Criminal Careers and "Career Criminals,": Volume II (1986)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

4 Co Offender Influences on Criminal Careers Albert I. Reiss, Jr. INTRODUCTION Offender histories ordinarily are char- acterizec! by a mix of different types of offenses and by a mix of offenses commit- ted alone and with accomplices. Group or accomplice offending is more characteris- tic of juvenile than adult careers. This paper reviews the current state of knowI- ecige about co-offending in juvenile and adult criminal careers to illuminate how and when co-offencling is relevant to strat- egies of intervention in criminal careers, particularly strategies to reduce crime rates. The paper begins with a brief cle- scription of the major policy questions that will be acIdressec3 and then summa- rizes the knowledge about co-offending that is relevant to them. Selective Incapacitation of Offenders One of the major strategies proposer! to reduce the crime rate is to incapacitate Albert J. Reiss, Jr., is the William Graham Sumner professor of sociology and in the Institution for Social and Policy Studies, Yale University, and lecturer in law, Yale Law School. The reader is referred to the Acknowledgments section at the end of this chapter. 72z career criminals selectively, i.e., to re- move from society those offenders who have high individual rates of offending. In theory one expects to reduce the num- ber of crimes, and correlatively the num- ber of victims involvecI in those crimes, by the amount of crime the incapacitates] offender wouIcT have committed were he or she not incarcerated. Within limits, that seems to be a reasonable presump- tion, provi(led the crimes are committee! by a single offender. But whether one actually prevents those crimes from oc- curring when incapacitating an offender will (lepenc3 also on the group status of the offender and the behavior of co-offend- ers and their affiliatecl offending groups. For unless an offender's accomplices are (leterre(1 from offending by the offen~ler's incapacitation, no crimes may be saved. The accomplices may continue to commit the offenses alone, with one another, by recruiting new accomplices from within their group, or by recruiting new mem-  bers to their membership network either as new participants in offending or at increased rates of offending. Group or- ganization and affiliation will facilitate the search for accomplices, and indeed,

122 Reiss (1980:1~16) has argued that inca- pacitation can increase the crime rate if it leads to a marginal increase in the num- ber of offenders and their individual rates of offending. It is important, conse- quently, to know to what extent patterns of recruitment into offending as well as changes in individual rates of offending may limit the capability of an incapacita- tion policy to reduce the crime rate. Age of Intervention It is well known that a substantial pro- portion of all offending, including offend . . . . 1ng in serious crimes, occurs at young ages and that the age of onset of offending is quite young. Moreover, a serious juve- nile career record appears predictive of high-rate offending in serious crimes as an adult (Chaiken ant] Chaiken, 1982:871. It is also recognized that, while a great many young people participate in crime, many drop out fairly early in their career (Wolfgang, Figlio, and Sellin, 1972:~), often without any official intervention to deter them from offending. Yet, at an early age a sizable minority of youthful offenders have high indiviclual rates of offending (Wolfgang, Figlio, and Sellin, 1972:1041. This raises the question of whether high-rate offenders can be iden- tifiect at an early age so that they can be selected for special treatment in a juve- nile or criminal justice system. Related to the issue of the early cletec- tion and selection of career offenders is whether group affiliation is critical in on- set, persistence, and clesistance from of- fending. Of particular interest is the role that groups play in these phases of a juvenile criminal career. Were one able, for example, to identify high-rate offen(l- ers who recruited a large number of per- sons into committing clelinquent acts or persons who hac] a substantial effect on the individual crime rate of a large nu~n- ber of offenders, one might want to select CRIMINAL CAREERS AND CAREER CRIMINALS those offender-recruiters for special treat- ment in a criminal justice system. Target of Intervention Strategies Many intervention strategies aim to in- tervene in the life of an offender to pre- vent offending either by incapacitating the offender or encouraging Resistance when the offender is allowed to remain in the free society. The latter are usually called individual-change strategies. Yet, other strategies are possible, such as al- tering group and other social structures or intervening in collective activity. If  groups are important to the onset, persis- tence, and Resistance from offending, one may want to intervene directly in group relationships or alter group structure so as to reduce the propensity for delinquent or criminal behavior. Or, recognizing that networks facilitate the search for accom- plices, one may want to intervene in those networks to increase the costs of search behavior. The discussion that fol- lows of the role of group offending in criminal careers will address the implica- tions for intervention strategies. THE NATURE OF GROUP OFFENDING There is no commonly agreed-on defi- nition of a group in research on delin- quent and criminal behavior. Offending groups often are treated in writings on delinquency as synonymous with gangs, the gang being a territorially organized, age-graded peer group engaged in a wide range of activities and having a well- defined leadership (Miller, 1975:9~. Em- pirically, most persons who engage in group delinquency are not members of such highly structured groups (Klein and Crawford, 1967; Morash, 1983:329), and in most aggregates of 20 or more peers, persons are only loosely associated with one another, leadership is unclear, and

CO-OFFENDER INFLUENCES ON CRIMINAL CAREERS membership turnover is fairly high. Yablonsky (1959) refers to these peer ag- gregates as "near-groups." Following Lerman (1967:63), group offending is treated in this paper from the perspective of social networks macle up of pairs, tri- ads, and constellations of four or more persons. The extent to which networks and relationships within them are bounded organizationally, behaviorally, and territorially is left problematic. SOME PRELIMINARY ISSUES Before examining the basic parameters of group offending and their relationship to criminal careers, a few issues that affect concepts and measures of group offencI- ing should be consiclered. These are the relationship of offending to crime events; defining ant! measuring lone, as con- trasted with group, offending; and the effects of criminal justice processing on parameters of group offending. Crime Incidents, Their Victims, and Their Offenders Any crime event or incident involves one or more crimes or offenses, one or more offenders, and, consenting and pub- lic-order crimes excepted, one or more victims. One expects to find a number of relationships among a population of crime events and their offenses, victims, and offenders. For example, the ratio of victims to events clepencis on the rate of multiple victimization in events and the rate of victimization ofthe same person in different events. And the population of offenders relative to the population of events depends on the size of the of- fending group in an event and individual rates of offending by group members. It is ordinarily assumed, moreover, that the prevalence or participation rate of offend- ers is well below the aggregate incidence of their offenses and charges and that the |23 differences are clue primarily to variation in both individual rates of offending and the size of offending groups in crime events. The above relationships between of- fen~lers and events can be illustrated quite simply using data on residential burglary and robbery. The Peoria (Illi- nois) Crime Reduction Council (1979), for example, undertook a study of all ju- veniles taken into custody for residential burglary from 1971 to 1978. There were 467 juveniles who accounted for 306 sep- arate burglaries during this period. The group composition of both the burglary incidents and the offenders involved in  them is shown in Table 1. In regard to burglary offenses commit- tec3 by apprehended juveniles cluring this period, single offenders accounted for one-half of all residential burglaries. In regard to offenders involves! in these bur- glaries, however, the one-half of the resi- clential burglaries with two or more offenders involved two-thirds of all of- fenders, and the great majority of multi- ple-offender burglaries involved two of- fenders. How the size of offending groups af- fects the size of an offender population is seen even more dramatically for robbery offenses. Just over one-half of all robbery TABLE 1 Group Composition of Burglary Incidents and Offenders Group Composition Number Percent Burglary Incidents With one offender only With two or more offend- ers Total burglary incidents Burglary Offenders Single offender in incident 155 151 306 155 Two or more offenders in incident 312 Total burglary offenders 467 50.6 49.4 100.0 33.2 66.8 100.0 SOURCE: Peoria Crime Reduction Council (1979).

124 CRIMINAL CAREERS AND CAREER CRIMINALS TABLE 2 Group Composition of Robbery Incidents and Offenders Percent Distribution by Size of Offending Group Total Four or Number One Two Three More Robbery Incidents 1,149,000 51.5 24.2 14.3 10.0 Robbery Offenders 2,215,272 26.7 25.1 22.3 25.9 SOURCE: Bureau of Justice Statistics (1984:Tables 52 and 62). victimizations in the United States in 1982 involved a single offender (Bureau of Justice Statistics, 1984:Table 621. Nonetheless, as seen from the clistribu- tion in Table 2, only one-fourth of all robbery offenders in those events of- fended alone, and there were about equal proportions of groups of two, three, and four or more offenders. Note that for about one-half of all rob- beries, the same robbery is part of the criminal history of more than one of- fender. Although there are on the average two offenders per robbery, in about one- fourth of all robberies, the same incident enters the criminal history of three or more offenders. If one knows the number of co-offenders in each offense in a crim- inal history, one can weight offenses ac- corclingly and estimate more precisely the "crimes saved" by incapacitating that offender. The larger the size of any par- ticipant's offending group, the less, on the average, that individuaT's absence should diminish the population of events. Lone and Group Offenders in Criminal Careers There is a firmly held view that most offenders are group offenders and that the career Tone offender is uncommon. These views are not based on the analysis of criminal careers or histories, however, but rather on the group composition of offending, especially of juvenile offencI ing. The statistical basis for the conclu- sion ordinarily is the individual and group composition of crime events rather than histories of offending for the offend- ers involved in those events. Quite commonly the group-size distri- bution for a population of events is used to estimate the participation rate of Tone offenders in a population. What is re- ported is a distribution of events by num- ber of offenders, and the proportion of single-offender events is taken as an esti- mate of the prevalence of lone offenders in a population. Such estimates are mis- leacling because events rather than per- sons are counted. Corrections can be made by weighting the events by the  number of offenders reported for them and using an appropriate population as the base for the rate. Percentage distributions of the size of offending groups by crime type (the up- per halves of Tables 1 and 2) reflect the aggregate risk of being victimized by a group of a given size, including a single individual. By way of contrast, percent- age distributions of the number of of- fenders involved in those incidents by the size of the offending group (the Tower halves of Tables 1 and 2) are offender- based statistics and state the probability that a randomly selected offender will commit that crime alone or with a given number of associates. The latter statistics are more appropriate in relating events to criminal careers or to offenders to be

CO-OFFENDER INFLUENCES ON CRIMINAL CAREERS processed. When the interest lies in the rate of Tone or co-offencting crimes, state- ments should be made about a population of crime events, but when the interest lies in Tone- or group-offender participation rates, statements must be macle about a population of offenders. When all the events in a criminal his- tory are considered, a large proportion of offenders exhibit neither exclusively Tone nor exclusively group offending. Rather, most offender histories are characterized by a mix of offending alone and with accomplices. Although information on the number of accomplices in each event of an offender's criminal history is gener- ally lacking, some idea of the mix of Tone and accomplice offending can be gained by examining the criminal history of the juvenile offenders in the stucly by the Peoria Crime Reduction Council (1979~. The 467 Peoria juveniles apprehended for at least one burglary during a 7.5-year period were involved in 2,820 offenses for which 3,426 charges were filed. Con- sidering only co-offending among the 467 juveniles, 79 (16.9 percent) always com- mitted their offenses without accomplices from the study population, and 91 (19.5 percent) only actec] with accomplices from the study population. The large ma- jority (63.6 percent) sometimes actecl alone and sometimes with others from the study population.) Unfortunately, good estimates are not available of the variation among criminal histories in the mix of Tone and group offenses, of variation in the number of accomplices, and of the consistency and variation of co-offending in an individu- al's history. With that information, one might weight individual-offencler rates by their partial contributions to crime iSince the co-o£ending data in the criminal his- tories were tabulated only for offenders included in the study population, these are probably underesti- mates for lone offending. 12S events, especially when estimating the expected crimes saved by incapacitation. One might wish to use such a weighted indiviclual rate as a criterion for selective incapacitation. Effects of Criminal Justice Processing on Estimates of Group Offending Self-report studies of mate delinquent behavior disclose a higher rate of Tone offending than clo official records of ap- prehension for the same offenders (Erik- son, 1971; Hindelang, 1971, 1976b). The question arises, therefore, as to whether  an apprehension hazard is associated with violating the law win others (Erikson, 1971:1211. For if violating in groups increases the likelihood of being apprehended, the prevalence of Tone of- fencling will be underrepresented in of- ficial records. Several attempts have been made to test whether there is a group-apprehen- sion hazard by comparing self-reported and officially reported offenses in an of- fender's history. Erikson (1971:125) con- cludecl that, although there is a greater risk of apprehension for offenses that are officially known rather than self-reported, the selection bias is consiclerably less for the most serious offenses. Subsequent re- search by Hinclelang (1976b:121) casts doubt on the group-hazarcl hypothesis. While exclusively group offenders have a higher risk of apprehension per crime than do those who always offend alone, those with a mix of Tone and accomplice offending have comparable risks in both types of events. What does seem apparent from Hinde- lang s work is that those engaging in illegal behaviour in groups are likely to engage in this behaviour more frequently than those engaging in the illegal behaviour alone" (1976b: 1221; incleed, the largest proportion of solitary offenders was found in the group with the lowest

126 individual rates of offending and the smallest proportion of solitary offenders in the group with highest indiviclual rates. This supports Eriks on' s (1971) con- clusion that among offending youths, sol- itary offenders are more likely to engage less frequently in crimes and to commit less serious crimes than group offenders. Unfortunately, data are unavailable to determine whether adult solitary offend- ers are also disproportionally involved in the less serious offenses or have lower individual rates of offending. Since major crimes against persons involve higher rates of solitary offending, it is possible that adult solitary offending, in contrast to that of juveniles, is disproportionally con- centrated on the more serious offenses. BASIC PARAMETERS OF GROUP OFFENDING Despite the fact that a number of major longitudinal studies of criminal careers have followed samples of youths into their adult years (Glueck and Glueck, 1930, 1937, 1940, 1943, 1946; Shannon, 1978; Wolfgang, 1978; Elliott et al., 1983), none has examined patterns of co-offend- ing into adulthood. Thus, this discussion of the parameters of group offending must depend primarily on research on juvenile delinquency. Even that research, how- ever, pays relatively little attention to individual rates of offending or the role of co-offending in careers so that often a single study or source of data must be relied on. Offending and Group Size Breckinridge and Abbot (1917) were among the first to point out that most delinquent offenses are committed with at least one other person and that even most youths regardecl as Tone offenders occasionally engage in clelinquency with a companion. Somewhat later, Shaw and CRIMINAL CAREERS AND CAREER CRIMINALS Meyer (1929) and Shaw and McKay (1931) estimated in juvenile court sam- ples that less than 20 percent of the juve- nile offenders before the court committed offenses alone, that the modal size of an offending group is small (two or three participants), and that most delinquency is not committed by well-organizec3 groups. Shaw and McKay concludecI, moreover, that while such offending two- somes and threesomes commonly are combinations from a larger group, a whole group is rarely involved in the same delinquent or criminal act. Large networks that link by association up to 200 youths ordinarily consist of  fewer than 30 or 40 active members orga- nized into smaller cliques of ~10 mem- bers (Klein and Crawford, 1967~. In a cohort study of delinquency in a Swedish community, a youth gang was defined as a group of juveniles who were linker] to- gether because the police suspected them of committing crimes together (Samecki, 1982: 144-145~.2 There were well over 100 of these gangs in the study. Membership varied from 2 to 30 boys; the mean size was 5 boys (Samecki: 151~. What was most evident, however, is that groups were constituted by co-offending relationships. These relationships con- sisted of links of co-offen(ling to form chains of association. One such chain in this Swedish city involved 260 boys and a few girls and constitutes! about 45 percent of the study population. Together they accounted for 86 percent of the crimes 2The study population was located in a southern Swedish industrial community of about 50,000 in- habitants. The records of the local police on all crimes whose suspects were under 15 years old and from the police register (PBR) for all juveniles 15 years and older were the main source of information on offenses. Additional data sources included re- ports from police hearings, from police interviews with juveniles suspected of crimes, and from social service authorities (Sarnecki, 1982:54~5~.

