Predictive Sentencing Research Paper

This sample Predictive Sentencing Research Paper is published for educational and informational purposes only. If you need help writing your assignment, please use our research paper writing service and buy a paper on any topic at affordable price. Also check our tips on how to write a research paper, see the lists of criminal justice research paper topics, and browse research paper examples.

Prediction research in criminology has, by and large, focused on characteristics of offenders: various facts about criminals are recorded – their age, previous arrests and convictions, social history, and so forth. It is then statistically determined which of these factors are most strongly associated with subsequent offending (see Gottfredson 1967). The result is a “selective” prediction strategy: among those convicted of a given category of offense, some will be identified as bad risks and others will not. Under an incapacitative strategy, those convicted offenders who constitute bad risks would be imprisoned or imprisoned for longer periods.

Traditional Prediction Methods

Traditional statistical prediction techniques pursued this selective approach. Generally, they found that certain facts about an offender (principally, his previous criminal history, drug habits, and history of unemployment) were to a modest extent indicative of increased likelihood of recidivism on his part (Gottfredson 1967).

These techniques did not, however, distinguish between serious and lesser forms of reoffending. Both the offender who subsequently committed a single minor offense and the individual who committed many serious new crimes would be lumped together as potential recidivists. Moreover, the technique offered no promise of generating reduced net crime rates, as they did not attempt to estimate such incapacitative strategies’ aggregate crimepreventive effects. Locking up the potential recidivist thus assured only that he or she would be restrained; since other criminals remained at large, it did not necessarily diminish the overall risk of victimization. These limitations eventually reduced penologists’ interest in traditional prediction techniques.

 “Selective Incapacitation”

Surveys of imprisoned offenders conducted in the USA in the early 1980s found that a small number of active offenders admitted responsibility for a disproportionate number of serious offenses. If that minority of dangerous offenders could be identified and segregated, perhaps this could reduce crime rates after all. These surveys thus generated a renewed interest in prediction research.

The most notable product was a 1982 RAND Corporation study by Peter W. Greenwood (1982). He named his prediction strategy “selective incapacitation.” His idea was to target high-rate, serious offenders – namely, those offenders likely to commit frequent further acts of robbery or other violent crimes in the future. For that purpose, he took a group of incarcerated robbers, asked them how frequently they had committed such crimes, and then identified the characteristics of those reporting the highest robbery rates. From this, he fashioned a predictive index, which identified potential high-rate offenders on the basis of their criminal records and histories of drug use and unemployment. Greenwood’s predictive factors were (i) prior convictions of instant offense type, (ii) incarceration for more than half the preceding 2 years, (iii) conviction before age 16, (iv) time served in a state juvenile facility, (v) drug use during the preceding 2 years, (vi) juvenile druguse, and (vii) employment for less than 50 % of the preceding 2 years. He defined “high-risk” offenders as those for whom at least four of these seven factors were present (Greenwood 1982).

Greenwood also devised a novel method of projecting the aggregate crime reduction impact of this prediction technique. On the basis of offender self-reports, he estimated the average annual rate of offending of those robbers who were identified as high risks by his prediction index. He then calculated the number of robberies that would be prevented by incarcerating such individuals for given periods. By increasing the prison terms for the high-risk robbers while reducing the terms for the others, he concluded it would become possible to reduce the robbery rate by as much as 15–20 % – without causing prison population to rise (Greenwood 1982).

Queries About Effectiveness

While Greenwood’s study initially attracted much interest, certain difficulties became apparent. One difficulty was making the predictions hold up when official data of the kind that sentencing courts have available are relied upon. The objective of selective incapacitation is to target the potential high-rate serious offenders and to distinguish them from recidivists who reoffend less frequently or gravely. To make this distinction, the RAND studies, including Greenwood’s, relied upon offender self-reports. A sentencing court, however, is seldom in the position to rely upon defendants’ willingness to supply the necessary information about their criminal histories.

The court will need to rely instead on officially recorded information about offenders’ adult and juvenile records, and such documentation is too meager to make the distinction adequately. When Greenwood’s data were reanalyzed to see how well the potential high-risk serious offenders could be identified from the information available in court records, the results were disappointing. The officially recorded facts –arrests, convictions, and information about offenders’ personal histories – did not permit the potential high-rate robbers to be distinguished from (say) the potential car thieves. The factors in the self-report study that had proved the most useful – such as early and extensive youthful violence and multiple drug use – were not reflected in court records (Chaiken and Chaiken 1984). To make the predictions work, the courts would need to obtain and rely on information in school and social-service file – with all the problems of practicability and due process that would involve.

