Click for next page ( 13


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 12
1 Introduction: Studying Criminal Careers THE CRIMINAL CAREERS CONCEPT Although widely stucliecI, crime is one of the most elusive subjects of social sci- ence research as well as a major public policy issue. It has been addressed by many scholars from a wide variety of disciplines; there has been extensive the- orizing about its etiology; the public has consistently ranker! crime as a serious problem; and there is intense public de- bate, even about facts, when considering alternative policies for dealing with crime. Yet, despite the importance of the issues and the amount of research, little definitive knowledge has been devel- oped that could be applied to prevent crime or to develop efficient policies for reacting to crime. Over the past clecade, however, some significant progress has been made, ant! much of that has focused on the "criminal career" paradigm. A criminal career is the characterization of the longitudinal sequence of crimes committed by an individual offender. A previous National Research Council panel called for "research . . . directed at characterizing the patterns of incliviclual criminal careers" (Blumstein, Cohen, ant! Nagin, 1978:781. This call was motivated in large part by the realization that crime is committed by inclivicluals, even when they organize into groups, and that incli- viduals are the focus of criminal justice decisions. Thus, a paradigm focusing attention on indivicluals might be most appropriate both for probing the causes of criminal behavior and for developing crime control policies intencled to inter- rupt or modify criminal careers. Much research on crime has been fo- cused on aggregate crime rates, that is, crimes per capita in the general popula- tion. The criminal career approach parti- tions the aggregate rate into two primary components: participation, the (listinc- tion between those who commit crime and those who do not; and frequency, the rate of activity of active offenclers. This partition is important not only because the two components may be subject to very different influences, but also be- cause different authorities have different levels of responsibility for control of the two components. In particular, education, social service, and mental health profes 12

OCR for page 12
INTRODUCTION: STUDYING CRIMINAL CAREERS signals all provide services that may re- duce participation in crime, while con- trolling the rate of criminal activity is more central to the decisions of the crim- inal justice system. A decision on impris- onment, for example, often involves some concern for incapacitating a convicted of- fender, and so it is a matter of prime interest to estimate how many crimes in the community might be avoided by the removal of that offender. In addition to the primary components of participation and frequency, two other dimensions that affect aggregate crime rates can be incorporated in the criminal career approach: duration, the length of an individual career (the time from first to last offense), and seriousness, which in- cludes both the offenses committed and patterns of switching among offenses. Separate consideration of participation, frequency, duration, and seriousness pro- vides a finer resolution in the search for the factors associates! with crime than c30 other common approaches. Traclitional research on criminal behavior has often relied on aggregate arrest rates or recidi- vism statistics to develop or to test various causal models. These conventional mea- sures of criminality, however, confound different aspects of individual offending patterns. For example, a drop in the ag- gregate crime rate may reflect a drop in the proportion ofthe population engaging in crime (level of participation), in the average number of crimes committee! by active offenders in a given time (inclivicI- ual frequency), or in the average number of years over which offenders commit crimes (cluration of activity). A causal fac- tor couIcI strongly affect one of these di- mensions, but variation in the others might mask that relationship if the aggre- gate crime rate is used as the primary measure of crime. Thus, as in most re- search, when different factors affect dif- ferent dimensions of a phenomenon of interest, it is extremely important to iso 13 late those dimensions in order to assess the influence of any factor. Different sets of"causes" may influ- ence individuals' decisions to initiate criminal (or delinquent) activity, the fre- quency with which they commit crimes, the types of crimes committed, and their decisions to stop committing crimes. At- tention to these separate dimensions of the criminal career can thus help to refine theories of criminal behavior, since some theoretical explanations may account for the initiation of deviant acts by teenagers, while very different theories may be more relevant to the termination of seri- ous criminality by adults or to fluctuations in rates of offending cluring an active criminal career. Thus, basic knowledge about each component of individual crim- inal careers is fundamental for an under- standing of how various factors and gov- ernment policies may encourage or inhibit criminal activity. For example, past research on the im- pact of unemployment on crime has re- sultec! in inconsistent conclusions. How- ever, the inconsistent results may be a consequence of the use of aggregate crime rates as the depenclent variable. If the relationship between unemployment and criminal behavior were to be studied at the individual level, the effects might become more focused ant! consistent. One might hypothesize, for example, that variation in unemployment is not likely to have much influence on the initiation of criminal activity, which usually occurs in the teenage years, before most people have begun working, but that the oppor- tunity for full-time employment leads to early termination of teenagers criminal activity. Separation of the dimensions of individual careers might permit more precise tests of these hypotheses. Criminal careers may vary substantially among offenders. At one extreme are of- fenclers whose careers consist of only one offense. At the over extreme are "career ,% , . . .

OCR for page 12
14 criminals" also variously characterized as dangerous, habitual, or chronic offend- ers who commit serious offenses with high frequency over extended periods of time. Both of these extreme career types are of policy interest. Offenders with short careers who commit very few of- fenses are of interest for identifying exter- nal factors that foster early termination of careers and that can be affected by public policy. Such knowledge will be espe- cially useful in the design of crime con- trol policies, including prevention achieved through the imposition of sanc- tions for deterrent or rehabilitative pur- poses. Career criminals are of interest for formulating policies that can identify these offenders early in their careers and selectively target criminal justice re- sources at preventing their crimes, pri- marily through incarceration. Individual-level data are fundamental to the analysis of criminal careers. An ideal approach for collecting data on the individual offender would involve get- ting a representative sample of offenders to maintain accurate logs of their daily criminal activity. In light of the obvious difficulties in mounting such an effort, two principal approaches have evolved for estimating the information that might be derived from those logs. One approach uses retrospective self-reports, typified by the work of Short and Nye (1958), West and Farrington (1977), Chaiken and Chaiken (1982a), and Elliott et al. (1983, 1985), which involves asking individuals to report on their offenses over a recent period. The other approach uses official records, typified by the work of Wolfgang, Figlio, and Sellin (1972), Blumstein and Cohen (1979), and Shannon (1982a, b), in which arrests are viewed as a sample of the actual criminal activity that led to the arrests. Each of these approaches relies on different assumptions and introduces different distortions into the resulting de- scriptions of the underlying crime pro cess. CRIMINAL CAREERS AND CAREER CRIMINALS In principle, offender self-reports could provide a complete picture of offending over time, but inaccuracies are intro- duced by respondents' selective nonre- sponse, intentional distortion, and inabil- ity to recall precise counts and sequences of previous crimes. In contrast, an of- fender's official arrest record provides counts and sequences of offenses, but only of those offenses recorded by police. The records therefore reflect only a sam- ple of all the offenses committed by an individual and may occasionally include notations of crimes that did not occur as charged.) As a sample of the underlying crime process, arrest records are distorted by patterns in victims' willingness to re- port crimes to the police, by differential police attention to different crimes, and by police discretion in deciding which suspects to arrest, which arrests to record, and what charges to file. Arrest histories used in research may be incomplete be- cause of failures to forward notations of all arrests or to include arrests recorded in jurisdictions over than the study site. Only recently has attention turned to iIn addition to the distortion that may result from differences between actual offenses and arrests, the issue of false arrests is an important concern when inferring crimes from arrests. It is well established that a majority of arrests fail to end in conviction even for serious crimes. This differential raises questions about the validity of assuming that every arrest is properly associated with a crime and sug- gests using only those arrests that are followed by conviction. Two types of error are involved: using arrests as indicators of crimes probably involves some errors of commission because of false arrests; using only convictions is more likely to involve errors of omission. In adjudicating specific individ- uals, of course, the presumption of innocence makes the error of commission unacceptable. In assessing the relative validity of data for research purposes, however, there must be a relative weighing of these two types of error. Reports by criminal justice prac- titioners indicate that the errors of commission as- sociated with using arrest records are far smaller than the errors of omission that would occur if only convictions were used.

