The Research Director Perspective on the Design, Implementation, and Impact of Risk Assessment and Offender Classification Systems in USA Prisons

A National Survey

Authored by: James M. Byrne , Amy Dezember

Handbook on Risk and Need Assessment

Print publication date:  October  2016
Online publication date:  November  2016

Print ISBN: 9781138927766
eBook ISBN: 9781315682327
Adobe ISBN:

10.4324/9781315682327.ch3

 

Abstract

Several challenges exist for corrections managers across the United States as they identify and implement “best practices” in the classification, treatment, and control of the prison population under the authority of the state. There have been several national commissions formed over the past decade to investigate the causes and consequences of incarceration, the nature and extent of prison violence, and most recently, the overuse of segregation to control offenders. As a result of the recommendations of these commissions, state corrections managers are being asked to critically examine their correctional management and control strategies, focusing primarily on the development of policies and procedures that make prisons safer, support positive offender change, and prepare individuals for reentry to the community. These recommendations for changes in the area of prison classification, control, treatment, and reentry can be viewed as essential features of a new treatment-focused crime control strategy. Corrections managers and public officials now recognize the inherent limitations of the four decade long punishment-focused crime control strategy, referred to by some commentators as the “great prison experiment” (Byrne, 2013; Clear and Frost, 2014).

 Add to shortlist  Cite

The Research Director Perspective on the Design, Implementation, and Impact of Risk Assessment and Offender Classification Systems in USA Prisons

Introduction and Overview

Several challenges exist for corrections managers across the United States as they identify and implement “best practices” in the classification, treatment, and control of the prison population under the authority of the state. There have been several national commissions formed over the past decade to investigate the causes and consequences of incarceration, the nature and extent of prison violence, and most recently, the overuse of segregation to control offenders. As a result of the recommendations of these commissions, state corrections managers are being asked to critically examine their correctional management and control strategies, focusing primarily on the development of policies and procedures that make prisons safer, support positive offender change, and prepare individuals for reentry to the community. These recommendations for changes in the area of prison classification, control, treatment, and reentry can be viewed as essential features of a new treatment-focused crime control strategy. Corrections managers and public officials now recognize the inherent limitations of the four decade long punishment-focused crime control strategy, referred to by some commentators as the “great prison experiment” (Byrne, 2013; Clear and Frost, 2014).

In recent years, policy makers across the country have been rethinking their approach to crime, and developing strategies designed to expand sentencing options that serve to reduce the use of incarceration as our primary crime control strategy for drug related crimes. These efforts have not yet resulted in significant overall reductions in the U.S. prison population. The United States still has one of the three largest prison systems in the world. Half of all prisoners worldwide are housed in prisons in the United States, China, and the Soviet Union (Byrne, Pattavina, & Taxman, 2015). In the United States, recent efforts to reduce our reliance on prison as the sanction of choice have certainly slowed the rate of increase, but prison and jail populations in 2014 were still higher than the 2000 totals. In 2014, there were 1,561,500 adult inmates housed in state prisons across the United States, with an additional 744,600 inmates serving time in our jails. While the state prison population declined slightly in 2014 compared to 2013 (1% decline), the jail population increased slightly (up 1.8%). It is critical to think strategically about how best to manage these offenders, because downsizing strategies—by definition—are not focused on what happens in prison.

For offenders sentenced to prison, decisions are made every day on where to house them, how to protect and treat them, how to keep them safe and healthy, and when to release them back to the community. The U.S. currently has over 5,000 adult prisons and jails, each with its own unique features, staffing ratios, design and operational capacity, offender population resource level, and reentry protocols. Another characteristic of prisons to add to this list: prison violence. There is variation in the levels and rates of prison violence and disorder by type of facility (maximum, moderate, minimum security) but the overall level and rates of prison violence are actually low. According to Byrne and Hummer’s review of violence in prison between 1995 and 2005, “A review of the official data on the extent of the prison violence problem (murder, rape, and assault) suggests that the most serious forms of violence are rare in federal and state prisons and that the rate of violence in federal and state prisons is actually slightly on the decline, despite the doubling of our prison population in the last decade (Useem & Piehl, 2006)” (2007, p. 79). With a low overall base rate for prison violence, it is more difficult for classification experts to correctly predict who the likely violent prisoner is, but there are new risk assessment tools designed to make these predictions (see Berk, Kriegler, & Baek, 2006; McGuire, 2016). For example, one forecasting model designed and tested by Berk and colleagues on a sample of California inmates identified the following violent prisoner profile: “The high risk inmates tend to be young individuals with long criminal records, active participants in street and prison gangs, and sentenced to long prison terms” (2006, p. 9). The researchers point out that the predictive accuracy of the models they tested was low, resulting in a high false positives rate (10/1). Questions have also been raised about the kinds of data that can be accessed to develop these forecasting models, including the race or ethnicity of the inmate (Byrne & Hummer, 2007b). According to the Commission on Safety and Abuse in America’s Prisons:

Reducing violence among prisoners depends on the decisions corrections administrators make about where to house prisoners and how to supervise them. Perhaps most important are the classifications decisions managers make to ensure that housing units do not contain incompatible individuals or groups of people: informants and those they informed about, repeat and violent offenders and vulnerable potential victims, and others who might clash with violent consequences. And these classifications should not be made on the basis of race or ethnicity, or their proxies.

