U.S. Department of Justice
National Institute of Corrections

A Guide for Correc tional Agencies

U.S. Department of Justice
National Institute of Corrections
320 First Street N.W.
Washington, DC 20534

Morris L. Thigpen
Director
Larry Solomon
Deputy Director
George M. Keiser
Chief, Community Corrections/Prisons Division
Madeline Ortiz
Project Manager

National Institute of Corrections
World Wide Web Site
http://www.nicic.org

Objective Prison Classification:
A Guide for Correctional Agencies

James Austin, Ph.D.
Patricia L. Hardyman, Ph.D.

July 2004
NIC Accession Number 019319

This document was funded by the National Institute of Corrections under cooperative agreement number
00P13GIN8 with The Institute on Crime, Justice and Corrections at The George Washington University.
Points of view or opinions stated in this document are those of the authors and do not necessarily reflect
the official position or policies of the U.S. Department of Justice.

Acknowledgments
Much of this report is based on the significant contributions of a diverse number of
state and local correctional agencies with which we have had the privilege of working on a number of classification studies over the past decade. Without their help
and assistance, much of the progress made in prison classification would not have
occurred.
We would also like to offer special thanks to Dr. Susan Hunter, former Chief of the
Prisons Division at the National Institute of Corrections (NIC), Madeline Ortiz, former NIC Corrections Program Specialist, and Georgette Walsh, Editor at NIC. In the
true spirit of a cooperative agreement, all three worked closely with us to guide our
research and refine our analysis. Their participation, guidance, and feedback during
our research and fieldwork were valuable contributions to this report.
Finally, we would like to acknowledge the long-term support of Sammie Brown,
formerly a Corrections Program Specialist with NIC and now back home in South
Carolina. She has been a strong and persistent advocate of objective prison classification and NIC’s technical assistance program.

iii

Contents
Acknowledgments ...............................................................................................iii
Executive Summary............................................................................................ix
Chapter 1. Introduction ......................................................................................1
Chapter 2. Objective Prison Classification: A Brief History...........................3
Chapter 3. What Is Objective Prison Classification?.......................................7
Key Terms and Concepts ............................................................................7
Essential Components of an Objective Prison Classification System ......14
Chapter 4. Effective Strategies: Guidelines for Implementing
Classification Systems........................................................................................25
Phase I: Mobilization—Determining the Need for Change .....................25
Phase II: Comprehensive Assessment of Current Classification
System .......................................................................................................30
Phase III: Planning....................................................................................32
Phase IV: Implementation.........................................................................36
Lessons Learned: Common Problems, Issues, and Solutions ..................38
Chapter 5. Evaluations of Prison Classification Systems ..............................41
Establish That the Classification System Is Working as Intended ...........41
Assess the Impact of the Classification System .......................................47
Chapter 6. Classification of Women Prisoners ...............................................49
Common Themes ......................................................................................50
Implications and Future Steps...................................................................56
Chapter 7. Other Special Topics and Issues in Classification .......................59
The Impact of Environmental Factors and Prison Management
on Prisoner Behavior.................................................................................59
The Need To Link Prison Classification and Risk Assessment
With Release Decisions ............................................................................60
v

Contents

Notes ...................................................................................................................63
References ...........................................................................................................65
Suggested Readings............................................................................................67
Appendix: Glossary of Key Terms ...................................................................73

vi

List of Exhibits
Exhibit 1. Overview of External and Internal Classification Systems ................10
Exhibit 2. Typology of High-Risk and Special Management Prisoners..............13
Exhibit 3. Nationwide Estimated Prisoner Classification Levels for the
Average Daily Population ....................................................................................14
Exhibit 4. Examples of Agency Mission Goals and Classification
Objectives.............................................................................................................16
Exhibit 5. Time-Task Line for Effective Classification Implementation ............26
Exhibit 6. Suggested Outline for a Classification Initiative Summary
Report...................................................................................................................36
Exhibit 7. Summary of Evaluation Standards .....................................................42

vii

Executive Summary
Classification systems help minimize the potential for prison violence, escape, and
institutional misconduct. During the past two decades, professionals in prisons and
those in other correctional systems worked assiduously to improve their approaches to classifying offenders according to custody, work, and programming needs. As
a result of these efforts, criteria for custody decisions have been validated, custody
decisions are more consistent, overclassification has been reduced, prisoner program needs are assessed more systematically, and institutional violence has
declined. Since the 1980s, objective prison classification systems have been widely
adopted in the United States, Canada, Europe, New Zealand, and Australia.

Objective Prison Classification: A Brief History
During the nineteenth and early part of the twentieth centuries, prisoners in the
United States were separated by age (adult versus juvenile), gender (male versus
female), number of offenses (first versus repeat), and special needs (mentally ill).
Humane treatment was the basis for these early and most basic classification systems. The Reformatory Movement and the Progressive Era gave rise to the progressive treatment and rehabilitation of confined offenders. The so-called medical
model, which included identifying the problems of individuals and prescribing
appropriate treatments to address their individual needs, was the essence of this
diagnostic approach to classification. Moreover, the treatment and rehabilitation of
confined offenders were expected to become the foundation of corrections.
Gradually, the emphasis on diagnosing the causes of offenders’ antisocial behaviors
was replaced by an emphasis on the types of treatment programs available. Unfortunately, early classification systems were largely grounded in subjective criteria or
clinical assessments that produced arbitrary and unreliable results. The lack of standardization in the assessment processes used in making custody-level decisions led
to well-founded claims that the classification systems were inherently flawed in
their ability to produce reliable and valid results. Several influential court cases in
the United States found that prison classification systems must demonstrate that
they are not arbitrary and capricious.
The use of reliable and valid criteria to assess a prisoner’s custody level is one of the
core distinguishing features of an objective classification system. Beginning in the
1970s, prison classification systems began to experiment with objective criteria. In
the United States, the California Department of Corrections and the Federal Bureau
of Prisons developed the first objective classification systems. These systems used
numerical scoring that sought, among other things, to improve the consistency and
objectivity of the assessment process.
A number of developments drove this shift in classification. In the 1970s, policymakers began a 30-year effort to “get tough on crime” that dramatically increased
the prison population. The burgeoning prison population gave rise to a series of
volatile and violent prison riots in many of the nation’s largest prisons in the early
1970s and placed new burdens on government spending as policymakers demanded

ix

Executive Summary

new and better methods for identifying and controlling high-risk prisoners and a
more scientific process for forecasting future prison bedspace requirements. Allegations of prisoner abuse and mistreatment stemming from overcrowding triggered
a wave of lawsuits that challenged traditional prison operations in general and classification procedures in particular. Finally, all these changes increased pressures on
correctional staff to deliver an intensive analysis, diagnosis, and prognosis for each
prisoner.
During the 1980s and the 1990s, public safety concerns led to the development of a
variety of risk assessment and prisoner treatment systems. Competing demands and
restraints within correctional agencies have resulted in a variety of prison classification systems, all of which promised to deliver reliable and valid assessments to manage, treat, and ultimately return prisoners to the community, although risk assessment
often places greater emphasis on measuring a prisoner’s likelihood of reoffending
than it does on how the prisoner should spend the time while incarcerated.

What Is Objective Prison Classification?
Understanding objective prison classification requires a grasp of several key terms
and concepts and an understanding of the essential components of a classification
system.

Key Terms and Concepts
Three pairs of terms appear regularly in any discussion of objective prisoner classification: “reliability” and “validity,” “external” versus “internal” prisoner classification systems, and “general” versus “special” prison populations:
Reliability and validity. Two core distinguishing features of an objective classification system are its accuracy, or validity, and its consistency, or reliability.
Validity refers to the accuracy of the classification system in predicting a prisoner’s behavior and assigning him or her to an appropriate risk level. Reliability considers whether the same decision will be rendered if the assessment is
repeated by the same or a different staff member.
External versus internal prisoner classification systems. External classification determines a prisoner’s custody classification and facility assignment. Internal classification governs facility-level decisions such as where and with whom
the prisoner will be housed, the types of programs and services to which the prisoner should be assigned, and the prison industry or work assignment most appropriate for the prisoner. Most often, the intent of internal classification systems is
to ensure that prisoners who are at risk for placement in a special management
population are supervised accordingly. Minimizing classification errors requires
both external and internal classification systems.

x

General population versus special prison populations. It is important to differentiate between prisoners housed in the general population—that portion of

Objective Prison Classification: A Guide for Correctional Agencies

the prison population with no special security-related restriction on access to
basic programs and services—and those who require placement in a special
prison population such as administrative or disciplinary segregation, protective
custody, specialized mental health, or medical units. In general, approximately 80
percent of prisoners are in the general population; these prisoners are classified
as minimum, medium, close, or maximum custody.

Essential Components of an Objective Prison Classification
System
The following components are essential to the success of an objective classification
system: a mission statement, classification goals and objectives, a dedicated classification unit and classification staff, centralized control over all prison transfers and
housing decisions, reliable and valid classification instruments that have been tested, appropriate use of overrides, timely and accurate classification, a formal housing plan and security/custody designation for each housing unit, adherence to the
housing plan, accurate prisoner data, an automated data system, continuous monitoring and process evaluation, and an impact evaluation.
Mission statement. As the foundation of the classification system, the mission
statement should be consistent with the agency’s mission, goals, and resources.
It should reflect both the formal process for classifying and managing prisoners
within the prison system and prisoner risks and service needs. The mission
statement should also include a clear and comprehensive statement of the core
values and philosophy of the agency in accordance with national and local correctional standards.
Classification goals and objectives. These must directly support the core values set forth in the mission statement and be formulated so that their outcomes
can be measured and evaluated. At a minimum, classification objectives should
be reviewed annually to verify that old challenges have been met and to identify new ones to be addressed.
Dedicated classification unit and classification staff. An established classification unit is necessary for a prison classification system to function properly.
The unit must have a sufficient number of dedicated, well-trained, and experienced staff to ensure that all prisoners are properly classified in a timely manner.
Centralized control over all prison transfers and housing decisions. Classification staff must have the sole authority to assign prisoners to housing units
according to the classification system. This does not mean that other staff members cannot recommend transfers in emergencies. However, even in emergency
circumstances, classification staff must review and approve all transfers. The
most effective way to ensure that all staff comply with this requirement is for
the prison to institute and enforce a written policy that clearly delineates the
authority and powers of the classification unit.
xi

Executive Summary

Reliable and valid classification instruments. Objective classification systems use well-structured instruments (i.e., forms) designed to produce reliable
and valid assessments of the risks posed by the prisoners. Three basic forms
guide the classification process: the initial screening form, used to help identify a prisoner’s emergency needs when he or she is first admitted to the prison
system, the initial classification form, used to determine the prisoner’s scored
custody level on admission to the prison system, and the reclassification form,
used to reevaluate the prisoner’s scored custody level throughout his or her
incarceration based on his or her institutional conduct.
Classification instruments that have been tested. A prison system must conduct a pilot test of the classification system’s instruments and procedures on the
prison population to which they will be applied before implementing the system. The pilot test measures two things: how well the proposed classification
instruments will perform on a given prison population and the likely impact of
the classification policy and procedures on prison operations.
Appropriate use of overrides. The general standard is that 5–15 percent of a
prison population’s custody levels are based on discretionary overrides rather
than the original initial classification or reclassification scores. In general,
approximately 50 percent of discretionary overrides should assign the prisoner
to a custody level lower than that derived from his or her original score and 50
percent should assign the prisoner to a higher custody level.
Timely and accurate classification. All prisoners entering the prison system
must be screened immediately to determine if they need to be housed apart from
other prisoners, either for their own protection or that of other prisoners and
staff. Custody level should be assigned within 30 days of admission. The agency
should also set standards for when and under what circumstances prisoners
must be reclassified—usually after 6 months of incarceration and at least annually thereafter.
Formal housing plan and security/custody designation for each housing
unit. To house prisoners according to their classification status, the prison facility must be classified as well. Classifying the prison facility determines the
number and types of bedspaces available, which, in turn, establishes the basis
for the facility’s housing plan—the blueprint for how prisoners will be housed
within the prison—and designates units for intake and assessment and for special management prisoners.
Adherence to the housing plan. Ideally, prisoners assigned to different custody
levels will not be housed together. For example, minimum-custody level prisoners should never be housed with maximum-custody level prisoners. A housing
unit’s structure and its supervision, programming, and privilege levels should be
well suited to the custody levels of the prisoners housed within the unit.

xii

Objective Prison Classification: A Guide for Correctional Agencies

Accurate prisoner data. The prison must have a comprehensive and orderly
recordkeeping system to ensure prisoner data are available to the classification
staff. The prison must also have access to automated criminal record and criminal court data systems so staff can retrieve the most current and complete information about the prisoner’s prior and current charges.
Automated data systems. Storage in the prison’s management information system (MIS) database of all information obtained from prisoners’ initial screening, initial classification, and all reclassification forms reduces the likelihood of
scoring errors and allows for systematic, ongoing monitoring of the classification system. An electronic database increases the accuracy and efficiency of the
classification process by reducing the need to reenter the same basic data at each
stage of the classification process.
Continuous monitoring and process evaluation. A classification system
should be monitored continuously to ensure that it was implemented as
designed and continues to work as designed in relation to the prison’s current
population. In some state prison systems, a process evaluation is equivalent to a
continuous audit of the classification system. As such, the evaluation should
answer two fundamental questions: 1) Are all prisoners classified according to
existing agency policies and procedures? and 2) Are prisoners being housed
according to the classification system?
Impact evaluation. An impact evaluation assesses the positive and/or negative
effects of the classification system on the prison system as a whole and on its
various components (e.g., prisoners, staff, risk assessment, administration, and
prison operations). To ensure an objective assessment of the system, the impact
evaluation should have a rigorous experimental or time-series research design
and should be conducted by persons trained in research methods and statistical
analysis.