OCR for page 12
INTRODUCTION: STUDYING CRIMINAL CAREERS IS ward synthesis of criminal career data from both self-reports and official records. Because of the limitations on the clata from both sources, moclels are needed to convert the observed information into es- timates of the principal dimensions of criminal careers. Such models have emerged only in the last decade, largely stimulated by the stochastic-process ap- ~ preach introduced by Avi-ItzEak and intensive supervision may well be effec Shinnar (1973), which was followed sev eral years later by empirical estimates of various career dimensions. Later, atten tion shifted to modeling the distributions of those dimensions and identifying their covariates (Chaiken and Chaiken, 1982a; Greenwood, 1982~. This panel has con tributed to that body of work by commis sioning extensions of the modeling from the perspectives of economics (Flinn, Volume II) and stochastic processes (Lehoczky, Volume II). CRIME CONTROL POLICIES The criminal justice system's efforts to control crime take three forms: deter rence, rehabilitation, ant! incapacitation. Deterrence is the symbolic threat broad cast to actual and potential offenders by imposing punishment on identified of fenders. Incapacitation is the removal of a convicted offender from the community, usually through imprisonment, to prevent the offender from committing further crimes. Rehabilitation is the modification of an offender's criminal behavior. These efforts to control crime are carried out in conjunction with efforts to achieve other goals of the criminal justice system, such as imposing"deserved" punishment, en hancing public confidence in the justice system, and maintaining order in penal facilities. Knowledge for Policy Knowledge about criminal careers may be especially helpful in cleveloping effec tive crime control policies. The appropri- ate response to crime will differ depend- ing on whether the aggregate crime rate is the result of a small group of high-fre- quency offenders or a large group of of- fenders who commit crimes infrequently. In the former situation- Tow participation combined with high individual fre- auenov strategies of incapacitation or five and feasible. But if participation is high, incapacitating many offenders would be impractical, and other crime control strategies involving efforts to pre- vent participation wouIcl be more effec- tive and feasible. Distinguishing among the various di- mensions of criminal careers may also serve to enhance the crime control poten- tial of alternative strategies by more effec- tively targeting various crime control ef- forts. The effectiveness of incapacitation as a crime control strategy, for example, depends not only on how frequently of- fenders commit serious crimes, but also on the duration of an offender's career. From the perspective of incapacitation, prison capacity is used inefficiently if of- fenders are imprisoned beyond the time at which their criminal activity would have terminated if they were free on the street. Therefore, it is reasonable to ask whether "habitual-offender" laws, which mandate very Tong sentences, may result in incarceration of offenders well after they have ceases! to be serious risks. Effective policy strategies may also emerge from knowledge about individual offending patterns and the relationship of different attributes of offenders to the di- mensions of their criminal careers. For example, early indicators of likely partic- ipation in criminal activity, such as edu- cational difficulties or disruptive school behavior, suggest preventive strategies like preschool education or family-orien- ted treatment, which may reduce partici- pation in crime. In contrast, career modi- fication or rehabilitation strategies seek to

OCR for page 12
16 terminate or reduce criminal activity through counseling, group therapy, job training, or similar programs. If heavy drinking and drug use are found dispro- portionately among high-rate offenders, intervention strategies might focus on treatment for substance abuse. If fre- quent, serious offenders can be identified by using characteristics of the offense or the offender, the criminal justice system could use incapacitation strategies effec- tively. Knowledge of the factors that are asso- ciated with variations in criminal careers can also be helpful in assessing existing policies and practices. For example, if a particular criminal justice system rou- tinely imposes long sentences on offend- ers who commit serious crimes at a high rate, then it is already enhancing crime control by using incarceration to incapac- itate offenders. But if incapacitation deci- sions implicitly invoke characteristics that are irrelevant to offenders' careers or use characteristics inappropriately-such as imposing long sentences on all young black offenders then wicler use of career knowledge could contribute to incapac- itative effectiveness. At the aggregate level, relationships between criminal careers and demo- graphic variables can be used to forecast changes in crime volume or in the size and composition of prison populations arising from demographic changes in the general population. When combined with data on criminal justice system perform- ance (e.g., arrest probabilities, judge's time required per case, conviction-to- arrest ratios, average time served per incarceration), criminal justice system re- source requirements can also be pro- jected. Knowledge of crime-specific of- fending frequencies and their correlates could also be user] to estimate the inca- pacitative crime-reduction effects of alter- native incarceration policies. Thus, in a number of ways, knowledge of criminal CRIMINAL CAREERS AND CAREER CRIMINALS careers and their correlates can refine the analysis of criminal justice policies. Classification of Offenders Selective treatment of offenders in the criminal justice system is one way to direct criminal justice discretion more ef- fectively toward reducing the level of crime. Recent research findings have re- kindlecl interest in basing selection ex- plicitly on classifications derived from empirically demonstrated relationships between specific variables and criminal career dimensions, especially the fre- quency of serious criminal activity. In various ways, such information has long been used as a selection criterion: in formal ways in parole release decisions and in structuring prosecution priorities for "career criminal units" and more often informally in setting pretrial release con- ditions and sentencing. The recently dis- cussed crime control strategy of"selec- tive incapacitation" at sentencing (e.g., Greenwood, 1982) is based on selection rules defined explicitly in terms of fre- quency classifications. More generally, classifications defined in terms of avail- able, reliable, and legally usable informa- tion about offenders and their offenses could be used as one of the bases for selection at such criminal justice decision points as arrest, pretrial release, prosecu- tion, sentencing, and parole. Selective treatment involves implicit or explicit use of classifications related to the frequency and duration of serious offending. Since those career dimensions cannot be directly observed for individu- als in the context of high-volume criminal justice decision making, offender cIassifi- cations are operationally defined in terms of other variables that are routinely mea- surable and that are correlatecl with Me frequency, seriousness, or duration of the criminal career. Criminal justice process- ing based on these classifications is, in