(Johnson v. California, 2005). (2006, p. 29)

There has been considerable debate among policymakers regarding the proper management of our prison population, not only in terms of how (or for some categories of offenders, whether) risk assessment should be used to make initial placement decisions, but also regarding how risk assessment can be used to inform prison management strategies designed to proactively reduce violence and further risk of recidivism while in prison and in the community upon release from prison. To gain a better appreciation of how risk assessment is currently employed by corrections managers across the United States, we surveyed the directors of research in state departments of corrections regarding current risk assessment policies and practices. The results of this survey reveal significant variation in how risk is conceptualized, assessed, and incorporated into institutional management and subsequent offender reentry strategies. Recommendations for changes in the design/utilization of risk assessment technology are offered in three areas: (1) the need to redesign risk assessment (and reassessment) instruments using the prediction of violence and disorder in prison as the sole outcome of interest; (2) the need to develop risk assessment instruments that use a full array of individual and community level risk variables that more accurately predict recidivism during reentry to the community; and (3) the need for prison researchers to advance the use of sound methodologies to ensure that risk assessment tools are meaningful in prison environments.

Research Design

The brief Survey of Department of Corrections Research Directors on the Current Status and Future direction of Risk Classification Systems consists of three sections with a total of 32 survey items covering: (1) the current status of risk assessment and classification system in each responding state, (2) the research evaluating the effectiveness of risk systems used, and (3) the future developments in risk classification in corrections (see survey instrument in Appendix A). The web survey was sent out to 50 individuals who were identified as research directors in state-level department of corrections agencies from each state. The respondents were notified that the survey was part of ASC’s Division on Corrections & Sentencing’s upcoming edition of the Handbook on Risk and Needs Assessment: Theory and Practice and that the results would inform the researchers on how departments use risk assessment and classification tools.

The survey invitation was emailed out on December 3, 2015 to all 50 potential respondents along with a short description of the survey and an individual link directly to the online survey. They were notified that the survey would take approximately 10 minutes to complete. Two weeks later, the first reminder was sent out on December 17, 2015 asking respondents to complete the short survey and reiterated the description of the study and the value of their response to the study. A third and final email reminder was sent out on January 5, 2016 to all individuals that had not yet completed the survey. Additionally, follow up phone calls were made on January 11–13, 2016 to encourage respondents to fill out the survey. After six weeks in the field, data collection was closed with 64.0% (32 out of 50) of respondents completing all or part of the survey 3 . Table 3.1 lists the state directors who participated in this survey.

Table 3.1   States Included in the National Survey of Research Directors

Responding states

Alabama

Michigan

Alaska

Minnesota

Arizona

Missouri

Arkansas

Nevada

California

New Hampshire

Colorado

New York

Connecticut

Oklahoma

Delaware

Pennsylvania

Florida

Rhode Island

Hawaii

South Carolina

Idaho

Utah

Iowa

Vermont

Kansas

Virginia

Kentucky

Washington

Louisiana

Wisconsin

Maine

Wyoming

Findings

Current Status of Risk Assessment and Offender Classification System

There has been a dramatic change in the policies and procedures used for initial assessment of offenders entering our prison system. Less than 15 years ago, a review of classification systems by James Austin and colleagues (2003) identified the Quay system—also known as the adult internal management system (AIMS)—as the most commonly used internal classification system used in state prisons. However, only one survey respondent indicated they used the Quay system while three others indicated some other hybrid method. The Quay classification system is focused on five personality types, such as inadequate-immature, neurotic-conflicted, unsocialized aggressive or psychopathic, sociable or subcultural offenders, and subcultural-immature offenders (Kratcoski, 1981). The goal is to ensure that the prison units are balanced within these five characteristics to improve safety in prison. However, it appears that many states have invested in a wide range of new classification systems, utilizing a combination of internal expertise and external NIC consultants to develop hybrid and/or new assessment systems.

When asked about the primary factors driving placement in minimum, medium, or maximum security housing units, 11 of 20 respondents (55%) indicated that the risk classification score was the primary factor, 4 (20%) noted risk level, 2 (10%) referenced conviction offense, and 1 (5%) simply said available space. The respondents noted a variety of secondary factors that influenced the initial placement decision, including conviction offense, prior institutional conduct, risk level, interpersonal or gang-related issues, and available space. Clearly, there are multiple considerations in play when making the initial security level determination but the tendency is not to use personality traits such as those deployed in the Quay management system. While risk of prison violence is not the main driver at this initial decision point, the majority of respondents still view it as the primary purpose of initial classification, particularly regarding examining prior institutional conduct and interpersonal or gang-related issues.

