-offending study, for example, offending with a violent co-offender in the first offense was a significant predictor of future violence, even when the joint offense itself was not violent (Conway and McCord 2002). In another study, offenders who could identify a criminal mentor (approximately = of whom were also accomplices) tended to report enhanced criminal earnings and reduced periods of incarceration, a finding which led the authors to conclude that “tutelage guides the criminal maturation process” (Morselli et al. 2006: 19). Further evidence for the impact of co-offending comes from a recent study in which juveniles who were tied into large, nonredundant co-offending networks exhibited more versatile criminal careers than solo offenders (McGloin and Piquero 2010). This finding is consistent with other aggregate-level evidence that, compared to solo offending, the distribution of co-offending is more varied across crime types (Andresen and Felson 2012). Together, these findings suggest that offending with others has the potential to enhance exposure to diversified criminal skills and deviant contacts and that the specific content of these exposures matters for the development of the criminal career.

Various offender and criminal career attributes may also impact the role one adopts within co-offending groups. In a recent investigation of self-reported instigation among a sample of incarcerated male offenders in the USA, both early age of onset and crime-specific expertise predicted instigation within crime type, a finding that highlights the importance of considering how individual and situational predictors may work together to shape co-offending behavior in particular offenses and across the criminal career (McGloin and Nguyen 2012).

In sum, the current evidence suggests that co-offending can act as both an important feature, and determinant, of the criminal career. Framing key dimensions of the criminal career as offender attributes and linking them to co-offending behavior over time may hold promise for development of both the co-offending and criminal career literature. Unfortunately, focused studies of the criminal career/co-offending relationship are rare at present. Better longitudinal data on co-offending, and more information about variations in the selection and impacts of co-offenders across the life course, will be particularly important for future development in this area.

Implications And Future Directions

Taken together, the current evidence provides strong support for Waring’s (2002: 321) skepticism that “co-offending can or should be treated as a unitary phenomenon.” To a greater or lesser degree, sex, ethnicity, age, and several key criminal career features have all been shown to vary with co-offending prevalence and form. This is significant for a number of reasons.

From a theory-development perspective, the fact that at least some of these person-based variables maintain simultaneous and independent effects on the prediction of co-offending, even when controlling for well-established situational-level variables (such as crime type), highlights the importance of integrated models of co-offending. Offender attributes, although important, do not exist in a vacuum and are only one piece of the co-offending puzzle. Explanations which are sensitive to the effects and interactions of individual offender attributes, routine activities, social networks, and situational contexts are likely to shed the best light on co-offending patterns and processes. Further, given that observed variations in the extent of co-offending are sometimes greater (and sometimes smaller) than differences in the form of co-offending, theoretical frameworks that allow for conceptual distinctions between the factors that determine whether a crime is committed with others and the factors that determine the selection of those others may be particularly powerful.

The co-offending variations outlined here are also important from a policy perspective. First, because co-offending prevalence appears to be higher for some social groups (i.e., females, ethnic minorities, young people, and individuals exhibiting particular criminal career trajectories), these groups will look “worse” in criminal statistics that do not take account of co-offending (for recent discussion of this issue see Andresen and Felson 2009; McCord and Conway 2005; van Mastrigt and Farrington 2009). As outlined earlier, these offender groups will be more affected by the multiple counting procedures inherent to many offender-based crime measures, and their share of crime will be distorted accordingly. Secondly, the fact that accomplice selection processes appear to vary somewhat across offender groups (and also, perhaps, within criminal careers) means that one-size-fits-all efforts to disrupt co-offending networks may not provide the greatest returns. To give but one example, efforts to hamper co-offending via situational crime prevention will likely be effective only to the extent that these efforts are sensitive to identifying how “offender convergence settings” vary across age and gender (Felson 2003). Similarly, targeted interventions aimed at individuals with particular types of co-offending careers may be more effective than those aimed at accomplice networks in general. Individuals who consistently exhibit recruitment or instigation behavior may be of particular interest (although even interventions targeted at these “extreme” co-offenders must recognize the importance of situational variables). In sum, to treat co-offending as invariant across and within individuals may not only obscure important features of the phenomenon from an academic perspective, it may also hinder attempts to address it in practice.