OCR for page 12
INTRODUCTION: STUDYING CRIMINAL CAREERS turn, subject to statutory anc! constitu- tional constraints. Within these con- straints, classification and selection rules involve articulating a balance between the competing objectives of protecting the community against harm by offenders anc! protecting the presumer! offender from unjust treatment by the state. Ethical Considerations Invoking precliction-basec] cIassifica . . . . . . . tons In cr~m~na Justice c recisions raises a number of important ethical concerns. When, if ever, is it appropriate to invoke preclictions of future criminal activity in criminal justice decisions? Is a difference in preclictec! future criminal activity a legitimate basis for imposing different sanctions for the same crime? What level of predictive accuracy is necessary to jus- tify using predictions? What balance is appropriate between the harm clone to offenders anc! that clone to the community by different types of prediction errors? ShouIc! some effective predictor variables be explicitly ignorer! for ethical reasons? These ethical concerns have a direct bearing on the policy uses of scientific knowledge about criminal careers. For this reason, their implications for criminal justice policies are important consider- ations in both Chapters 5 and 6, in which Me crime control uses of criminal career knowledge and the efficacy of precliction- based decisions are examined. Many of the ethical issues considered are matters of long-stancling legal and philosophical debate. The resolution of some issues may rest on absolute princi- ples, but more often it involves a balanc- ing of competing ethical concerns. It is thus reasonable to anticipate that the principles governing the use of predic- tion-based classification rules as an aid to criminal justice decision makers will evolve, changing as public concerns about crime ebb ant! flow ant! as im 17 proved knowledge about inclividual crim- inal careers is developed. DIMENSIONS OF CRIMINAL CAREERS The panel's characterization of crimi- nal careers partitions crime into four di- mensions: one dimension describes the fraction of a population becoming offencI- ers by virtue of participation in crime; three others describe active offenders in terms of the frequency, seriousness, and duration of their activity. The criminal career paradigm presumes that offending is not pervasive throughout a population, but is restricted to a subset of the popula- tion. This subset consists of active of- fenders those who commit at least one crime during some observation period. It also presumes that the composition of active offenders varies over time as some criminal careers are initiated and others terminate. Participation The measurement of participation the fraction of a population that is criminally active depends on the scope of criminal acts considered and the length of obser- vation periods.2 Including minor infrac- tions in the scope of crime types greatly increases the level of participation in of 2In the literature reviewed by the panel, the term "prevalence" is often used to refer to the current involvement of some portion of a sample in criminal activity, but occasionally to other measures of par- ticipation. The term "incidence" is usually used to refer to the per capita crime rate for the sample but occasionally to refer to individual frequency. In short, these terms have been used inconsistently by many researchers for decades. To avoid confusion, the panel adopted the vocabulary being introduced here. The relationship between the panel's frame- work and other individual and aggregate statistics on criminal behavior is discussed in the section, "Basic Definitions and Symbols."

OCR for page 12
18 fencling in a population. Longer obser- vation periods also increase participa- tion measures, since more offenders who commit offenses only rarely will be in- cluded as will more offenders who are initiating or terminating their criminal activity. Conversely, restricting the crite- ria to focus only on serious offenses and using short observation periods will re- sult in lower measures of current partici- pation. In any observation period, active of- fenders include both new offenders whose first offense occurs cluring the ob- servation period and persisting offenders who began criminal activity in an earlier period and continue to be active during the observation period. Participation in any observation period thus depends on the number of indivicluals who become offenders ant! on how long offenders re- main active. The longer the duration of offending, the greater the contribution of persisters to measured participation in successive observation periods. Individual Frequency Rates, Seriousness, and Duration Individual frequency rates the num- oer of crimes per year per active of- fender vary substantially among offend- ers, with some having very high rates and others low rates of offending. For any indiviclual, frequency rates may vary over time. Because they commit more crimes per unit of time, high-rate offenders con- tribute clisproportionately to the total measured number of crimes. Many different offense types may con- tribute to an individuals frequency rate. The scope of offending for individual of- fenders may vary from "specialists" (who engage predominantly in only one of- fense or a group of closely relater! offense types) to "generalists" (who engage in CRIMINAL CAREERS AND CAREER CRIMINALS a wide variety of offense types). The degree of specialization may also vary across offense types, with some offense types committed predominantly by spe- ciaTists ant] other types routinely commit- ted by generalists. The mix of offense types committed by offenders may also vary over the course of their careers, with offenders becoming either more or less speciaTizecI, and with the mix of offense types escalating or de-escalating in seri ousness. The duration of criminal careers can also be expected to vary across inclividual offenders. It is likely that many criminal careers are very short, ending in the teen- age years. However, some offenders con- tinue to commit crimes into their 30s and even older ages. Thus, it is important to understand average total career length and the factors that distinguish offenders with Tong careers from offenders with short careers. But to understand the im- pact of decisions made about individual offenders, it is also extremely important to understand residual career length, the expected time remaining in an offender's criminal career at the time of the deci sion. Basic Definitions and Symbols To use these concepts to organize re- search results that have been developed in a variety of paradigms, the panel found it necessary to adopt consistent vocabu- lary and symbols for labeling the key dimensions that characterize individual criminal careers. This need arises from a confusion in the literature over specific terminology for individual and aggregate crime data. The confusion can be seen clearly, for example, in the variety of referents of the tend "rate" in the litera- ture: to a population (e.g., prevalence rate, the fraction of mates ever participat