Questions became apparent, also, in the projections of preventive impact. Greenwood based his crime reduction estimates on the self-reported activities only of incarcerated robbers and then extrapolated those estimates to robbers generally. Incarcerated robbers, however, are scarcely a representative group: they tend to offend more frequently than robbers generally in the community. (It is like trying to learn about the smoking habits of smokers generally by studying the self-reported smoking activities of patients in a lung cancer ward.) When this kind of extrapolation is eliminated, the projected crime reduction impact is reduced by about half (see von Hirsch 1985, Chap. 10).

Other problems with the projections also became apparent. Greenwood assumed, for example, that his high-rate robbers would continue offending for a long time. When shorter (and perhaps more realistic) residual criminal careers are assumed instead, the estimated preventive effect shrinks dramatically. The accuracy of the forecasts was also disappointing: comparing predicted with actual offending rates, Greenwood’s analysis showed that only about half of the respondents were categorized accurately the by predictive scale.

These doubts were confirmed by a report of a National Academy of Sciences’ Panel on criminal careers that appeared 4 years after the Greenwood study (for text of the Panel’s report, see Blumstein et al. 1986). The Panel included several noted advocates of predictively based sentencing – and the report endorsed the idea of predictive strategies (within certain limits) so long as these could be shown to be effective. Nevertheless, the Panel’s conclusions regarding the crime-preventive effects of selective incapacitation were quite skeptical. After recalculating Greenwood’s results and thus scaling its initial preventive estimates down considerably, the Panel noted that even those revised estimates would shrink further were the scale drawn from a broader and potentially more heterogeneous population than persons in confinement and-were to utilize officially recorded rather than self-reported information. Even those reduced projected effects would largely disappear, moreover, if the estimated length of the residual criminal career were scaled down (see von Hirsch 1988).

Prospects For Improving Predictive Techniques

Could these difficulties be overcome? Greenwood’s research has been only a beginning, and future selective incapacitation research may eventually produce more convincing results. The obstacles are considerable, however. If the aim is to distinguish potential high-rate, serious offenders from lesser potential criminals, this task remains difficult to achieve using the scant official records courts have at their disposal. Records of early offending might become somewhat more accessible with a change in the law concerning the confidentiality of juvenile records – but such records notoriously suffer from incompleteness and inaccuracy. Social histories, such as drug use and unemployment, will be even more difficult to ascertain accurately.

Estimation of the impact of selective incapacitation on crime rates also involves difficult problems of sampling. Analyses of convicted or incarcerated offenders’ criminal activities suffer from the difficulty mentioned already: it is not clear to what extent these person’s activities are representative of the activity of offenders in the community. Samples drawn from the general population are free from such bias but may contain too small a number of active offenders.

The most troublesome issue, however, remains that of estimating the length of offenders’ residual criminal careers. The serious offenders who are the targets of selective incapacitation policies ordinarily would be imprisoned in any event; the most salient policy issue is that of the length of their confinement. The strategy is to impose longer terms on the supposed high-risk offenders, but that assumes they will continue their criminal activities for a protracted period. Little prevention is achieved if the bad risks who are confined are those whose active careers will terminate fairly soon. This means that selective incapacitation, to succeed, needs not merely to pick out high-risk offenders during their initial periods of high-rate offending, but rather those who are likely to go offending for substantial time. But how much do we know about forecasting residual careers?

A number of recent criminal-career studies have attempted to make estimates of the duration and intensity of offenders’ residual criminal careers (see, e.g., Kazemian and Farrington 2006). These studies confirm that residual career length and frequency of offending decline significantly with age. Moreover, offenders’ scores on risk-assessment indices – when based mainly on information included in official records – were significantly but only modestly associated with the extent of their remaining criminal careers. These findings also suggest that incapacitative benefits will decline significantly during the remainder of a predictively based sentence – thus limiting the crime-preventive payoff of selective incapacitation strategies.

Even with improved predictive efficacy, there would be no assurance as to how much confining the higher-risk offenders will prevent offenses from taking place. With group crimes and network offenses, for example, the offender may be replaced easily by other criminal participants. Moreover, it also needs to be born in mind that individuals who are admitted into prisons must eventually be released. Sampson and Laub’s (1997) theory of cumulative disadvantage, for example, emphasizes the “indirect role of incarceration in generating future crime,” which may occur through severance of bonds to conventional social institutions.