Effective Strategies: Guidelines for Implementing
Classification Systems
One key component of the National Institute of Corrections’ technical assistance
efforts in the mid-1990s was developing a model that would guide correctional
administrators in either designing a new classification system or modifying an existing one. The basic implementation process of this model includes four phases—
mobilization, assessment, planning, and implementation—each of which involves
several steps. To effectively address specific local questions such as budget cuts,
new laws and legislation, prisoner and population trends, or available prison data
systems, a correctional agency may need to modify the classification process. However, the four phases and the steps and tasks within each phase are critical for
designing and updating classification systems.

xiii

Executive Summary

Phase I: Mobilization—Determining the Need for Change
Step 1. Identify the classification issues. It is recommended that an agency revalidate its classification system every 3 to 5 years. Revalidation ensures that the classification system is operating as intended and that the risk factors and custody scales
are appropriate for the current prison population. Past issues that have strongly
influenced states to reassess and modify their prisoner classification system in terms
of policies, procedures, risk factors, relative weight of the risk factors, custody
scale, and mandatory and discretionary override factors include the following:
Changes in the characteristics and behavior of the prisoner population.
“Get-tough-on-crime” legislation.
Budget cuts, overcrowding, and reductions in programming and services.
Questions about the validity of the classification system for women prisoners.
Identification and management of high-risk offenders.
Before undertaking any initiative to change its classification system, the agency
should assess its commitment and readiness to change. All levels and divisions within the agency—the commissioner, the wardens and superintendents, the supervisors,
correctional officers, case managers, and, in particular, the research and information
system staff—should be committed to the classification initiative before it is begun.
Step 2. Designate a steering committee. The next step is to identify a project
leader who will be responsible for overseeing all tasks associated with the initiative
and ensuring they are completed. The project leader’s first task is to organize a steering committee to manage or direct the initiative throughout the change process. The
steering committee should include representatives from all of the agency’s operational areas, such as security, medical and mental health services, programming,
classification, research and planning, information systems, budget, training, and
legal counsel.
Step 3. Review the current written classification policies and procedures. When
the steering committee has been established, it should conduct a comprehensive
review of the agency’s current classification procedures and practices to identify
problems and issues in the classification process and examine the links between the
internal, external, and needs assessment processes. This review should consider the
following questions:
What current policies, practices, and issues are potentially affecting the classification system?
What trends are associated with these policies and practices?

xiv

What outcomes would the steering committee like to achieve with the revised
classification system?

Objective Prison Classification: A Guide for Correctional Agencies

Phase II: Comprehensive Assessment of Current Classification
System
Step 1. Conduct an onsite assessment of the classification system. The assessment should include site visits to correctional facilities and the research and information system units. The tasks of this phase of the initiative will vary according to
organizational structure and classification issues; however, the following activities
should always be completed:
Interview the central office and facility-level classification staff. The team
should conduct face-to-face interviews with key supervisory and line classification staff concerning their specific classification tasks and should identify any
concerns about the current classification process. The purposes of these interviews are to clarify the agency’s current approach to classification, ascertain the
pragmatic issues of concern to line staff, and review current classification policies, procedures, and instruments.
Observe the classification process. The site visit team should have access to
case files, observe the classification process, and identify the criteria for making
custody decisions. These data, combined with the staff interviews, should provide insight into both the availability and quality of the data required to score the
current initial and reclassification instruments, new risk factors suggested by the
steering committee, the degree of discretion associated with the classification
process, and the reliability of the classification scores and custody designations.
Review the results of the site assessment. The site visit team should present
their findings to the steering committee. During this meeting, any unresolved
issues should be addressed and a consensus reached on the project tasks, the
specific responsibilities of the steering committee members, and the feasibility
of the project’s time-task line.
Step 2. Compile baseline data. Baseline data should be collected for each of the
outcomes identified by the steering committee and stakeholders. These may include
rate of institutional violence, number of housing transfers, number of discretionary
overrides at initial classification and reclassification, rate of institutional misconduct
by custody level, and the custody distribution at initial classification and reclassification. Each statistic should be compiled separately by gender, facility, and custody
level.
Step 3. Prepare the classification assessment report. Within 2 weeks of completing the assessment, the team should prepare a draft report documenting its activities, the baseline data, and the findings of the assessment. The report should
describe the agency’s current practices, identify the classification issues to be
addressed by the initiative, and document the agreed-upon project time-task line.
The draft report should be circulated to the commissioner, steering committee,
director of classification, and key staff who participated in the assessment.
xv

Executive Summary

Phase III: Planning
Step 1. Learn about promising systems, models, strategies, and best practices.
Obtaining this information by reviewing the literature or obtaining technical assistance from a provider with expertise in classification will help avoid some of the unanticipated pitfalls in trying to reinvent a system that already has been developed or
refined by another jurisdiction. An important strategy for learning about models and
promising approaches is to contact both comparable state agencies that have implemented the model and agencies that may have considered the model but rejected it.
Step 2. Design or modify the classification system. The specific tasks at this stage
will vary according to the findings of the assessment. However, the following substeps should be completed whether a new system is being developed or the existing
system modified:
Substep 2.1: Develop a preliminary classification system. At this stage, the
committee must consider how to proceed to achieve its goals and objectives. To
develop a preliminary classification system, the committee must reach consensus regarding the classification process (instruments, schedule, staffing, quality
control, MIS, strengths, and weaknesses), and the content and format of the preliminary classification forms (risk factors, risk criteria, offense severity, institutional disciplinary code, factor weights, scale cut points, and overrides).
Substep 2.2: Design prototype instruments and manual. Based on the decisions reached in substep 2.1, the steering committee must develop prototype
instruments and a training manual to document the new or revised system. The
operational definitions and instructions for each classification factor should be
specified.
Substep 2.3: Pilot test the prototype instruments and manual. This key task
requires drawing representative samples of key prisoner populations, developing a supplemental data collection instrument and coding instructions, and then
collecting and analyzing the data.
Step 3: Develop an action plan. Developing a comprehensive plan for the implementation of the new system is critical to the success of the classification initiative.
The steering committee should consider the implementation process at each point in
designing and testing the system. The action plan must consider questions regarding staffing, training, timing, stakeholders, materials, the MIS, evaluation, budget,
and automation of the system. It should also include goals, objectives, and specific
timelines for implementing the system.

Phase IV: Implementation
Step 1: Reengage, reorient, and reeducate stakeholders. The steering committee
should present an overview of the new or modified system to the commissioner and
all stakeholders. This presentation should review the issues and trends precipitating
xvi

Objective Prison Classification: A Guide for Correctional Agencies

the initiative, the initiative’s goals and objectives, the design and pilot test activities,
the baseline and pilot test data, and the committee’s recommendations.
Step 2: Train staff. All correctional staff should receive training in using the new
system, and classification staff should receive specialized training that covers topics
such as instrument use, information management, resource allocation, and program
development decisions. All training sessions should include an overview of how the
new system was developed to acquaint staff who were not members of the classification committee with the background of the system. Ongoing in-service training
should supplement the initial orientation and implementation training.
Step 3: Implement the new/revised system. A detailed time-task line that accounts
for the complexity of implementation across facilities and staffing patterns must be
developed to serve as the guide for all implementation-related activities. It should
be fully explained to all stakeholders.

Evaluations of Prison Classification Systems
Agencies that have implemented a new or modified prison classification system
must submit the system to comprehensive and rigorous evaluation to establish that
the classification system is working as intended, reconfirm its validity, and assess its
impact on the intended aspects of the prison system. The main goal of the impact
evaluation is to examine whether the new system (the independent variable) may be
regarded as the cause of selected desired outcomes (dependent variables). Process
evaluations are used to determine whether a system is functioning as planned. A
process evaluation of a classification system should assess the system’s reliability
and validity and can be viewed as an audit procedure that should be an ongoing feature of any classification system.
Much of the work of a process evaluation can be accomplished easily if the system
is fully automated. An automated classification system facilitates the taking of periodic statistical “snapshots” of the prisoner population that are needed to produce
management reports and answer questions such as whether prisoners are being classified in a timely manner and housed according to the classification system and
whether overrides are being used properly.

Classification of Women Prisoners
Because U.S. correctional systems—both facilities and policies—were originally
designed to accommodate male prisoners, they are based on risk factors that have a
tenuous relationship, at best, to the behavior of women prisoners. The constellation
of characteristics and needs commonly seen in prisoners manifests itself differently
in women than in men. Whereas women generally pose little threat of institutional
violence or escape, their significant substance abuse and mental health needs can produce behaviors that are difficult to predict. These differences are particularly important to institutional classification systems but remain underresearched.

xvii

Executive Summary

In 1994, the National Institute of Justice (NIJ) sponsored a study to assess programming needs and promising approaches for incarcerated women.1 The identified
management issues included shortcomings in classification procedures. Administrators reported that their systems did not collect and compile adequate information on
the risk factors and needs of women offenders and that, therefore, the systems were
not useful in matching the women to appropriate custody levels or programming.
Furthermore, the classification and screening instruments were often unrelated to
where the women were housed or to which programs they had access. Lack of bedspace and constant movement of large numbers of women were cited as specific
operational barriers.
Dissatisfied with current classification systems, correctional administrators have
been faced with three basic options: 1) use the current instruments and override the
scored custody levels; 2) modify the current risk factors and/or scale cut points; or
3) discontinue use of the current instruments and classify women prisoners based on
a subjective, intuitive process. NIC-sponsored initiatives have provided some
insight into the viability of these three options.
Use the current instruments and override the scored custody levels. This is
the most popular strategy for making classification systems more responsive to
women prisoners. It is used as an interim solution by many states until they can
undertake a validation study and incorporate the changes necessary to make
their classification system gender specific. Although useful as a short-term
means for addressing a state’s concerns about overclassification, this strategy is
problematic because the classification decisions are based on subjective overrides rather than statistically validated risk factors.
Modify the current risk factors and/or scale cut points. This is the most common strategy employed by systems that have undertaken a validation study and
found a statistically significant difference in the predictive power of the risk factors for their male and female prisoners. Across jurisdictions, the research findings have been somewhat inconsistent; however, the most common risk factors
modified to better assess the risk posed by women offenders are age, criminal
history, current offense, and stability factors such as institutional adjustment,
education, history of substance abuse, and relationships and mental health.
Discontinue use of the current instruments and classify women prisoners
based on a subjective, intuitive process. The most common variant of this
rarely chosen option is to classify women prisoners using the agency’s standard
instruments and process but use this classification only to determine a woman’s
eligibility for work assignments outside the security perimeter and the level of
supervision she requires when leaving the facility grounds for court hearings or
medical appointments. That is, the woman’s custody level has little impact on
the facility, housing unit, program(s), or institutional jobs to which she is

xviii

1
M. Morash, T.S. Bynum, and B.A. Koons, Women Offenders: Programming Needs and Promising
Approaches, Research in Brief, Washington, DC: U.S. Department of Justice, Office of Justice Programs,
National Institute of Justice, 1998.

Objective Prison Classification: A Guide for Correctional Agencies

assigned. These decisions are based on subjective factors such as bed or program availability, the relationships between the women and/or staff members,
and/or efficiency. Regardless of their custody levels, all women prisoners have
access to the same housing units, programs, institutional work assignments,
recreation activities, visitation privileges, and so forth.
NIC has long advocated for the validation of any classification system within the
population to which it is to be applied. Although more research is needed before
final conclusions are drawn, modifying current risk factors and/or scale cut points
is the best option for making classification systems more responsive to the risk and
needs of women prisoners because it refines risk factors and tests them in the population to which they will be applied. This strategy also provides the opportunity to
develop and test new factors to assess the risk posed by women prisoners. The other
two strategies, which do not rely on objective, reliable assessments, are at best shortterm options for managing women prisoners during the development and pilot testing of a more gender-responsive system.