OCR for page 12
INTRODUCTION: STUDYING CRIMINAL CAREERS ing in crime); to a time period (e.g., fre- quency rate, the inclivicluaT's annual frequency of offending); or to both simul- taneously (e.g., crime rate, the annual crime index per capita).3 The symbols and definitions adopted by the panel are displayed in Table 1-1. Symbols in the seconc! column describe the crime process, i.e., the actual offend- ing process; symbols in the third column clescnbe the arrest process from which 3Crime itself is a heterogeneous phenomenon, and a variety of terminology has developed to dis- tinguish more serious or less serious crimes. This report is primarily concerned with serious crimes, which are classified herein in several ways. Per- sonal crimes encompass homicide, aggravated as- sault, and rape. Rape and homicide are relatively infrequent, and data are not usually reported sepa- rately by researchers for these crimes. The FBI includes these three crimes and robbery in their definition of violent crimes. Safety crimes include violent crimes and burglary. The five safety crimes plus larceny and auto theft are frequently referred to as index crimes, because they are included in the Part I crime index reported in the FBI's annual Uniform Crime Reports. In 1981, arson was added to the FBI's crime index; however, because report- ing of arson was sporadic before that year, it is rarely included in the research reviewed here. Property crimes typically include burglary, larceny, and auto theft. In addition to the Part I index crimes, the FBI also uses the category of Part II crimes, which are recorded separately, are less often reported to police (e.g., white-collar crimes), are less serious (e.g., public order crimes), and include "victimless" crimes (e.g., prostitution, drug use). Some research- ers also distinguish between a felony and a misde- meanor: the former is usually defined in statutes as any crime carrying a sentence length of at least 1 year. Delinquency has a less precise definition: it can refer to traditional youth crimes, such as truancy or underage drinking (sometimes called status of- fenses because they are only offenses as a result of the status, i.e., age, of the offender), or it can refer to any crime committed by someone under the age of majority. Virtually all the research discussed in this report is on male offenders; hence, the pronoun "he" is used exclusively in referring to offenders. 19 much of the available information about the unclerlying crime process is clerived. We begin with the aggregate annual crime rate per capita (C) that is normally reported to the public in the press, and partition it into its component parts: one that focuses on the inclividual offender and his individual offending frequency measurer! in terms of crimes per year, A, and another that describes the current participation rate of offenders in the pop- ulation at any time All. These three key variables are linked by the relationship C = Acl. Thus, for example, an aggregate crime rate of 1,000 crimes in a population base of 100,000 people could be a conse- quence of 10 criminals each committing an average of 100 crimes (d = 10/100,000, and A = 100) or 100 criminals committing 10 crimes each (c! = 100/100,000, and A = 101. The two situations present different problems in the clevelopment of crime control strategies. Much of the literature is not concerned! with the current rate of participation in crime (cl) but with the question of what fraction of a population has ever been involved in crime. This is simply the accumulation of new participants over time. We denote cumulative participa- tion by the relater! symbol D. (For further discussion of definitional issues relating to participation, see Appendix A.) Comparable symbols are cleaned for the arrest process, which is often the only source of information about inclividual criminal careers. The counterpart to the crime rate (C) is the number of arrests per capita per year, A. The average number of arrests experienced per year by active offenders is denoted by A. The percent of a population that is arrested (or "busted") within some observation period is de- noted by b. A relationship similar to that in the crime process exists for the arrest process, A = ,ub. Finally, as an analogue

OCR for page 12
20 CRIMINAL CAREERS AND CAREER CRIMINALS TABLE 1-1 Glossary of Basic Symbols and Relationships . Measure Measure Based on Crimes and Arrests Aggregate crime rate per capita per year Individual frequency per active offender Current participation rate Cumulative participation rate Aggregate crime rate per capita: frequency per active offender times participation rate Symbol and Definition Crime Process C Crimes per capita per year A Crimes per year per active of- fender d Percent of a population com- mitting a crime within a year ,``, . .. ~ doing D Percent of a population ever committing a crime C = Ad Arrest Process A Arrests per capita per year ,u Arrests per year per active of- fender b Percent of a population ar- rested for crimes within a year ("busted") B Percent of a population ever arrested A = ,ub Other Measures Arrest probability q = Probability of arrest following a crime (q = ,u/A) Career length T = Total criminal career length Residual career length TR = Average time remaining in a criminal career Career dropout rate ~ = Fraction of a criminal population whose careers terminate during an observation period to the cumulative participation rate in crime, we denote the fraction of a popu- lation that is ever arrested as B.4 The arrest process can be viewed as a sampling from the crime process, since not all crimes result in arrest. This linkage is reflected in the "sampling probability" or the probability that a crime will result in an arrest, denoted by q. The two key individual rates, ,u and A, are thus linked through q by the relationship ,u = Act. Thus, for example, if an offender commits crimes per year and there is a 10 percent chance that commission of a crime will lead to an arrest, then the average arrest rate, A, should be 840.1) or 0.8 arrests per year. The final principal construct for which we use symbols is the total criminal ca 4Many studies use an alternative official-record indicator of an event, such as referral to juvenile court or conviction. For notational simplicity, we symbolize fractions involving all these events by B and b, but recognize in discussion the implications of these more stringent thresholds for empirical results. reer length, T. denoting the number of years over which an offender is criminally active. In many cases, we are more inter- ested in the residual career length, the number offuture years a currently active offender is expected to remain active, and we denote this length by TR. The fraction of careers that terminate during an obser- vation period is denotecl by the career dropout rate, B. Obviously, estimates of the criminal ca- reer dimensions represented by these sym- bols will vary by crime type, across dig ferent population groups, and in different settings. In the text of the report, we occa- sionally distinguish these different varia- tions of the basic criminal career dimen- sions. The glossary presented in Table 1-1 should therefore be helpful as a continuing reference in reading that material. Basic Mode} of a Criminal Career The basic criminal career paradigm pro- vides a framework for organizing knowI- edge about the dimensions that describe