At the same time, attribute-based variations in co-offending should not be overstated. Many of the between-group differences identified in the current literature are small and should not deter efforts to identify general risk factors for co-offending. There is tremendous room for growth in the co-offending literature and need for additional research at all levels of analysis (individuals, offenses, environments). With respect to offender attributes specifically, future directions are endless. Only a handful of the individual-level variables most readily available in official records have been discussed here; however, studies based on other data sources are already beginning to identify other potential correlates of co-offending, including economic adversity and cooperative orientation (McCarthy et al. 1998), being unmarried and undereducated (Alarid et al. 2009), and exhibiting lower levels of psychosocial maturity and higher levels of anxiety and psychopathy (Goldweber et al. 2011). As the literature in this area continues to develop, an even greater number of attribute-level and situational-level variables are likely to emerge as significant correlates of group crime. The challenge for future co-offending researchers, theorists, and policy makers will be to evaluate the role of these variables and the person-by-situation interactions that tie them together. To navigate this complexity will be no small feat.

Take Home Message

There is notable individual-level variation in the extent to which offenders commit crimes alone and with others, and the form that their co-offending takes. Offender attributes such as sex, ethnicity, age, and criminal history have all been shown to impact co-offending, but questions remain regarding the underlying reasons for these effects, and potential interactions with situational and structural factors. There is tremendous room for development in this field; as a first step, it is recommended that information on a wider range of offender, offense, and situational correlates are collected in future co-offending research.