Other Special Topics and Issues in Classification
Two issues have been identified as of special concern with regard to the management of prison populations: the impact of environmental factors and prison management on prisoner and staff behavior and the need to link prison classification and
risk assessment with release decisions.

The Impact of Environmental Factors and Prison Management
on Prisoner Behavior
Very little is understood about how the physical environment of the prison and style
of prison management influence prisoner and staff behavior. It would be difficult to
find a correctional official, warden, superintendent, or line officer who does not
agree that a facility’s architectural design influences prisoner behavior or that similarly designed facilities with similarly situated prison populations can produce very
different rates of prisoner misconduct, both within and across state prison systems.
Unfortunately, few if any studies have assessed the impact of architecture on suppressing or controlling prisoner behavior, and except for a few recent evaluations of
the use of internal classification systems, no studies have substantiated the observation that variations in misconduct rates for prisons equivalent in design and prison
population are likely related to differences in the management style adopted by each
prison administrator.
Also needed are formal assessments of the often advocated, but still highly controversial, super maximum-security facilities. Specifically, basic research is needed to
determine how best to identify prisoners who require this level of segregation, how
long they should remain segregated from the general population, what interventions
should be used to control their high-risk behavior, how to safely reintegrate them
into the general population, and how they behave after release from these units.
xix

Executive Summary

Without such basic research, it will be difficult to propose new methods for identifying high-risk prisoners and to apply interventions that will help control and manage them.

The Need to Link Prison Classification and Risk Assessment
With Release Decisions
The past three decades have witnessed an unparalleled increase in the nation’s
prison population, yet some states are showing signs of beginning to reduce their
prison populations. Several states are taking administrative and legislative actions to
either divert offenders or reduce their period of incarceration. Interest in reentry—
the transition from incarceration to the community—is also increasing. Accompanying this growing commitment to reduce the number of offenders who are
rearrested and/or reincarcerated is a trend toward releasing prisoners with no form
of parole or community supervision. These trends raise concerns about the lack of
programming and services for incarcerated prisoners as well as for those who have
been released to the community.
State correctional agencies facing increasing pressures to control or reduce prison
populations will need to use classification and risk assessment instruments to inform
the following key decisions regarding prisoners:
What custody level and type of programs are appropriate for the prisoner during incarceration?
When should the prisoner be released and under what forms of supervision and
services?
Answering these two basic questions will require well-coordinated and virtually
seamless classification and risk assessment processes from the time the prisoner is
admitted to the prison system through the prisoner’s eventual release from parole or
other forms of postincarceration supervision. Improving the ability to assess and
manage the level of risk posed by the millions of persons who pass through the
nation’s probation, prison, and parole systems each year is a goal correctional agencies can no longer afford to ignore or neglect.

xx

one

Chapter

Introduction
During the past two decades, professionals in prisons and those in other correctional systems worked assiduously to improve their approaches to classifying offenders
according to custody, work, and programming needs. Fueled by litigation, overcrowding, growing expenditures, and public safety concerns, they made classification systems the principal management tool for allocating scarce prison resources
efficiently.
Classification systems help to minimize the potential for prison violence, escape,
and institutional misconduct. These systems are expected to provide greater
accountability and to help forecast future bedspace and prisoner program needs.
Considered the “brain” of correctional system management, classification systems
are essential for projecting the future resource needs of a correctional agency. A
properly functioning classification system governs many important decisions,
including those that heavily influence fiscal matters such as construction, staffing
levels, bedspace, and programming expansion.

Classification systems
help to minimize
the potential for
prison violence,
escape, and institutional misconduct.

Since the 1980s, objective prison classification systems have been widely adopted
in the United States, Canada, Europe, New Zealand, and Australia. These systems
for managing a rapidly expanding prison population have produced significant
results. Overclassification has been reduced, institutional violence has declined, criteria for custody decisions have been validated, custody decisions are more consistent, and prisoner program needs are assessed more systematically.
This report summarizes for correctional administrators the current state of the art in
prison classification. The report’s principal focus is the use of prison classification
instruments for custody or security rating purposes. Chapters 2 and 3 of the report
provide a brief history of the evolution of classification systems and define the elements of objective prison classification, including key terms, concepts, and essential components. Chapter 4 lays out detailed guidelines for implementing a
classification system, and chapter 5 reviews methods of evaluating a system’s effectiveness once it has been implemented. The final two chapters of the report address
the classification of women prisoners and other special topics in classification.

1

two

Chapter

Objective Prison Classification:
A Brief History
Rudimentary forms of segregating prisoners have been in use for as long as incarceration has been used as a punishment for convicted offenders. During the nineteenth and early part of the twentieth centuries, prisoners in the United States were
separated by age (adult versus juvenile), gender (male versus female), number of
offenses (first versus repeat), and special needs (mentally ill). Humane treatment
was the basis for these early and most basic classification systems. The Reformatory Movement and the Progressive Era gave rise to the progressive treatment and
rehabilitation of confined offenders. The so-called medical model, which included
identifying the problems of individuals and prescribing appropriate treatments to
address their individual needs, was the essence of this diagnostic approach to classification. Moreover, the treatment and rehabilitation of confined offenders were
expected to become the foundation of corrections.
Gradually, the emphasis on diagnosing the causes of offenders’ antisocial behaviors
was replaced by an emphasis on the types of treatment programs available. Unfortunately, early classification systems were largely grounded in subjective criteria or
clinical assessments that produced arbitrary and unreliable results. The lack of standardization in the assessment processes used in making custody-level decisions led
to well-founded claims that the classification systems were inherently flawed in
their ability to produce reliable and valid results. Courts issued consent decrees and
other orders because of the capricious and discriminatory nature of subjective classification systems.

Early classification
systems were largely
grounded in subjective criteria or clinical
assessments that produced arbitrary and
unreliable results.

Claims regarding humanity and equitable prison procedures have been common
themes in litigation cases on behalf of prisoners. Several influential court cases in
the United States found that prison classification systems must demonstrate that
they are not arbitrary and capricious.1 That is, classification systems must demonstrate that they are reliable and valid by using factors known to be associated with
prisoner misconduct, escape, and risk to public safety.
The use of reliable and valid criteria to assess a prisoner’s custody level is one of the
core distinguishing features of an objective classification system. Beginning in the
1970s, prison classification systems began to experiment with objective criteria. In
the United States, the California Department of Corrections and the Federal Bureau
of Prisons developed the first objective classification systems. These systems used
numerical scoring that sought, among other things, to improve the consistency and
objectivity of the assessment process.
3

Chapter 2

A number of developments drove this necessary shift in classification. In the 1970s,
policymakers began a 30-year effort to “get tough on crime.” Various legislations
were passed (e.g., truth-in-sentencing and “three-strikes-and-you’re-out” laws and
changes in sentencing practices) that dramatically increased the prison population.
As the nation’s prison population grew, prison disruptions became more frequent.
The early 1970s witnessed the most volatile and violent era of prison riots in many
of the United States’ largest prisons. Consequently, the need to gain better control
of a rising and increasingly disruptive population accelerated.

During the 1980s
and the 1990s, public
safety concerns led to
the development of a
variety of risk assessment and prisoner
treatment systems,
all of which promised
to deliver reliable and
valid assessments.

The rapid surge in the prison population placed an added burden on government
spending. Policymakers demanded new and better methods for identifying and controlling high-risk prisoners and a more scientific process for forecasting future
prison bedspace requirements. As crowding began to characterize most of the states’
prison systems, allegations of prisoner abuse and mistreatment triggered a wave of
lawsuits. These lawsuits challenged traditional prison operations in general and
classification procedures in particular. All these changes increased pressures on correctional staff, who were already working under time and resource constraints, to
deliver an intensive analysis, diagnosis, and prognosis for each prisoner.
During the 1980s and the 1990s, public safety concerns led to the development of a
variety of risk assessment and prisoner treatment systems. Quite often, risk assessment places greater emphasis on measuring a prisoner’s likelihood of reoffending
than it does on how the prisoner should spend the time while incarcerated. Competing demands and restraints within correctional agencies have resulted in a variety of prison classification systems, all of which promised to deliver reliable and
valid assessments to manage, treat, and ultimately return prisoners to the community. Proper classification and housing of all prisoners, including high-risk and special
management prisoners, are the product of objective assessment methods and sound
professional judgment, both of which must meet legal requirements. Despite a shift
toward actuarial instruments, prison populations cannot simply be scored. Rather,
data and input from a diverse array of correctional professionals representing fields
such as security, medicine, mental health, education, and the prison industry are
required to make the right classification designation.
The classification decision is followed by equally important decisions concerning
proper housing and treatment. Providing proper housing and treatment affords the
prisoner the opportunity to live in a less restrictive correctional environment and,
ultimately, to return to the community. This decision course requires timely and
appropriate reviews of the prisoner’s classification status from admission to the
prison system to release back to the community.
Finally, prison classification instruments should measure the risk a prisoner poses
both inside and outside the prison walls. Security risk assessments measure the likelihood of a prisoner engaging in high-risk behavior or attempting to escape while

4

Objective Prison Classification: A Brief History

incarcerated. Conversely, public risk assessments measure the likelihood of a prisoner engaging in criminal activities when released to the community. To safeguard
staff and prisoners inside the prison walls and the public outside, it is imperative that
both security and public risk measurements use appropriate factors and variables.
Unfortunately, if rehabilitative programs, treatments, and services are not available
to assist offenders when they reenter the community, then offenders are more likely to recidivate, placing communities at risk.

5

three

Chapter

What Is Objective Prison
Classification?
Classification systems fulfill a wide range of correctional purposes, including preserving order in an institution; sustaining prisoner discipline; assessing prisoners’
needs; assigning prisoners to appropriate programs; providing equitable treatment;
protecting staff, prisoners, and the public; allocating prison resources; and planning
for prison management. Many correctional agencies believe they have an objective
classification system but are unable to identify the basic elements of such a system.
This chapter defines key terms and concepts necessary to understanding classification systems and outlines the essential components of an objective classification
system. This information will help agencies conduct a self-assessment to determine
whether their classification system meets the requirements for an objective system.

The subjective nature
of early classification
systems led to several
influential cases in
which the courts
ruled that classifica-

Key Terms and Concepts

tion systems must

Three pairs of terms appear regularly in any discussion of objective prisoner classification: “reliability” and “validity,” “external” versus “internal” prisoner classification systems, and “general” versus “special” prison populations. The following
sections define these key terms and explore the concepts they embody.

use valid criteria,

Reliability and Validity

prisoner misconduct.

i.e., factors known
to be associated with

Two core distinguishing features of an objective classification system are that it
must use accurate, or valid, criteria to assess a prisoner’s custody level and that it
must ensure that these criteria produce consistent, or reliable, decisions.2 The subjective nature of early classification systems led to several influential legal cases in
which the courts ruled that classification systems must use factors known to be associated with prisoner misconduct. That is, systems used to classify prisoners must
produce valid results.
Validity. Validity refers to the accuracy of the classification system in predicting a
prisoner’s behavior and assigning him or her to an appropriate risk level. Risk can
be defined and measured in several ways. Custody and public risk instruments are
designed to predict a prisoner’s risk to self, other prisoners, and staff; risk of escape;
and risk to the public. They classify prisoners according to their propensity to comply with institutional rules and regulations, commit violent acts, and/or to attempt
escape while incarcerated. Because prison escapes and violence are quite rare, it is
difficult to develop an instrument that accurately predicts these events.

7

Chapter 3

The term “validity” generally pertains to face and predictive validity. Face validity
refers to whether the terms used to classify prisoners make sense to the staff using
them: Do the factors appear to be, or are they logically associated with, the predicted behavior? Predictive validity refers to whether the factors used to classify prisoners demonstrate a capacity to predict risk based on a statistical test of association.
A form of face validity relates to the general policies of the prison. Administrators
often do not want certain prisoners—for example, sex offenders and high-profile or
celebrity prisoners—placed in low-security settings, regardless of their risk of
escape or of committing violent crimes. In these situations, the classification system
helps to maintain prison security by minimizing the potential for a well-publicized
negative incident.
Reliability. Reliability refers to consistency in making classification decisions, both
among classification staff (interrater reliability) and by a single staff person (intrarater
reliability). Reliability considers whether the same decision will be rendered if the
assessment is repeated by the same or a different staff member. In general, the more
complicated the classification process, the less reliable it will be. Agencies must evaluate the reliability of their classification systems because, by definition, a classification process or risk instrument that is unreliable is not a valid system. Chapter 5 of
this report provides guidance on evaluation of classification systems.