OCR for page 12
INTRODUCTION: STUDYING CRIMINAL CAREERS ~ ~ Criminal Am_ Events a, a: ._ C' ~5 ._ ._ X- - 43)= Arrest ~ = .~ = I Incarceration FIGURE 1-1 An individual criminal career. individual criminal behavior. It permits specification of relationships among the dimensions and computation of statistics that describe offending in an observed sample, such as a cohort of individuals followed over time, a cross-section of ac- tive offenders arrested during some time interval, or matched controVexperimental groups being studied to assess the impact of some intervention. Figure 1-1 presents a highly simplified framework that introduces the essential concepts of the criminal career. The top of the figure represents a sequence of events during the criminal career of an active offender. On the line, the symbol x denotes the times at which the offender committed crimes. Symbols of crimes for which the offender was arrested are cir 21 Period of I ncarceration Crime aO ~Criminal Career Length T Conviction , Age of Onset I,,,-------i_ Residual Criminal-Career Length Following Release, TR End of I Criminal Career ~ AGE aT cled, and the crimes for which the arrest led to conviction are enclosed in a square. The shaded area indicates a period when the offender was incarcerated following conviction. In theory, all incidents marked x could be reported by the of- fender in a self-report survey, while only the circled ones could appear in an of- ficial arrest record. A person initiates criminal activity at some time. That first offense may involve a conscious personal choice, it may follow from the development of a new set of associations developed, or it may be an inadvertent consequence of other changes in the individuaT's life. Once the offender has begun his criminal involve- ment, it continues for some period of time, perhaps increasing or decreasing in

OCR for page 12
22 frequency. Finally, the person terminates his criminal career, possibly because of death, but more typically at a relatively young age, after which his probability of offending within any observation period is small enough to be ignored. The minimum representation, which clearly omits many of the complexities of a real career, is represented in the lower portion of Figure 1-1. The offender is assumed to begin criminal activity at some "age of onset," aO, but his official record does not reflect onset until some later time, at the point of his first arrest. Once begun, the offender continues to commit crimes at a constant rate A during any time that he is not incarcerated. The career ends when the last crime is com- mitted, represented at age aT in Figure 1-1. The representation in Figure 1-1 in- vokes three primary elements of informa- tion: the frequency or mean individual crime rate, A, the age at career initiation, and the duration of the criminal career, T. Each of these three dimensions of a ca- reer varies across offenders. This varia- tion may be influenced by personal events associated with the individual or by broader forces such as sanction levels or other community characteristics. Extensions of the Basic Mode! This simple representation can be ex- tended to more richly describe a criminal career. Possibilities include a"start-up" time during which an offender's fre- quency increases, a decrease in fre- quency toward the end of a career, and sporadic spurts or intermittent recesses from criminal activity. If the intervals be- tween spurts are much shorter than a typical sentence, the intermittent pattern may be ignored and the average A used as a consideration in setting sentencing pol- icy. But if the intervals of inactivity are long, then separate estimates are needed CRIMINAL CAREERS AND CAREER CRIMINALS for high- and low-rate periods and for the duration of these periods, to adequately estimate average A. In addition, spurts in activity make it difficult to obtain informa- tion from offenders for use in estimating average annual rates. Distinctions among different offense types can also be made, permitting attention to single offenses (e.g., "robbery careers") or to patterns of switching among offense types during a career. In this context, it is important to know whether offenders are more likely to be "specialists" (who engage in only one or a small group of offenses) or "gen- eralists" (who switch more widely among a range of offenses). Last, extensions of the basic mode] can address whether of- fending patterns typically "escalate" in the seriousness of successive events so that crimes later in the career are more serious, or whether they peak in serious- ness in mid-career and then begin to ^~ v ~ ^. ,,, ~ _ _ decline in seriousness as a career nears its end. USING THE CRIMINAL CAREER PARADIGM Interpreting Aggregate Crime Rates The criminal career paradigm adopted here represents a departure from analyses that have focused on aggregate measures, such as incidence rates of crimes per capita or arrests per capita. Applying to active offenders and nonoffenders alike in a population, these aggregate measures confound the combined contribution of the extent of participation and the fre- quency of offending by active offenders. Despite this confounding of different as- pects of individual offending, variations in aggregate measures have served as the basis for much of the broadly accepted current knowledge on the causes and cor- relates of crime. Perhaps the most widely accepted view of crime is that it varies substantially

OCR for page 12
INTRODUCTION: STUDYING CRIMINAL CAREERS 100 90 80 . . o ~ A, 70 ~ Cal Q ~ O ~ 60 a, O ~ 8< 50 o Y a, a, ~40 Q o In 4' 4 - ~ ~0 20 10 Age 1 7A~ "Age 21 il 11. 11. i\ \1\ \\ \\ Age 21 \ Age Z ~ `\ Peak Arrest Rates {arrests Offense per 100,000 Type population) Robbery ---Burglary ~ ~ Aggravated : assault . Ace 24 G'. Age 36 "art... O 1 ~1 ,1 317 1 ,1 26 340 a, 0 10 20 30 40 50 60 Age (years) FIGURE 1-2 1983 U.S. age-specific arrest rates (arrests per 100,000 population of each age). The curve for each offense type is displayed as a percentage of the peak arrest rate. The curves show the age at which the peak occurs (at 100 percent) and the age at which the rate falls to 50 percent ofthe peak rate. Source: Federal Bureau of Investigation (1984~. with age. While varying in absolute mag- nitude, aggregate population arrest rates display a very consistent pattern in- creasing rapidly during the juvenile years to reach peak rates in the late teens and then steadily declining.S Figure 1-2 illus- trates this pattern for the FBI index of- fenses of robbery, aggravated assault, and burglary in 1983. Analyses of recidivism rates for identified offenders have also Historically, arrest rates peaked in an offender's early twenties. This pattern was observed as early as 1831 in France (Quetelet, 1984~; recent reviews of this research can be found in Greenberg (1983), Hirschi and Gottfredson (1983), and Farrington (1986~. 23 found a decline in criminal involvement for older offenders.6 The declines in of- fending observed in aggregate data and recidivism measures have been charac- terized as "maturing out of crime" and have been attributed to physiological and social changes with age that lead to grad- ual reductions in criminal involvement. These analyses, however, do not tell the full story. The distinctive age patterns 6See, for example, Glueck and Glueck (1937, 1940), Sellin (1958), and more recently, studies of differential success on parole release (e.g., Hoffman and Beck, 1980; Rhodes et al., 1982; Bureau of Justice Statistics, 1984c).