CO-OFFENDER INFLUENCES ON CRIMINAL CAREERS reported cluring the study period (Sar- necki: 14~1451. Although accomplices in delinquency are drawn more often from smaller cliques than from their larger network, the number of accomplices in any delin- quent act is much smaller than the num- ber in the clique (Short and Stro~tbeck, 1965; Klein and Crawford, 19671. Consic3- ering only those offenses committed by two or more offenders, the mocial size of an offending group from age 12 onward is two participants. Groups with four or more participants are relatively uncom- mon after age 14 or 15. The majority of offenders have accomplices until their early 20s, after which the majority commit their offenses alone (Hood and Sparks, 1970:87-881. There is some variation in Tone and group offending rates by country and considerable variation by type of of- fense (Sveri, 19651. The distribution of group size within an offending history has consequences for intervening in that career. A majority of most common crimes are committed by two or three offenders, and most offend- ers will have a substantial and continuing history of offenses in which their associ- ates change from crime to crime. The larger the offending group, the more likely it is to be confined to a single event unless the group specializes in specific targets of offending, as in terrorism or in some kinds of white-colIar offending in which continuing use of organizational power is integral to committing offenses. Age of Onset There is much controversy concerning whether the onset of delinquency is a consequence of induction by co-offencl- ers. Glueck and Glueck (1934a, b, 1937, 1943, 1950) contended that pre-clelin- quent and delinquent behavior began at an early age in family and school social- ization. They claimed that those with de 127 linquent tendencies associate with one another rather than being led into their joint clelinquent behavior by their associ- ations. Their definition of delinquency is quite broacI and includes behavior la- beled as "antisocial" and also "clelin- quent tendencies" (Glueck and Glueck, 1950:421. Moreover, their conclusion is not based on collecting information about the group composition of specific behav- ioral acts committed by offenders in their sample. Rather, their contention is based on their observation (Glueck and Glueck, 1950:41) that the onset of delinquency for their institutionalized delinquents oc-  curred before age 10 and consequently, in their view, before the time that groups play a significant role in a boy's life. Challenging this conclusion, Eynon ant! Reckless (1961) founct that the me- dian age of first contact with juvenile authorities was 13 for incarcerated juve- nile offenders and that there were no significant differences in age of onset or the presence or absence of companions at the first officially recorded (lelinquent act (Eynon and Reckless:169~. The meclian age of onset for the first self-reported delinquent act ranged from 11 to 14 years clepencling on the type of offense. Self- reports on whether one hac3 companions for each of seven offense types ranged from 56 percent of those who ran away from home to a not surprising 100 percent of those engaging in gang fights (Eynon and Reckless: 170~. They concluclec! that the "presence of companions is a major component of male delinquency, regarcI- less of the age of delinquency onset" (Eynon and Reckless:1681; companion- ship is present at early as well as at late onset. But, as they note, what we still lack is information that tells us whether com- panionship experience relevant to the on- set of delinquency causes delinquent be- havior (Eynon and Reckless:168~. Most investigators seem to have missec! this obvious point that companionship

128 begins among children at a fairly early age. Although Thrasher (1927), in his classic stucly of what he called "youth gangs" in Chicago, was well aware of the fact that the kinds of groupings in which he was interested were found at quite young ages, he focused on territorial groupings. Subsequent work focused on "organizer! gangs" rather than on "com- panionship." These groups were largely territorially organized anct regarded as a phenomenon of aclolescent and young aclult ages. Recruitment and induction into these peer groups were macle prob- lematic but without any explicit attention to prior delinquent histories. This was partly because Thrasher devoted a great deal of attention to official records of de- linquency, ant] a clelinquent act was rarely enterer] as an official record prior to age 12. In any case, what is at issue is whether the onset of delinquency involves primar- iTy Tone offending or whether it is linked primarily to companionship or accom- plice relationships. With answers to this key question, one may begin to unravel the problem of the role of groups in launching a continuing offending career, since there must be consiclerable clesis- tance from offending, even at very early ages. It is difficult to assess the role that companionship plays in the onset of cle- linquency because of the weak cross- section designs of most etiological studies and the failure to specify a testable causal model. A longitudinal design that follows members of a birth cohort as well as all their accomplices who are not members of the cohort is necessary to test causal hypotheses about onset. Unfortunately, up to now cohort studies have not exam- inec3 the role of co-offending in criminal careers. Olson's (1977) multiple regres- sion analysis of the personal interview data for a subset ofthe Racine birth cohort study (Shannon, 1978) found that first police contact at a very young age was CRIMINAL CAREERS AND CAREER CRIMINALS associated only with being mate and hav- ing friends in trouble with the police. Yet those two variables accounted for at most one-fourth of the variance, inclicating that the personal interview variables did not include the most important determinants of age at first police contact (Petersilia, 1980:349~. Group Affiliation and Individual Rates of Offending Individuals vary considerably in their rates of offending. Most young offenders  also have co-offenclers in their offending and associate in other group activities with still other offenders. An interesting question is how an indiviclual's rate of offending is related to the offending rate of the affiliatec3 group. Morash (1983:319) concludes that the clelinquencY rate of one's peers is a strong predictor of an individuaT's rate of delinquency. Boys who belonged to peer groups with a be- low-average rate of clelinquency had be- Tow-average rates of clelinquency and boys with peers who had high indiviclual rates of clelinquency hacT an above-ave , ~ , rage indiviclual rate of delinquency (Morash:3211. Juveniles with high rates of offending typically commit those offenses with a large number of accomplices. Sarnecki (1982) found that the 35 most delinquent juveniles in the Swedish community he studied were linkecl to one another by membership in the largest gang and two smaller gangs. These were among the most criminally active gangs in the com- munity. The 35 juveniles were involves! with 224 accomplices in crime. Almost 3 in 10 of the 799 other delinquents in the community committed at least one crime win 1 of these 35 high-rate offenders (Sarnecki: 171,2091. The accomplices of these 35 were usually selected from the criminally more active part of the total offending population (Sarnecki:144-1451. High-rate juvenile offenders affiliate

CO-OFFENDER INFLUENCES ON CRIMINAL CAREERS with one another.in peer groupings. The 35 most active members were but 6 per- cent of the delinquents in Sarnecki's study population. They belonged to the three largest membership gangs, in which the mean number of suspected crimes per gang member was 22.3 during a 3-year period a rate that was three times that ofthe delinquent population as a whole. The members of the largest gang made up only 13 percent of the popula- tion of offenders, yet they accounted for 42 percent of all suspected crime (Sarnecki: 1741. Duration of Accomplice Relationships Most pairings in committing delin- quencies are of short duration. Sarnecki (p. 140) found that only 13 percent of the 1,162 juvenile clelinquency pairings in an offense in the Swedish community per- sisted beyond 6 months. Only 4 percent of the pairs were still committing crimes together after 1.5 years, but 1 of the 1,162 offending pairs was committing offenses together after 3.5 years, roughly half the 6 years the stucly tracked offenders. The short life of any pairing in delinquency partly reflects the fact that most delin- quent careers are short. Almost 6 in 10 youths in the Swedish cohort were known to the police cluring only one 6- month period ofthe 6-year study (Sarnec- ki:l411. Among those persisting in delin- quency, the modal pattern was to change associates in committing offenses. Sarnecki (pp. 142-143) also found that the most criminally active usually commit their crimes in pairs and that they pre- serve particular pairings longer than (lo those who are less active. The most active and seriously delinquent juvenile sus- pects were 45 times more likely to com- mit crimes with the same associates than were less active juveniles from the study population. Quite clearly, accomplices in offending change quite frequently in juvenile ca 729 reers. The larger the number of offenses committee! by an offender, the larger the number of different accomplices linked to that offender's career. One's accomplices as a juvenile are likely to be drawn from cliques or constellations of cliques with which one is affiliatecl. Or(linariTy, these cliques are part of a network. Adults are perhaps more likely than juveniles to be Inked in loose networks, ones in which they are linked by weak rather than strong ties (Granovetter, 19731. Stability of Group Affiliation  An important issue is how shifts among cliques of peers and co-offenclers occur en c] how affiliation with cliques and larger networks is stabilized, on the one hand, and disconnected and terminated, on the other. These issues are not clealt with systematically in the research on crime and clelinquency.3 Thus, empirical studies on clelinquent gangs and street- corner groups are used here to explore these issues. It was quite evident in the Swedish community stucliecl by Sarnecki (1982) that most gangs, as well as pairings, ex- isted for only short periods of time. Only one gang persisted for the entire 6-year period and towards the end of that period it split into two separate groups. Yet a third was spun offin the final 6 months of Similarly, little is known about the stability of group affiliations among a population of nonoffend- ers. The stability of pair and group affiliations seems to vary by age and to be more stable during the pre- adolescent than the adolescent years. Still, it seems that all youths frequently change companions for conforming acts, such as walking to school, dating, going to the movies and shopping. Much more needs to be known about pair and group affiliations for nondelinquent or all youths to assess the relative stability of delinquent affiliations. It is possible that delinquent pairings show greater stability than non- delinquent pairings. Choosing different compan- ions for different activities, moreover, may simply be a characteristic of both youthful and adult behav ~or.

130 the stucly. The original gang was still the largest of the three at the close of the study period, however, despite its losing members to form two others (Sarnecki: 153~. Although gangs can (livi(le by schism, merge with other gangs, or join a network of gangs, their durability has consequences for individual careers. Ca- reer termination may result simply from the short duration of groups and the loose links that bins! members, especially for those who rely primarily on group affili- ation for accomplices and support in of- fending. Suttles's (1968) study of street-corner groups in Chicago allows us to explore the stability of affiliations with o~encling groups. He identifiecI 32 named street- corner groups in the Aciciams area of Chi- cago (Suttles:1571. With one exception, they were made up exclusively of mates and averaged from 12 to 15 members in size. Some had as few as 8 members and one had 29. Each of the groups had some members who lived outside the Addams area ranging from 4 to 17 percent; 12 percent of the members of all the groups lived outside the area (SuttIes:157-1671. Outsiders were youths who either for- merly lived in the area or were relatecI by kinship to one of the group members (Suttles: 1651. Although Suttles (1968) floes not pro- vicle detailed information on the duration of the 32 groups, he reports that one laster! but 1 year and that some lasted 2 or 3 years (pp. 161, 1661. All were subject to considerable turnover in membership, partly because of resi(lential mobility.4 Of those known to have quit membership in 4Klein and Crawford (1967) similarly reported high turnover in the black Los Angeles gangs they studied. They reported that "many members af- filiate with the group for brief periods of from a few days to a few months, while others move out of the neighborhood or are incarcerated for periods some- times exceeding a year" (p. 661. CRIMINAL CAREERS AND CAREER CRIMINALS groups, 41 percent moved to another area of Chicago (SuttIes: 1671. One additional fact is worth noting. A substantial proportion of all boys in an area never affiliate with these larger groups (Short and Stro(ltbeck, 1965:56- 57; Suttles: 1731. Suttles reports that most of the unaffiliated are boys who regularly "bane! together" in small cliques (p. 1691. Few boys, then, are isolates. The role of affiliation with territorial delinquent groups in accounting for differences in individual rates of offending is unclear, however. SuttIes (p. 220) reports that ar-  rest rates were about equal for the affil- iated and unaffiliated boys. Yet others, especially Short and Strodtbeck, report much higher individual rates of offending for gang-affiliated boys. The stability of territorially based groups is threatened by three major con- tingencies: transiency, incarceration of members, and shifts to conventional ca- reers. Slum neighborhoods especially are characterized by high residential tran- siency of families. That transiency has three major consequences for territorial youth organizations. For one, it makes the membership of any group volatile. From the perspective of the group, to survive substantial annual turnover in member- ship, it must obtain new members. From the perspective of the individual mem- ber, it means transitory affiliations with some group members and adapting to the exodus of former members and an influx of new members. Recruitment and re- placement are age graded, and there is some preference for older rather than younger boys (Suttles:163~. Even associ- ations between gang boys appear age graded. Klein and Crawford (1967:74) re- port that younger gang members are sel- dom seen in the company of older gang members. Another consequence of transiency is that it spreads the network and influence of the group beyond the confines of its