Reliability refers to
consistency in making
classification decisions, both among
classification staff
(interrater reliability)
and by a single staff
person (intrarater
reliability).

8

Overrides. There are two types of overrides: nondiscretionary and discretionary.
Nondiscretionary overrides are static in that they reflect an agency’s policy, which
typically restricts the placement of certain prisoners in minimum- or low-security
facilities. Agency policy decisions usually apply in all cases. For example, a correctional agency may prohibit a sex offender assessed as a minimum-custody level
prisoner from residing in a minimum-security unit until within 6 months of the prisoner’s projected release date and/or until the prisoner has completed sex offender
treatment. Likewise, an agency’s policy may prohibit a prisoner of any classification
status who has more than 2 years to serve from residing at a minimum-security level
unit. In these examples, the classification staff would have no discretion over the
prisoner’s designated custody level.
Discretionary overrides are dynamic in that they reflect the professional judgment
of trained classification staff to account for other factors not explicitly used in the
scoring process. The classification staff have the discretion to recommend a higher
or lower custody level based on other pertinent factors associated with the prisoner.
For example, the classification staff may recommend a lower custody level because
of the prisoner’s exceptional behavior during a prior incarceration. Alternatively, the
staff may override the custody level of a prisoner classified as medium custody
because the nature of the prisoner’s current offense is more serious than indicated
by the initial classification form. In this case, the prisoner’s medium-custody classification would be overridden to place the prisoner at a maximum-security unit.

What Is Objective Prison Classification?

External Versus Internal Prisoner Classification Systems
External classification determines a prisoner’s custody classification and facility
assignment. Internal classification governs facility-level decisions such as where
and with whom the prisoner will be housed, the types of programs and services to
which the prisoner should be assigned, and the prison industry or work assignment
most appropriate for the prisoner. Most often, the intent of internal classification
systems is to ensure that prisoners who are at risk for placement in a special management population are supervised accordingly. Minimizing classification errors
requires both external and internal classification systems.
External classification systems, which focus on interinstitutional placement, are
well established within most state prison systems. Conversely, well-structured internal classification systems, which focus on intrainstitutional decisions, are in their
beginning stages. Even so, both external and internal classification systems are
essential for minimizing classification errors, housing and managing prisoners, and
safeguarding prison staff, prisoners, and the public. Exhibit 1 illustrates how external and internal classification systems function within a prison system.
External prisoner classification. Shortly after a prisoner is admitted to the prison
system, the external classification system places the prisoner in one of several custody levels that directly affects the type of facility to which the prisoner will be
assigned and, once there, the level of supervision the prisoner will receive. Custody
levels are defined in different manners; however, they generally conform to the broad
categories of minimum, medium, close, and maximum security. Some classification
systems designate these custody levels numerically (e.g., from I through IV).
The external prison classification system consists of both an initial and a reclassification instrument. Based on the prisoner’s attributes at the time of admission, the
initial instrument uses a predictive model to estimate the prisoner’s custody risk
while incarcerated. Since the institutional adjustment of a prisoner who has no history of incarceration is unknown, the initial instrument places greater emphasis on
the prisoner’s current offense, criminal history, and a number of socioeconomic factors known to be associated with institutional conduct. Professional staff use the initial screening process and initial classification form to determine the most
appropriate level of custody for the prisoner.

Both external and
internal classification
systems are essential
for minimizing classification errors, housing and managing
prisoners, and safeguarding prison
staff, prisoners, and
the public.

After a custody level has been defined, the prisoner is transferred to a correctional
facility that best meets his or her security and program needs. For external classification, the decision to place a prisoner in the general prison population and the subsequent decision regarding the appropriate level of supervision for the prisoner within
the general population are critical to the efficient operation and safety of the facility.
A prisoner’s external custody level is often modified through a reclassification
process, which places greater emphasis on the prisoner’s behavior and conduct while
incarcerated than does the initial classification instrument. No later than 12 months
after the initial classification process, a reclassification instrument is used to score the
prisoner on factors such as the type and number of misconduct reports lodged against

9

Chapter 3

Exhibit 1. Overview of External and Internal Classification Systems

Admission to prison

Initial external classification
Custody assessment
Program needs assessment
Facility designation

Transfer to facility

Initial internal classification
Housing assignment
Program assignment
Work assignment

Transfer to designated housing area

Reclassification
External reclassification
❖ Custody

Internal reclassification
❖ Programs ❖ Facility
❖ Housing
❖ Community
programs
❖ Work

Source: Internal Prison Classification Systems: Case Studies in Their Development and Implementation
(Hardyman et al., 2002).

the prisoner, the prisoner’s participation in a variety of programs offered by the
prison system, and the prisoner’s work performance. As a result, some of the initial
classification factors may be deleted or reduced in their scoring importance.
By emphasizing the prisoner’s conduct while incarcerated, the reclassification
process represents a “just desserts” model in which the classification system
10

What Is Objective Prison Classification?

rewards prisoners for good behavior by allowing them to work their way to lower
custody levels over time. A reclassification process that does not allow prisoners to
work their way to lower custody levels results in significant overclassification of
prisoners who were originally assigned higher custody levels because they were
convicted of serious crimes. These prisoners may establish good institutional conduct records, but in the absence of a reclassification process, they will remain in the
custody level to which they were initially assigned—that is, a higher custody level
than required to ensure the safety and security of the institution.
Internal prisoner classification. To improve institutional security, some prison
systems use a second layer of classification, which is the internal classification system. When a prisoner arrives at the facility, the internal classification staff determine
the prisoner’s housing unit and cell, facility program needs, and work assignment.
Like external classification systems, formal internal classification systems may
include structured scoring instruments, formally trained classification specialists,
and a reclassification process to update previous classification designations.

It is important to differentiate between
prisoners housed in
the general popula-

Approaches to internal classification systems vary as to their purpose, process,
instruments, and level of automation. The internal classification system identifies
prisoners according to personality or behavioral typologies. Psychologists designed
some of the early offender typology systems, which use reasonably well-researched
criteria for internal classification. Other offender typology systems simply use specialized criteria applied by individual institutions to augment the external classification system.

tion and those who

Internal classification systems strive to improve prisoner management at the facility level by addressing housing, programming, and compatibility issues. Most often,
the intent of internal classification systems is to ensure that prisoners who are at risk
for being placed in a special management population are supervised differently than
other general population prisoners. By design, internal classification systems complement objective custody classification or external classification systems. The task
of an internal system is to devise appropriate housing plans and program interventions within a particular facility for prisoners who share common custody levels,
whether minimum, medium, close, or maximum.

tive custody, special-

require special placement in administrative or disciplinary
segregation, protec-

ized mental health,
or medical units.

General Population Versus Special Prison Populations
It is important to differentiate between prisoners housed in the general population
and those who require special placement in administrative or disciplinary segregation, protective custody, specialized mental health, or medical units. National estimates of the average daily population (ADP) show that approximately 80 percent of
all prisoners are initially assigned to the general prison population (Austin and
McGinnis, 2004). These data show that assigning a prisoner to a special prison population often occurs after the prisoner has been classified, when, because of disruptive behavior or other problems, it becomes obvious that he or she needs to be
removed from the general population.
11

Chapter 3

Furthermore, there is a pool of general population prisoners who are classified as
maximum or close custody because of disruptive behavior patterns or the suspicion
that they may be involved in disruptive behavior. Appropriately viewed as high risk,
these general population prisoners warrant special observation, monitoring, housing, and programming. Such preemptive measures by staff may negate the need for
more restrictive interventions.
In general, the vast majority of prisoners never become disruptive or difficult to
manage. The most serious forms of disruptive behaviors within a prison, such as
homicide, escape, aggravated assault on other prisoners or staff, and rioting, are rare
events. Prison staff and prisoners seldom become the victims of prison violence.
Moreover, it is difficult, if not impossible, to predict when an act of prison violence
will occur.
California is one of the few states that publicly report the rate of serious incidents
within their prison systems. The California Department of Corrections (CDC)
defines serious incidents as homicide, assault and battery, attempted suicide, suicide, possession of a weapon, and possession of controlled substances. In 2001, the
CDC housed 155,000 prisoners, and the rate of serious incidents was approximately 8 per 100 prisoners per year.3 About half of these incidents were assaults and batteries. The assault rate was 0.4 per 100 prisoners. There were also 30 suicides and
13 homicides. If one were to compute a homicide rate for the CDC population, it
would be approximately 8–9 per 100,000—just slightly higher than the state homicide rate of 6.4 per 100,000. Given the demographics of the CDC population, which
is mostly young males with criminal histories, one can argue that the homicide rate
for those incarcerated in a CDC facility is actually lower than it would be had they
remained in the community.
Based on the CDC’s categorization of serious incidents, those of other state prison
systems, and known high-risk behaviors, a typology of high-risk and special management prisoners has been established. These categories, shown in exhibit 2, are
not mutually exclusive. For example, a prisoner may be categorized as “chronic
misbehavior—assaultive,” but the prisoner’s conduct may be related to a mental
condition. In practice, most of these designations are based on subjective judgment;
however, staff members are sufficiently trained to make such assessments and use
additional sources of objective data to reach a reliable determination.
Although the rate of violence in most prisons is low, approximately 10–15 percent
of the nation’s prison population is classified as special management, based on the
typology shown in exhibit 2, and another 10–15 percent is classified as maximum
or close custody within the general population, as shown in exhibit 3 (Austin and
McGinnis, 2004). Although these high-risk prisoners reflect a minority of the
national incarcerated population, a disproportionate amount of staff and agency
resources must be allocated to them to safeguard prisoner safety and institutional
security. As a result, an agency’s ability to fund more services and programs for the
larger general population is diminished.
12

What Is Objective Prison Classification?

Exhibit 2. Typology of High-Risk and Special Management Prisoners
Category and Assessment Method

Placement

Security threat group
Subjective assessment based on at least three sources of independent objective data as applied to well-defined agency criteria.

Administrative segregation or general
population—high custody.

Likely victim
Subjective assessment based on at least three sources of independent objective data as applied to well-defined agency criteria.

Protective custody or restricted general
population facilities.

Mentally ill
Standardized psychometric tests and clinical judgment by mental
health staff.

Mental health unit and/or administrative
segregation.

Chronic misbehavior—assaultive
Objective external classification.

General population—high custody, administrative segregation, or mental health unit.

Chronic misbehavior—nonassaultive
Objective external classification.

General population—high custody, administrative segregation, or mental health unit.

Nonsexual predator
Subjective assessment based on at least three sources of independent objective data as applied to well-defined agency criteria.

General population—high custody, administrative segregation, or mental health unit.

Sexual predator
Subjective assessment based on at least three sources of independent objective data as applied to well-defined agency criteria.

General population—high custody, administrative segregation, or mental health unit.

Developmentally disabled
Standardized psychometric tests and clinical judgment by mental
health staff.

General population (all custody levels) or
mental health unit.

Source: Classification of High-Risk and Special Management Prisoners: A National Assessment of Current Practices (Austin and McGinnis,
2004).

Improving treatment programs and services for some prisoners and ensuring that
prisoners are not overclassified may help to reduce the administrative segregation,
medical, mental health, and special management populations, which, in turn, could
free up funds that could be allocated for services and programs for the larger general prison population.
To date, correctional officials have relied on reactive rather than proactive management strategies to control high-risk or special management prisoners. In the 1980s,
the reactive approach to controlling special management prisoners (e.g., organized
gangs within the prison systems) produced larger, more secure, and more heavily
staffed administrative segregation and super maximum-security housing units in
many of the states’ prison systems. Unfortunately, much less attention has been
given to the use of classification tools and aggressive management techniques that
would serve to prevent potentially dangerous incidents and thwart prison violence.

13

Chapter 3

Exhibit 3. Nationwide Estimated Prisoner Classification Levels for the
Average Daily Population
Custody Level

Percent

General population

80

Minimum/community

35–40

Medium

35–45

Maximum/close

10–15

Special populations

15

Administrative/disciplinary segregation

5–6

Protective custody

1–2

Severe mental health

1–2

Severe medical

1–2

Unclassified

5

Source: Classification of High-Risk and Special Management Prisoners: A National Assessment of Current
Practices (Austin and McGinnis, 2004).