OCR for page 12
24 in aggregate measures may be due either to changes in participation or in individ- ual frequency rates for active offenders. In the former case, the peak rates of crim- inal activity would result from growing participation in crime during the late teen years, followed by declining participation as increasing numbers of offenders end their criminal careers. In the latter case, peak rates would arise from variations in the intensity of offending by a fairly fixed group of active offenders, with indiviclu- als' frequency rates increasing during the juvenile years and then gradually declin- ing with age. Distinguishing between these alterna- tive processes has both theoretical signif- icance for understanding the causes of crime, as well as operational implications for efforts to control crime. From the the- oretical perspective, the participation and frequency alternatives have implications for whether causes should be sought in broad social processes affecting the gen- eral population and people's movement into and out of criminal careers or in more CRIMINAL CAREERS AND CAREER CRIMINALS arrest rates also show large differences in levels of criminal activity between whites and blacks. As shown in Table 1-2, the differences between races increase as one focuses on more serious offense types, and have decreased in recent years. How- ever, large differences between races are generally not fount! when recidivism rates for black and white offenders are compared (Wolfgang, Figlio, and Sellin, 1972:28~289; Blumstein and Graddy, 1982:28~284; Bureau of Justice Statis- tics, 19846:Table 10~. The differences in aggregate measures by sex and race re- flect the combined contributions of differ- ences in levels of participation in crime and differences in frequency for active offenders~istinct phenomena with very different policy implications. Demographic Correlates of Criminal Careers Demographic variables have received considerable attention, primarily because they are widely available in data on the isolated processes affecting only active general population, are easily observed offenders and their activity patterns. ' From a policy perspective, distinguishing between these alternatives has implica- tions for Me relative effectiveness of ef- forts to reduce peak levels of offending in the late teens by preventing participation by new offenders or by targeting inter- vention more narrowly at already active offenders. Similar concerns about disentangling the relative contributions of participation, frequency, and duration emerge when considering other offender attributes. The sex of offenders is potentially an impor- tant factor in characterizing individual offending. Very large differences be- tween mates and females are observed both in aggregate arrest rates (see Table 1-2) and in recidivism rates (Bureau of Justice Statistics, 19846), with substan- tially higher rates for males. Aggregate by crime Scums ana police, ana are rou- tinely recorded for identification pur- poses in data from administrative and operational agencies. Data on dimensions of criminal careers are therefore wiclely available for different demographic sub- ~rouns defined in terms of age, sex, and race. The strong empirical associations observed between the demographic vari- ables and aggregate arrest rates (see Ta- ble 1-2 and Figure 1-2, above) have gen- eratec3 substantial debate about the causes of subpopulation differences. But much ambiguity surrounds the underly- ing theoretical meaning of differences in criminal behavior between mates and fe- maTes (see PolIack, 1950; Adler, 1975; Nagel and Hagan, 1984), among racial and ethnic groups (e.g., Moynihan, 1965; Berger and Simon, 1974), and most re- cently, across age (Greenberg, 1983;

OCR for page 12
INTRODUCTION: STUDYING CRIMINAL CAREERS TABLE 1-2 U.S. Sex- andRace-SpecificArrestRates in 1970 and 1980 25 Offense Type Arrests per 1,000 Population Ratiosa Sex-Specific Arrest Rates Males 1970 1980 Females 1970 1980 Male/Female 1970 1980 All (except traffic) 76 81 12 14 6.3 5.6 All index 14 17 3 4 5.2 4.6 PropertyC 11 14 2 3 4.6 4.0 Violent 2 3 0.2 0.3 8.0 8.3 Robbery 1 1 0.1 0.1 16.2 13.6 Race-Specific Arrest Rates Whites 1970 1980 Blacks 1970 1980 Black/White 1970 1980 All (except traffic) 35 40 107 97 3.1 2.4 All index 6 8 28 29 4.7 3.7 Property 5 7 20 21 4.0 3.2 Violent 0.6 1 4 5 7.3 4.6 Robbery 0.2 0.3 3 3 15.4 10.3 NOTE: Rates are estimated from the number of reported arrests in 1980 (Federal Bureau of Investigation, 1981) and 1980 population figures (Bureau of the Census, 1983:Table 33~. Similar data are available for adults and juveniles separately. Arrest rates for 1970 are estimated from the number of arrests reported in 1970 (Federal Bureau of Investigation, 1971) and 1970 population figures (Bureau of the Census, 1983:Table 331. To adjust for agencies not reporting to the FBI, reported arrests are increased by the ratio of total population available from the Bureau of the Census to the population covered by reporting agencies. This assumes that the arrest rates and population distribution in nonreporting agencies are similar to those available for reporting agencies. aThese ratios were computed before the arrest rates were rounded. bIndex rates include arrests for murder, rape, robbery, aggravated assault, burglary, larceny, and motor vehicle theft. CProperty rates include arrests for burglary, larceny, and motor vehicle theft. Violent rates include arrests for murder, rape, and aggravated assault. Hirschi and GottEredson, 1984; Green- berg, 1985; Wilson anal Herrnstein, 1985; Farrington, 1986~. While demographic differences ac- count for large portions of the variability in aggregate measures of criminal in- volvement, these differentials reflect rela- tionships with other variables that are not yet well understood. For example, over the years, explanations offered for the differences between males and females have variously been biological differ- ences, differences in moral training, dif- ferences in socialization experiences, and fewer criminal opportunities for girls be- cause they are more closely supervised. These changes over time in interpreta tion of demographic differentials reflect the changing social contexts in which this research has been carried out. Recently, Peterson and Hagan (1984) argued that criminal justice research on differences between racial and ethnic groups should be viewer! in an historical context. In the first half of this century, American research on criminal behavior focused on the criminality of newly ar- rivecI ethnic groups, the Irish ant! Italian immigrants in large urban areas (e.g., Glueck and Glueck, 19401. Researchers associated high crime rates among urban ethnic groups with a vast array of social problems, including Tow-paying jobs, poor housing, and weak ties to the "dom