CO-OFFENDER INFLUENCES ON CRIMINAL CAREERS territory, always linking some former res- idents to it. Chicago appears similar to Sarnecki's (1982) Swedish community in that some youths who move into other areas affiliate with other territorial groups (Keiser, 1969; Short and Moland, 1976:1681. Transiency thus both expands the choice of accomplices and links terri- torially based groups. lust how widely such individual contacts spread across territorially based groups is unclear but some of the most active and serious of- fenclers are transient and use this larger network to search for accomplices. Transiency also has important conse- quences for the temporal continuity of territorially based groups. To survive, the group must continually invest in replac- ing members. The more formally orga- nized a group and the more provision it makes for replacement of members, the more likely it is to survive, as studies of Chicago's black conflict gangs demon- strate. The Vice Lords, which incorpo- rated as a not-for-profit organization, gained as many as 8,000 members in 26 divisions (Sherman, 1970), while the in- formally organized Nobles barely sur- vivec] (Short and Moland: 1681. The higher the residential transiency of an area, the more likely its youths are to be organized into a loose confederation and the fewer and less stringent are the crite- ria for entry and continuing affiliation with its territorial groups. Moreover, the combination of transiency and the aging of members, with replacement confiner] to a narrow range of age-mates, suggests that recruitment is limited to newly en- tering residents or, in a few cases, by mergers among gangs (Sherman, 1970; Short ant! Moland: 168~. The combination of these contingencies suggests that un- less youth groups are formally structured to deal with turnover, they should have fairly high death rates. The second important source of insta- bility in gang membership is the rate of ]3] incarceration of gang members. The more seriously delinquent the members of a gang, the more likely they are to be incar- cerated for substantial periods of time. Short and Molanct (p. 168) report that nearly all the Vice Lords had been in correctional institutions at one time or another, and Short and StroUtbeck (1965) draw attention to the disruptions causer] by incarcerating gang leaclers. Unfortu- nately, they do not provide estimates of the rate of incarceration for any time in- terval. A third important source of turnover in  street-corner group membership is the shift to more conventional career affilia- tions by some members. At least one-half of Suttles's (p. 167) Addams-area group dropouts left because Key married, went into the service, joined a job training program, or worked regularly in a job. Relatively few were lost to jail. The general impression is that delin- quents are organized into loose federa- tions rather than highly organized groups. The federation is characterized by loose ties among inclivicluals and cliques or clusters. Members are linked by a variety of activities in addition to delinquent of- fencling. Inclivicluals are not tightly bound either to large groups or to partic- ular pairings within groups. Accomplices change quite frequently. According to Olofsson (1971), Swedish male juvenile delinquents are less selective of compan- ions ancl their choices are less stable than are those of youths in the general popu- lation. It perhaps is reasonable to con- clucle that, while there are some highly structured territorial gangs that persist for periods of time (Miller, 1958, 1975; Clinard and Ohlin, 1960; Spergel, 1964; Klein and Crawford, 1967), associates in most delinquency offenses are drawn from much less structured networks in which nuclei of offenders are linked as nodes in that network. Territorial gangs (Thrasher, 1927; Whyte, 1943) are per

132 haps more like nodes in networks than like independent groups in organized conflict relations with others (Bordua, 1961, 1962; Yablonsky, 19621. Over time, networks are durable while particular groups and pairings are transitory and any individuaT's affiliation is of short duration. One expects to find a large number of accomplice pairings in any individual's offending history of some duration, but ordinarily the same associates are in- volvect for only a short period of time. Moreover, the higher an indivicluaT's rate of offending and the more serious the crimes committed, the more likely that person selects accomplices from a net- work. Accepting the importance, even domi- nance, of networks in delinquent behav- ior should draw attention to the role of networks in offending ant! to the role that pairings, as contrasted with individual offending, play in the commission of crime. PATTERNS IN GROUP OFFENDING There is considerable variation in the extent to which offenders commit crimes alone and with others when examined in terms of the characteristics of the offense, the offender, and the victim. The bulk of research on the role of groups in offend- ing unfortunately is done only with young offender populations. There are relatively few studies of the group behav- ior of youths in transition to adult status or of adult offenders at different ages. This makes it necessary to rely on single stud- ies or case studies to infer patterns of group offending for adult offenders. Sex Somewhat over 1 in 10 Tone offenders in crimes of personal violence in the United States are female offenders (Bu- reau of Justice Statistics, 1984:Table 401. CRIMINAL CAREERS AND CAREER CRIMINALS The proportion of female Tone offenders varies by type of violent crime, being negligible for the crime of rape and great- est for the crime of assault. Female of- fenclers in violent crimes select females as their victims far more often than males select mates. When females select mate victims in violent crimes, they are most likely to assault men and least likely to rob them (Hindelang, 1976a: 1781. Participation in most voluntary activi- ties varies by age. Youths are more likely than aclults to limit their choices to per- sons of the same sex. Opportunities for informal contacts with the opposite sex are more limiter! among youths. Not sur-  prisingly, the associates of young persons in delinquency are almost always of the same sex. Shaplanct (1978:262) reports that no boy in her sample of 11- and 12-year-olds, when interviewer] at age 13 or 14, admitted committing any offense with girls. The pattern is somewhat dif- ferent when older offenders are included. Among violent, multiple-offender victim- izations reported to the National Crime Survey (NCS) in 1982, 19 percent in- volvecT women as offenders, either with other women only (7 percent) or with men or men ant! women (12 percent) (B JS :49~. Only limited clata are available to esti- mate the incidence of Tone offending for both personal and property crimes by sex. The Federal Republic of Germany (1982) reports the size of arrested groups for a large number of offenses; the data show that the aggregate mate rate of solo of- fending is somewhat below that of fe- males but that the rate varies by offense. The West German police statistics for 1982 disclose that 68 percent of all mates suspected of offenses, compared with 76 percent of all females, were solo offend- ers. That difference is not large, but women are disproportionally found in of- fenses that have high proportions of lone offenclers, such as assault, shoplifting,

CO-OFFENDER INFLUENCES ON CRIMINAL CAREERS prostitution, and petty theft without con- tact. Moreover, given the low incidence of offending among women and the gen- eral absence of organized women's delin- quent groups or gangs, it seems reason- able to conclude that women more commonly than men engage in offenses that make them more likely to offend alone. Some confirmation of this is also found in the fact that multiple-offender incidents in which all offenders are fe- maTe are fewer in absolute numbers and proportionally (7 percent) than are those involving women as Tone offenders (13 percent) (compare Tables 40 and 45, BIS: 1984). Women are less likely to be associated exclusively with other women than they are with men in committing violent crimes. Only 36 percent of the violent criminal victimizations involving one or more female offenders reported to the NCS in 1982 were made up entirely of female offenders; 64 percent involved association with men in the offense (BIS :Table 451. Correlatively, only 12 percent of the violent, multiple-offender incidents involving mate offenders also involved female associates. Women are least likely to offend only with women in the more serious violent crimes of robbery and aggravated assault. Among crimes of violence towards per- sons' substantial involvement of women offenders with women victims is largely confined to simple assault. Race Blacks are somewhat less likely than whites to be solo offenders. Although data are lacking for a population of adult of- fenders, the NCS data for 1982 disclose that 72 percent of all violent criminal victimizations by whites, compared with 60 percent of those by blacks, were sin- gle-offender incidents (BIS:caTculated from Tables 44 and 491. 733 Mixed-race offending is infrequent in multiple-offender victimizations. Less than 6 percent of all 1982 crimes of vio- lence reporter! to the NCS involved a mix of black, white, or other races of offenders (BIS:Table 491. There was little variation by type of violent crime. Comparing mixed-sex with mixecI-race offencling, multiple-offender groups ap- pear about twice as likely to include per- sons of the opposite sex (12 percent) as another race (6 percent) (BIS:Tables 44 and 491. Thus, accomplices from a cTif- ferent race and sex are uncommon and most accomplices are of the same race  and sex. Age Group offenders are, in the aggregate, younger than solo offenders (B IS :Ta- bles 41 and 461. In the NCS, offenses in which the offencler's age was perceived by victims to be un(ler 21 years were found more often among multiple- than single-offender victimizations (Hinde- lang, 1976a: 172; Bureau of Justice Statis- tics:Tables 41 and 461. A study of apprehender] burglars in the Thames Valley, Englan(l, fount! that some- what over three-fourths of the aclult bur- glars, compared with one-half of the juve- nile burglars, acted alone in the offense for which they were arrested. Consider- ing only those offenses in which there were accomplices, adult burglars were more likely than juvenile burglars to act in pairs (Macguire and Bennett, 1982: 1841. Information is lacking on the size of offending groups by age for a population of U.S. offenders. Hoot] and Sparks (1970:87~9), however, report data on size of offending group by age for appre- hencled offenders in London boroughs and for offenders convicted of theft in Norway. They concluded that, as offend- ers grow oilier, they are more likely to be

134 apprehended or convicted for an offense committed alone. Not until the mic3-20s, however, are a majority of those appre- hencled or convicted, Tone offenders. The age curve of solo offending is relatively flat until age 16, when the proportion of Tone offenders begins to rise rather sharply. This shift is primarily accounted for by a rapid decline in apprehensions involving three or four or more partici- pants, especially the latter. The propor- tion of apprehensions or convictions in- volving two participants fluctuates far less over time than does that for Tone offend- ers or for groups with three or more of- fenclers. The proportion with two or more offenders perhaps peaks at ages 16-18, but there is no substantial decline with age. Lone offending exceeds pair offend- ing by the late teens. OIcler offenders thus ordinarily commit offenses alone or with a single co-offender. Still, at least 1 in 10 offenders in their mid-20s is appre- henclec3 or convicted for offending with three or more offenders. Type of Offense Certain offenses are iclentified as typi- cally inclividual or group offenses. With the exception of robberies and some as- saults, the modal size of offending groups reporter] by victims of major crimes against persons is a single offender; pair offenders are the next most common (Reiss, 1980:Table 11. But, with the ex- ception of homicides and rapes without theft, the majority of offenders in crimes against persons commit their offenses with accomplices (Reiss, 1980:Table 21. The mean number of offenders per major common crime is between two and three for robbery and assaults and about one and one-half for all other offenses against the person and property (Reiss, 1980:Ta- ble 21. Self-report studies show considerable variation by types of crime in the propor CRIMINAL CAREERS AND CAREER CRIMINALS tion of young persons committing of- fenses alone or with others. Much cle- pends on how the self-report specifies the offense, especially as to the conditions of its occurrence. Shapland (1978:Table 2) found, not surprisingly, that all the 13- and 14-year-oIcis she studier] took money from home as a single offender. At the other extreme, more than 9 of 10 boys said they vanciaTized public property with oth ers. Most boys, nonetheless, report acting alone and with others for a variety of offenses. Shapland (1978:262) calculated that the mean percentage involvement in  group offending was 59.68 percent (stan- ciard (leviation of 15.6) for all boys aged 13-14, whereas that for solitary offending was 30.02 percent (stanclar(1 deviation of 16.4~. From this we can conclude that although the typical offender history in- clucles both Tone and group offending in the same and different kinds of offenses, the ratio of group to solitary offenses in a youthful offencler's career is, on the aver- age, 2:1. Detailed information on the clistribu- tion of solo and group offending for spe- cific offenses is unavailable for the adult offender population of the Unitecl States. Although the rate of solo offending ap- pears to be higher for a population of offenders in the Federal Republic of Ger- many (Kaiser, 1982:103) and the German Democratic Republic (KraeupI, 1969:63) than in the rest of Europe or in North America, police statistics for adult offend- ers in West Germany disclose rather marked variation in solo offending by type of offense. Among the major offenses in which at least 8 of every 10 offenders committed the offense alone were mur- cler; sexual offenses, such as exhibition- ism and sexual murder; drug abuse in- volving heroin and cocaine; and the white-colIar offenses of embezzlement, forgery, and fraud. By contrast, most of the common crimes had much Tower solo

CO-OFFENDER INFLUENCES ON CRIMINAL CAREERS offending fractions. At least 8 in every 10 suspects in common thefts, robberies, breakings and enterings, and breaches of the peace had at least one co-offencler (Federal Republic of Germany, 19821. The significance of group affiliation for young offenders is reinforced when the variation in the group composition of of- fending is examined by age ant] type of offense. Even for homicide, which in the aggregate is a solo-offender crime and an infrequent offense among young offend- ers, Zimring (1984:91) found that the younger the homicide offender, the more likely he was to have killed as part of a group and to have done so when engag- ing in a collateral felony, such as robbery- murder. In brief, young homicide offencl- ers are far less likely than oIcler ones to commit murder ant! felonious homicide alone. Relationship Between Victims and Offenders It is commonly observed that a prior relationship between victims and offend- ers is more characteristic of some crimes than others. Domestic assaults, for in- stance, are characterized by cohabitation of victim and offender, whereas assaults involving theft ordinarily occur among strangers. There is a moclest relationship between the size of an offending group and the relationship of offenders to their victims (Bureau of Justice Statistics, 1984:Tables 52 ant! 62~. The modal of- fencler in crimes of violence against strangers is a co-offender, whereas single offenders predominate when there is a prior relationship between victim ant! of- fenclerks). Even the modal offender in robberies of nonstrangers is a Tone of- fencler, whereas it is a co-offencler in rob- beries of strangers. The larger the of- fending group in robberies and assaults of strangers, the more likely victims are to be injured; however, a larger proportion ]35 of lone offenders inflict serious when the victim is a nonstranger than a stranger. Territorial Concentration Perhaps the single most noteworthy aspect about common crime is the territo- rial concentration of offenses, offenders, ancl, to a substantial degree, victims. A substantial majority of both personal and household crimes occur close to the resi- clences of the offenders and their victims (Reiss, 1967; M. W. Smith, 1972; Pyle et al., 19741. Juvenile offenders commonly belong to  territorially based groups and typically select their co-offen(lers from those groups or the territory where they reside (Straw and McKay, 1931; SuttIes, 1968; W. G. West, 1974, 1977, 1978; Sarnecki, 1982~. Shaw's (1938) tracing ofthe accom- plices in the criminal careers of five Mar- tin brothers illustrates this territorial con- centration of career offenders. In the course of their careers these five brothers were implicated in theft with at least 103 other delinquents and criminals. The other offenders resided, for the most part, within seven-tenths of a mile of the Mar- tin residence. Of the 103 co-offenders, 28 were adjudicated in delinquent and crim- inal proceedings, and all but 3 of the 28 served adult as well as juvenile institu- tional sentences (Shaw:11~1161. The geographic concentration of the Martin brothers' accomplice network was charac- teristic of both their juvenile and young adult years, although some geographic diversification occurred through accom- plices met during periods of incar- ceration. Recent work on the neighborhood de- terminants of criminal victimization sheds additional light on the territorial concentration of offenders and on pat- terns of group offending. D. Smith (1986) reports that the larger the proportion of