Nevertheless, one cannot and should not rely on classification and risk assessment
instruments as the sole means for identifying high-risk prisoners. Any prisoner classification system is subject to two critical types of errors: false-positives and falsenegatives. A false-positive, or overclassification, error occurs when a prisoner is classified as positive for high-risk behavior but does not exhibit such behavior. A prisoner
placed in administrative segregation for alleged drug distribution activities who has not
been involved in such activities is an example of a false-positive classification error.

Nevertheless, one
cannot and should
not rely on classification and risk assessment instruments as
the sole means for

A false-negative, or underclassification, error occurs when a potentially disruptive
prisoner is classified as low risk (i.e., negative for high-risk behavior) but exhibits
high-risk behavior. An example of a false-negative error would be a prison’s failure
to identify and segregate an active gang member who, as a result, is better able to
continue to control and coordinate gang activities.
Classification errors can lead to serious incidents, including assaults on staff and
prisoners, suicides or suicide attempts, and escapes. Such episodes are extremely
disruptive and costly to a facility’s operation and expose the agency to expensive litigation. Reducing classification errors by using better risk assessment procedures
affords staff the opportunity to intervene proactively with more appropriate prisoner
supervision strategies and programs.

identifying high-risk
prisoners.

14

Essential Components of an Objective Prison
Classification System
A classification system is a formal process used to categorize and manage prisoners. Classification systems rely on specially trained staff, the use of reliable and

What Is Objective Prison Classification?

valid criteria, objective and accurate offender data, and a means for monitoring and
evaluating their impact on the correctional system. The various aspects of these
essential components are discussed in the following sections.

Mission Statement
A mission statement is the foundation of an objective classification system. However,
a classification system does not stand alone. Rather, it is an integral part of the much
larger prison system and, as such, its mission statement should be consistent with the
agency’s mission, goals, and resources and reflect prisoner risks and service needs.
The mission statement of the classification system should also reflect the formal
process for classifying and managing prisoners within the prison system. Finally,
the mission statement should include a clear and comprehensive statement of the
core values and philosophy of the agency in accordance with national and local correctional standards. These values should include a commitment to:

At a minimum, classification objectives

Adhere to all federal, state, and local laws and regulations regarding the prison’s
operations.

should be reviewed

Ensure public safety.

old challenges have

Ensure the safety of both correctional staff and prisoners.

been met and to

Provide essential medical, mental health, educational, and program services.

identify new ones

Manage agency resources effectively.

to be addressed.

annually to verify that

Classification Goals and Objectives
After an agency has formulated its mission statement, it is ready to articulate its
classification goals and objectives. These must directly support the core values set
forth in the mission statement and be formulated so that their outcomes can be
measured and evaluated.
Evaluating and measuring the outcomes of classification objectives at regular intervals enables agencies to better gauge their success over time. At a minimum, classification objectives should be reviewed annually to verify that old challenges have
been met and to identify new ones to be addressed. Some examples of how these
goals and objectives can be stated to support the mission statement are shown in
exhibit 4.

Dedicated Classification Unit and Classification Staff
Prisons are labor-intensive operations. An estimated 60–70 percent of a prison’s
operational budget is allocated to salaries, benefits, and other labor-related costs. An
established classification unit is necessary for a prison classification system to function properly. The classification unit must have a sufficient number of dedicated,
well-trained, and experienced staff to ensure that all prisoners are properly classified
15

Chapter 3

Exhibit 4. Examples of Agency Mission Goals and Classification Objectives
Mission Goal

Classification staff
must have the sole
authority to assign
prisoners to housing

Classification Objective

Comply with all federal,
state, and local laws and
regulations regarding the
prison’s operations.

Reduce the amount of litigation pertaining to conditions
of confinement.
Reduce the number of grievances and amount of litigation regarding staff working conditions and harassment.

Ensure public safety.

Reduce incidents of escapes and walkaways.
Reduce the number of crimes committed by prisoners
who have been released from prison to communitybased programs.

Protect staff and ensure
prisoner safety.

Reduce the number of prisoner-on-staff assaults.
Reduce the number of prisoner-on-prisoner assaults.

Provide essential medical,
mental health, educational,
and program services.

Ensure that qualified medical, mental health, and
classification staff properly screen new prisoners.
Based on the initial assessment, ensure that all prisoners
needing medical, mental health, education, and other
program services receive such services in a timely
manner.

Manage agency resources
effectively.

Reduce the amount of overtime.
Ensure that security staff are properly deployed in the
prison’s housing unit.

in a timely manner. Even in small agencies, this is a strict requirement. Understandably, classification staff may have other duties that prohibit them from dedicating 100 percent of their time to prisoner classification. However, as part of their
primary responsibilities, designated staff members must be held accountable for
administering classification policies and procedures. For example, conducting initial prisoner screenings, which includes identifying severe medical and mental
health needs, is one of the most important classification functions exercised by
intake center staff. Although these staff may not be directly assigned to the classification unit, they must be able to conduct initial screenings in a professional, reliable, and accurate manner.

units according to the
classification system.

Centralized Control Over All Prison Transfers and Housing
Decisions
Classification staff must have the sole authority to assign prisoners to housing units
according to the classification system. This does not mean that other staff members
cannot recommend transfers in emergencies. However, even in emergency circumstances, classification staff must review and approve all transfers. The most effective way to ensure that all staff comply with this requirement is for the prison to
institute and enforce a written policy that clearly delineates the authority and powers of the classification unit.

16

What Is Objective Prison Classification?

Reliable and Valid Classification Instruments
Objective classification systems use well-structured instruments (i.e., forms) to
guide the assessment process. These forms are designed to produce reliable and
valid assessments of the risks posed by the prisoners, which are essential for a wellfunctioning classification system.
Three basic forms guide the classification process: the initial screening form, the
initial classification form, and the reclassification form. The initial screening form
is used to help identify a prisoner’s emergency needs when the prisoner is first
admitted to the prison system. The initial classification form determines the prisoner’s scored custody level on admission to the prison system, and the reclassification
form is used to reevaluate the prisoner’s scored custody level throughout his or her
incarceration based on his or her institutional conduct.

Three basic forms
guide the classification process: the initial screening form,
the initial classification form, and the
reclassification form.

Initial screening form. Completed immediately after the prisoner is admitted to the
prison system, the initial screening form helps the intake center staff to identify his
or her emergency needs. Unlike the initial classification and reclassification forms,
the initial screening form is not a scored instrument. Instead, it is often a standard
checklist that includes a series of questions regarding the prisoner’s medical needs,
mental health needs, drug use, substance abuse, suicide risk, and other information
that may be used to assign the prisoner to one of the special housing units. Prisoners who are separated from the general prison population can be further observed by
the classification staff and receive initial treatment by qualified medical or mental
health professionals.
Initial classification form. Professional classification staff use the initial classification form to determine the most appropriate custody level for the prisoner. This
process usually takes place after the initial screening process is completed but within 30 days of admission to the prison system.
The initial classification form consists of a set of scored items that are weighted
according to their reliability and ability to predict institutional adjustment. In general, scores on this form are the prisoner’s current charges or convicted crimes, previous criminal record, escape history, history of serious institutional misconduct,
and other factors known to be related to institutional misconduct, such as the prisoner’s age, education level, and employment history.
Reclassification form. The reclassification form is used to reevaluate the prisoner’s
custody classification throughout his or her incarceration. The reclassification
process focuses on the prisoner’s institutional behavior during the past 6–12
months. By placing less emphasis on the prisoner’s current offense and criminal history, the reclassification process represents a “just desserts” management philosophy that modifies the prisoner’s custody level based on his or her institutional
behavior. Because many prisoners do not remain in custody for a full 12 months,
their custody levels may not be reclassified, although some prison systems reclassify prisoners with short sentences at 6 months to ensure they have an opportunity to
17

Chapter 3

progress to a lower custody level. Nevertheless, medium- and close-custody prisoners, who occupy the greatest portion of a prison’s bed capacity and who will remain
in custody for 12 months or more, are the most likely to be reclassified. The reclassification process allows some prisoners to be placed in lower custody levels, thereby freeing up precious high-security bedspace.

Classification Instruments That Have Been Tested

A pilot test must be
completed before a
classification system
is implemented.

Before implementing any classification system, a prison system should test the classification system’s instruments and procedures on the prison population to which
they will be applied. A pilot test of a classification system measures two things:
How well the proposed classification instruments will perform on a given prison
population.
The likely impact of the classification policy and procedures on prison operations.
A pilot test must be completed before a classification system is implemented. There
are no exceptions to this general rule. Agencies that implement classification instruments prematurely are likely to endanger staff and prisoners. The following paragraphs discuss some of the major tasks in conducting a pilot test. Chapter 4 of this
report, which presents guidelines for implementing classification systems, includes
additional information on pilot tests.
The first task associated with a pilot test is to draw a representative number of prisoners from the average daily population, or “sample,” to test the levels of reliability and validity of the classification instruments. Using a representative sample of the
current prison male and female populations helps to determine the effect that the
new classification system will have on how prisoners are classified. The overall
sample should be divided into two subsamples: 1) prisoners who have been in custody for less than 6 months (or whatever the period is for the reclassification process
set by the agency) and 2) prisoners who have been in custody for 6 months or more.
The two subsamples should be stratified by gender to allow for separate testing of
the instruments for the male and female prisoners.
Dividing the overall samples allows the reliability of both the initial classification
and reclassification instruments to be tested. The classification system’s ability to
produce reliable results is directly related to the classification staff’s ability to use
the classification instruments consistently. Having several staff recompute the initial
classification or reclassification scores for a random sample of 25–50 prisoners will
test the staff’s overall ability to use the instruments effectively.

18

If the classification instruments have been well designed and the classification procedures clearly written, then different staff members should be able to complete the
forms and derive similar scores for the prisoners selected for the reliability test. If the
staff agree on the scores for at least 80 percent of the sample and on the overall custody level for at least 90 percent of the sample, then the system can be deemed reliable. Any percentages below these levels are unacceptable and indicate that additional

What Is Objective Prison Classification?

training and/or clarification of the classification policy and procedures should be
undertaken at once to improve the reliability of the system.

Appropriate Use of Overrides
As noted earlier, both initial classification and reclassification instruments allow
discretionary overrides to be applied to the scored custody level. No classification
instrument is capable of producing the most accurate or appropriate custody level
decision for all prisoners. Consequently, classification staff must retain the discretionary authority to deviate from the prisoner’s scored custody level at both initial
classification and reclassification.
One danger inherent in using discretionary overrides is that, if not used properly,
they can undermine the consistency of the decisionmaking process. Conversely,
classification staff who rarely use overrides may not be exercising their professional judgment to supplement the custody decisionmaking process. As a result, a percentage of the prisoners may be misclassified.
The general standard is that 5–15 percent of a prison population’s custody levels are
based on discretionary overrides rather than the original initial classification or
reclassification scores. In general, approximately 50 percent of discretionary overrides should assign the prisoner to a custody level lower than that derived from his or
her original score and 50 percent should assign the prisoner to a higher custody level.

No classification
instrument is capable

Timely and Accurate Classification

of producing the

Prisoner classification should be timely and accurate. Agencies can ensure the accuracy and timeliness of their classification system or process by establishing specific standards and requiring classification staff to adhere to them.

most accurate or

All prisoners entering the prison system must be screened immediately to determine
if they need to be housed apart from other prisoners, either for their own protection
or that of other prisoners and staff. A custody level should be assigned within 30 days
of admission, so that the subsequent decision of whether to place the prisoner in the
general population or in one of the special management populations—administrative
segregation, protective custody, mental health, or medical/infirmary—can be made.

appropriate custody
level decision for
all prisoners.

Similarly, the agency should set standards for when and under what circumstances
prisoners must be reclassified. Custody reassessment is similar to the initial custody
assessment but places greater emphasis on the prisoner’s actual behavior during
incarceration. All prisoners should be reassessed after 6 months of incarceration and
at least annually thereafter. Prisoners with long sentences who comply with the
institution’s requirements should have the opportunity to progress to a less restrictive custody level. Conversely, if a prisoner is found guilty of a major disciplinary
infraction or staff receive new information that potentially affects the prisoner’s custody level (e.g., conviction of a new charge, detainer, sentence reduction), then the
prisoner’s custody level should be reassessed within 48 hours.
19

Chapter 3

Formal Housing Plan and Security/Custody Designation for
Each Housing Unit
For a prisoner classification process to function as designed, the prison facility must
be classified as well. From a security standpoint, classifying the prison facility
determines the number and types of bedspaces available, which, in turn, establishes
the basis for the facility’s housing plan.
The housing plan is the blueprint for how prisoners will be housed within the prison.
It must also include units for intake and assessment and for special management
prisoners (e.g., a protective custody unit, administrative and disciplinary segregation
units), and for medical and mental health units.

