Handbook on Risk and Need Assessment

Theory and Practice

Faye Taxman, Faye S. Taxman

• Publisher: Routledge
• Year: 2016
• eBook ISBN: 9781317402824
• Topic: Social Sciences
• Subtopic: Criminology
• Index: Social Sciences

The Handbook on Risk and Need Assessment: Theory and Practice covers risk assessments for individuals being considered for parole or probation. Evidence-based approaches to such decisions help take the emotion and politics out of community corrections. As the United States begins to back away from ineffective, expensive policies of mass incarceration, this handbook will provide the resources needed to help ensure both public safety and the effective rehabilitation of offenders.

The ASC Division on Corrections & Sentencing Handbook Series will publish volumes on topics ranging from violence risk assessment to specialty courts for drug users, veterans, or the mentally ill. Each thematic volume focuses on a single topical issue that intersects with corrections and sentencing research.

1
THE VALUE AND IMPORTANCE OF RISK AND NEED ASSESSMENT (RNA) IN CORRECTIONS & SENTENCING

An Overview of the Handbook

Faye S. Taxman with Amy Dezember¹

Missouri’s Chief Justice Ray Price (2010) in his State of the Judiciary speech: “There is a better way. We need to move from anger-based sentencing that ignores cost and effectiveness to evidence-based sentencing that focuses on results—sentencing that assesses each offender’s risk and then fits that offender with the cheapest and most effective rehabilitation that he or she needs.”

(cited in Casey, Warren & Elek, 2012: pg. 12)

Policymakers’ need to know the subsequent strategies for public safety and recidivism reduction might begin with a simple question: Do risk assessment instruments reliably predict recidivism? The short answer, according to years and volumes of research, is resoundingly: yes. But we must be mindful of what saying yes may mean. Adoption of a risk assessment tool goes hand-in-hand with fundamentally altering approaches to reentry and correctional management, supervision, services, and more broadly criminal justice practice. Ultimately, the process of implementing risk assessments within an agency should consist of more than simply adding a tool to the agency portfolio; it should result in a shift of corrections culture, practices, and policies.

(cited in Desmarais & Singh, 2013)

The Public Safety Performance Project at Pew put out an issue brief titled:
“Risk/Needs Assessment 101: Science Reveals New Tools to Manage Offenders”

(The Pew Center on the States, 2011)

Most advancements in managing individuals in the criminal justice system begin with a discussion of the need for the adoption of a standardized risk and need assessment (RNA) tool. Standardized, valid tools are research-based in that they are built using statistical methods to predict desired (or undesired) outcomes. RNA tools are generally recommended at all decision points along the criminal justice system—from booking to pretrial release to sentencing to release from prison or jail to services. Standardized RNA tools, in theory, offer a science-based approach to regulate decision-making to avoid or minimize biases, decrease unnecessary discretion, improve proper use of resources, and/or increase fairness. As noted above in the various headlines of recent policy pieces, RNA tools are considered the panacea to better practice. Great promises are tied to the use of RNA tools, primarily better performance of the justice system at the system level to deliver fairness and justice, at the program level to assign individuals to appropriate controls and programs, and at the individual level to improve outcomes. The rapid growth and expansion in risk assessment tools—in various domains of the justice system—illustrates a growing appreciation for integrating science into practice in order to improve operational practices.

The Division of Corrections and Sentencing of the American Society of Criminology is devoting the first edition of our Handbook series to this timely topic of RNA tools. The interest in RNA tools is dear to the hearts of practitioners as well as researchers and academics. The rapid expansion of RNA tools in various justice settings—generating a growth in both public domain and proprietary tools, even for special offenders—requires more attention to better understand how RNA tools are developed, implemented, and impact justice. In fact, given the current body of tools being employed for various goals, it is a worthy endeavor to assess the “state of the art”. This review of both theory and policy is devoted to critically analyzing the issues surrounding the methods to develop RNA and then how this RNA tool is used to fulfill the grand vision of having more precision in decision-making among justice actors as part of the process of diagnosis, prediction, and linkages to better outcomes. We are grateful that members of the American Society of Criminology were willing to critically assess the state of RNA, and to help identify a research agenda that advances the feld. Given that RNA is tied to the linkage of science influencing practice, the analysis includes a research agenda to advance the field of RNA tool development and utilization, which is needed to garner better implementation and utilization of RNA tools.