136 single-parent households with children between the ages of 12 ant] 20 in a neigh- borhoocI, the higher the neighborhood's perceived risk and actual rate of victim- ization by crime. Sampson (1985:25) finds that the greater the proportion of femaTe- heaclec3 households and the higher the density of settlement in a neighborhood, the greater the rate of victimization by crime. Sampson (1983: 172) similarly finds that juvenile offenders in neighbor- hoods with both high-clensity settlement and a large proportion of femaTe-headec] households commit a larger proportion of their offenses with others than JO juve- niles in areas with Tow density and a small proportion offemaTe-heacled house- holds. Sampson's finclings hoist inclepen- dent of the racial composition of the neighborhoods. Additionally, Bottoms and WyTe (1986) report that clelinquency rates in public housing projects in English industrial communities are correlated with the de- gree of concentration of single-parent households with youths of an age to of- fencI. They conclude that high clelin- quency rates in public housing projects are partly a function of managers' concen- trating single-parent households in se- lectec3 public housing projects. These studies lenc] support to the hy- pothesis that it is the territorial concentra- tion of young males who lack firm con- trols of parental authority that leads them into a peer-control system that supports co-offencTing ant! that simplifies the search for accomplices. Siblings in Offending Brothers in Crime, the classic study by Clifford Shaw (1938), traces the criminal careers of five Martin brothers over a period of 15 years. At the close of that time the brothers ranged in age from 25 to 35 years. Four of the five had by then CRIMINAL CAREERS AND CAREER CRIMINALS terminated their criminal careers. Shaw (p. 4) describes the crime and criminal justice consequences of their careers in the following way: The extent of their participation in delinquent and criminal activities is clearly indicated by the fact that they have served a total of approx- imately fifty-five years in correctional and pe- nal institutions. They have been picked up and arrested by the police at least 86 times, brought into court seventy times, confined in institutions for forty-two separate periods and placed under supervision of probation and parole officers approximately forty-five times. These are but the official statistics oftheir five criminal careers. Autobiographical  reports accounted for more than 300 bur- glaries, the theft of 45 automobiles, ant! a host of other crimes involving theft, re- ceiving stolen property, and armed rob- bery (Shawl. The autobiographical accounts of these five brothers offer evidence of ogler brothers' recruiting their younger broth- ers into offencling, thereby focusing on the role that siblings play in co-offending and how common sibs are as initial and continuing co-offenclers. Whether one of- fends with siblings can be expected to depencl on such characteristics as family size and sibling composition by sex and birth order. There is no reliable research on how much the co-offencling of siblings accounts for group offending. We can gain some notion of the role of siblings in criminal events from the Peoria study of residential burglary (Peoria Crime Reduction Council, 19791. Of the 151 burglaries involving two or more offenders for which a juvenile was apprehended, roughly 24 percent in- volved two or more siblings (and some- times nonsibling offenders as welI). Of these, about two-thir(ls involved nonsib- lings and one-third involved only sib- lings. What is apparent from these Peoria data

CO-OFFENDER INFLUENCES ON CRIMINAL CAREERS is that more than 1 in 10 adjuclicatec3 burglary events involved at least two members from the same family. Thus, among a population of Tone and group offenders in residential burglary, a small proportion of families accounts for a dis- proportionate amount of the adjudication decisions. Little is known about the transmission of antisocial and criminal behavior within families anct across generations of kin. It has been recognized for some time that male delinquents come from larger fami- lies than JO male nondelinquents of the same age and socioeconomic status (Fer- ~uson, 1952; D. I. West ancI Farrington, 1973; Blakely, Stephenson, and Nichol, 19741. Jones, Offorcl, and Abrams (1980) founc! that, in their comparisons of mate probationers and controls, this difference in the size of sibling groups was entirely clue to an excess of brothers. Probationers and controls clid not differ in number of sisters (Iones, Offorc3, and Abrams: 1401. Of special interest was the finding that the antisocial scores ofthe brothers of the delinquent male probationers were sig- nificantly higher than those of the broth- ers of their matched controls (Iones, Of- forcI, and Abram s :141~. By contrast, among female probationers, both broth- ers and sisters were more antisocial than the siblings of their control counterparts (Iones, Offorc3, anal Abrams: 1441. Even more striking was the discovery that the average antisocial score of the probationers' brothers increased with the number of brothers in the family when the number of sisters was held constant; it decreased with the number of sisters, holding the number of brothers constant (Tones, Offord, and Abrams:1421. Tones and his colleagues interpret these results to mean that sisters suppress antisocial behavior in their brothers, whereas broth- ers respond! to one another in ways that stimulate their potential for antisocial be c7 , _, _ ~ 137 havior, i.e., it is due less to learning anti- social behavior from siblings than to their mutual participation. Evidence on this point appears lacking. ROLE OF GROUPS IN RECRUITMENT FOR CO-OFFENDING A substantial proportion of offending involves two or more offenders. there Is no evidence that very many offenclers, if any, keep the same accomplices over Tong periods of time. Incleed, quite the oppo- site appears to characterize the criminal careers of high-rate offenders. Most of-  fenders have a substantial number of dif- ferent accomplices. Most offenders appear to select dif- ferent accomplices, especially as adult offenders. Not too much is known about how and why accomplices are selected. In this section the structure of peer and aclult networks that facilitate the search for co-offenders is examined, followed by an examination of what is known about the search for and active recruitment of associates in offending. Structure of Delinquent Peer Networks There is some disagreement about how youth groups and their peer networks are structural. Most young mates do not ap- pear to belong to tightly bounclecT groups that have a constant membership and from which they select their accomplices, if indeed selection is the appropriate mocleT for describing how persons come to offend! together. This exposition links youths in a web of affiliation or network of contacts and exchanges. Typically, sociometric techniques are used to clefine a large group that has a much smaller core membership that gathers together with some frequency (Yablonsky, 1962; Short and Stro~tbeck, 1965; Gannon, 1966;

138 CRIMINAL CAREERS AND CAREER CRIMINALS TABLE 3 Percentage Comparison of the Types of Group Contacts for Boys in Four Gang Clusters with Different Rates of Delinquency Higher Delinquency Clusters 42 81 s4 82 Lower Delinquency Clusters 43 72 35 73 Type of Delinquency Contact Percent of all members in cliques Percent of all contacts that are mutual contacts among members of their cluster Percent of all contacts that occurred only once Percent of all clique member contacts made within their clique A B C D 16 - 15 20 73 47 32 77 40 SOURCE: Adapted from Klein and Crawford (1967:Table 2); Hood and Sparks (1970:Table 3:1). Klein and Crawford, 1967; Sarnecki, 1982~. The core of active members within the larger youth group ordinarily is no more than one-fifth the total aggregate of affiliated youths and ranges in size from 20 to 40 or so members (Hood and Sparks, 1970:~901. That core of firm or active members, in turn, often subdivides into clusters or cliques of ~10 members (Klein and Crawford, 19671. These aggre- gations are best described as a loose web of affiliations because most persons have  most of their contacts with others in their clique and little, if any, contact with oth- ers in the network. Nein and Crawford (1967) studied the extent to which 32 youths in one gang had contacts with one another during a 6- month period. The observed number of interactions of any member with all oth- ers during the 6 months ranged from 1 to 202. Of the 486 possible pair relation- ships among the 32 members, two-thirds involved no contact and for three-four~s contact was limited to no or only one contact. Most of the contacts occurred within two cliques-one of nine mem- bers and the other of five, with the nine- member clique having a much greater frequency of contact than the five- member one (Klein and Crawford:72~. - Klein and Crawford (pp. 7~75) di- vided the gang members into four clus- ters according to their irldividual rates of delinquency. Those labeled A and B in Table 3 had relatively higher rates than those labeled C and D. Those in A and B clusters also had continued gang relation- ships for longer than those in clusters C and D. Higher delinquency group mem- bers were more likely to be members of the two major cliques and to have more of their delinquency contacts within the clique, as Table 3 makes readily appar- ent. We may conclude from these delin- quency and group-affiliation data that de- linquent contacts do not result from a stochastic process. Indeed, a substantial amount of delinquency occurs within rel- atively small clusters that form within a larger network of affiliations. A minority of the contacts, nevertheless, are within the larger network although they occur much less frequently. Structure of Adult Networks Surprisingly little is known about how adults make contacts and decide to offend together. Apart from the highly organized criminal activity that is conducted by syn

CO-OFFENDER INFLUENCES ON CRIMINAL CAREERS dicates, it seems clear that most adult co-offending does not arise from partici- pation in groups of which they are mem- bers. Although some adults may form co- offencling relationships that are stable over time, the typical co-offending rela- tionship appears to be transitory ant] there is a continual search for new co- offenders. Among the career thieves stud- ied for the President's Commission on Law Enforcement and the Administration of Justice, the search for opportunities and co-offenclers was self-styled as "hus- tTing." Hustling led to "connecting" with other individuals who were similarly searching, "scouting" for opportunities to commit offenses, and looking for buyers for their stolen goods (GouIc3 et al., 1965:2~26~. The particular set of accom- plices often varies from crime to crime and offenders must work with people ac- cording to the requirements of particular crimes. This is especially the case for what Sparks, Greer, and Manning (1982) . . ~ . ,, . ~ c Descry Be as crime as wor ~ in crimes for gain." GouIc3 anct his co-workers (pp. 51~2) concluded that While a few professional criminals work for extended periods of time with the same ac- complices, most work from day to day, or week to week, with whomever they can put together for a particular job. Each job requires different personnel, different plans, different resources, and even a different working schedule. Searches for accomplices and for means to dispose of illegal gain are facil- itated by using networks as well as par- ticular organizations. There are gathering places, organizational settings, and kinds of encounters that facilitate locating ac- complices. Aclditionally, offenders are symbiotically linkecl to these settings by common residence in a community that is host to these organizations; the accom- plice-offenclers themselves comprise a loose web of affiliations and a resource for referrals in searches. The local pool halls, ]39 bars, and all-night restaurants (Gould et al.:25; Polsky, 1969), the fences (Klockars, 1974), the chop-shops (shops that dis- mantle motor vehicles), and the legiti- mate businesses that deal in some stolen goods, such as parts shops, auto dealers, second-hand stores, and pawn shops (Gould et al.:2~27; 37~39) all are points of contact to search for accomplices and to dispose of illegal goods for gain. For some, the syndicate facilitates the search.  What seems to characterize the net- work among adult offenders is that adult offenders patronize the same places, make the same kinds of transactions, and often reside in the same area. The casual encounter can conclude the search as much as "putting out the word." "Hus- tling" is not a passive activity but an active search for connections (Gould et al. :251. Indeed, excepting recruitment for the more sophisticated crimes, which re- auire a variety of highly specialized skills, the daily round suffices to select accomplices. Dual Processes of Recruitment in Groups Dual processes of recruitment go on In many delinquent and criminal cliques or groups. One process recruits members to a group, either to a loose affiliation with an informal group or, in the limiting case, to a structured position in an organized criminal syndicate. The recruit may be a "raw recruit"-one who is being initiated into delinquent or criminal activity or more commonly, one who has a history of offending. A second kind of recruitment process involves group members recruiting ac- complices in crimes. Although a member is more likely to recruit an accomplice from among fellow group members than from outside, offenders cross group boundaries to select accomplices.

140 Recruitment into Offending Groups Almost all groups that endure for any period of time experience turnover in membership, with some leaving and oth- ers joining. Even size of membership orctinariTy does not remain constant over time (SuttIes:161~. Sarnecki (p. 148) con- cludes that juveniles come into ant] leave gangs quickly; the boundaries of gangs are quite permeable. Normally, member- ship is shorter than 6 months, especially for the less clelinquent. Members also appear to leave easily, generally without resistance or resentment by the group (Suttles: 1671. This suggests that there is a kink! of sifting and sorting going on in peer groups of delinquents. Yet, it is not necessarily high-rate offenders who re- main. A number of studies, e.g., Sarnecki (p. 148), report that a substantial number of high-rate offenders may leave after committing only a few crimes with ac- complices drawn from the group. High- rate offenders often have contacts among a number of such groups, recruiting many of their accomplices from outside their own group. The stable core members of a group during the period it persists, thus, are generally neither the highest nor the lowest rate offenders. Recruitment of Accomplices Selection of accomplices is facilitated by the fact that some groups are linker] with others. Groups with older members, for example, are linked to groups with younger members, ant] groups of the same ethnic origin are linked with one another. This loose linkage facilitates se- lection of accomplices from different groups. The most active offenders, more- over, belong to several groups, ant! so themselves serve as links among their groups. They are the most likely to select accomplices from outside a group in which they may be re garde c3 as a core CRIMINAL CAREERS AND CAREER CRIMINALS member. What is not known from any of the studies, however, is whether the highly active members within these groups frequently are selectecl by others as accomplices. It would appear that they usually are not. It is no simple matter to disentangle the effect of recruitment on offending rates from the selective recruitment of accom- plices on the basis of prior record since longitudinal studies have not addressed recruitment of accomplices. There is no certain answer to the question: How do previously unacquainted offenders find one another and become accomplices? The answer is less obvious than it may  seem. Empey (1970), in a postscript to the account of an ex-offencler, notes that ex- offenders continue to face blandishments from old friends to return to the old ways since friendship networks are one vehicle for entering into complicity to commit an offense. But he also reports (personal communication) that, when he moved de- linquents from one school setting to an- other to provide them with the anonymity to change to a conventional life-style, within a few weeks they had formed as- sociations with all the "local hoods in the new school." This suggests that active delinquents are continually signaling their interest in locating others with whom they may engage in offending. Such signaling is readily picked up by others who are similarly searching. We need to understand how people search for accomplices in offending and how successful those searches are, espe- cially among strangers. W. G. West (1974) concluded that local networks that uniquely identify offenders and their skills facilitate recruitment of accom- plices in theft. Initial contacts are made when relevant information is passed about a named individual. But just how strangers search out one another to com- mit an offense is unclear. They may be- gin, as do cletectives, by asking individu

CO-OFFENDER INFLUENCES ON CRIMINAL CAREERS als they know to help them finct a particular kind of accomplice. However, much searching probably is done by fre- quenting places where cues direct one to potential accomplices. From his study of juvenile thieves who were serious offenders, W. G. West (1974, 1977, 1978) reports that it wouIc3 be an exaggeration to say that most thieves learner! how to practice thieving as a member of a group into which they were recruited. Rather, based on his interviews and observations of high-rate juvenile thieves, he reports that theft is both en- demic to and a highly visible occupation in many Tower class or working-cIass communities. Theft is a readily available activity for almost everyone in the com- munity. Recruitment to high-rate offend- ing in theft takes place, however, in loosely structured peer settings. As cle- scribed elsewhere in this paper, W. G. West (1978:177-178) reports that Groups coalesce and disperse, individuals drift in and out of them, alone or in pairs. Almost all of the neighborhood youths have committed some petty varieties oftheft during childhood. They are usually caught and la- beled as secondary deviants. A great many teens know of the existence of theft through at least one peer who is thieving as an occupa- tion. The symbiotic relationship between some older and younger clusters facilitates contact between peer groups and the potential recruits are able to meet the already initi- ated.... most teens do not need to befriend somebody to find a "partner in crime": they already have one. West goes on to describe how serious thieving involves training through re- peated contacts with experienced per- sons in loosely structured social settings or encounters. Some learn by observing older, experienced thieves; for some there is anticipatory apprenticeship by modeling after an oIcler thief and learning hi s s kill s . S ome may e ve n b e s e if-taught as in the case of one who took a locksmith ]4] course by mad! and . . . ~ was in demanc! as an accomplice because he could pick locks. Especially important, W. G. West (1978: 179) points out, is the cultivation of contacts and relationships that provide information: He needs colleagues, in most cases, who will "cut him in" on jobs, angles, or "hot tips," warn him when "the heat is on," and lend him money when he gets down on his luck. He  needs reliable fences and customers, who are aware of constrictions on his work and know how to "play his game" or interact with him to . . . . 1 . . minimize risks anc ~ maximize gain. Many jobs are carried out in partnerships and some require an elementary division of labor (Shover, 19731. All of these rely on supporting contacts and networks. W. G. West's description can be re- statecl in the following form. Minor delicts and theft are part of class culture and its organization in American society. Hence many youths have engaged in such behavior while growing up. Theft is common among pre-adolescent peer groups as well as adolescent ones. Minor clelinquencies, such as truancy, theft, and vandalism, are committed at early ages. What is critical is how some few get channeled into becoming a high-rate or a specialized offender, such as in occupa- tional theft. Loosely structured groups in networks are critical perhaps in leading to high-rate offending careers. To develop an occupational specialization in theft, one needs oIcler offenders who serve as moclels and who inculcate necessary skills, whether for shoplifting or breaking and entering. Once the skills are ac- quired, however, whether one commits the offense with accomplices depends on whether a division of labor is required to commit the crime or whether one seeks to reduce risks by taking accomplices. One can shoplift alone, for example, but one's risk may be reduced if an accomplice distracts clerks or is alert to security per- sonnel. On the other hand, accomplices . ,