A housing unit’s
structure and its
supervision, programming, and
privilege levels should
be well suited to the
custody levels of the
prisoners housed
within the unit.

To ensure that prisoners are housed according to their custody designations, the
prison system’s management information system (MIS) database should include
information on the custody level and type of prisoner (e.g., general population, special management population, medical, mental health) that can be assigned to the
housing unit, cell, and bed. In short, prisoners should be housed according to their
classification status. For example, a prisoner classified as maximum custody, general population, should be assigned to the maximum-security, general population
housing unit.

Adherence to the Housing Plan
Ideally, prisoners assigned to different custody levels will not be housed together.
For example, minimum-custody level prisoners should never be housed with
maximum-custody level prisoners. A housing unit’s structure and its supervision,
programming, and privilege levels should be well suited to the custody levels of the
prisoners housed within the unit.
Only prisoners who exhibit a willingness and ability to obey the rules of the facility should be allowed to remain in the general prison population. A prisoner in the
general population whose presence disrupts the orderly operation of the facility
should be removed and placed in disciplinary segregation, administrative segregation, protective custody, or the mental health unit.
All prisoners housed in the general population, regardless of their custody levels,
should be afforded privileges and living conditions sufficient to convince them that
it is in their best interest to remain in the general population. Prisoners should know
what privileges would be lost if they were reclassified to a higher custody level.
They should also know what privileges they would gain if they were reclassified to
a lower custody level.
Finally, prisoner behavior should be documented. A prisoner’s documented behavior can be used to determine housing placement, program eligibility, future prison
classification, and release conditions.

20

What Is Objective Prison Classification?

Accurate Prisoner Data
Most classification systems rely on the data listed below to determine a prisoner’s
custody level. These data must be maintained by the prison system and be available
to the classification staff:
Current charges or offenses, including a description of the crimes.
Juvenile and adult criminal records.
Prior classification records, if the prisoner was previously incarcerated in the
prison.

Electronic storage of

Any active warrants.

prisoners’ classifica-

Escape history.

tion data reduces the

Stability factors such as age, education level, employment history, and residency.

likelihood of scoring

Special medical and mental health needs.

errors and allows for
systematic, ongoing

Special management factors such as gang affiliation and orders for separation
from other prisoners.

monitoring of the

Disciplinary records.

classification system.

To ensure these data are available to the classification staff, the prison must have a
comprehensive and orderly recordkeeping system. The prison must also have access
to automated criminal record and criminal court data systems so staff can retrieve the
most current and complete information about the prisoner’s prior and current
charges. Court data on incarcerated populations are extremely critical because oftentimes when prisoners appear in court, charges are modified or dropped. Unless the
prison staff are informed of these changes, prisoners will be improperly classified.
Aside from using a prisoner’s criminal record as a basis for classification, a face-toface interview with the prisoner can provide critical classification data. Studies have
shown that a well-conducted interview provides a greater amount of high-quality
data than the information stored by law enforcement and correctional agencies. Thus,
current data contained in official documents should be verified in an interview.

Automated Data System
Related to the requirement for accurate data is the need to automate the classification system. All information obtained from the initial screening, initial classification, and all subsequent reclassification forms must be stored in the prison’s MIS
database. Electronic storage of prisoners’ classification data reduces the likelihood
of scoring errors and allows for systematic, ongoing monitoring of the classification
system. An electronic database increases both the accuracy and efficiency of the
21

Chapter 3

classification process by reducing the need to reenter the same basic data at each
stage of the classification process.
One of the most perplexing and frustrating problems for correctional staff is how to
maximize the use of the data stored in their computer systems. A common complaint is that it is very difficult to retrieve the information stored in a prison’s computer system and to use it to monitor the classification system and other aspects of
prison operations.

Continuous Monitoring and Process Evaluation
A classification system should be monitored continuously to ensure that it was
implemented as designed and continues to work as designed in relation to the
prison’s current population. In some state prison systems, a process evaluation is
equivalent to a continuous audit of the classification system. As such, the process
evaluation should answer two fundamental questions:
Are all prisoners classified according to existing agency policies and
procedures?
Are prisoners being housed according to the classification system?
In general, both quantitative and qualitative methods are used to answer these two
questions. The quantitative data analyses verify the use of the classification system
and the housing of prisoners according to their custody level and the housing assignment plan. Of primary concern when housing prisoners properly is whether highcustody prisoners are intermingling with low-custody prisoners. Some degree of
interaction between low- and medium-custody prisoners and between medium- and
high-custody prisoners is to be expected. However, the housing plan should prevent
the intermingling of low-custody prisoners and high-custody prisoners.
Qualitative data are derived from observations of components of the classification
system, such as the intake or admission process and classification interviews conducted by staff. These data should be documented in a report that records the number of performance standards observed and describes the results of the observations.
A second key resource used to measure quantitative data is the MIS’s data analysis
software and reporting generators, which can analyze and transform raw data into
management reports, graphics, and security alerts. Analysis of statistical trends and
patterns is critical at all management levels. This need is likely to escalate as prison
administrators, policy teams, and line staff require a relevant, realistic, and timely
analysis of the classification system’s performance for a variety of policy problems,
planning, and management decisions.
To date, the MISs used by most prison systems have had only basic data analysis
capacities. An MIS’s ability to analyze or transform data relevant to prisoner classification into useful management reports is limited by the software. Therefore, the
22

What Is Objective Prison Classification?

reports generated with this software, which are used for monitoring purposes, security alerts, classification evaluations, and decisionmaking, are also limited. Most
prison MISs have fixed graphics capabilities and are restricted to simple counts of
the current prison population and prisoners’ individual characteristics.
Ideally, the prison’s MIS database should provide reports that management can use
to routinely monitor the impact of key performance criteria on the classification system. The reports should indicate trends for key performance indicators and whether
any of these indicators have fallen below a minimum standard. If a performance
indicator falls below the minimum standard, the MIS software should alert the
appropriate manager. However, few of the current prison management software systems have the capability to alert staff or to detect an emerging trend within the classification system.

Impact Evaluation
A classification system is expected to affect the prison system in various ways. An
impact evaluation assesses the positive and/or negative effects of the classification
system on the prison system as a whole and on its various components (e.g., prisoners, staff, risk assessment, administration, and prison operations).

Ideally, the prison’s
MIS database should
provide reports that
management can use
to routinely monitor
the impact of key
performance criteria
on the classification
system.

To ensure an objective assessment of the system, the impact evaluation should have
a rigorous experimental or time-series research design and should be conducted
by persons trained in research methods and statistical analysis. Furthermore, highquality data are required to assess the validity, reliability, and objectivity of the system. Finally, because the impact evaluation almost inevitably will not conform to
strict experimental design requirements (e.g., randomized treatment and control
groups), the findings and conclusions regarding whether the classification system is
working will be limited. In general, a classification system always has aspects that
require change and reform. The purpose of an impact evaluation is to identify such
weaknesses and correct them in a timely manner.

23

four

Chapter

Effective Strategies: Guidelines
for Implementing Classification
Systems
In the mid-1990s, the National Institute of Corrections (NIC) created numerous
technical assistance programs to assist state and local correctional agencies with
developing and improving their classification systems. One key component of NIC’s
technical assistance efforts was developing a model that would guide correctional
administrators in either designing a new classification system or modifying an existing one. The overall objectives in developing this model were as follows:

As “the brain” of the

To identify the basic steps in designing a classification system.

correctional system,

To determine when an existing classification system should be modified.

the classification sys-

To identify the steps necessary to implement the new or modified system.

tem must respond
to the many-sided

The basic implementation process includes four phases: mobilization, assessment,
planning, and implementation. To effectively address specific local questions such
as budget cuts, new laws and legislation, prisoner and population trends, or available prison data systems, a correctional agency may need to modify the classification process. However, the four phases and the steps and tasks within each phase are
critical for designing and updating classification systems. Exhibit 5 summarizes
these steps and the estimated time required to complete each one.

issues that challenge
the correctional
system as a whole.

Phase I: Mobilization—Determining the Need for
Change
Step 1. Identify the Classification Issues
Classification systems are well established in virtually every state correctional system. However, not all prisoner classification systems work as well as expected. Quite
often, changes in the overall sociopolitical environment affect the classification
process. As “the brain” of the correctional system, the classification system must
respond to the many-sided issues that challenge the correctional system as a whole.
Past issues that have strongly influenced states to reassess and modify their prisoner classification system in terms of policies, procedures, risk factors, relative weight
25

Chapter 4

Exhibit 5. Time-Task Line for Effective Classification Implementation

Month of Project

Phase and
Steps

1

2

3

4

5

6

7

8

9

10

11

12

Who Is
Responsible

Phase I: Mobilization

Stakeholders

1. Identify classification
issue(s)

Project leader

2. Designate a steering
committee

Steering
committee

3. Review current
classification policies

Steering
committee

Phase II: Assessment
1. Conduct onsite assessment

Steering
committee

2. Compile baseline data

Research/MIS

3. Prepare assessment report

Project leader

Phase III: Planning
1. Learn about promising
systems

Project leader

2. Design and pilot test
the system

Steering
committee

3. Develop action plan

Steering
committee

Phase IV: Implementation
1. Reengage stakeholders

Project leader

2. Train staff

Project leader,
steering
committee

3. Implement the system

Steering
committee

of the risk factors, custody scale, and mandatory and discretionary override factors
include the following:

26

Changes in the characteristics and behavior of the prisoner population.
Since 1990, the number of younger, more impulsive, more difficult to manage
prisoners in the U.S. prison population has been increasing. Changes in the characteristics and behavior of the prison population are evident in the average age
of prisoners at admission, the severity of their current offenses, the extensiveness
of their criminal histories, and the length of their sentences. In response to the
changing prison population, correctional administrators questioned whether the
risk factors used to assess prisoners’ demographic and criminal histories were

Effective Strategies: Guidelines for Implementing Classification Systems

valid and reliable predictors of their current threat to institutional safety and
security.
“Get-tough-on-crime” legislation. Since 1990, numerous states have passed
legislation to abolish parole and good-time credits and require mandatorysentencing, truth-in-sentencing, and “three-strikes-and-you’re-out” laws. These
laws and similar legislation have increased the average daily population of the
states’ prison systems. Additionally, the types of crimes for which offenders are
incarcerated and the length of time prisoners must now serve for their crimes
have also increased. Given the increased sentence lengths, mandatory minimums, and more conservative parole decisionmaking processes, many jurisdictions have questioned the validity of risk factors that consider the prisoner’s
sentence or time to serve. Also, because of the increased sentence lengths, the
intervals normally set for reclassifications and needs assessments have been
reconsidered.
Budget cuts, overcrowding, and reductions in programming and services.
In recent years, many correctional agencies have seen their budgets cut dramatically. At the same time, public pressures to take away or substantially lower the
availability of prisoner programming and work assignments have reduced prisoners’ access to institutional jobs, education, and treatment programs. A reduction in prisoner programs and services undermines the foundation of the
classification system because the reclassification process is designed to encourage and reward prisoners for their participation in work and treatment programs.
These programs provide an active and productive way for prisoners to serve
their sentences. When prisoners do not have access to jobs or programs, the
power of the risk factors used to assess a prisoner’s threat to the safety and security of the institution is diminished. Recent state and federal reentry initiatives
emphasized the importance of institutional programming and planning to the
offender’s return to the community. As a result, the purpose of the classification
system may need to be modified to more accurately assess the prisoner’s criminogenic tendencies and identify the risk he or she poses to the community.
Questions about the validity of the classification system for women prisoners. Several jurisdictions have questioned the validity of their classification system for women prisoners.4 The primary concern among correctional agencies is
the overclassification of women prisoners. Traditional risk factors were based
on the criminal history data of the male prison population. These traditional risk
factors do not adequately assess the risks and needs posed by women prisoners.
This finding indicates the need to identify another set of risk factors, modify the
operational definitions of the traditional risk factors, and adjust the custody
scale cut points to better account for the risks posed by women prisoners.
Identification and management of high-risk offenders. Jurisdictions also
have struggled with how to differentiate among close-custody general population prisoners and maximum-custody and administrative segregation populations. This struggle has prompted changes in states’ external and internal
27

Chapter 4

classification systems. For the external system, jurisdictions either developed or
refined their risk factors. By adjusting the external classification risk factors,
jurisdictions are now better able to differentiate between predatory institutional
behavior (e.g., aggravated assault) and behavior that simply disrupts the smooth
and orderly management of the facility (e.g., disobeying a direct order or interfering with count). These same issues have prompted other states to develop formal internal classification systems to identify and place aggressive and
vulnerable prisoners in the appropriate housing unit.
Regardless of the immediate pressures or issues a jurisdiction is facing, it is recommended that an agency revalidate its classification system every 3 to 5 years. Revalidation ensures that the classification system is operating as intended and that the
risk factors and custody scales are appropriate for the current prison population.
Whatever the salient issues driving the modification of the classification system,
before undertaking any initiative to change its classification system, the agency
should assess its commitment and readiness to change. This is true regardless of the
extent of the anticipated change.