Proportionality Problems

Selective prediction strategies – whether the traditional sort or more sophisticated methods such as Greenwood’s – must confront also an important ethical question: their apparent conflict with the requirements of proportionality. The conflict stems from the character of the factors relied upon to predict. Those predictive factors tend to have little relation to degree of reprehensibleness of the offender’s criminal choices (which is the basis of proportionality judgements – see more fully von Hirsch and Ashworth 2005, Chap. 9).

Proportionality requires that penalties be based chiefly on the seriousness of the crime for which the offender currently stands convicted. The offender’s previous criminal record, if considered at all, should have a secondary role and the offender’s social status is largely immaterial to the penalty he or she deserves. With selective prediction, the emphasis necessarily shifts away from the seriousness of the current offense. In fact, Kazemian and Farrington found that offense type was not a significant predictor of residual career length or residual number of offenses. Since the aim is to select the high-risk individuals from among those convicted of a given category of crime, the character of the current offense cannot have much weight.

Traditional prediction indices largely ignored the gravity of the current offense and concentrated on the offender’s earlier criminal and social histories. Selective incapacitation techniques have similar emphasis: of Greenwood’s predictive factors, three do not measure criminal activity of a significant nature at all but, instead, measure the offender’s personal drug consumption and lack of stable employment. Of the four other factors, only two reflect the offender’s recent criminal record, and none measures the heinousness (e.g., the degree of violence) of the offender’s current offense (see discussion above).

When aggregate preventive effects are taken into account, the proportionality problems become more worrisome still. Selective incapacitation techniques, by their own proponent’s reckoning, could promise significant crime reduction effects only by infringing proportionality requirements to a very substantial degree. Greenwood’s projection of a significant reduction on the robbery rate was made on the assumption that robbers who score badly on his prediction index would receive about 8 years imprisonment, whereas better-scoring robbers would receive only 1 year in jail (Greenwood 1982). This means a great difference in severity – in Greenwood’s study, about 800 % – in the punishment of offenders convicted of the same type of offense, and one that can scarcely begin to be accounted for by distinctions in the seriousness of the offender’s criminal conduct. When this punishment differential is narrowed – when high-risk robbers receive only modestly longer terms than robbers deemed lower risks – the crime reduction payoff (even by Greenwood’s methodology) shrinks to slender proportions.


Where does this leave us? A limited capacity to forecast risk has long existed: persons with criminal records, drug habits, and no jobs tend to recidivate at a higher rate than other offenders, as researchers have known for decades. However, the limitations in that forecasting capacity must be recognized – for selective incapacitation as well as more traditional forecasting techniques. Identifying high-risk, serious offenders will be impeded by the quality of information available (or likely to become available) to sentencing courts. The potential impact of selective incapacitation on crime rates is far below proponents’ initial estimates and is likely to be modest. Considerations of proportionality limit the inequalities in sentence that may fairly be visited for the sake of restraining high-risk offenders, and limiting these permissible inequalities will, in turn, further restrict the technique’s impact on crime. In order for this sentencing model to be effective, some important empirical and ethical caveats associated with selective incapacitation policies must be addressed.


  1. Blumstein A, Cohen J, Roth J, Visher C (eds) (1986) Criminal careers and career criminals. National Academy of Sciences, Washington, DC
  2. Bottoms AE, von Hirsch A (2010) The preventive effect of criminal sanctions, Ch. 3. In: Cane P, Kritzer H (eds) Oxford handbook of empirical legal studies. Oxford University Press, Oxford
  3. Chaiken M, Chaiken J (1984) Offender types and public policy. Crime Delinq 30:195–226
  4. Gottfredson D (1967) Assessment and prediction methods in crime and delinquency, in president’s national commission on law enforcement and criminal justice, task force report: juvenile delinquency and youth crime. Washington, DC
  5. Greenwood PW (1982) Selective incapacitation. RAND Corporation, Santa Monica
  6. Kazemian L, Farrington DP (2006) Exploring residual career length and residual number of offenses for two generations of repeat offenders. J Res Crime Delinq 43:89–113
  7. Sampson RJ, Laub JH (1997) A life-course theory of cumulative disadvantage and the stability of delinquency, Ch. 4. In: Thornberry T (ed) Developmental theories of crime and delinquency: advances in criminological theory. Transaction Publisher, New Brunswick
  8. von Hirsch A (1985) Past or future crimes: deservedness and dangerousness in the sentencing of criminals. Rutgers University Press, New Brunswick
  9. von Hirsch A, Ashworth A (2005) Proportionate sentencing: exploring the principles. Oxford University Press, Oxford
  10. Zimring FE, Hawkins G (1995) Incapacitation. Oxford University Press, New York

See also:

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


Always on-time


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