142 may Tower skill levels and increase the risk of apprehension. From this perspective, the role of groups in recruiting members and incluc- ing persons to offend with them in com- mitting clelicts is overstated. For, as W. G. West, SuttIes, en c] others contend, the culture, group organization, and networks that constitute claiTy life in many neigh- borhoocis create the necessary ecological environment for the development of of- fending careers. All too little is known about how incli- vidual offending careers intersect one an- other and how such intersections are de- termined by individual and collective patterns of group affiliation and recruit- ment. Studies are needed that examine the intersection of careers for a cohort of offenders to determine the extent to which the members of a group contribute to one another's offending. But, since de- mographic processes and individual choice leacI to selection of accomplices from outside the cohort, to stucly their contribution to offending, the study pop- ulation must include all co-offenclers of the cohort. Some indication of how sub- stantial that external set of accomplices can be is found in Sarnecki's (1982) Swedish cohort study. The original co- hort consisted of all persons born be- tween 1957 ant! 1968 who were resident in the community and who the police concluded had committed at least one crime in the community between January 1, 1975, and December 31, 1977. There were S75 such individuals in that cohort. But two additional study populations were acIded as the cohort's offending his- tories were followed. The first comprised those individuals who were born before 1957 but who police concluded hac3 com- mitted a crime during the study period with someone in the 1957-1968 cohort or who had committed a crime with a juve- nile who, in turn, had committed a crime with someone in the original cohort. The CRIMINAL CAREERS AND CAREER CRIMINALS second group included persons who po- lice concluded had committee! a crime with someone in the 1957-1968 cohort during the follow-up perioc] from January 1, 1978, to December 31, 1980. These two additional populations included 259 indi- vicluals, so that the total population stud- ied was 834 individuals (Sarnecki: 5~51~. It may be that high-rate offenders are youths who are highly susceptible to overtures from any offender to join in committing a delinquent act. Their high rate of offending with many clifferent of-  fenciers could thus result from their being a "joiner" rather than a recruiter. While further research is needed to explicate the mechanisms of selection of co-offend- ers, both the Peoria (Peoria Crime Reduc- tion Council, 1979) and the Swedish stud- ies provide some evidence that it is primarily a recruiter effect. It is possible that high-A offenders who frequently change co-offen(lers may actually be com- posecT of subpopulations of "joiners" and cc . ,, recruiters. Tracing the web of these relationships, Sarnecki (1982) became aware that the high-rate offenders were linked to other offending groups and inclivicluals outside the city as members of groups extencled their territorial range in the search for accomplices. He concluclecT that by age 1~20 the cohort members still active in crime became part of new gangs that represented larger ant! larger areas ant! increasingly were composed of more high-rate offenclers. The most active juve- niles with high rates of offending thus became affiliated with groups whose members encompassed an entire city and, eventually, several cities (Sarnecki: 2411. Sarnecki even suggests at one point that were one to trace the web of affilia- tions over the careers of the most active members of this small group of gangs in one city, one would find that they were linked to a network that encompasses! all

CO-OFFENDER INFLUENCES ON CRIMINAL CAREERS the major gangs in Sweclen. These re- marks emphasize again the importance of looking at offending networks and the roles that individuals play in them. Some of the most effective intervention may lie in selecting inclivicluals in terms of their place in offending networks. Each offender potentially recruits oth- ers to offend together in a criminal inci- clent, since each may search for an accom- plice. Yet some offenders may be more open to recruitment and others to be re- cruiters. Below, evidence is examined that strongly suggests that some offenders actively recruit co-offenclers. Given their high individual rate of offending, recruit- ers seem more likely to recruit persons whose individual rate of offending is be- Tow theirs. These recruitments may well account for a consiclerable portion of the recruitee's offending. Evidence that there are recruiters comes from the Peoria juvenile residen- tial burglary study. Recall that there were 306 residential burglaries, 151 of which involved at least two co-offenclers (Peoria Crime Reduction Council, 1979~. Of these, 74 or 49 percent involved at least one offender who was previously apprehended and at least one never pre- viously apprehended. This result is con- sistent with the view that an experienced offender selects a less-experienced one. These juvenile "recruiters" the recidi- vists apprehen(lec] with a first-time ar- restee-could logically be targets for spe- cial treatment. Substantial evidence that there are re- cruiters is also found in the stu<ly of a Swedish community by Sarnecki (19821. The 35 delinquents with the highest in- diviclual rates of offending had a total of 224 accomplices in crime, or one-thir(1 of all juveniles in the study population (Samecki:2091. These high-rate offenders generally clid not offend with the same accomplices in very many offenses, which strongly suggests they actively re ]43 cruitec3 other offenders. Moreover, Sarnecki found that, when high-rate of- fenders joined less criminally active groups, they recruited members to of- fencling and appeared to introduce new members to a criminal career (Sarnecki: 2361. Finally, the most criminally active committed offenses with co-offenclers from a larger territorial area (Sarnecki: 171), which suggests that they were re- cruiting from a larger network.5 There is evidence, then, that individu- als with high rates of offending often commit offenses with accomplices, most  of whom have Tower offending rates. Most of these high-rate offenclers keep the same accomplice for only a short period of time. Thus, high-rate offenders commit offenses win a substantial number of dif- ferent accomplices and so must continu- ally search for new accomplices. Still, they are not precluded from also commit- ting some sizable proportion of offenses alone, and that often seems to be the case, increasingly so, perhaps, as they get oIcler.6 Recruiter Elects on Offending The review of patterns of Tone ant! group offending and their relationship to incliviclual rates of offending leacls to a number of conclusions that have policy implications. First, it is apparent that a relatively small number of very high rate youthful offenders can be identifiecl retrospec- tively at a fairly young age. Their preva- lence in a population of offenders will vary by age and place of residence. Were sSarnecki was well aware that limiting the study to juveniles with police contacts and relying on official reports of delinquency limited the size of the network and its population and may have biased results in other ways (Sarnecki, 1982:235). 6Research is needed on the group composition of offenses in offending histories by age of offenders and accomplices.

144 one able to select these high-rate offend- ers prospectively and isolate them, one couIcI, in the short run, avert a substantial amount of juvenile crime, both those they commit alone ant! those attributable to recruiting others. Moreover, to the degree that their incarceration deters or reduces the offending rate of a sizable proportion oftheir accomplices, consider- able additional reductions in crime might be expected since it seems likely that these high-rate offenders seek accom- plices who might otherwise not be as active in offending. Second, although just under two-thirc3s of all crimes against persons and their property are estimated to have but a sin- gle offender (Reiss, 1980:Table 1), an es- timated two-thirds of all offenders com- mit their crimes as members of groups (Reiss, 1980:Table 21. Put another way, although the mean size of offending groups in major crimes against persons ancI their property is just over two of- fenclers, roughly a third of all offenders offend in groups of four or more persons. Where group offending is involved, one would have to incapacitate a substantial proportion of the offending population since there is on the whole very little overlap in accomplices from one offense to the next, especially among oIcler youth- fuT and young adult high-rate offenclers. Third, the most efficient gains in recluc- ing the crime rate might be made by incapacitating high-rate offenders who commit most of their offenses alone. But their number may be trivial even in the population of high-rate offenclers. Incleecl, since high-rate youthful offenders involve a substantial number of accomplices and since the size of the offending group cle- creases with the age of offenders, it may be more efficient to incapacitate young high-rate offenders who are also recruit ers. Fours, the chaining of accomplices in offending provides a means to identify such high-rate individual offenders for CRIMINAL CAREERS AND CAREER CRIMINALS intervention strategies. The use of net- work information may substantially in- crease the capacity to select high-rate offenders who account for the offending of others. Such identification requires that records be kept that uniquely iden- tify all persons engaging in offenses. The Swedish police information system pro- vides opportunities to trace such offend- ers as does that of the police in Japan. Typically, U.S. juvenile courts uniquely identify offenders by name, family and personal characteristics, and address.  They commonly identify co-offenders by name only, making little systematic effort to link offender records to indicate group- offending careers. Similarly, the adult po- lice arrest record typically lists co-offend- ers arrested. Yet, no effort is made to select those individuals who may be sys- tematic recruiters of accomplices. Atten- tion needs to be devoted to identifying high-rate offenders who are also recruit- ers, akin to those in either the Peoria residential burglary study or the Swedish community study. Such recruiters appear logical targets for selective incapacitation or other intervention strategies to reduce their recruitment as well as their offend- ~ng. These studies of juvenile offending and recruitment effects suggest that early in- tervention in the careers of high-rate of- fenders is possible by selecting those youths who fit the recruiter pattern char- acterized by a high individual rate of offending with groups involving a large number of different accomplices. CHANGES IN THE GROUP COMPOSITION OF OFFENDING One can postulate three kinds of crim- inal careers characterized by distinct types of offenders and patterns of offend- ing. The first type of offender always offends alone and can be designated as having a solo offending career. The sec- ond always offends with others. The

CO-OFFENDER INFLUENCES ON CRIMINAL CAREERS third, and by far the most common career, is characterized by both solo and group offending. The solo offender is relatively rare. For some, such as murderers, their career is very short, usually being limited to a single offense. Others begin their career as a solo offender around a particular offense and never commit any other. This is characteristic of many sex offenders and is especially so for certain kinds of sex offending, such as voyeurism and peclo- philia. Just how common a Tong career of solo offending is cannot be determined from existing investigations. Apart from sex offending, confidence forms of fraucI, family assaults, and certain white-colIar offending (e.g., embezzlement), most solo offending careers are probably of short duration. Little also is known about careers char- acterized exclusively by group Fencing. Political criminals, such as terrorists, may be exclusively group offenders. A sub- stantial proportion of very young clelin- quents who have short careers offend only with others. Incleed, their Resistance rates are probably greatest at the young ages following a first apprehension (Wolf- gang, Figlio, and Sellin, 1972:873 889~. Generally, however, most criminal ca- reers that endure are characterizes] by a diverse mix of individual and group of- fending. It has Tong been known that most aclult criminal careers in common crime begin with a juvenile delinquency career. And it is commonly assumed that most careers begin with at least a predom- inance of group offending and that the rate of solo offending increases with age. This conclusion is based on the following evidence previously cited in this paper: · Solo offending is relatively uncom- mon at young ages and does not become the modal form of offending until the late teens or early 20s. · The mean size of offending groups declines with age; groups of three of ]4S fenders become relatively uncommon af- ter age 20; groups of four or more become infrequent at an earlier age, perhaps by age 17. · Solo offending begins to rise sharply at ages 1~16, shortly before the peak age of juvenile offending, and becomes the dominant form of offending at about age 20. Yet, information from longitudinal studies of criminal careers is lacking to determine whether these aggregate statis-  tics support the contention that offenders move from predominantly group to pre- dominantly solo offending in the course of a criminal career. Before alternative explanations for these aggregate changes are offered, several other facts merit at- tention because they are consistent with some alternative explanations: ~ The peak participation rate in offend- ing occurs around age 17 or 18; the abso- lute size of the offender population de- clines rapidly thereafter. · There is a Resistance from offending at every age but especially so in the early teens. · There are many careers of one or two offenses and these predominate at the very young ages. · The participation rate declines mark- ec3ly in the early 20s, suggesting either substantial clesistance or declining aver- age inclivi<1ual rates of offending, or both. · The proportionate increase in solo offending is largely at the expense of a decrease in offending for groups of three or more offenders; the proportion of of- fenses committed by pairs remains fairly constant from late juvenile years through at least the mid-20s. ~ Juvenile offending networks are rela- tively unstable and few are linked to adult networks. On the basis ofthese observations, a num- ber of models of co-offending in criminal careers can be postulated that are consis

146 tent with most, if not all, of the observa- tions. MODELS OF CO-OFFENDING IN CRIMINAL CAREERS A stochastic mode' might provide a reasonable fit. One would expect that as the size of the offending population cle- creases, especially in local areas, there would be fewer offending groups of large size and more pair and solo offending. The rather steep rise in solo offending in the mict-teens, however, casts some doubt on how well such a model would fit if appropriate data were available to test it. A dynamic population model assumes offenders have considerable residential mobility, especially as they reach the point of establishing inclependence from their families. Residential and occupa- tional mobility weaken group ties. Group offending based on prior acquaintance with others in the group should decline as age increases. Those offenders who re- main active must either commit offenses alone or search for similar unaffiliated co-offen(lers. Since groups larger than two persons generally are unnecessary, the search can be truncated when one co-offencier is found. A functional mode' assumes that co- offencling is necessary to commit at least some crimes. One can advance several reasons why offending may need more than a single offender. Whenever a clivi- sion of labor is required to commit a particular offense, co-offending is neces- sary. The necessity for specializes] skills, such as picking a lock, and for collabora- tion, such as driving an escape vehicle, are examples. Co-offending also may re- cluce the risk of apprehension, as when an accomplice diverts attention from the crime scene. Not uncommonly, more- over, offenders seem to require social support to plan and commit an offense. Social support appears to be more charac CRIMINAL CAREERS AND CAREER CRIMINALS teristic of juvenile than adult offending since juvenile offending seems more closely linked to daily routines anct activ- ities. Most juvenile groups provide social support for characteristically juvenile crimes, such as vandalism or shoplifting. This functional model is consistent warn tne selective attrition of low-offen- ding persons, the decrease in the size of offending groups with the age of offend- ers, and the increase in solo offending wherein adults seem to commit offenses that usually do not require a division of labor. Yet, it floes not seem to account very well for the mix of solo and group  offending characteristic of many criminal careers. A selective attrition of group offenders or a solo survivor model tries to account for the sharp decline in co-offencling with age, especially of large groups, anal, cor- relatively, the marked increase in solo offending. More specifically, the more general explanatory problem can be posed of whether the sharp rise in solo offending is (lue to a greater survival of solo offending ant! the selective attrition of group offenders in a population of of- fenders or to a gradual shift of persisters from group to solo offending. Several ex- planations seem worth exploring. First, certain kin(ls of high-rate career offenders may shift towards solo offend- ing because they require a substantial cash flow. This is characteristic, for exam- ple, of drug acicTicts who must commit a number of crimes each day to support their addiction, especially at the peak of dependence. The necessity to acquire cash and to obtain it quickly to make a buy to satisfy an individual need can lead to solo offending. The addict may con- sume search time in locating a buy rather than in locating co-offenders or be unwill- ing to take the time to commit enough crimes to split the income wit-in a co- offender before making a buy. One ex- pects drug dependence to create a sub