OCR for page 12
26 inant" culture. Slowly the emphasis in racial/ethnic studies of criminal behavior has shifted to comparisons of white and black Americans. This shift in attention probably reflects the changing demo- graphic composition in large cities from a variety of ethnic minorities to blacks as the dominant minority and the most re- cent arrivals to large urban areas. Even more recently, other racial/ethnic groups- Hispanics and Southeast Asians have begun to receive special attention (see Moore et al., 1978; Zatz, 1984; LaFree, 1985). The pane! did not attempt to resolve the theoretical debates concerning age, sex, and racial/ethnic differences in crim- inal behavior. Empirical relationships are reported because so much of the research results are presented in these terms, and because the differences are often large in bivariate comparisons, robust in multi- variate analyses, and stable across a vari- ety of geographical and temporal settings and data collection methods. However, by allowing for different causal structures for the initiation, persistence, and termi- nation of offending, the partitioning ofthe aggregate differentials into separate ca- reer dimensions should facilitate better theoretical interpretation. As suggested above, one finding of research invoking the criminal career paradigm has been that the demographic differentials arise more from differences in criminal partic- ipation patterns, rather than in the fre- quency or duration of individual offend- ing. Incapacitation, Rehabilitation, and Deterrence Analyses that partition the effects of sanctions among participation, frequency, and duration may provide valuable in- sights for improving the effectiveness of alternative crime control strategies. To date, for example, evaluations of deter- rent effects have relied almost exclusively on aggregate crime rates, and evaluations CRIMINAL CAREERS AND CAREER CRIMINALS of rehabilitative interventions have relied primarily on recidivism rates. To the ex- tent that participation, duration, and fre- quency are differentially affected by de- terrence, rehabilitation, or incapacitation policies, important relationships may be obscured in aggregate measures. Analy- ses of the separate effects may provide valuable insights for improving the crime control effectiveness of deterrence and rehabilitation policies. Crime control ef- fectiveness may be improved by targeting some strategies at reducing participation, very different ones at encouraging career termination among active offenders, and still others at reducing frequency among active offenders. Incapacitation, which is usually a- chieved by incarcerating active offenders, is frequently presented in terms of the criminal career paradigm. In Figure 1-3, the shaded area indicating crimes pre- vented during the period of incarceration represents the "incapacitative effect." Crime control effects through incapacita- tion increase with the magnitude of indi- vidual frequency, with the length of in- carceration, and with the expected duration of the criminal career. More spe- cifically, higher frequency means more crimes averted for each unit of time incar- cerated. And longer career duration means less likelihood of wasting incarcer- ation on offenders who would have ended their careers during the time they were incarcerated and would therefore not be committing any additional crimes whether incarcerated or not. The incapacitative effect actually achieved will depend on the effective- ness of the criminal justice system in identifying and incarcerating offenders, especially those with the highest rates of offending. The incapacitative effect is re- duced if criminals are not arrested, if arrested criminals are not convicted, if convicted criminals are not incarcerated, if sentences are short, or if parole is early. The incapacitative effect is further re- duced if the crimes of an incarcerated

OCR for page 12
INTRODUCTION: STUDYING CRIMINAL CAREERS _y~ Criminal Events ~-~ . _ Cat ~5 ._ ._ - X = Crime By= Arrest ma= Conviction ~ = I Incarceration Period of I Incarceration I Incapacitation Effect Age of Onset C."'-2.--."2. Hi-..''....... ,.... .~ .2.. f ~ So Residual Crimin al Career _; T Length Following Release, TR Criminal Career Length, T Criminogenic Effect Rehabilitation Effect End of Criminal Career - AGE FIGURE 1-3 Incapacitation, rehabilitation, and criminogenic effects in an individual criminal career. offender are replaced by crimes by other offenders. This might occur, for example, if the offender is part of an organized illegal economic activity, like drug sales or burglaries organized by a fence; in this event, a replacement might simply be recruited from the available "labor mar- ket" to continue the crimes that would otherwise be committed by the incarcer- ated offender. If the offender is part of a crime-committing group, the remaining members of the group might continue their crimes, with or without recruiting a replacement. Group offending ant! its im- plications for crime control are acIdressect by Reiss (Volume II). The criminal career construct dis- cussed in this section is most directly relater! to incapacitation, with its funcla- mental objective of interrupting a crimi 27 nal career, but it also relates (Erectly to the other two modes of crime control user! by the criminal justice system, reha- bilitation and deterrence. In most evalu- ations of rehabilitation programs, effec- tiveness is measurer! in terms of the recidivism rate. Recidivism is usually measured as the fraction of a release group that is rearrested, usually within an observation period! of 1 to 5 years.7 If the 7Recidivism is often also measured by reconvic- tion or recommitment to an institution. Indeed, there is considerable debate over the "true" value of recidivism, which is unwarranted since measure- ment at these different stages will provide different numerical values, but usually not different relative values when all measure the same process (see Blumstein and Larson, 1971~. This discussion fo- cuses on arrest, but it applies equally to any other recidivism event.

OCR for page 12
28 offender depicted in Figure 1-3 received effective rehabilitative treatment in prison, its effect would be reflected as a change in his criminal career: as a reduc- tion in his offending rate, as a shift to less serious crimes, or as a reduction in his career length. The crimes avoided by rehabilitation are depicted in Figure 1-3 by the dark gray area, "rehabilitation ef- fect." These crimes are averted in adcli- tion to those averted through incapaci- tative effects during the period of impris- onment. However, incarceration can also be counterproductive, criminogenic, if it leacis to increases in postrelease criminal- ity. This effect could be reflected by an increase in frequency or career length, depicted by the dotted area labeled "criminogenic effect," in Figure 1-3, or by an increase in the seriousness of crimes committed.8 A criminogenic effect may result from an offender's enhanced identification as a"criminal," his learning new criminal techniques from fellow prisoners, or his strengthening ties to other offenders. The net rehabilitation effect is the rehabilitation effect less the criminogenic effect. No empirical com- parison of these two effects has yet been carried out in the criminal career context. However, major reviews of evaluations of offender rehabilitation programs suggest that the net aggregate rehabilitation and criminogenic effect is small (Lipton, Martinson, and Wilks, 1975; Sechrest, White, ancl Brown, 19791.9 8A special case ofthis effect would be a postpone- ment of the offender's criminal career, so that the residual career length after release assumes the same value it had when incarceration began. Thus, the delay in the career is equal to the period of incarceration, and the crimes avoided through inca- pacitation are exactly offset by the same number of crimes committed after the career would otherwise have ended. If this effect occurs, incarceration would delay crimes, but not reduce their number. 9Sechrest, White, and Brown (1979:34) caution that the poor quality of the evaluations, and the CRIMINAL CAREERS AND CAREER CRIMINALS The third principal mode of crime con- troT by the criminal justice system, in addition to incapacitation and rehabilita- tion, is deterrence. Deterrence is the crime-reduction effect achieved from the symbolic threat communicated to other offenders and to potential offenders by sanctions imposed on identified offend- ers.~ The most common mode for exam- ining the deterrent effect has been cross- sectional studies of jurisdictions with diverse sanction practices to examine the effect on aggregate crime rates of the sanction variation (controlling for other sources of variation in crime rate and in sanction policy). This research has yet to provide good estimates of the magnitude of deterrent effects (see Blumstein, Cohen, and Nagin, 1978~. To determine the extent to which the deterrent effects of sanctions work by inhibiting career initiation, decreasing individual offend- ing frequency, or encouraging career ter- mination, it would be desirable for deter- rence research to focus more specifically on the effects of community sanction lev- els on participation, and on the careers of active offenders. SCOPE OF THE PANEL'S REPORT The above discussion has presented the basic structural concepts involved in the criminal career paradigm. Aside from narrow range of options explored, militate against excessive pessimism concerning the potential of rehabilitation. But despite these qualifications, the failure of hundreds of evaluations to find demonstra- ble effects suggests that, given the current state of research methodology and correctional manage- ment, the net effect is fairly small. iThis effect is often characterized as "general deterrence" to distinguish it from "special deter- rence," which refers to the effect of punishment on later behavior of the person punished. In this report special deterrence is considered an aspect of reha- bilitation and the term deterrence refers to the broader crime-reduction effects achieved through general sanction policies.