Additionally, we asked the research directors to indicate areas where the initial classification of prisoners upon entry to prison could be improved. Of the 18 research directors who responded to this question, 5 (27.8%) recommended a greater initial focus on offender needs and 2 (11.1%) suggested that appropriate offender treatment programs be identified at this initial assessment point. It appears that for a number of respondents, more attention to a third generation risk and need assessment tool that links risk to needs is important. One research director stated, “We’re using our static risk assessment to determine initial level (intensity) of supervision, and then the DRAOR [Dynamic Risk Assessment for Offender Re-entry] scores modify (increase/decrease) level of supervision as time progresses.” Another research director indicated that the risk assessment used in their state was selected because, “COMPAS is a 4th generation actuarial assessment tool.” Lastly, a research director indicated, “We are in the process of implementing a combined static and dynamic risk assessment, which also factors in correctional events, such as programming, infractions, violations and visiting.” With many research directors selecting 3rd and 4th generation tools, there does appear to be a general acceptance of the notion that there is value in the assessment of both risk and needs for classification and treatment decisions. This points to the need to develop classification systems designed for the dual purpose of violence reduction while in prison and offender change upon release to the community.

Many directors highlighted the need for subsequent development and implementation of risk reduction strategies. For a small number of respondents (4) risk reduction was viewed as a function of deterrence-based strategies. Additionally, three respondents indicated that current classification systems could be improved by identifying treatment programs for the individual assessed. This shows that there is some division on how research directors believe that prisons want to use a classification system to fit their needs based on the outcomes they desire. Some institutions may want to have greater emphasis on deterrence and treatment, while others may be more focused on incapacitation and improving safety through reduced prison violence. Based on survey responses, it certainly appears that the tripartite deterrence vs. incapacitation vs. treatment debate is ongoing among state research managers.

Table 3.2   Which Risk Assessment Instrument Is Being Used in Your System?

N

%

COMPAS

2

9.5

ORAS

4

19.0

LSI-R/LSCMI

6

28.6

Developed own risk tool

7

33.3

Other

2

9.6

Total

21

100.0

No Response

11

We asked respondents about the specific assessment instruments used in their prison system. Of the 21 state research directors who responded, 7 (33.3%) indicated that their state decided to develop their own risk assessment tool, rather than employ a proprietary risk assessment instrument (see Table 3.2). Two state directors discussed how their state uses a static risk tool that they developed since they feel that the static component is the only important factor for initial classification decisions. These states then use other instruments to measure special dynamic needs such as substance abuse, criminal thinking/cognitions, or dynamic needs. For example, one director stated, “We use [a static risk screening tool created in house] to assess risk and the LS/CMI to assess need among the higher-risk offenders.” Among the remaining states, a variety of proprietary risk instruments were identified, including the LSI-R/LSCMI (6 states), ORAS (4), COMPAS (2), and PAI or other dynamic/static assessment tool (2). The reasons given for selecting a particular assessment tool varied, but cost, ease of use, and accuracy were the three most common rationales for adoption. The states that use a short-static risk tool indicated that the research team developed their own instrument to reduce the cost of using the standardized tools. Research directors noted that the proprietary instruments as well as the short static tools were originally designed to predict general recidivism after release or during the period of community supervision. They emphasized a need for validating the use of the risk assessment tools in prison to predict prison violence and disorder. Research directors stressed that many of the current tools have not been adequately scientifically reviewed, which raises concerns regarding their current use.

Research Evaluating the Effectiveness of Risk Assessment and Offender Classification Systems

There is a small body of research available for review that allows us to assess the accuracy of risk assessment instruments used in prison settings to predict prison violence (Byrne and Hummer, 2007a, 2007b; McGuire, 2016, in press). Perhaps the most widely accepted technique for assessing the accuracy of risk predictions is AUC (area under the curve) analysis. According to McGuire’s (2016) review of available studies conducted since 2000 using this validation technique, our ability to accurately predict prison violence is modest at best:

The majority of AUC statistics show a significant improvement over chance, though there is marked variation and the predictive success is modest in most cases. The most widely used scales for risk assessment in other contexts (e.g. VRAG, PCL:R and HCR-20) do not emerge especially well from this set of studies, in some cases having no significant association with physical violence. Better results were found using the Risk Assessment Scale for Prison (RASP) and the Risk Assessment for Violent Nonsexual Victimization (RVNSV). The highest figure shown (0.831) is for a simplified model its authors called the Risk Assessment Scale for Prison-Reduced Burgess (RASP-RB) in respect of its level of accuracy in predicting serious assaults (Cunningham, Sorensen, Vigen and Woods, 2011). However, in the study in which that scale was originally developed, the observed AUC was less impressive at 0.687 (Cunningham and Sorensen, 2006b).

(McGuire, 2016, p. 37)

Table 3.3   What Type of Validation Was Conducted of the Risk Assessment Tool?