CO-OFFENDER INFLUENCES ON CRIMINAL CAREERS stantial rise in solo offending for property crimes that provide cash, especially rob- bery. Nonaddict offenders, moreover, may shun offending with addicts because they perceive an increaser] risk of their apprehension because of the higher risk of apprehension an addict in neec] of a fix may take. Second, the mix of offenses in an of- fending career usually changes substan- tially over time. The offending pattern of very young offenders is to commit in groups such offenses as vandalism and gang fighting. Later careers are character- ized largely by offenses that can be com- mitted alone-burglaries, thefts, armed robberies, and assaults. Thirst, group offenders may be more likely to desist because their networks anc! group affiliations change substan- tially with age. There is considerable ev- idence that the groups to which juveniles belong are quite unstable and do not persist for Tong periods of time. There is also evidence (Suttles, 1968) that many youth groups do not persist across the transition from juvenile to young adult years. Incleed, it seems reasonable to as- sume that the network structure of local communities is age graclecl so that as of- fenclers age, they move into aclult net- works, which are less likely to provide informal support for offencling.7 Attrition can be expected in any transitional proc- ess. We need to learn more about how such network transitions are macle and how they affect offending behavior. It is likely that adult networks on the whole facilitate inctiviclual rather than group search behavior. Moreover, it is likely that the aclult networks that facilitate in 7Exc~luding formally organized group behavior in which participation increases with age, it can be speculated that the propensity to do almost any activity alone or in pairs increases with age. But just what causes these changes is not well understood; maturation is a description, not an explanation, of what is taking place with age. 147 clividual searches for co-offenclers are less cohesive ancI clique oriented] than are those of juveniles. They also are more covert. Research on criminal careers is needed to determine which ofthese explanations accounts for the aggregate shift towards solo offending. If it turns out that juve- niles who shift early to high-rate, solo offending are most likely to have high- rate careers as adults, they are candidates for early intervention. What may be of special concern for timing incapacitative  forms of intervention is identifying such individuals and detecting when the shift to predominantly solo offending occurs. GROUP OFFENDING AND DESISTANCE FROM CRIMINAL CAREERS Earlier it was conjectured that the shift towar(ls solo offending with age might be accounted for by the selective attrition of group offenders, i.e., group offenders cle- sist at an early age. More generally, the question arises of whether group proc- esses account for Resistance from crimi- nal careers so that the nature of one's affiliation with groups of offenders affects one's Resistance probability. This section begins with a review of several empirical studies of Resistance from offending that take into account group characteristics of offending in crim- inal careers. This is followed by specula- tion on how group processes may enter into selective attrition from offencling. Empirical Studies of Desistance from Group Offending There clearly is Resistance from of- fencling at every age. The reasons for desisting may well vary with career sur- vival time. Some criminal histories are of very short duration, especially so for those who enter at a very young age;

148 others offend over a long period and cle- sist or reduce their rate of offending sub- stantially only at an advanced age (Blum- stein, Farrington, and Moitra, 19851. How these points in an offender's history cle- termine the rate and form of Resistance is explored below. Since most offenders be- gin offending by age 16, and since desis- tance rates appear to be greatest soon after entry, one expects greater aggregate desistance before age 20 than after. Knight and West (1975) report on tem- porary an ct continuing clelinquents in a Tong-term survey of a cohort of 411 boys in a working-cIass London neighborhood (see also D. I. West, 19691. They selected 83 boys who constituted the most delin- quent fifth of their cohort on the basis of prior convictions or admissions of delin- quency (Knight and West:43~. The 83 delinquents were divided into two groups. One group of 33 was labeled temporary delinquents because each had no official record of delinquency since turning age 17 and denied committing any offenses at age 1~19. The second group of 48 was labeled continuing delin- quents because each had one or more criminal convictions since turning 17 or ac3mittec3 to committing one or more ma- jor offenses.8 Temporary delinquents were conviction-free for a period of at least 3 years; continuing delinquents had a continuing record of convictions or ac- knowledged offenses. The largest single difference between , c7 O CRIMINAL CAREERS AND CAREER CRIMINALS that the continuing delinquents were more likely to commit their offenses alone. None of the official records for the temporary (lelinquents stated a boy was convicted alone, compared with 14 per- cent of the convictions for the continuing delinquents, a difference that was signifi- cant even when the larger number of con- victions for continuing delinquents was taken into account. Although it is clear that the large majority of convictions for continuing offenders involved group of- fenses, temporary delinquents had been involved only in group offending (Knight and West:45). Some explanation for this difference may be found in offenders' reported in- volvement in adolescent peer groups. Knight and West (1975:45), like Scott (1956), found that involvement with peer groups declines with age and can be short-lived. Temporary delinquents re- ported greater abandonment of their ado- lescent male peer groups than did con-  tinuing delinquents. Although all but one of the temporary delinquents had re- ported going about in an adolescent peer group of four or more boys between the ages of 15 and 17, somewhat less than half of them said they were doing so at age 17~/2. The involvement of roughly 80 per- cent of the continuing delinquents in groups of this size remained unchanged during these years (Knight and West: 45~6~. In disengaging from peer groups, the youths in these two groups was their involvement with adolescent peer groups in offending (Knight and West:45~. Both official records and self-reports disclose Two youths are not included in the analysis because they were not interviewed at ages 1~19. One was killed in an accident at age 17, and the other was a fugitive from justice and untraceable (Knight and West, 1975:43). the temporary delinquents did not be- come social isolates. Rather, they gener- ally began to go about with only one or two companions in contrast to their ear- lier participation with four or more. It is apparent once again that the more serious and higher rate offending youths who continue in delinquency include some who exhibit solo offending while continuing to associate with the members of rather large groups with which they are

CO-OFFENDER INFLUENCES ON CRIMINAL CAREERS affiliated. In some sense the large group perhaps serves as a reference group and perhaps also as a resource for recruiting accomplices for offending, since persist- ing offenders maintain a mix of indiviclual and group offending. From Suttles's (1968) work, it seems that dropping out from the larger partici- patory group is a function of both mobility to other neighborhoods and a shift to more conventional roles, such as work, marriage, and military service, that bring new forms of association. In addition to the 41 percent of the Addams-area youth who quit their neighborhood group be- cause they moved to another area of Chi- cago, 26 percent left Chicago for jobs or military service. Another 30 percent ei- ther married or went to work in Chicago. Only one left to serve a prison term and only one left because he decicled he ctid not want to belong to the group. Although dropouts were not looked down on for leaving their group "involuntarily," the one who left because he wanted to was reported to have been ostracized for it (Suttles:1671. The effect of leaving on offending behavior is unclear in Suttles's study, but the development of conven- tional bonds (e.g., marriage, regular em- ployment, and getting adclitional ecluca- tion) anct the severing of oIc! ties by moving out of the neighborhood or enter- ing military service could account for a sizable proportion of Resistance in the late teens and early 20s. The importance of environmental as well as status kansi- tions on Resistance rates is buttressed by a major study of aclults paroled directly to the U.S. Army in World War II. Mattick (1960:49-50) found that the 1-year parole violation rate was 5.2 percent for parolees in the army, compared with 22.6 percent for civilian parolees. Moreover, only 10.5 percent of the army parolees had commit- tec! an offense within 8 years of discharge ]49 from prison, compared with an expected recidivism rate of 66.6 percent (Mattick, 1960:54). W. G. West (1978) found that just over one-half of 40 high-rate juvenile thieves had retired from active criminality by age 20. Comparing these "reformed" crimi- nals with those who were still "active," West fount! that 86 percent of the re- formed criminals, compared with only 11 percent of those still active, had formed conventional bonds with a woman, a job, anchor schooling (W. G. West, 1978:186~. Of course, it is difficult from these behav- ioral ciata to determine whether the form- ing of the bond led to the Resistance or  whether the bond was former! as a means of extricating oneself from a pattern of offending in a social network. Moreover, one does not know whether those who were still active and had not formed such bonds hac3 tried and failed in doing so. The causal ordering of variables account- ing for Resistance is not easily resolved either theoretically or empirically. The effect of leaving on offending be- havior also is illuminated by the work of Knight and West (19751. They concluded that the delinquency of temporary delin- quents is dominated by group solidarity rather more than by inclividual motiva- tion to offend. As group offenders mature out of adolescence, many shift towards conventional work and family roles and going about with one or two of"the boys" in sports and bars or pubs. The "true" transient group clelinquent in this sense is more likely to be a temporary delin- quent whose affiliation is broken by transiency or by maturation into more conventional roles. The continuing delin- quent who moves from juvenile into adult offending while maintaining network and large-group affiliations more and more offends alone or offends with one or two accomplices with whom affiliation is tran

150 sitory. Some confirmation for this pattern of shifting to lone offending with aging was found in a major study by Sveri in Sweden (19651. Group Processes in Selective Attrition from Criminal Careers The concern in this section is to ac- count for the selective attrition of group offenders. In doing so, explanations of how group and network processes ac- count for Resistance from criminal careers are posited. Three archetypes of desis- tance are presented. Desistance of group offenders can be attributed to specific deterrence, to status transactions, and to disruption of group affiliations. Specific Deterrence This archetype assumes that punishing an offender not only leacis the offender to desist but has consequences for other group members as well. It assumes. moreover, that an individual is inducted into offending by others, as a member of a group, and continues to offend with one or more group members until appre- hendecI. The sanction of being caught (particularly if reinforcer] by parental or other sanctions) leads that individual to desist from offending. The experience of being sanctioned has a specific deterrent effect on the offender. Apprehending and punishing a member of a clique may increase the likelihood that other mem- bers, especially accomplices who are not punishecl, will also desist. The percep- tion of the risk of being caught is in- creased by apprehension followed by punishment. This is a special kind of specific (leterrence whereby the punish- ment of some specific and significant other leads to Resistance. What is critical in this deterrence is that one knows the person who is being punished and conse- quently has a more (lirect basis for vicar CRIMINAL CAREERS AND CAREER CRIMINALS iously experiencing the punishment and calculating one's risk. This archetype probably accounts for much of the Resistance at young ages ant! early in an offending career. The (lesis- tance rate is greatest following the first apprehension (Wolfgang, Figlio, and Sellin, 1972:871. Status Transitions Adolescence is characterized by the substitution of peer for parental relation- ships and control, especially for males, in American society. As boys grow up, peer groups lose some of their influence and control over those affiliated with them.  This Toss of influence is closely tied to status transitions, particularly those con- necte(1 with the transition to adult status. But from early adolescence on, indiviclu- als make transitions to more conforming affiliations. Such transitions decrease the influence of nonconforming peers. This archetype can account for the cle- sistance of in(livicluals with low rates of offencling. Such offenders usually offend! with accomplices. They are recruited by others to participate in offenses others initiate, an(1 they rarely, if ever, initiate offending. They are at most peripheral members in a group and offend as occa- sional recruits. With marginal affiliation to a group, they desist after one or a few such experiences because they find con- forming activities more rewarding and less risky or because they are not selectecI as accomplices. These low-rate offenders normally clesist in the early adolescent years. This second archetype is especially germane, however, for persons who par- ticipate in delinquency as part of cliversi- fied peer activity. This is generally char- acteristic of groups of young mates, such as athletic teammates and street-corner gangs. Here one gradually is drawn into the cultural life-styTe of one's class

CO-OFFENDER INFLUENCES ON CRIMINAL CAREERS wherein delinquency is part of that life- styTe. With aging, members turn to court- ship, marriage, and having children as well as securing regular employment to fulfill family responsibilities. They may withdraw from their peer group to as- sume these adult roles. Women and fam- iTy may play a role in weaning the mate delinquent from his gang. Desistance in this archetype is a function of movement to an adult status. This pattern probably explains the Resistance of many Tower- cIass ethnics in the United States in their young adult years. The Resistance of group offenders may result not only directly from the cleclining influence of peers on a member's behav- ior but also from the indirect effect that has on group cohesiveness and the selec- tion of co-offenders. With Reclining cohe- siveness, groups become vulnerable to dissolution. There appears to be some discontinuity between the individuaT's maturation and group adaptation to the changing requirements of its members. Since such transitions do not occur at the same rate for all members, the group experiences selective attrition that gradu- ally weakens it. Most cannot recruit re- placements since the pool of those eligi- ble also declines with transition to adult status. Failing to recruit enough members to maintain its status quo, the group dis- integrates. Much Resistance of group offenders and offending may be tied to the fate of the particular groups to which they be- Tonged. Individuals who depend primar- iTy on co-offending with accomplices af- fiTiatect with a particular group or network are particularly vulnerable to its clemise. Inasmuch as the demise of delinquent peer groups is most commonly associated with the transition of their members from juvenile to aclult status, group clemise should account for at least some of the Resistance of group offenders in their late teens. Desistance in this second arche 151 type does not clepencl on apprehension and punishment. The archetype is consis- tent with explanations of selective attri- tion of group offenders. Disruption of Group Affiliation Indivicluals are particularly vulnerable to the disruption or dissolution of group ties. Where such ties are primarily re- sponsible for or play a major role in rein- forcing one's delinquent or criminal be- havior, breaking them may lea(1 to Resistance. This is especially likely to occur when the incliviclual is unable to  replace those bonds with others that per- mit continuation of the behavior. The three major ways that group bonds and affiliations that reinforce offending are dissolved are through residential mobil- ity of an offender, affiliation with a total institution, and the dissolution of the re- inforcing group itself. Resi(lential mobility breaks group ties en c] makes problematic reincorporation into a new group in the area to which one moves. Those who are unlikely to offend without group support drop out if they cannot affiliate with a new offending group or find accomplices. Residential mobility often is combined with other forms of social mobility, such as getting aclvanced schooling or a job or joining a military organization. Total institutions also may affect clesis- tance from careers. Two types of total institutions can have important conse- quences for Resistance from criminal ca- reers: prisons and the military. Incarcera- tion affects one's position in a group and, if it still exists on release, one may have (lifficulty reentering. By (lisrupting pat- tems of search for co-offen~lers, incarcer- ation may also shift offenders from pre- clominantly group to predominantly solo offending. Apart from (lisrupting co- offending patterns, incarceration also may have specific deterrent effects, especially