Regardless of the
immediate pressures
or issues a jurisdiction
is facing, it is recommended that an
agency revalidate its
classification system
every 3 to 5 years.

Classification design and implementation tasks are time consuming and are likely to
require substantial agency resources. These tasks include, but are not limited to, collecting the data, designing the classification instruments, revising the information
system, rewriting the procedures manual, and implementing the revisions. If the
agency has sufficient resources to participate in the planning tasks and pilot test, but
the commissioner and/or facility administrators are not committed to changing the
classification system, then the initiative should not proceed. Initiatives that are started and then shelved for lack of commitment waste limited agency resources, negatively affect staff morale, and diminish the staff’s willingness to participate in future
classification projects. Developing the agency’s commitment is the first task, but it
is also important to nurture this commitment throughout the initiative.
All levels and divisions within the agency—the commissioner, the warden and
superintendents, supervisors, correctional officers, case managers, and research and
MIS staff—should be committed to change. The research and MIS staff are of particular importance to the success of the process. Revisions to the classification system, for example, frequently entail modifications to the automated information
system screens and monitoring reports. Including the research and information system staff in the change process increases their understanding of the initiative and of
the importance of implementing the changes.

Step 2. Designate a Steering Committee
The next task in the process of changing a classification system is to identify a project leader. The project leader is responsible for overseeing all the tasks associated
with the initiative and ensuring they are completed. The project leader must have the
full support of the agency’s commissioner and access to all resources necessary for
the design, pilot test, and implementation activities.
28

Effective Strategies: Guidelines for Implementing Classification Systems

The project leader’s first task is to organize a steering committee to manage or direct
the initiative throughout the change process. The steering committee should include
representatives from all of the agency’s operational areas, such as security, medical
and mental health services, programming, classification, research and planning,
information systems, budget, training, and legal counsel. The project leader must
identify the appropriate steering committee members, solicit their commitment, and
outline their respective roles. The steering committee’s roles are to:

The most critical, and

Analyze current classification practices to identify issues, trends, and questions.

perhaps the most dif-

Develop practical solutions that address the current issues, questions, and
trends.

ficult, roles required

Develop preliminary classification instruments.

mittee are to define

Draft new policies.

the specific problems

Pilot test the preliminary instruments and policies.

to be addressed, set

Develop and execute implementation and evaluation plans.

realistic goals, and

Develop a time-task chart to reflect the goals of the initiative, the required activities, and the responsibilities of each member.

of the steering com-

define measurable
objectives.

Step 3. Review the Current Written Classification Policies and
Procedures
When the steering committee has been established, it should conduct a comprehensive review of the agency’s current classification procedures and practices to identify problems and issues in the classification process and examine the links between
the internal, external, and needs assessment processes. The assessment should be
carefully planned, as it will lay the foundation for the classification reform, and
should address the following questions:
What current policies, practices, and issues are potentially affecting the classification system?
What trends are associated with these policies and practices?
What outcomes would the steering committee like to achieve with the revised
classification system?
The most critical, and perhaps the most difficult, roles required of the steering committee are to define the specific problems to be addressed, set realistic goals, and
define measurable objectives. The steering committee should agree on the relevant
issues associated with the classification system, desired outcomes, and how the outcomes will be achieved and measured. Although each committee member may
assign a different priority to one or more of the current issues, goals, and objectives,
29

Chapter 4

all members should understand the importance of their contributions and what is to
be accomplished by the classification initiative overall.
To ensure that the assessment generates sufficient high-quality information, the steering committee should develop a fairly structured assessment plan with specific
assignments for each member of the site visit team. However, before visiting the site,
designated members should review any classification and needs assessment data
included in the agency’s information system to determine relevant data and resources.
Time should be spent at the central intake facility, for example, to assess institutional processes and data requirements. During the early stages of the assessment, before
any site visits, the team should compile and review the following information:
All relevant written classification policies and procedures.
Agency annual reports.
Current classification instruments.
Relevant classification-related and offender population statistics.
Agency staffing and budget.
Any recently enacted or pending legislation or administrative policies that may
affect the classification system.

Phase II: Comprehensive Assessment of Current
Classification System
Step 1. Conduct an Onsite Assessment of the Classification
System
The assessment of the current classification system should include site visits to correctional facilities and the research and information system unit. Site visits allow the
steering committee to more closely examine the various sites’ current classification
policies, procedures, practices, issues, data, resources, and limitations. Site visits also
provide the opportunity to identify potential data sources and initiate data collection.
A primary purpose of a visit to the agency’s intake center, for example, may be to
identify and assess current initial classification practices. However, the purpose and
itinerary for any site visit must be tailored to the facility’s classification issues.
Because of the myriad organizational structures and possible classification issues,
the specific tasks associated with an assessment of the classification system cannot
be fully articulated here. However, the following activities should be completed during this phase of the classification initiative:
Interview the central office and facility-level classification staff. As needed,
the team should conduct face-to-face interviews with key supervisory and line
30

Effective Strategies: Guidelines for Implementing Classification Systems

classification staff concerning their specific classification tasks. During the
interview process, any concerns about the current classification process should
be identified. Because interviewing all or even a majority of the staff may not
be feasible, the team should carefully select a cross-section of the staff to interview to capture the broadest possible range of perspectives. The purposes of the
interviews are to clarify the agency’s current approach to classification, ascertain the pragmatic issues of concern to line staff, and review current classification policies, procedures, and instruments.
Observe the classification process. As necessary, the team may augment the
interviews by reviewing a random sample of recently completed classification
instruments. The site visit team should have access to case files, observe the
classification process, and identify the criteria for making custody decisions.
These data, combined with the staff interviews, should provide insight into both
the availability and quality of the data required to score the current initial and
reclassification instruments, new risk factors suggested by the steering committee, the degree of discretion associated with the classification process, and the
reliability of the classification scores and custody designations.
Review the results of the site assessment. The site visit team should present
their findings to the steering committee. During this meeting, any unresolved
issues should be addressed and a consensus reached on the project tasks, the
specific responsibilities of the steering committee members, and the feasibility
of the project’s time-task line.

Baseline data provide
a more complete
understanding of the
scope of the anticipated changes to the
system and should
be collected for each

Step 2. Compile Baseline Data
Compiling baseline data that describe the classification system at the beginning of
the initiative is a foundation for sound planning. These data provide a more complete understanding of the scope of the anticipated changes to the system. Baseline
data should be collected for each of the outcomes identified by the steering committee and stakeholders. These may include rate of institutional violence, number of
housing transfers, number of discretionary overrides at initial classification and
reclassification, rate of institutional misconduct by custody level, and the custody
distribution at initial classification and reclassification. Each statistic should be
compiled separately by gender, facility, and custody level. If the initiative is to
develop an internal classification system, “maps” of the various cellblocks, housing
units, programs, and work assignments will illuminate any questionable patterns to
be addressed by the new system.

of the outcomes
identified by the
steering committee
and stakeholders.

Step 3. Prepare the Classification Assessment Report
Within 2 weeks of completing the assessment, the assessment team should prepare a
draft report documenting its activities, the baseline data, and the findings of the
assessment. The draft report should be circulated to the commissioner, steering committee, director of classification, and key staff who participated in the assessment.
The report should describe the agency’s current practices, identify the classification
31

Chapter 4

issues to be addressed by the initiative, and document the agreed-upon project timetask line.
The project leader or author of the report should poll the commissioner, steering
committee members, director of classification, and key stakeholders to identify
questions, errors, and/or omissions from the draft report. Based on these discussions, the report should be finalized.

Phase III: Planning
Step 1. Learn About Promising Systems, Models, Strategies,
and Best Practices
A thorough review of the literature or technical assistance from a provider with
expertise in classification should inform the development of a new system or even
modifications to an existing system. Obtaining this information will help avoid
some of the unanticipated pitfalls in trying to reinvent a system that already has been
developed or refined by another jurisdiction. An important strategy for learning
about models and promising approaches is to contact both comparable state agencies that have implemented the model and agencies that may have considered the
model but rejected it. The steering committee should supplement these conversations with visits to the facilities or agencies that have implemented the models under
consideration.

Step 2. Design or Modify the Classification System
The specific tasks at this stage will vary according to the findings of the assessment.
However, the following substeps should be completed whether a new system is
being developed or the existing system modified.
Substep 2.1: Develop a preliminary classification system. After the assessment
has been completed and the steering committee is fully informed on the current
process, trends, and issues, the committee must consider how to proceed to achieve
its goals and objectives. To develop a preliminary classification system, the committee must reach consensus regarding the classification process and the content and
format of the preliminary classification forms.
Classification process:
❖ Instruments: What instruments are required (e.g., initial, reclassification,
and/or needs assessment)?
❖ Schedule: When will the instruments be completed?
❖ Staffing: Who completes the respective instruments?

32

❖ Quality control: Who reviews overrides, conducts reliability checks, and
so forth?

Effective Strategies: Guidelines for Implementing Classification Systems

❖ MIS: When and how are classification data entered into the MIS database?
❖ Strengths: What are the strengths of the current classification system?
❖ Weaknesses: What are the weaknesses of the current classification system?
Content and format of the preliminary classification instrument(s):
❖ Risk factors: What risk factors need to be created, redefined, added, or
deleted?
❖ Risk criteria: Do the categories within the risk factors need to be defined
or redefined?
❖ Offense severity: Is an offense severity scale required or does the current
scale need to be updated?
❖ Institutional disciplinary code: Does the institutional disciplinary code
meet classification misconduct standards?—that is, does the steering committee need to differentiate predatory from major infractions? Are the disciplinary codes mutually exclusive and exhaustive?
❖ Factor weights: Are the appropriate scores assigned to the respective risk
factors and criteria?
❖ Scale cut points: Are the cut points of the close/maximum-custody scales
(if applicable) and overall custody scales appropriate?
❖ Overrides: Are the appropriate nondiscretionary and discretionary factors
for modifying the scored custody level identified and defined?
A consensus-based decisionmaking process in which all committee members participate equally in the development of practical resolutions to the issues is recommended. If the committee encounters an impasse, pilot testing of alternative risk
factors and operational definitions offers an opportunity to retain a rational decisionmaking process while maintaining a cohesive work group. A consensus-based
process is critical for generating ownership and buy-in to the new system.
Substep 2.2: Design prototype instruments and manual. Based on the decisions
reached in substep 2.1, the steering committee must develop prototype instruments
and a training manual to document the new or revised system. The operational definitions and instructions for each classification factor should be specified. As necessary, scales for ranking the severity of convictions and institutional misconduct
should be developed and included in the manual. The prototype instruments and
manual should be developed by the project leader or a designee and distributed to
the entire steering committee to ensure their comments and suggestions were captured accurately. Based on the committee members’ comments, the prototype instruments and manual should be finalized.
33

Chapter 4

Substep 2.3: Pilot test the prototype instruments and manual. A key task of any
classification system design or modification process is a scientific pilot test. The
pilot test includes several steps:
Draw representative samples of key prisoner populations. Sampling procedures must be tailored to the agency’s institutional populations and the capabilities of its MIS. To allow for separate analyses of the male and female prisoner
populations, separate samples of male and female prisoners should be generated
that take into consideration the respective ADP, number of admissions per year,
and average length of stay of each group. In addition, according to the specific
issues to be addressed by the initiative, stratification or oversampling of special
populations may be necessary. For example, if the steering committee is concerned about the validity of the instruments for prisoners with mental health
problems, it may need to oversample this population to ensure an adequate number of cases for statistical analyses. The size of the samples required for the statistical analyses should be adjusted according to the jurisdiction’s ADP. For
example, for the statistical analyses, a random sample of 1,200 cases should be
taken that includes data from a minimum of 600 initial classification forms (300
male and 300 female prisoners) and 600 reclassification forms (300 male and
300 female prisoners). If the jurisdiction has fewer than 300 male or female prisoners under its supervision, then preliminary initial and reclassification forms
should be completed for 100 percent of the prisoner population. For example,
jurisdictions holding 200 women prisoners should complete classification and
reclassification forms for 100 percent of their women prisoner population.
Develop a supplemental data collection instrument and coding instructions. Depending on the sophistication, reliability, and accuracy of the data
stored in the agency’s MIS, the information system or research staff will need
to generate electronic data files with the prisoners’ criminal history, demographics, and institutional disciplinary records. A detailed data request that provides the specific data to be included in the electronic files is critical to avoiding
misunderstandings and spurious conclusions.
Collect data. If any data are to be collected manually, members of the steering
committee should serve on the data collection team because they are familiar
with the prototype instruments and manual. Participation in the data collection
effort also facilitates their understanding of the results of the analyses and ensuing recommendations. To minimize the time required to collect the data and
ensure its accuracy, an independent reviewer should check the validity and
integrity of the data collection instruments while the case file is still available to
resolve inconsistencies and missing data.
Analyze data. The preliminary risk factors, scale cut points, and override factors
will need to be assessed to determine if they are valid and reliable for identifying
prisoners who pose a threat to the safety and security of the institution. Separate

34

Effective Strategies: Guidelines for Implementing Classification Systems

analyses by gender are important for determining whether the system is appropriate for both male and female prisoners and whether separate instruments, scales,
and/or risk factors are required for specific populations (e.g., women prisoners,
aging prisoners, or prisoners with severe medical and/or mental health conditions).