N

%

Simple correlation

4

23.5

AUC analysis

1

5.9

Recidivism study

3

17.6

Other

5

29.4

All of the above

4

23.5

Total

17

100.0

No Response

15

Notes:

★  Respondents who selected other provided the following responses: multiple regression, norming/validation study, or validation was conducted by an outside consultant (2).

One interesting finding from our survey is that the accuracy of the risk assessment tool—when validated—was determined by conducting research on the predictive accuracy of the instrument after release, instead of during the period of incarceration. That is, the tool was validated on subsequent recidivism in the community during a specific follow-up period as the criterion variable. Prison-specific outcomes, such as infractions or prison violence, were seldom used to validate a risk assessment tool. While developing and validating an accurate risk assessment tool for offenders about to leave prison and reenter the community certainly is an important component of risk management in today’s corrections system, it would seem that this is more appropriate for reentry uses than risk assessment during the period of confinement. In fact, given the importance of prison violence, it would seem that there is a need for developing and validating risk instruments designed to predict violence and disorder in prison rather than the community. For example, such instruments might suit the classification decision regarding which housing unit might minimize security issues in prison settings. Since 7 of 20 respondents indicated that they do not currently use a risk instrument designed to predict prison violence, this underscores the need to rethink the purpose of prison risk classification systems. If there is agreement on purpose, new instruments targeting prison violence can be field tested, and decisions can be made about the timing of these risk assessments and how they can be integrated with risk assessments used during the reentry phase of an offender’s prison experience.

Future Developments in Risk Assessment and Offender Classification in Corrections

The majority of survey respondents (13/17 or 76%) indicated that improvements in prison classification procedures would reduce the level and rate of prison violence and disorder. Similarly, most respondents (13/18 or 72%) stated that improvements in reentry classification policies and procedures would reduce recidivism among offenders released to the community. A common sentiment was that risk and need assessments are beneficial for “better release planning, transition services, responding to needs early on in community placement/supervision.” Another respondent recognized the importance of making improvements to the classification process, “If we assume we have valid instruments and if the instruments are used to effectively identify high risk/high need inmates for programs and if we are able to place the inmates in the needed programs and if the programs are effectively delivered, Then it would be logical to conclude a potential reduction in recidivism.” Many research directors identify reentry classification systems as a separate need from initial classification systems.

When we asked respondents to consider what other changes they think would reduce prison violence and/or reduce recidivism upon reentry, the most common answer given was that more effective treatment programming was needed (12), followed by increased incentives for offender change (10). Other strategies included ongoing performance measurement (9), increased staffing (8), better supervision and deterrence (8), and new technology (6). One respondent indicated a need for “better coordination, data collection and resources for analyses among and between CJ agencies.” Another research director emphasized the need for “greater availability of meaningful and effective evidence based programs.” One respondent stated,

The tools currently used to assess risk can be improved, some more than others, but the area in greatest need of improvement involves a stronger focus on implementation science. In short, a greater emphasis needs to be placed on how best to bridge the gap, which is rather large, between what we know (i.e., what works) and what we do. Implementation science doesn’t get much attention within corrections, but when it does it’s usually within the context of program delivery.

Research directors recognize the need to use evidence-based programs and improve implementation of such efforts to reduce recidivism and bridge the gap between corrections and successful reentry.

Concluding Comments

The results of our survey of research directors in state corrections systems underscore the need for a new approach to prison classification, one that views the primary purpose of initial classification as prison violence risk reduction through accurate risk assessment and evidence-based risk reduction planning, program design, and implementation. A major issue identified by the research directors in our survey is that the goal of using a risk and need assessment tool is unclear. The question becomes: how do we design a prison classification system that minimizes the level of prison violence while maximizing the opportunity for long term offender change upon release to the community?

The first step in this direction involves the development and proper validation of a risk assessment instrument that targets prison violence as the outcome of interest. According to the recent review by McGuire (2016), only a small number of jurisdictions currently focus on prison violence, and when they do, they typically target a subgroup of violence—prison sexual violence. The importance of focusing on overall security issues such as any prison violence is that it is tied to the health and well-being of the prison environment. A safe prison environment is important, not only for those confined but also for the correctional staff. Generally, classification systems are designed to address these prison environmental issues by ensuring that the mix of individuals in a housing unit will minimize security problems. Risk and need assessment tools that are used for the purpose that they are developed for are more likely to achieve their possible outcomes. As this survey highlights, prison classification systems are designed with multiple purposes in mind, including offender change upon release to the community. If reduction in recidivism is a critical performance measure for institutional corrections, then it will be necessary to identify the range of static and dynamic risk factors that are linked to recidivism (Via, Dezember, & Taxman, 2016). At the same time, research directors recognized that more research is needed to identify accurate predictors of violence in prison settings. The challenge facing these directors is to develop risk assessment tools that address both outcomes: prison violence reduction and recidivism reduction.