1S2 for young group offenders. If so, incapac- itation at young ages may be a viable strategy to bring about Resistance from offending.9 Another major source of Resistance through entry into a conventional total institution is the military. Some criminal careers obviously go on in the military but many appear to be broken. As noted previously, Mattick (1960) found that 87 percent of the Illinois inmates paroled to the army were given honorable dis- charges and their recidivism rates were far below those of civilian parolees. Re- search is necessary to examine the effect of military, merchant marine, and other forms of service on interrupting and ter- minating contemporary criminal careers since Mattick's research was on parole to the World War II army (Mattick, 1958, 19601. Service in conventional total insti- tutions is a potential alternative source of intervention in criminal careers. This third archetype suggests that breaking group bonds and relationships may be a critical factor in Resistance from offending. The breaking of ties is facili- tated by forming new relationships or movement to new environments. Such ties are most likely to be broken in the late teens or early adult years. Offenders who continue to offend after these years may be those who usually commit of- fenses alone or who select different ac- complices, often strangers, for each of- fense. The adult career offender, then, is characterizes] by transient relationships with other offenders. Relationships with co-offenders are instrumentally contrived 9Impr~sonment also introduces one to new of- fenders and networks. Just how influential such ties are on one's offending after leaving prison is not known. Such ties may not be as important as com- monly assumed because inmates win close ties are usually not released at the same time and such ties and prison attitudes become less influential as re- lease nears (Wheeler, 1961~. CRIMINAL CAREERS AND CAREER CRIMINALS rather than by-proclucts of group affili- ation. INTERVENTION ISSUES AND GROUP OFFENDING Group offending is most characteristic of what we think of as juvenile delin- quency and characterizes juvenile ca- reers. Characteristically, the juvenile court clears with a group of co-offenders rather than a solo offender when consid- ering a particular offense for which the juvenile is apprehended. Moreover, the juvenile career is more likely to be char-  acterized by a predominance of group offending when compared with adult ca- reers. Ad~litionally, the juvenile court is better able than is a criminal court to consider co-offenders in dispositions be- cause it is less bound by many of the proceclural safeguards attending criminal proceedings and by the obligation to try fact without knowledge of the prior record and current status of the offender and co-offenders. The juvenile court has greater latitude to investigate ant] dispose of matters involving joint offending and joint careers. When siblings are involved in juvenile offending, there is greater op- portunity for family intervention. Given this greater latitude in investigating, ex- amining, and disposing of juvenile cases and the fact that most adult criminal ca- reers are initiated in the juvenile years, serious consideration must be given to identification of and intervention in those juvenile careers most likely to lead to adult careers. A number of issues relating to such interventions are examined be low. Probability of Being Caught in Co-Offending Earlier, attention was drawn to a possi- ble apprehension selection bias in con- junction with membership in a group

CO-OFFENDER INFLUENCES ON CRIMINAL CAREERS (Erikson, 1971). Specifically, the question is whether there is evidence that group offenders are more likely to be caught than are solo offenders ant] whether one's risk of apprehension increases with the size of an offending group. Also, the prob- ability of being apprehended as either a juvenile or an adult may well be a func- tion of the prior record of a current co- offender. It seems reasonable to conclude that one's risk of apprehension is a function both of one's indiviclual rate of offending ant] that of one's co-offenders. And it is apparent that, at least for juveniles, high- rate offenders with accomplices are more likely to engage in crimes than are those who commit the same acts alone (Hindelang, 1976b:1221. One may also expect interaction effects between one's rate of offending, the group mix of one's offenses, and the offending rate of one's co-offenders. Inasmuch as one of the major ways of estimating an individual's rate of offend- ing is from official records of apprehen- sions, it is important to know whether the probability of apprehension is substan- tially greater for group than lone offend- ers who have the same indiviclual rate of offending. One expects, however, that this could vary by age so that the risk of apprehension is a function of whether one offends alone or with others at older as well as younger ages. It is important also to know whether apprehension is more likely to occur for a co-offending than a solo offense, even for those careers characterized by a substantial proportion of solo offenses. If so, using official rec- ords of apprehension will underestimate substantially the individual rate of of- fencling for those who engage predomi- nantly in solo offenses and overestimate the rate for group offenders if the two are not distinguished. Just how much the individual rate of offending, in comparison with the group 753 status of an offense, increases one's risk of apprehension is most problematic for of- fenses against the person. A predomi- nantly solo pattern of offending may be disproportionally constituted of non- stranger crimes against the person, since that arrest rate is a function of victim identification et the time ofthe complaint. In brief, the rate of apprehension for solo offenses against persons who are strang- ers as well as against property may be substantially lower than that for others at risk. This suggests that it is important to separate solo from group offending rates  by type of offense in assessing apprehen . sion r1s Is. Early Identification of Career Criminals One of the barriers to early identifica- tion of adult criminal careers has been the high apprehension rate of juveniles such that the population of offenders for dispo- sition is large. Moreover, it is commonly presumed that at young ages it is clifficult to distinguish high-rate offenders from low-rate offenders. The Wolfgang, Figlio, and Sellin (1972) retrospective cohort study supports this conclusion by report- ing probabilities of committing a next offense by offense number and transition probabilities based on number and prior offense type. These probabilities were based on the entire juvenile career with- out respect to annualized indiviclual rates of offending or of the time between of- fenses. This may be important informa- tion that may help to separate high-rate juvenile offenders from the population of all offenders at a fairly early age. There is, of course, the possibility that many early high-rate offenders desist from a criminal career well before adult- hood. There is no strong evidence that this is the case, however, and there is evidence that high-rate adult offenders can be identified by their juvenile of

154 fencling rates (Chaiken and Chaiken, 1982). It has been suggested here that there is an important subset of juvenile offenders that should be identified for special adju- ctication. These are the high-rate offend- ers who can have high rates of recruiting co-offenders. They would appear to be easily identified by both their high incli- viclual rates of offending and the number of offenses they commit with a large num- ber of different co-offenders. One shouIc] be able to distinguish between those high-rate co-offenders whose rate is pri- marily dependent on affiliation with a small number of co-offenders ant] those who are continually involved with new associates. Moreover, there is some evi- dence that these recruiters affect the par- ticipation rate of their co-offenders as well as the co-offenders' rates of offend- ing. If that is the case, they are especially opportune targets for early intervention. In any event the evidence presented sug- gests that the recruiter population for ju- venile offenders is a small enough subset to warrant considering early intervention in their careers, perhaps even with strat- egies of incapacitation, since their inca- pacitation may have a marked effect not only on the aggregate crime rate but on participation and Resistance rates of co- offenders as well. Investigation into these early intervention possibilities and their effects would add appreciably to our un- derstancling of early criminal careers ant] the possibilities for intervening in them. Sanctioning Group Offenders Our adult system of justice is to a sub- stantial degree based on preserving the individual integrity of co-offenders as they are processed in the criminal justice system. This means that not only must their individual history of offending be disregarded when trying a current set of charges but also inclividuals involved in CRIMINAL CAREERS AND CAREER CRIMINALS the same offense can be treated cliffer- ently based on their role in the charged offenses and that differences in their prior offending or personal histories can be taken into account at sentencing. The particular doctrine of sentencing will de- termine what will or may be taken into account, but there is a general presump- tion that the disposition for one offender need not be contingent on that of another so Tong as equals in all respects under law are treated equally. As noted above, the juvenile court has not been bound as tightly by these considerations and hence may have considerably more latitude to  consider alternative strategies for sanc- tioning accomplices in an offense. Unfor- tunately, very little evidence about the effects of differences in sanctioning co- ofFenders is available to guide such choices. One ofthe important issues in juvenile sanctioning is the extent to which early sanctioning may be a specific and a gen- eral deterrent to offending. It has been suggested here that punishment of one's co-offenders increases the sense of risk. Worth considering is whether early sanc- tioning of all co-offen(lers increases the Resistance rate. Of special interest also is whether differential sanctioning for of- fenders in the same offense has different specific deterrent and Resistance effects for co-offenclers. Where offenders are linked in the same networks, one expects to find overlap in offending careers. Each of these careers can be treated inclependently to deter- mine the extent to which sanctioning in- terventions affect each career. Yet, each may also be regarcled as affected by the interventions in the careers of co-of- fenders. Desistance may be influenced as much by the sanctioning of co-offenders as ofthe individual offender himself. Sim- ilarly, one's pattern of offending with oth- ers may be influencer! by interventions in their careers or by their clesistance. Is one

CO-OFFENDER INFLUENCES ON CRIMINAL CAREERS more likely to desist if one's co-offenders have desisted? These seem questions worthy of investigation. Intervention Strategies and Group Offending Intervention strategies disrupt the lives of individuals ant! affect their criminal careers. Although many interventions are individual-change strategies, some aim to change the external conditions believed to cause the behavior of individuals. Klein ant] Crawford (1967:6~66) con- cluded that "elimination of external sources of cohesiveness of gangs, in most cases, wouIcI be followed by clissolution of a relatively large proportion of gang membership." They reached this position by observing that the internal sources of gang cohesion are weak and that the gang is maintained by strong external pres- sures (Klein and Crawford:661. From this perspective, juvenile gangs are rather fragile entities. They lack the internal stability of internally generated group goals, have high turnover in membership, generate few unique group norms, and lack a lasting identity with a group name (Klein and Crawford:66~. There is also considerable evidence that these inter- nally unstable gangs are kept together by external pressures, such as conflict with other groups (Klein and Crawford:65~6) or political activity ( lacobs, 1976; Short and MolancT, 19761. Disrupting the intemal structure of such gangs to diminish their members' delinquency is not likely to be successful. Klein and Crawford (1967) concluded that interventions by group workers in black clelinquent gangs in Los Angeles tencled to increase the social cohesive- ness of the gangs, thereby resulting in increaser! indiviclual rates of offending. Similarly, Short and StroUtbeck (1965) concluded that disrupting the gang lead- ership in the black gangs they studier! l5S failed to disrupt the gang and may actu- ally have contributed to increased rates of minor delinquent acts. Although the group workers' interven- tion strategies that aimec] at disrupting group structure and processes may not have achieved their intended result, other strategies of intervention aimed at disrupting networks may nonetheless be consequential for individual careers. Klein and Crawford (1967) suggested that strategies that weaken group cohesion may have that effect. Incapacitation of offenders similarly does not necessarily disrupt gang struc-  ture. Jacobs (1976, 1977, 1983) described the U.S. prison of the 1960s and 1970s in our most urbanized states as organized around an inmate system that was an extension of the gang organization and the conflict relations from which the prison population was drawn. He sug- gested, moreover, that the prison society tencled to increase the cohesion of those gangs and their importance to their mem- bers because they performed a wicle va- riety of functions for them. The gangs also recruited members within the prison, es- pecially as the inmate system became politicized (Jacobs, 1976, 1977: 14~1491. On release a substantial minority of the recruits and many of the former members of the supergangs (which include juve- nile and adult divisions in the community and in juvenile and adult correctional institutions) retained ties with at least some former gang members. The prison of this period appeared to extend the network of a(lult co-offenders through in- mate recruitment into the gangs while incarcerated and sustained at least some of them on release. Group Recruitment and Replacement Processes When more than one incliviclual is in- volvecl in offending, incapacitation of one

156 of the offenders will not necessarily affect the crime rate. Much depends on what happens to the individual rate of offend- ing of his co-offenders. The current prac- tice is to discount the "crimes saved" through incapacitation of an offender by weighting the individual's contribution to the incident, using the mean size of of- fending groups for that offense. If the mean size of offenders is two, for exam- ple, an individuaT's crime saved is caTcu- lated as one-half. Yet, there is no empiri- cal evidence to conclude that individuals recluce their crime rate, on the average, in this way. There is some reason to believe they might not. One wouIc! expect that the incapacitation of recruiters, for exam- ple, would have a far greater impact on the crime rate than would the incapacita- tion of followers or many of their co- offenclers, even granted equal rates of offending. What is neecled is a comparison of over- lapping criminal careers to determine whether incapacitation in the career of any offender has any effect on the indivicl- ual rate of offending of his co-offenders subsequent to incapacitation. Given the fact that the more organized gangs of career criminals are more adept at recruit- ing replacements for incarcerated mem- bers, incarceration may save only those crimes that are solo offenses. If so, one should not only aim to incapacitate incli- viduals with high solo offending rates but one should expect little impact from inca- pacitating individuals who have high co- offencling rates as accomplices. If this were to be verified empirically, one would also expect to gain less from inca- pacitating juveniles than adults, since their solo offending rates are on the aver- age Tower. One might also expect to gain more by disrupting processes of recruitment and replacement of co-offenders in juvenile than in adult populations, since juvenile CRIMINAL CAREERS AND CAREER CRIMINALS networks that facilitate offending are seemingly based more on soliciaristic and personal relationships than on rational choice and impersonal contact. For most juveniles, the selection of co-offenders is far more restricted than it is for adults, being limited to a territorial neighbor- hood and its environs. We need to know more about how adults recruit their co- offenders and the extent to which their co-offenders are dispersed rather than concentrated in space and time. We clearly need both ethnographic and net- work studies of adult offending popula- tions to explore these issues of co-offend-  ing and their effects. CONCLUDING NOTES It should be abundantly clear that re- search on group offending not only is disproportionally concentrated on juve- niles but that it has focused almost exclu- sively on documenting how pervasive it is and on speculating on its role in the etiology of delinquency. The etiological question therefore remains murky and the consequences of groups for criminal career development remain unexplored. We need, therefore, to devote far more attention to detailed studies of offending careers and to pay special attention to the group composition of offending in those careers, treating each individual's career in terms of its intersections with others. Not only do such investigations provide sociometric information so that such tech- niques as block-modeling (White, Boor- man, and Breiger, 1976) can isolate par- ticular networks, but also they permit us to examine how criminal justice interven- tions have consequences for the offend- ing of all individuals in the same network. Examination of the consequences of solo and co-offending is but part of a larger need to order over time the life events of offenders with their offending history so