The Generations of RNA

Risk and need assessment tools (RNAs) represent a research-based application relevant to the operations of many justice agencies. RNAs are standardized tools that apply scientific principles to develop and test tools that are presumably valuable to the field. The onset of the movement to build and use standardized risk and need assessment tools highlights the need to improve decision-making in the justice system as compared to unstructured, or even semi-structured, interviews. The first generation of tools involves the use of justice actors to make decisions using information that is available on an individual. In this first generation, the approach was to rely on clinical and professional judgment that does not have explicit or objective scoring rules (Brennan, Dieterich, & Ehret, 2009). The first generation essentially dominated institutional and community corrections for several decades, and remains preferred by many decision makers (Boothby & Clements, 2000; Wormith, 2001). Unstructured or semi-structured interviews depend on the quality of the staff and their ability to use the appropriate information in making a decision. The general concerns with the interview approach is that it is characterized by excessive subjectivity, inconsistency, bias and potential stereotyping, which raises the issues of vulnerability in decision-making (Brennan, 1987; Grove & Meehl, 1996; Hastie & Dawes, 2001). Individual actors can select the information they consider to be relevant, which may vary across actors, and the information they find relevant may not be empirically related to individual outcomes. In general, the use of unstructured information relies on the skill set of the individual actor, which can create the possibility of bias or fluctuations in discretionary decision-making.

The second generation of RNA tools represents the beginning of structuring information in the tools based on the criminal history and case file. The selected information items draw from research findings on the factors that are linked to the outcome, which is generally recidivism (in whatever form it is measured such as arrest, reconviction, incarceration, etc.). The selection of the items requires statistical analyses to identify those factors that predict the outcomes as well as the weight given to any item. Second generation tools rely on risk prediction and are usually characterized by their brevity and efficiency (Brennan, Dieterich, & Ehret, 2009). The relatively simple point scales associated with the second generation tools were surprisingly effective in terms of predictive validity and generally outperformed professional judgment or the opinions of trained experts (Dawes, 1979). Dating back to the 1920s, the idea of a parole release decision tool was conceived to reconcile a person’s past criminal history with the decision to be released early from prison. Risk assessments transformed into a tool to help parole examiners sort through information about the individual and identify the relevant factors associated with reincarceration or other recidivism measures. The process of tallying up past behavior resulted in the creation of the second generation of risk assessment tools, which in this case assisted in providing better guidance to the Parole Board on an individual’s likelihood of having further involvement with the justice system through systematic identification of factors related to “failure”. The emphasis on using past information to inform present choices was recognized as a means to facilitate difficult decisions, and to reduce the risks associated with making that decision. Peter Hoffman, a scientist at the U.S. Parole Commission, developed the salient factor instrument for parole supervision as another example of using key factors that predict success in a standardized tool to collect information about individuals (Hoffman & Beck, 1997). The process of using statistics to identify the key set of factors related to success and failure was considered an advancement for the field.

In the mid-1960s, the Vera Institute introduced risk assessment tools for bail decisions, a landmark event in the risk assessment field given the use of the tool at an earlier stage in the justice system decision-making, and the components of the tool. The Vera Institute’s tool expanded the list of relevant information in the instrument to include factors related to stability (i.e. housing, employment, etc.) some of which are dynamic risk factors and some are indicators of social and community stability. These dynamic risk factors were linked to the outcome of failure during the pretrial period either in terms of failure to appear for court or rearrest (Mamalian, 2011). Since the Vera Institute tool included criminal history and stability factors in the community, this commenced the beginning of the third generation of RNA tools with the added information on needs that affect outcomes. Third generation tools introduced a more explicit, empirically based, theory guided approach, with a broader selection of dynamic factors that were more sensitive to change (Brennan, Dieterich, & Ehret, 2009). By incorporating a host of need factors (i.e. those related to recidivism), third gene ration tools became more widely used to assess offenders’ risk for recidivism and their treatment needs (Campbell, French, & Gendreau, 2009). Over several decades, the use of standardized decision making tools slowly crept into existence in justice organizations with scientists and academics working on tools for various uses.

