Showing papers in "Crime and Justice in 1987"

Prediction and Classification in Criminal Justice Decision Making

Abstract: "Prediction and Classification: Criminal Justice Decision Making," a collection of commissioned essays by distinguished international scholars, is the ninth volume in the Crime and Justice series. Like its predecessors, "Prediction and Classification" is essential reading for scholars and researchers seeking a unified source of knowledge about crime, its causes, and its cure.

Classification: An Overview of Selected Methodological Issues

Abstract: Classification methods have two fundamentally different but related goals. The first is to build new classifications. The second is to assign unknown cases to existing classifications. Many traditional offender criminological classification systems are of questionable value, partly because of confusion concerning purposes and roles of classification. The link between purpose and method has recently become more important because of the emergence of a vast range of quantitative methods. The multidisciplinary origins of these methods have meant that some of their embedded concepts and logic may be inconsistent with criminological theory or data. No single method is best for all purposes. Used correctly, new methods may dramatically upgrade the quality of our classifications. A framework for creating and validating classifications in criminology is urgently needed. The framework should include interacting stages for choosing a content domain, constructing the classification, checking internal validity, and es...

Predicting Individual Crime Rates

Abstract: There are great individual differences between offenders in their crime rates The chronic offenders who commit the most crimes tend to commit them at the highest rates High individual crime rates are predicted by an early age of onset of offending, a serious first offense, and a high past crime rate Longitudinal surveys show that high individual rates are predicted by early antisocial behavior, convicted parents and siblings, low family income, and school failure Cross-sectional surveys indicate that they are also predicted by a poor employment record and drug use The prediction of individual crime rates should be investigated in prospective longitudinal surveys, and predictor and criterion variables should be selected on theoretical grounds rather than on availability in records A standard index of predictive efficiency is needed in studies of individual crime rates, and this should be measured in validation samples Acceptance or rejection of a penal policy based on predictions of high-rate offending depends on the choice of a utilitarian or just deserts approach to criminal justice decision making and on a comparison of social costs and social benefits More research is warranted, rather than the adoption of a new penal policy Language: en

Prediction: An Overview of Selected Methodological Issues

Abstract: A variety of statistical methods are available for making criminal justice predictions None possesses a clear-cut advantage over the others, and this may be due to the poor quality of available data Factors that affect the accuracy of predictions, regardless of the method used, include measurement reliability, base rates (the relative frequency of an event in the study population), selection ratios (the proportion of events identified by the prediction method as belonging to the outcome class), sample representativeness, and cross validation The present ability to predict the decisions of criminal justice functionaries or offender behavior is modest Predictions of decisions are more accurate than are predictions of behavior, and empirical studies indicate that statistical predictions outperform intuitive or clinical predictions Predictive accuracy can be enhanced by improving data reliability, by improving measurement of prediction and outcome variables, by using both statistical and subjective boots

Classification for Risk

Abstract: Officials assessing the risks to be taken with alleged or convicted offenders have traditionally been guided only by their subjective deliberations on each case. Subcultural norms shape such risk determinations, and psychological processes result in the decision makers judging carelessly or in a biased fashion but being overconfident about the wisdom of their decisions. Statistical tabulations that provided ways of identifying relative risk from past experience were extolled at first as remedies for the defects of subjective risk assessment. They were not much used in criminal justice agencies, however, until they were developed collaboratively by researchers and officials, and they dealt not only with risk assessment but also with other central concerns in the decisions for which they were employed. They now have very diverse forms and applications but are not used nearly as much or as well as they could be.

Prediction and Classification: Legal and Ethical Issues

Abstract: Recent moves toward increased use of preventive detention, selective incapacitation, and predictions of dangerousness have paralleled vigorous debates concerning the legal and ethical implications of those developments. People's views about these matters are often firmly held and not easily changed by empirical evidence or logical argument. In general, those who believe in limited state power and attach high importance to civil liberties tend to oppose these trends; those who believe in extensive state powers and attach high importance to public safety favor them. The major controversies concern the appropriateness of increased punishment or state intrusion into the lives of those predicted to be dangerous, the disparities in outcome that result from use of predictions, the low levels of accuracy of such predictions, and their disparately harsh impact on minorities and the poor. Recent convergence of views among writers supports a system in which the scope for increased punishment or intrusion because of ...