Additionally, with only about half of the respondents indicating that they have conducted validation studies of the risk and/or need assessment tools used in their state, the results of this survey indicate that many tools currently should be validated to ensure that the risk and need assessments are accurate and reliable. AUC analysis is one of the better methods to evaluate predictive validity of risk assessment tools, yet only five respondents indicated they used this method when validating their tool. The predictive validity of risk assessment tools is vital to the success of proper placement and classification within a facility. This remains an outstanding issue for the field, which is to have accurate current tools that can predict the desired outcome. A meta-analysis of 47 studies examined how predictive validity is analyzed and reported in studies of instruments used to assess risk and found many inconsistencies in the AUC methodologies and in how validity is measured (Singh, Desmarais, & Van Dorn, 2013). Often, validity studies are not using the correct outcome measures and thus the results may not be accurate. Relatively new to the field are machine learning techniques to help improve the predictive validity of tools available and reduce the cost of using these tools. But, there are cautionary notes about using machine learning models since it is unclear how much they improve the predictive accuracy of the tool, and on the appropriateness of the predictor variables included in these analyses (Hess & Turner, 2017; Kim & Duwe, 2017; Schwartz et al., 2017; Brennan, 2017). Future research should focus on validating different scales and measures that will predict risk of institutional violence and other more accurate outcome measures for use inside of prisons.

The research directors recognize that assessing an individual’s risk level is important to address an individual’s dynamic needs to lower their potential for future violence and offending. Many of the research directors surveyed for this study responded that they currently use a proprietary tool that was adopted for their needs. These assessment tools are taken “off-the-shelf” and adopted on populations that they were not originally designed to serve (Hamilton et al., 2017). This can cause the users of the tool to inaccurately measure outcomes, resulting in the misclassification of inmates. It is important for risk and need assessment tools to be customized to measure the desired outcome (Hamilton et al., 2017). Part of the customization process is ensuring that the selected tool will best serve the agency’s needs, including the staff who will administer it. In many cases, risk assessment tools are not implemented properly, and are not accurately assessing risk and classification levels (Miller & Trocchio, 2017; Rudes, Viglione, & Meyer, 2017). It is important for future research to address how research directors of departments of corrections can select the best risk assessment tools for their needs, customize them for their specific population, and ensure staff buy in for successful implementation (including higher levels of inter-rater reliability).

For those offenders beginning the institutional phase of reentry, the challenge for correctional managers will be to link institution-based risk reduction efforts to a community based outcome: the prediction of recidivism upon return to the community. This effort will necessarily require the utilization of a different risk assessment process, since the outcome of interest will shift from institutional behavior to community protection. Perhaps the most important development in this area is the recognition of the link between individual and community level risk factors. According to a recent review:

Until we address the underlying community factors that social ecologists have long argued are associated with crime—including location in high risk neighborhoods, culture, resource availability, jobs, poverty, and a breakdown of informal social control mechanisms—even high-quality, resource-rich rehabilitation programs are not likely to result in broad-scale desistance from crime among individual offenders. Unless we design correctional strategies that (1) recognize the link between person environment interactions and recidivism, and then (2) attempt to change both individual offenders and individual communities, we will continue the cycling of these individuals from community to prison to community.

(Byrne, 2008, p. 270)

The results of our survey of research directors in state correction systems underscore the need for revisiting the approach to prison classification, utilizing a perspective that echoes the need to consider the primary purpose of initial classification as prison violence risk reduction through accurate risk and need assessment. The research directors view classification as being linked to evidence-based risk reduction planning, program design, case planning and implementation. For those offenders beginning the institutional phase of reentry, the challenge for correctional managers will be to focus on a different outcome: the prediction of recidivism upon return to the community. This effort will necessarily require the utilization of a different risk assessment process, since the outcome of interest will shift from institutional protection to community protection. The research directors clearly indicated that the focus on using validated risk and need assessment tools is an important movement, but the tailoring of these tools for the intended use is critically important. The implementation of the risk and need assessment in prison environments is essentially a work in progress, and one that can benefit from more attention to the goals of classification.

Notes

Professor, School of Criminology and Justice Studies, University of Massachusetts, Lowell.

Doctoral Student, Center for Advancing Correctional Excellence, George Mason University.

The draft survey was pre-tested by two research directors from state-level department of corrections agencies and their feedback was incorporated into the final survey. Since one individual was identified from each state, we received responses from 32 of the state-level department of corrections, while 18 states did not respond to the survey after being contacted four times.