The famous Wisconsin Risk and Needs Assessment (WRA) was developed in the late 1970’s for probation and parole agencies and featured items that used both static risk factors and dynamic risk (need) factors. The WRA concept generated an interest in RNA tools being used to support resource allocation decisions (see O’Leary & Clear, 1984). The resource allocation decision is to assign higher risk/need offenders with more frequent supervision and/or services to prevent recidivism. The notion that the risk score could be used to guide the level and types of supervision was born. Andrews, Bonta, and Wormith (2006) then developed the Level of Service Inventory (LSI) in the 1980’s and expanded the number of domains in the RNA tool. The LSI instrument includes a criminal history domain along with domains related to substance use, accommodation, attitudes and orientation, employment, education, peers, and so on. Beginning in the 1980’s, the RNA emerged as a dominant theme for institutional and community corrections, which contributed to an expanding number of tools in the public domain and proprietary tools. As the use of the tools expanded, so did the different types of predicted behavior (i.e., tools for specialized populations such as sex offenders, driving while intoxicated, females, violence, are a few examples). The expansion in the number and types of tools, as well as the different applications, raises a number of unanswered questions. Frequent concerns are whether the added items improve the predictive validity of the tool, whether the added items foster improved decisions, and whether the elements of the tool are responsive to the needs of the individual. There is also the question of whether proprietary tools outperform public domain tools, or increase the capacity of the system to better use the tools. In many ways, RNA tools continue to evolve and a greater emphasis is on increasing the complexity of the tools.

Don Andrews, James Bonta, and Steve Wormith (2006) crafted the need for a fourth generation of RNA tools to advance decision-making in the area of case management. Case management is different than typical recidivism-related outcomes in that the goal is to identify the factors that are dynamic and amendable to change—factors that should be targeted through treatment, programs, or controls to reduce recidivism. Essentially, the same factors that are used to predict recidivism are then considered in terms of their ability to define malleable behavior that can be altered as a result of programming and services. Some newer instruments expanded the third generation tools to highlight the greatest area of dynamic risk/needs or added items to the instruments. The unanswered questions about the fourth generation tools are: 1) what are the predicted outcomes, and 2) do the tools facilitate the ability to identify the dynamic needs for programming and service purposes?

A new movement is emerging in the field to use machine learning tools. Machine learning tools use “big data” with advanced statistical methods to calculate a predicted value. The algorithms are used with an assortment of criminal histories, prior justice involvement, and other data that are readily available through management information systems maintained by the justice agencies. The advantage of the machine learning algorithms is their ability to calculate the formula without having an individual (justice actor) complete an assessment tool, therefore removing the potential for human errors that come from administering an instrument. The machine learning models primarily focus on the static risk items since criminal history is maintained in automated records. In some way, this focus only on risk factors seems to revert back to second generation tools but is justified on the ease of administering the tool. Of course, many questions remain about these tools regarding whether they increase predictive accuracy, provide for greater utilization by staff and justice actors, and facilitate how agencies can accommodate the dynamic needs into risk assessment processes that have prefilled risk scores.

The generations of RNA illustrate the degree to which science has influenced the development of decision support tools for the field as well as advanced criteria for key decisions at various points in the justice system. On the surface, it appears that the evolution of risk and need assessment has been logical and progressive (see Table 1.1 above), and that this strategy influences practice going back nearly 100 years. But this is far from true—the emergence of risk and need assessment tools in the practice of criminal justice is actually slow and grinding. It is only within the past five years that there has been an upsurge in the use of risk and need assessment tools by justice agencies, now that the tools are classified as an “evidence-based practice” for most just...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Table of Contents
• Notes on Contributors
• Introduction
• 1 The Value and Importance of Risk and Need Assessment (RNA) in Corrections & Sentencing: An Overview of the Handbook
• Part I History of RNA
• Part II Methodological Issues in Creating and Validating RNA
• Part III Dynamic Risk Factors and Responsivity Toward Different Populations
• Part IV RNA Implementation and Practice
• Part V Special Issues Regarding the Conceptualization for RNA
• Index