Guideline-Based Justice: Prediction and Racial Minorities

Abstract: Although blacks compose only 12 percent of the national population, they account for almost 50 percent of the prison population. Many states have adopted the use of guidelines for sentencing, parole, and decisions concerning the level of probationer supervision. Some argue that use of certain factors in guidelines systematically adversely affects minority offenders. The extent to which commonly used guideline factors are correlated with race and recidivism was established using data on over 16,500 offenders convicted of felonies in California in 1980. Race and recidivism correlations were calculated for all convicted felons, for probationers, and for prisoners. When all factors in the data base were used, accuracy in predicting rearrests was seldom greater than a 20 percent improvement over chance. The use only of factors that were not racially correlated increased predictive accuracy from 3 to 9 percent above chance; including racially correlated factors increased predictive accuracy another 5-12 percent...

Classification for Treatment

Abstract: Treatment classifications for prisoner rehabilitation generally lack theoretical coherence. They tend to assign individuals to descriptive classes that possess little or no direct relation to treatment programs. The mismatch between classification systems and treatment programs can only serve to undermine achievement of rehabilitative goals. There are two kinds of treatment programs: those that offer services as a package and those from which services are selected from a menu of possible options. Much additional research is needed on the relation between offender needs, including postrelease needs, and treatment programs before the goal of rehabilitation can be realized.

Classification for Control in Jails and Prisons

Abstract: Classification systems provide means for maintaining institutional safety and order, for providing inmates protection and services, and for managing and allocating personnel and resources. A traditional "judgmental" approach is being largely displaced by objective data-based procedures. The two basic components of institutional classification are the selection of risk factors and the formulation of decision rules. Risk factors in use include consensus-based models, equity-based models, and selections based on empirical prediction and psychometric testing. Decision rules in use or under consideration include linear additive point scales, decision trees, matrix classifications, and attempts to integrate the objective and judgmental aspects of classification. Jail and prison classifications differ procedurally and substantively. Because their populations, resources, policy orientations, and architectures differ, no one standardized classification can fit both kinds of institutions. Weaknesses in management, ...

Prediction in Criminal Justice Policy Development

Abstract: Prediction methods have been employed to model criminal justice decisions from arrest to parole and less frequently for policy review and reformulation in areas marked by the exercise of broad discretion. The development of decision guidelines is a principal example. A recent bail-guidelines project in Philadelphia illustrates policy and methodological issues that have been raised about the use of predictions in decision making. Bail decisions under guidelines in Philadelphia, when compared with a nonguidelines control group, revealed clear-cut differences, particularly in disparity reduction. The seriousness with which some of the methodological critiques have been viewed is not commensurate with the likely impact of guidelines on actual cases. Examination of the criticism that modeling practices may continue or inadvertently generate biased decision making suggests that the impact of poor modeling procedures may be much less than feared. There is insufficient evidence to discredit the claims of the deci...

Causal Inference as a Prediction Problem

Abstract: Impact assessments of criminal justice interventions depend on predictions of the "what-if" variety rather than on predictions of future events. They require "predictions" of what would have occurred if the experimental units that were exposed to the intervention had not been exposed. These predictions, when compared with the responses of treated units, form the basis for all evaluations of program impacts. The statistical concept of "strong ignorability" provides a framework for addressing the internal validity of all impact assessments. When use of strongly ignorable assignment mechanisms is not possible or practicable, strategies are available for achievement of assignments that are ignorable but not strongly ignorable. Strongly ignorable assignment mechanisms are clearly desirable; some might say mandatory.