CO-OFFENDER INFLUENCES ON CRIMINAL CAREERS that we may determine their conse quences. Furler study of group offencI ing requires far more information on how juvenile and adult careers are linked. That information must be obtained from prospective longitudinal cohort studies that approximate more closely the design of the Swedish community study under taken by Sarnecki (19821. That design is based on a social-network approach and expands the study population to inclucle all co-o~enders who are not initially part of the cohort. It is well to bear in mind that in a dynamic society there is consicI erable movement into and out of commu nities and that individuals are drawn into different networks over time. The artifi cial divorcement of the cohort from a changing environment and its reduction to a population of individuals unrelated in time and space restrict considerably what can be learned about individual careers. ACKNOWLEDGMENTS The persons who have assisted me in different ways in the preparation of this paper number well over 100 scientists in this country and abroad who responded to my requests for information. I am much indebted to them. I owe special thanks also to those who took the time to read and provide help in revising one or more drafts. Numbered among them are some of my colleagues on the Panel on Re search on Criminal Careers. Those to whom I owe this special debt include Larry Baron, Alfred Blumstein, Shari Di amond, David Farrington, Malcolm 1970 Postscript. In A. Manocchio and J. Dunn Klein, Richard Lempert, Lloyd Ohlin, lames Parker, Elizabeth Piper, Brent Shea, lames F. Short, Ir., Marvin Wolf gang, Stuart Wright, and Frank Zimring. I owe a special debt also to Jerzy Sarnecki, who discussed his Swedish study win me, and to Denise Galarraga of Me Na 157 tional Criminal Justice Reference Service for a partial translation of Mat study. REFERENCES Blakely, R., Stephenson, P. S., and Nichol, H. 1974 Social factors in a random sample of juvenile delinquents and controls. International Journal of Social Psychiatry 20:20~211. Blumstein, A., Farrington, D., and Moitra, S. 1985 Delinquency careers: innocents, Resisters, and persisters. Pp. 187-219 in M. Tonry and N. Morris, eds., Crime and Justice: An An- nual Review of Research, Vol. 6. Chicago, Ill.: University of Chicago Press. Bordua, D. J. 1961 Delinquent subcultures: sociological inter- pretations of gang delinquency. Annals of  the American Academy of Political ~ Social Science 338:120-136. 1962 Some comments on theories of group delin- quency. Sociological Inquiry 32:24~246. Bottoms, A., and Wyle, P. 1986 EIousing tenure and changing crime careers in Britain. In A. J. Reiss, Ir., and M. Tonry, eds., Communities and Crime. Chicago, Ill.: University of Chicago Press. Breckinridge, S. P., and Abbot, E. 1917 The Delinquent Child and the Home. New York: Russell Sage Foundation. Bureau of Justice Statistics 1984 Criminal Victimization in the United States, 1982. A National Crime Survey Report. NCI-92820. Washington, D.C.: U.S. Govern- ment Printing Office Chaiken, I., and Chaiken, M. 1982 Varieties of Criminal Behavior. Santa Monica, Calif.: Rand Corporation. Clinard, R. A., and Ohlin, L. E. 1960 Delinquency and Opportunity. Glencoe, N.Y.: Free Press. Elliott, D. S., Ageton, S. S., Huizinga, D., Knowles, B.,andCanter,R.J. 1983 The Prevalence and Incidence of Delinquent Behavior: 197~1980. Boulder, Colo.: Be- havioral Research Institute. The Time Game. Beverly Hills, Calif.: Sage Publications. Erikson, M. 1971 The group context of delinquent behavior. Social Problems 19:11~129. Eynon, T. G., and Reckless, W. C. 1961 Companionship at delinquency onset. Brit- ish Journal of Criminology 2:162-170.

158 Federal Republic of Germany 1982 Police Statistics, 1982. Bonn. Ferguson, T. 1952 The Young Delinquent in His Social Setting. London: Oxford University Press. Cannon, T. M. 1966 Emergence of the defensive gang. Federal Probation 30:44 48. Glueck, S., and Glueck, E. T. 1930 Five Hundred Criminal Careers. New York: Alfred A. Knopf. 1934aFive Hundred Delinquent Women. New York: Alfred A. Knopf. 1934b One Thousand Juvenile Delinquents. Cam- bridge, Mass.: Harvard University Press. 1937 Later Criminal Careers. New York: Com- monwealth Fund. 1940 Juvenile Delinquents Grown Up. New York: Commonwealth Fund. 1943 Criminal Careers in Retrospect. New York: Commonwealth Fund. 1946 After Conduct of Discharged Offenders. London: Macmillan. 1950 Unraveling Juvenile Delinquency. New York: Commonwealth Fund. Gould, L., Bittner, E., Chaneles, S., Messinger, S., Novak, K., and Powledge, F. 1965 Crime as a Profession: A Report on Profes- sional Criminals in Four American Cities. President's Commission on Law Enforce- ment and the Administration of Justice. Washington, D.C.: U.S. Government Print- ing Office. Granovetter, M. S. 1973 The strength of weak ties. AmericanJournal of Sociology 78:1360-1380. Hindelang, M. J. 1971 The social versus solitary nature of delin- quent involvements. British Journal of Criminology 11:167-175. 1976a Criminal Victimization in Eight American Cities: A Descriptive Analysis of Common Theft and Assault. Cambridge, Mass.: Bal- linger. 1976b With a little help from their friends: group participation in reported delinquency. Brit- ish Journal of Criminology 16:10~125. flood, R., and Sparks' R. 1970 Key Issues in Criminology. London: World University Library. Jacobs, J. B. 1976 Stratification and conflict among prison in- mates.Journal of Criminal Law ~ Criminol- ogy 66:476 482. 1977 Stateville: The Penitentiary in Mass Society. Chicago, Ill.: University of Chicago Press. CRIMINAL CAREERS AND CAREER CRIMINALS 1983 New Perspectives on Prisons and Imprison- ment. Ithaca, N.Y.: Cornell University Press. Jones, M., Offord, D., and Abrams, N. 1980 Brothers, sisters, and antisocial behavior. BritishJournal of Psychiatry 136: 13~145. Kaiser, G.  1982 JugendEriminalitat: Rechtsbruche, Rechts- brecher, und Opfersituationen im Jugen- dalter. 3 Aufl., Basel: Weinheim. Keiser, L. 1969 The Vice Lords: Warriors of the Streets. New York: Holt, Rinehart, & Winston. Klein, W., and Crawford, L. Y. 1967 Groups, gangs and cohesiveness. Journal of Research in Crime and Delinquency 4:142- 165. Klockars, C. B. 1974 The Professional Fence. New York: Free Press. Knight, B. J., and West, D. J. 1975 Temporary and continuing delinquency. British Journal of Criminology 15:43~0. Kraeupl, G. 1969 Der Einfluss social fehlentwickelter Ju- gendlicher auf die Enstehung, Entwicklung, StruLtur und Funktion krimineller Grup- pen 14~25 jahriger. Stoat und Recht 18:6 75. Lerrnan, P. 1967 Gangs, networks, and subcultural delin- quency. American Journal of Sociology 73:63-72. Macguire, M., and Bennett, T. 1982 Burglary in a Dwelling. London: Heine- mann. Mattick, Hans W. 1958 Parole to the army. Department of Sociology. Unpublished M.A. thesis. University of Chi- cago. 1960 Parolees in the army during World War II. Federal Probation 14:49 55. Miller, W. B. 1958 Lower class culture as a generating milieu of gang delinquency. Journal of Social Issues 14:~19. 1975 Violence by Youth Gangs and Youth Groups as a Crime Problem in Major American Cit- ies. Washington, D.C.: U.S. Government Printing Office. Morash, M. 1983 Gangs, groups, and delinquency. British Journal of Criminology 23:30~331. Olofsson, B. 1971 Vad car det vi sa! Om kriminellt och konformt beteende bland skolpodhar. Stock- holm: Utbildningsforlaget.

CO-OFFENDER INFLUENCES ON CRIMINAL CAREERS Olson, M. R. 1977 A Longitudinal Analysis of Official Criminal Careers. Ann Arbor, Mich.: University Mi- crofilms. Peoria Crime Reduction Council 1979 Criminal Activity of Juvenile Residential Burglars. Peoria, Ill.: City of Peoria. Petersilia, J. 1980 Criminal career research; a review of recent evidence. Pp. 321~379 in N. Morris and M. Tonry, eds., Crime and Justice: An Annual Review. Vol. 2. Chicago, Ill.: University of Chicago Press. Polsky, N. 1969 Hustlers, Beats and Others. Garden City, N.Y.: Doubleday. Pyle, G. E., Hanten, E., Williams, P., Pearson, A., Doyle, J., and Kwofie, K. 1974 The Spatial Dynamics of Crime. Depart- ment of Geography. Chicago, Ill.: University of Chicago. Reiss, A. J., Jr. 1967 Studies in Crime and Law Enforcement in Major Metropolitan Areas. Vol. I: Measure- ment of the Nature and Amount of Crime: Field Surveys III. President's Commission on Law Enforcement and the Administration of Justice. Washington, D.C.: U.S. Govern- ment Printing Office. 1980 Understanding changes in crime rates. Pp. 11-17 in S. E. Fienberg and A. J. Reiss, Jr., eds., Indicators of Crime and CriminalJus- tice: Quantitative Studies. Washington, D.C.: Bureau of Justice Statistics. Sampson, R. J. 1983 The Neighborhood Context of Criminal Vic- timization. Unpublished Ph.D. thesis. State University of New York, Albany. Ann Arbor, Mich.: University Microfilms. 1985 Neighborhood and crime: the structural de- terminants of personal victimization. Journal of Research in Crime and Delinquency 22:7-40. Sarnecki, J. 1982 Brottslighet och Kamratrelationer: Studie an ungbrottsligheten i en svensk kommun. Rapport 1982:5. Stockholm: Brottsforebyg- gnderadet (National Council for Crime Pre- vention). Portions translated in 1984 by Denise Galarraga as Criminality and Friend Relations: A Study of Juvenile Criminality in a Swedish Community. National Criminal Justice Reference Service. Washington, D.C.: National Institute of Justice. Scott, P. D. 1956 Gangs and delinquent groups in London. British Journal of Delinquency 7:~26. ]59 Shannon, L. W. 1978 A longitudinal study of delinquency and crime. In C. Welford, ea., Quantitative Stud- ies in Criminology. Beverly Hills, Calif.: Sage Publications.  Shapland, J. 1978 Self-reported delinquency in boys aged 11 to 14. British Journal of Criminology 18:25 266. Shaw, C. R. 1938 Brothers in Crime. Chicago, Ill.: University of Chicago Press. Shaw, C. R. and McKay, H. D. 1931 Male juvenile delinquency as group behav- ior. In Report on the Causes of Crime. No. 13. Washington, D.C.: National Commission on Law Observance and the Administration of Justice. Shaw, C. R., and Meyer, E. D. 1929 The juvenile delinquent. In The Illinois Crime Survey. Peoria, Ill.: Illinois Associa- tion for Criminal Justice. Sherman, L. W. 1970 Youth Workers, Police, and the Gangs: Chica- go 1956-1970. Department of Sociology. Un- published M.A. thesis. University of Chicago Short, J. F., Jr., and Moland, J., Jr. 1976 Politics and youth gangs: a follow-up study. Sociological Quarterly 17:162-179. Short, J. F., Jr., and Strodtbeck, F. L. 1965 Group Processes and Delinquency. Chicago, Ill.: University of Chicago Press. Shover, N. 1973 The social organization of burglary. Social Problems 20:499~514. Smith, D. 1986 The neighborhood context of police behav- ior. In A. J. Reiss, Jr., and M. Tonry, eds., Communities and Crime. Chicago, Ill.: Uni- versity of Chicago Press. Smith, M. W. 1972 An Economic Analysis of the Intracity Dis- persion of Criminal Activity. Department of Economics. Unpublished Ph.D. dissertation. North Carolina State University. Sparks, R. F., Greer, A., and Manning, S. A. 1982 Crime As Work: Theoretical Studies. School of Criminal Justice. Final Report to National Institute of Justice, U.S. Department of Jus- tice under Grant # 80-IJ-CX-0060. Newark, N.J.: Rutgers University. Spergel, I. 1964 Racketville, Slumtown and Haulburg. Chi- cago, Ill.: University of Chicago Press. Suttles, G. D. 1968 The Social Order of the Slum. Chicago, Ill.: University of Chicago Press.

60 Sveri, K. 1965 Group activity. Scandinavian Studies in Criminology 1:17~185. Thrasher, F. 1927 The Gang. Chicago, Ill.: University of Chi- cago Press. West, D. J. 1969 Present Conduct and Future Delinquency. London: Heinemann. West, D. J., and Farrington, D. P. 1973 Who Becomes Delinquent? London: lIeine- mann. West, W. G. 1974 Serious Thieves, Lower Class Adolescent Males in a Short-Term Deviant Occupation. Ann Arbor, Mich.: University Microfilms. 1977 Serious thieves: lower class adolescent males in a short-term deviant occupation. In E. Vaz and A. Lodhi, eds., Crime and De- linquency in Canada. Toronto: Prentice- lIall. 1978 The short term careers of serious thieves. Ca- nadian Journal of Criminology 20:169- 190. Wheeler, Stanton 1961 Socialization in correctional communities. CRIMINAL CAREERS AND CAREER CRIMINALS American Sociological Review 26:699- 712. White, PI., Boorman, S. A., and Breiger, R. L. 1976 Social structure from multiple networks. I: blockmodels of roles and positions. Ameri- canJournal of Sociology 81:73~780. Whyte, W. F. 1943 Street Corner Society. Chicago: University of Chicago Press. Wolfgang, M. 1978 Overview of Research into Violent Behavior. Testimony to the Subcommittee on Domes- tic and International Scientific Planning, Analysis, and Cooperation of the House Committee on Science and Technology, 95~ Cong., 2d sess. Wolfgang, M., Figlio, R. M., and Sellin, T. 1972 Delinquency in a Birth Cohort. Chicago, Ill.: University of Chicago Press. Yablonsky, L. 1959 The delinquent gang as a near-group. Social Problems 7:108-117. 1962 The Violent Gang. New York: Macmillan. Zimring, F. E. 1984 Your homicide in New York: a preliminary analysis. Journal of Legal Studies 13:81-99.