Survey Instrument

Name:

Agency:

Survey of Department of Corrections Research Directors on the Current Status and Future Direction of Risk Classification Systems

Section 1  Current Status of Risk Assessment and Classification System in Each State

  1. Which internal classification system is used in your state corrections system? (Select one)
    • Quay Adult Internal Management (AIM) System
    • Prison Management Classification (PMC) System
    • Hybrid
    • Don’t Know
    • None
    • Other, please specify: ___
  2. What factors drive the initial placement of offenders into maximum, medium, or minimum security prisons? (Primary, select one / Secondary, check all that apply)
    • Conviction offense
    • Prior institutional conduct
    • Risk level
    • Space available
    • Interpersonal or gang related issues
    • Classification score
    • Don’t know
    • Other-Specify: ___
  3. What is the purpose of the classification system in your jurisdiction? (Primary, select one / Secondary, check all that apply)
    • Risk assessment to determine level of risk that the person presents in prison
    • Need/service assessment to determined needed programs
    • Both risk and need assessment
    • Screening tool
    • Classification
    • Inform access to treatment
    • Inform housing placement in prison (e.g., general population, administrative segregation, solitary confinement, etc.)
    • Case planning
    • Other-Specify: ___
  4. How can the current classification instrument be improved? (Primary v. secondary)
    • Assess client’s needs
    • Identify treatment programs
    • Need to include deterrence approaches
    • Change outcomes measures (e.g., shift focus towards treatment and away from recidivism)
    • Need a tool customized to department needs
    • Shorten assessment tools
    • Other, please specify: ___
  5. Which risk assessment instrument is being used in your system? (Select one)
    • COMPAS
    • ORAS
    • LSI-R
    • LSCMI
    • Wisconsin Risk and Needs
    • RASP
    • VRAG
    • PCL-R
    • RVNSV
    • HCR-20
    • PCL-R
    • PAI
    • Combination of the above (please specify):___
    • Developed own risk tool (please describe): ___
    • Other (please specify): ___
  6. Why was the risk assessment instrument (or instruments) selected? (Select one)
    • Cost
    • Ease of use
    • Accuracy
    • All of the above
    • Other, specify: ___
  7. What type of validation was conducted of the risk assessment tool? (Select one)
    • Simple correlation
    • AUC analyses
    • Recidivism study
    • All of the above
    • Other, specify: ___
  8. Which type of validation study of the risk assessment tool is the most valuable to your organization? (Select one)
    • Simple correlation
    • AUC analyses
    • Recidivism study
    • Other, specify: ___
  9. What is the dependent variable used in the validation studies of the risk assessment tool? (Select one)
    • Violent crime in prison
    • Disciplinary infractions in prison
    • Recidivism upon release to the community
    • Other, Specify: ___
  10. Have you conducted an evaluation of whether the risk assessment tool used can be scored reliably?
    • Yes
    • No
  11. When is the risk assessment tool administered? (Select one)
    • At intake or diagnostic center only
    • At intake/diagnostic center and then bi-annually until release
    • At intake/diagnostic center and then at release or near reentry
    • Other, specify: ___
  12. Who primarily administers the risk assessment tool? (Select one)
    • Intake staff
    • Private provider
    • Clinical staff
    • Other, please specify: ___
  13. What issues do you experience when using risk assessment tools? (Check all that apply)
    • Misclassification of offenders
    • Too time consuming
    • Tool is not customized to department needs
    • Requires too many resources to use properly
    • Tool is outdated
    • Tool only addresses risk and does not address needs
    • Not reliable
    • Inter-rater reliability concerns
    • Other, please specify: ___
  14. Which of the following do you use to predict prison violence? (Check one)
    • Actuarial risk instrument
    • Special violence instrument such as VRAG, or list the tool:___
    • Structured professional judgment tool (e.g. HCR-20)
    • None
    • Don’t know
    • Other, specify: ___
  15. Which of the following types of violence and/or disorder are distinguished in your risk assessment or classification system (Check all that apply):

    Risk Assessment Classification System
    a Inmate on inmate
    b Inmate on staff
    c Staff on inmate
    d Collective violence
    e Intra-personal violence
  16. Some systems have developed a separate process for assessing risk upon reentry or transition back into the community. Do you have a separate reentry-based risk assessment, classification system, or discharge planning process?
    • Yes
    • No If yes, how is this process different? ______ ______ ______ ______
  17. Upon release, when do you assess an inmate’s risk for re-offending? (Select one)
    • Within 3 months
    • Within 6 months
    • Within 1 year
    • We do not assess an inmate’s risk for re-offending after release
    • Other, please specify: ___
  18. If yes, which risk assessment instrument do you use? (Select one)
    • COMPAS
    • ORAS
    • LSI-R
    • Wisconsin Risk Needs
    • Combination of the above (please specify): ___
    • We do not assess an inmate’s risk for re-offending after release
    • Other (please specify): ___ Any additional comments on current use of assessment tools: ______ ______ ______ ______ ______ ______ ______