OCR for page 12
INTRODUCTION: STUDYING CRIMINAL CAREERS its intrinsic value, knowlecige about the dimensions of criminal careers, their clis- tribution in the population, ant] the fac- tors that affect them, is important for a variety of policy uses: identifying variables associates! with the most serious offenders (in terms of their criminal careers) so that such infor- mation may be user! by decision makers, within legal ant! ethical constraints, to anticipate future criminal activity by an offender about whom they must make a . ~ processing c .eclslon; identifying variables that are wiclely but erroneously viewer] as predictors of offenders' future criminal activity; improving identification of high-risk offenders ant] designing programs likely to be effective for them; better assessment ofthe magnitude of incapacitative effects uncler current or proposer! imprisonment policies, possi- bly leacling to more efficient use of lim- itec! prison space; ant] planning research programs that will built! on existing knowlecige ant! provide more effective policy directions over the next clecacle. In its work, the panel focuses] primar- ily, though not exclusively, on criminal careers that involve robbery, burglary, and aggravates! assault, inclucling inci- clents leacling to the victim's cleath. This decision was motivated by the scarcity of research on careers that involve only other offense types, by the priority given to those crimes by policy makers, and by the fear of those crimes expressed by the public (see, for example, Research and Forecasts, Inc., 19801. The following crime types were excluded from the pan- el's primary focus: arson, "white-colIar" crime (e.g., embezzlement, securities fraud, mad! fraud), and organized crime, for which neither extensive prior emp1rl- cal research nor the requisite data exist; "victimless" crimes (e.g., prostitution, 29 gambling, and drug use or possession); minor sex offenses and other deviant be- haviors, for which offenders are com- monly cliverted from criminal justice to mental health or community treatment agencies; minor property offenses; and status offenses (acts that are illegal only when committed by juveniles). Although these types of offenses were not of pri- mary interest, the pane! did review stucI- ies that related them to the major crime types considered by the panel. For exam- ple, the panel was particularly interested in the relationship of minor clelinquent acts to later adult careers involving rob- bery, burglary, and aggravated assault. The pane] also focused primarily, though not exclusively, on research that involves! incliviclual-leve! data on crimi- nal careers of large samples representing clearly defined populations. Examples in- clude analyses of arrest histories of an urban birth cohort, self-reports of offend- ing by high school students, self-reports of incarcerated offenders, and arrest his- tories of criminals who are active during some observation period. Overall, the re- search reviewer! by the pane! represents most of the literature that is relevant to the study of indiviclual criminal careers. Several categories of research literature that study criminal behavior using para- digms other than the criminal career as defined in this report were not systemat- ically included. Biographical and auto- biographical case studies of individual offenders are excluded because the sub - jects are not generally representative of specific offender populations. Most eco- logical studies (,urisdiction-leve] analy- ses) generally relate aggregate crime measures to community or jurisdictional characteristics and do not inclu(le infor- mation about dimensions of individual criminal careers. Most recidivism studies lack sufficient detail on the number and timing of postrelease arrests that would be necessary for estimating annual of

OCR for page 12
30 fending frequencies and te~ination rates; hence, many research reports and program evaluations that user] recidivism as the only measure of criminal behavior were not reviewed. Finally the research reviewed is primarily based on U.S. sam- ples; a few studies of British and Danish populations are also included. The panel limited its review almost exclusively to studies published in English. While much of the panel's work as with all research- is concerned with the eventual development of causal theory, the panel did not pursue any particular theoretical tradition. This approach re- flects the panel's recognition that synthe- sizing the available statistical descrip- tions of criminal careers wit! contribute to many theoretical approaches. However, development and refinement of causal theory about criminal careers should evolve from much more intensive inves- tigation of the various dimensions of criminal careers than the literature now contains. Ideally, this investigation should proceed with a mixture of ethno- graphic studies to generate detafled hy- potheses, longitudinal studies to explore temporal sequences, ant! field experi- ments to rigorously test hypotheses. In considering the effect of influences possibly associated with reducing crime, the criminal career approach with its focus on individual offenders naturally suggests various influences that operate at the individual level to prevent initia- tion of a career, reduce the frequency of offending, interrupt the career, or encour- age termination of a career. Individual- leve! influences on careers (e.g., matura- tion, family influences) and planned interventions grounded in knowledge l CRIMINAL CAREERS AND CAREER CRIMINALS about those influences (e.g., substance abuse treatment, incapacitation through incarceration) were reviewed by the panel. However, they are not the only ways in which individual careers are in- fluenced. Indiviclual criminal careers are also influenced in broader ways, such as through planned community-level inter- ventions and uncontrolled events in the community (e.g., a factory shutdown). These other influences may operate through the social, economic, political, or environmental structure of the commu- nity in which indivicluals grow up or live, the social networks with which they be- come involved, or the deterrence effects generated by local sanctioning practices. Over time, all of these may change and therefore alter the nature of criminal ca- reers. Therefore, their relationships to ca- reers should be objects of future research. But because so little research is currently available that links these community- level effects to inclividual criminal ca- reers, the panel did not address these broader relationships. With the framework (leveloped in this chapter for characterizing individual criminal careers, Chapters 2 and 3 review the existing empirical knowledge about criminal careers, and Chapter 4 reviews the methodological problems in studying inclividual careers. Chapter 5 considers the uses of this information in assessing different types of interventions intencled to modify criminal careers. Chapter 6 con- siders the effects of incorporating knowl- edge about criminal careers into explicit classification rules intended to aid crimi- nal justice decision making. The last chapter presents an agenda for future re- search on criminal careers.