  1. Alarid LF, Hochstetler AL, Burton VS (2009) Group and solo robberies: do accomplices shape criminal form? J Criminal Just 37:1–9
  2. Andresen MA, Felson M (2009) The impact of cooffending. Brit J Criminol 50:66–81
  3. Andresen MA, Felson M (2012) Co-offending and the diversification of crime types. Int J Offender Ther Comp Criminol 56(5):811–829
  4. Budd T, Sharp C, Mayhew P (2005) Offending in England and Wales: first results from the 2003 crime and justice survey. Home Office research study, vol 275. Home Office, London
  5. Carrington PJ (2002a) Group crime in Canada. Can J Criminol 44:277–315
  6. Carrington PJ (2002b) Sex homogeneity in co-offending groups. In: Hagberg J (ed) Contributions to social network analysis, information theory and other topics in statistics. A festschrift in honour of Ove Frank. Stockholm University Press, Stockholm, pp 101–116
  7. Carrington PJ (2009) Co-offending and the development of the delinquent career. Criminology 47:1295–1329
  8. Carrington PJ, van Mastrigt SB (2013) Co-offending in Canada, England and the United States: a cross national comparison. Global Crime. doi:10.1080/17440572.2013.787926 (Online First)
  9. Conway KP, McCord J (2002) A longitudinal examination of the relation between co-offending with violent accomplices and violent crime. Aggress Behav 28:97–108
  10. Emler N, Reicher S, Ross A (1987) The social context of delinquent conduct. J Child Psychol Psychiat 28(1):99–109
  11. Erickson ML, Jensen GF (1977) Delinquency is still group behavior! Towards revitalizing the group premise in the sociology of deviance. J Crim Law Criminol 68:262–273
  12. Felson M (2003) The process of co-offending. In: Smith MJ, Cornish DB (eds) Theory for practice in situational crime prevention, vol 16, Crime Prevention Studies. Criminal Justice Press, Monsey
  13. Goldweber A, Dmitrieva J, Cauffman E, Piquero AR, Steinberg L (2011) The development of criminal style in adolescence and young adulthood: separating the lemmings from the loners. J Youth Adolesc 40 (3):332–346
  14. Hakkert A (1998) Group delinquency in the Netherlands: some findings from an exploratory study. Int Rev Law Comput Technol 12:453–474
  15. Knight BJ, West DJ (1975) Temporary and continuing delinquency. Br J Criminol 15:43–50
  16. McCarthy B, Hagan J, Cohen LE (1998) Uncertainty, cooperation and crime: understanding the decision to co-offend. Soc Forces 77(1):155–184
  17. McCord J, Conway KP (2002) Patterns of juvenile delinquency and co-offending. In: Waring R, Weisburd D (eds) Crime and social organization. Transaction Publishers, New Brunswick, pp 15–30
  18. McCord J, Conway KP (2005) Co-offending and patterns of juvenile crime. Research in Brief. NCJ 210360. National Institute of Justice, Washington, DC
  19. McGloin JM, Nguyen H (2012) It was my idea: considering the instigation of co-offending. Criminology 50:463–494
  20. McGloin JM, Piquero AR (2009) I wasn’t alone: collective behaviour and violent delinquency. Aust N Z J Criminol 42(3):336–353
  21. McGloin JM, Piquero AR (2010) On the relationship between co-offending network redundancy and offending versatility. J Res Crime Delinq 47:63–90
  22. McGloin JM, Stickle WP (2011) Influence or convenience? Rethinking the role of peers for chronic offenders. J Res Crime Delinq 48(3):419–447
  23. McGloin JM, Sullivan CJ, Piquero AR, Bacon S (2008) Investigating the stability of co-offending and cooffenders among a sample of youthful offenders. Criminology 46:155–188
  24. Moffitt TE (1993) Life-course persistent and adolescencelimited antisocial behavior: a developmental taxonomy. Psychol Rev 100:674–701
  25. Morselli C, Tremblay P, McCarthy B (2006) Mentors and criminal achievement. Criminology 44(1):17–43
  26. Pettersson T (2003) Ethnicity and violent crime: the ethnic structure of networks of youths suspected of violent offenses in Stockholm. J Scand Stud Criminol Crime Prev 4:143–161
  27. Pettersson T (2005) Gendering delinquent networks. A gendered analysis of violent crimes and the structure of boys and girls co-offending networks. Young 13:247–267
  28. Piquero AR, Farrington DP, Blumstein A (2007) Key issues in criminal career research: new analyses of the Cambridge Study in Delinquent Development. Cambridge University Press, Cambridge
  29. Reiss A (1988) Co-offending and criminal career. In: Morris N, Tonry M (eds) Crime and justice, vol 10. University of Chicago Press, Chicago
  30. Reiss AJ, Farrington DP (1991) Advancing knowledge about co-offending: results from a prospective longitudinal survey of London males. J Crim Law Criminol 82:360–395
  31. Sarnecki J (1990) Delinquent networks in Sweden. J Quant Criminol 6(1):31–50
  32. Sarnecki J (2001) Delinquent networks. Cambridge University Press, Cambridge
  33. Shaw CR, McKay HD (1931) Report on the causes of crime. US Government Printing Office, Washington, DC
  34. Steffensmeier D, Terry RM (1986) Institutional sexism in the underworld: a view from the inside. Sociol Inq 56(3):304–323
  35. Stolzenberg L, D’Alessio SJ (2008) Co-offending and the age-crime curve. J Res Crime Delinq 45(1):65–86
  36. van Mastrigt SB (2008) Co-offending: relationships with age, gender and crime type. Ph.D. Dissertation, University of Cambridge
  37. van Mastrigt SB, Carrington PJ (2013) Sex and age homophily in co-offending networks: opportunity or preference? In: C Morselli (Ed.) Crime and networks. Routledge, New York, NY
  38. van Mastrigt SB, Farrington DP (2009) Co-offending, age, gender and crime type: implications for criminal justice policy. Brit J Criminol 49:552–573
  39. van Mastrigt SB, Farrington DP (2011) Prevalence and characteristics of co-offending recruiters. Justice Quart 28:325–359
  40. Waring EJ (2002) Conceptualizing co-offending: a network form of organization. In: Waring E, Weisburd D (eds) Crime and social organization: advances in criminological theory, vol 10. Transactions Publishers, New Brunswick
  41. Warr M (1996) Organization and instigation in delinquent groups. Criminology 34:11–37
  42. Warr M (2002) Companions in crime: the social aspects of criminal conduct. Cambridge University Press, Cambridge, MA
  43. Weerman F (2003) Co-offending as social exchange: explaining characteristics of co-offending. Brit J Criminol 43:398–416
  44. Zimring FE (1981) Kids, groups, and crime: some implications of a well-known secret. J Crim Law Criminol 73(3):867–885

See also:

Free research papers are not written to satisfy your specific instructions. You can use our professional writing services to buy a custom research paper on any topic and get your high quality paper at affordable price.


Always on-time


100% Confidentiality
Special offer! Get discount 10% for the first order. Promo code: cd1a428655