Section 2  Research Evaluating the Effectiveness of Risk System

  1. Has the overall effectiveness of the prison risk assessment system been evaluated?
    • Yes
    • No
  2. If yes, how was it evaluated? (Select one)
    • Internal agency assessment
    • External review
    • Other, please specify: ___
  3. If yes, what was the outcome of the evaluation? (Check all that apply)
    • Instrument worked well
    • Needed to adjust the instrument cut-offs
    • Reduced some items from the instrument
    • Other, specify: ___
  4. What is your overall assessment of the current classification system used to initially assess new admissions? (Select one)
    • We have staffing issues that affect the assessment system
    • The current instrument is not that useful for normal prison operations
    • The current procedures were updated and we still have problems of violence
    • Our procedures need improvement (please explain): ___
  5. Which of the following changes (if any) in the initial assessment process would you recommend? (Check all that apply)
    • New assessment instrument (please explain): ___
    • Validation of existing assessment tool(s) (please explain): ___
    • More and/or better qualified staff (please explain): ___
    • Other, please specify: ___
    • No changes
  6. When are inmates re-assessed in your system? (Select one)
    • Quarterly
    • Annually
    • Ongoing, based on behavior
    • No re-assessment
    • Other, please specify: ___
  7. How would you rate the effectiveness of the re-assessment process currently in place? (Select one)
    • Excellent
    • Good
    • Average
    • Poor
  8. If you use a separate reentry assessment process, how is the performance of the reentry strategy evaluated in your jurisdiction? (Select one)
    • Internal agency assessment
    • External review
    • We do not use a separate reentry assessment process
    • Other, please specify: ___
  9. If you use a separate reentry assessment process, has the reentry system performance been examined to address the following responsivity factors? (Check all that apply)
    • Age
    • Gender
    • Race
    • Offense type
    • Location
    • We do not use a separate reentry assessment process
    • Other, please specify: ___ If yes, briefly highlight results: ___ ______ ______ ______ ______ ______ ______

Section 3  Future Developments in Risk Classification in Corrections

  1. In your view, can the level of violence and disorder in prison be reduced through improved classification procedures?
    • Yes
    • No
  2. In your view, can we reduce the current rate of recidivism through improved reentry classification procedures (e.g. separate reentry risk assessments, classification systems, or discharge planning)?
    • Yes
    • No If yes, please explain how so: ___ ______ ______ ______ ______ ______ ______
  3. What else can we do to reduce recidivism? (Check all that apply)
    • Increased staffing (number/quality)
    • New technology
    • Increased incentives for offender change
    • Ongoing performance measurement
    • More effective supervision and deterrence-based approaches
    • More effective treatment programming availability
    • Other, Specify: ___ Any comments you would like to make about the state of risk assessment and its use in prison/community: ______ ______ ______ ______ ______ ______ ______ ______ ______

References

Austin, J. (2003). Findings in prison classification and risk assessment. Washington, D.C.: U.S. National Institute of Corrections.
Berk, R.A. , Kriegler, B. , & Baek, J. (2006). Forecasting dangerous inmate misconduct. Berkeley, CA: California Policy Research Center, University of California, Berkeley.
Brame, B. (2016). Static Risk Factors and Criminal Recidivism. In F.S. Taxman & (Ed.), Risk and Need Assessment: Theory and Practice (pages pending). Abingdon: Routledge.
Brennan, T. (2016). An alternative scientific paradigm for criminological risk assessment: Closed or Open Systems, or both? In F.S. Taxman (Ed.), Risk and Need Assessment: Theory and Practice (pages pending). Abingdon: Routledge.
Byrne, J.M. (2008) The social ecology of community corrections. Criminology and Public Policy, 7(4), 263–274.
Byrne, J. M. (2013). After the fall: Assessing the impact of the great prison experiment on future crime control policy. Federal Probation, 77(3), 3–14.
Byrne, J. M. & Hummer, D. (2007a). Myths and realities of prison violence: A review of the evidence. Victims and Offenders, 2, 77–90.
Byrne, J. M. & Hummer, D. (2007b). In search of the “Tossed Salad Man” (and others involved in prison violence): New strategies for predicting and controlling violence in prison. Aggression and Violent Behavior, 12, 531–541.
Byrne, J. M. & Hummer, D. (2008). Examining the impact of institutional culture on prison violence and disorder: an evidence-based review. In J. M. Byrne , D. Hummer & F. S. Taxman (eds.), The Culture of Prison Violence, pp. 40–90. Boston, MA: Pearson.
Byrne, J. M. , Hummer, D. , & Taxman, F. S. (eds.) (2008). The Culture of Prison Violence. Boston, MA: Pearson.
Byrne, J. , Pattavina, A. , & Taxman, F. (2015). International trends in prison upsizing and downsizing: In search of evidence of a global rehabilitation revolution. Victims and Offenders: An International Journal of Evidence-based Research, Policy, and Practice, 10(4), 420–451.
Clear, T.R. & Frost, N. A. (2014). The Punishment Imperative: The rise and failure of mass incarceration in America. New York, NY: New York University Press.
Cunningham, M. D. & Sorensen, J. R. (2006b). Actuarial models for assessing prison violence risk: Revision and extensions of the Risk Assessment Scale for Prison (RASP). Assessment, 13, 253–265.
Cunningham, M. D. , Sorensen, J. R. , & Reidy, T. J. (2005). An actuarial model for assessment of prison violence risk among maximum security inmates. Assessment, 12, 40–49.
Cunningham, M. D. , Sorensen, J. R. , Vigen, M. P. , & Woods, S. O. (2011). Correlates and actuarial models of assaulting prison misconduct among violence-predicted capital offenders. Criminal Justice and Behavior, 38, 5–25.
Endrass, J. , Rossegger, A. , Frischknecht, A. , Noll, T. , & Urbaniok, F. (2008a). Using the Violence Risk Appraisal Guide (VRAG) to predict in-prison aggressive behaviour in a Swiss offender population. International Journal of Offender Therapy and Comparative Criminology, 52, 81–89.
Endrass, J. , Rossegger, A. , Urbaniok, F. , Laubacher, A. , & Vetter, S. (2008b). Predicting violent infractions in a Swiss state penitentiary: A replications study of the PCL-R in a population of sex and violent offenders. BMC Psychiatry, 8–74. Doi:10.1186/1471-244X-8-74.
Gendreau, P. , Little, T. , & Goggin, C. (1996). A meta-analysis of the predictors of adult offender recidivism: what works! Criminology, 34(4), 575–608.
Gibbons & Katzenbach (2006). Confronting confinement. Washington, DC: Vera Institute of Justice, Commission on Safety and Abuse in America’s Prisons.
Hamilton, Z. , Tollefsbol, E.T. , Campagna, M. , & van Wormer, J. (2016). Customizing Criminal Justice Assessments. In F.S. Taxman (Ed.), Risk and Need Assessment: Theory and Practice (pages pending). Abingdon: Routledge.
Hanson, R. K. & Morton-Bourgon, K. E. (2009). The accuracy of recidivism risk assessments for sexual offenders: a meta-analysis of 118 prediction studies. Psychological Assessment, 21(1), 1–21.
Hastings, M. E. , Krishnan, S. , Tangney, J. P. , & Stuewig, J. (2011). Predictive and incremental validity of the Violence Risk Appraisal Guide scores with male and female jail inmates. Psychological Assessment, 23, 174–183.
Hess, J. & Turner, S.F. (2016). Accuracy of risk assessment in corrections population management: Where’s the value added? In F.S. Taxman (Ed.), Risk and Need Assessment: Theory and Practice (pages pending). Abingdon: Routledge.
Kim, K. & Duwe, G. (2016). In F.S. Taxman (Ed.), Risk and Need Assessment: Theory and Practice (pages pending). Abingdon: Routledge.
Kratcoski, P. C. (1981). Use of the quay system of Classification. California: Duxbury Press.
Kroner, D. G. & Mills, J. F. (2001). The accuracy of five risk appraisal instruments in predicting institutional misconduct and new convictions. Criminal Justice and Behavior, 28, 471–489.
Labrecque, R. M. , Smith, P. , & Wooldredge, J. D. (2014). Creation and validation of an inmate risk assessment for violent, nonsexual victimization. Violence and Victims, 9, 317–333.
McGuire, James (2016, in press). Prison violence: A review of research on models and Contributors. United Kingdom, National Offender Management Service (NOMS).
Miller, J. & Trocchio, S. (2016). Risk/Need Assessment Tools and the Criminal Justice Bureaucrat: Reconceptualizating the frontline practitioner. In F.S. Taxman (Ed.), Risk and Need Assessment: Theory and Practice (pages pending). Abingdon: Routledge.
Rudes, D.S. , Viglione, J. , & Meyer, K.S. (2016). Risky Needs: Risk entangled needs in probation supervision. In F.S. Taxman (Ed.), Risk and Need Assessment: Theory and Practice (pages pending). Abingdon: Routledge.
Schwartz, I.M. , York, P. , Greenwald, M. , Ramos-Hernandez, A. , & Feeley, L. (2016). Using Predictive Analytics and Machine Learning to Improve the Accuracy and Performance of Juvenile Justice Risk Assessment Instruments: The Florida Case Study. In F.S. Taxman (Ed.), Risk and Need Assessment: Theory and Practice (pages pending). Abingdon: Routledge.
Singh, J. P. , Desmarais, S. L. , & Van Dorn, R. A. (2013). Measurement of predictive validity in violence risk assessment studies: A second-order systematic review. Behavioral Sciences & the Law, 31(1), 55–73.
Via, B. , Dezember, A. , & Taxman, F.S. (2016). Exploring How to Measure Criminogenic Needs: Five Instruments and No Real Answers. In F.S. Taxman & A. Dezember (Eds.), Risk and Need Assessment: Theory and Practice (pages pending). Abingdon: Routledge.
Walters, G. D. , Duncan, S. , & Geyer, M. (2011). Predicting disciplinary adjustment in inmates undergoing forensic evaluation: a direct comparison of the PCL-R and the PAI. Journal of Forensic Psychiatry and Psychology, 14, 382–393.
Walters, G. D. & Heilbrun, K. (2010). Violence risk assessment and Facet 4 of the Psychopathy Checklist: Predicting institutional and community aggression in two forensic samples. Assessment, 17, 259–268.
Search for more...
Back to top

Use of cookies on this website

We are using cookies to provide statistics that help us give you the best experience of our site. You can find out more in our Privacy Policy. By continuing to use the site you are agreeing to our use of cookies.