Step 3: Develop an Action Plan
Developing a comprehensive plan for the implementation of the new system is critical to the success of the classification initiative. The steering committee should
consider the implementation process at each point in designing and testing the system. The action plan must consider the following issues:
Staffing: What new hires and reallocation of positions are essential to complete
the required tasks within the schedule outlined by the policy and procedures?

Regardless of how
simple the instru-

Training: What schedule, location(s), agenda, materials, and cases will be
required to prepare staff to complete their assignments? Who will conduct the
training? Should an orientation session be held to introduce the new/modified
system to all staff?
Timing: Will the implementation be systemic, by institution, or by type of
classification?
Stakeholders: Who are the stakeholders for the classification system? Will the
new system affect entities outside the department of corrections and, if so, what
changes or training are required to ensure linkages?

ments are to score
or how few criteria
have been built into
the system, ultimately automation is
essential.

Materials: Who will be responsible for updating and printing the new/revised
instruments, manuals, and policies?
MIS: Who will be responsible for revising/creating the information system
screens, writing the automated scoring programs, and updating the tickler systems that track the classification process? When will these modifications be
completed? Are adequate resources available to cover the programming costs?
Evaluation: What data are required for the process and impact evaluations to
determine whether the system was implemented as designed and whether it
accomplished the identified goals and objectives?
Cost estimate: What are the estimated fiscal and staffing costs associated with
each element of the implementation plan?
A critical element of the action plan that is often overlooked is automation of the system. Regardless of how simple the instruments are to score or how few criteria have
been built into the system, ultimately automation is essential. The action plan should
also include goals, objectives, and specific timelines for implementing the system.

35

Chapter 4

Phase IV: Implementation
Step 1: Reengage, Reorient, and Reeducate Stakeholders
The steering committee should present an overview of the new or modified system
to the commissioner and all stakeholders. This presentation should review the issues
and trends precipitating the initiative, the initiative’s goals and objectives, the design
and pilot test activities, the baseline and pilot test data, and the committee’s recommendations. This meeting provides an opportunity for feedback and clarification of
findings and may highlight the need for additional analyses to resolve questions.
If “document, document, document” are the three most important commands when
scoring an objective classification instrument, this message is even more critical
when completing a classification design or modification initiative. At the close of
the project, a written report that documents the tasks completed, methodology, statistical reports, recommendations, and implementation process is important. The
report should be written in nontechnical language and distributed to administrative
officials at facilities and to classification supervisory and line staff. The report
should give the history of the development and evolution of the classification system and provide baseline data for tracking the modifications to the system and
assessing their impact. Suggested sections for inclusion in the summary report are
shown in exhibit 6.
Exhibit 6. Suggested Outline for a Classification Initiative Summary Report
1. History of the Classification System.
2. Development of the Classification System.
3. Assessment of the Current System.
3.1

Classification Issues, Trends, and Questions.

3.2

Revisions to the Classification System: Issues and Resolutions.

4. Methodology for the Design/Modification of the Classification System.
5. Profile of the Prisoner Population by Gender.
6. Refinement of the Instruments and Manual.
6.1

Predictive Power of the Individual Risk Factors.

6.2

Predictive Power of the System.

7. Recommendations for Modifying the Classification System.
8. Conclusion.
Appendix A. Prototype Instruments.
Appendix B. Results of Detailed Statistical Analyses.
Appendix C. Prisoner Code of Conduct and Offense Severity Indexes.

36

Effective Strategies: Guidelines for Implementing Classification Systems

Step 2: Train Staff
A new or modified classification system cannot be implemented successfully if staff
are not adequately trained in using it. Training for all correctional staff is an important factor in building their confidence in the new system. Classification staff should
receive specialized training that covers topics such as instrument use, information
management, resource allocation, and program development decisions. All training
sessions should include an overview of how the new system was developed to
acquaint staff who were not members of the classification committee with the background of the system.

Training is key to
achieving reliability,
especially as newly

Ongoing in-service training should supplement the initial orientation and implementation training. Ongoing training will assist in problem solving, evaluating the
system, facilitating staff feedback, reinforcing the system’s objectives, and enhancing its management application potential.
Instructors may be drawn from within the agency—for example, from the classification staff or from administrative personnel—and from professional fields outside the
agency. Each has advantages and disadvantages. An instructor from the agency’s staff
will be familiar with the participants; however, filling the role of both co-learner and
instructor can be difficult. The system planners of the classification system also run
the risk of being unable to break out of their role of system developers and may be
seen by other staff as having a vested interest in the successful implementation of the
new system. Outside instructors, on the other hand, are more likely to be viewed as
experts but may not be completely familiar with the classification system and may be
out of touch with the realities of the workplace. Clear lesson plans, personal contacts
with staff, and last-minute briefings can help minimize these potential problems.

hired or transferred
staff are required
to use the system.

Step 3: Implement the New/Revised System
The final step in the process is implementing the action plan. A detailed time-task line
that accounts for the complexity of implementation across facilities and staffing patterns must be developed and fully explained to all stakeholders. It should serve as the
guide for all implementation-related activities. The importance of staff training and
automation cannot be overemphasized. Training is key to achieving reliability, especially as newly hired or transferred staff are required to use the system. As previously noted, regardless of how simple the system, automation ultimately is essential.
Full implementation will bring additional challenges and obstacles that may require
modifications to the instruments. Such challenges should be expected, and time and
resources should be built into the plan to allow obstacles to be addressed appropriately. The biggest obstacle is rarely the process of designing and validating the system, but rather organizational resistance to change. New systems for classifying,
making housing assignments, organizing the delivery of services, and holding prisoners accountable for their behavior generally threaten the various power structures
throughout the prison system. Even when an agency designs and implements its
objective classification system with integrity and appropriate methodologies, it must
be prepared to monitor and update the system periodically.
37

Chapter 4

Most states have
found that high levels of reliability and
validity are relatively
easy to achieve with
behavior-based systems that are quantitative, objective, and
automated.

Lessons Learned: Common Problems, Issues, and
Solutions
The complex and varied experiences of the states with which we have worked to
design, modify, and then implement internal and external classification systems provide substantial insight into the process, core elements, and critical barriers to
implementing classification systems. New lessons learned continue to add depth and
dimension to the understanding of the key issues. Some of the lessons learned have
been noted in previous NIC reports (Hardyman et al., 2002; Hardyman, Austin, and
Tulloch, 2002); some have lent new growth and complexity to the field of classification. This section summarizes the most significant lessons.

Plan to Plan, Then Double Your Resources
Design and implementation of any classification system requires strong commitment, considerable time, and very careful planning. This lesson was learned early in
our work with state correctional agencies and then reinforced throughout subsequent initiatives. Although states embarked on their respective initiatives with what
appeared to be ample commitment, time, staff, and resources and then carefully
selected pilot sites based on their resources, goals, and objectives, some initiatives
met significant challenges.
For example, at the outset of the Oregon Department of Corrections’ initiative to
develop and test an internal classification system within its primary women’s correctional facility, the level of overcrowding limited flexibility with bedspace and
staff had insufficient time to develop the model. The project did not move forward
until the agency allocated additional bedspace at another facility and the initiative
was designated as the primary responsibility of a member on the steering committee. With the adequate resources—staff time and bedspace—the initiative was able
to meet all timelines for the development, testing, and implementation of the internal classification for the female population.
Florida experienced a very different situation in the design and implementation of
its internal classification system. From the outset, the project was a top priority not
only for the Central Office Classification Bureau, but also for the entire agency.
Extensive computer, staff, and travel resources were tapped to design, automate, and
implement the system. Only with this significant and enduring commitment was
Florida able to carry out its very ambitious initiative to design, pilot test, and implement the system statewide.

Keep It Simple
A second lesson learned was that simplicity is paramount. Regardless of the type of
classification initiative—internal, external, or needs assessment—simplicity is key
for ensuring reliability and validity within the complex environment of a correctional system. Most states have found that high levels of reliability and validity are relatively easy to achieve with behavior-based systems that are quantitative and objective.
38

Effective Strategies: Guidelines for Implementing Classification Systems

Over time, the automation of classification systems has proven to enhance both reliability and validity. With proper planning, automated behavior-based systems that
appeared to meet the goals set by the agency have been implemented with relatively
few resources. The strength of these systems is in their simplicity and objectivity.
In contrast, for example, Missouri and South Dakota set out to jointly develop a
personality-based internal classification system. Although both states expended
tremendous time and resources to develop the model, they encountered significant
reliability problems with their instruments, which relied on multiple checklists composed of numerous subjective assessments of prisoner personality. The experiences
of Missouri and South Dakota suggest that states may struggle to meet the staff
training, monitoring, and fiscal resources demanded by personality-based systems.
Other states have had similar experiences with attempts to implement interviewbased needs assessment processes. A lengthy interview process requires extensive
staff training, increases staff workload, and decreases reliability of the assessment.

Automation Is Critical to Managing the Classification Process
MIS and customized computer programs are central to the design and testing of a
classification system, even for minor modifications within a single facility. Automation is critical to fully implement the system and ensure its reliability and validity.
Some states are able to access local resources to automate their classification systems; however, others require additional technical assistance and outside funding to
fully automate the system.

Management information systems and
customized computer

Classification Is Unique to Each System
The diversity of the classification systems considered and tested by the states with
which we have worked suggests that no distinct set of classification factors can be
applied across systems. The specific goals, resources, and prisoner population of a
given state’s prison system will determine the critical factors, operational definitions,
processes, timing, and other elements of its classification system. Even behavior-based
systems, which are clearly the most objective, vary from state to state. There is no best
model, nor should there be. This observation holds particularly true in designing and
implementing internal classification systems. As described in previous chapters of this
report, the external classification instruments and process must be tailored to the population for which they will be used and validated by pilot testing with this population.
In sum, classification systems should be responsive to individual correctional systems.
Because the expectations and needs of the department (central office), facility
administrators, and line staff determine the specific “make and model” of the classification system, much of the success of the system depends on how clearly its purpose, goals, and desired impact are defined. As stressed in the guidelines presented
in the preceding sections of this chapter, the process of designing and refining an
internal classification system requires each agency—and in some circumstances,
specific facilities—to be responsive to its unique problems, issues, goals, and
resources. Furthermore, each agency must implement its new system fully before it
can reach definitive conclusions about the model’s effectiveness.

programs are central
to the design and
testing of a classification system, even for
minor modifications
within a single facility.

39

five

Chapter

Evaluations of Prison Classification
Systems
As noted earlier in this report, agencies that have implemented a new or modified
prison classification system must submit the system to comprehensive and rigorous
evaluation. This is necessary for three reasons:
To establish that the classification system is working as intended.

Process evaluations
are used to determine
whether a system

To reconfirm the validity of the classification system.

is functioning as

To assess the impact of the classification system on the intended aspects of the
prison system.

planned and should

This chapter reviews the evaluation methods that should be used to address all three
of these objectives. Although correctional administrators, classification staff, and
many other readers of this report may not have been trained in research methods and
statistics, it is nonetheless important that they have a conceptual understanding
of how such studies should be undertaken and their value to sustaining a wellfunctioning classification system. Exhibit 7, which lists the major standards to be
followed in completing an evaluation, can help agencies and evaluators understand
the evaluation process and can also be used as a checklist in conducting process, validation, and impact studies. Definitions of statistical terms and other terms used in
evaluation are included in the appendix, “Glossary of Key Terms.”

be an ongoing
feature of any classification system.

Establish That the Classification System Is Working as
Intended
Process evaluations are used to determine whether a system is functioning as planned.
A process evaluation can be viewed as an audit procedure and should be an ongoing
feature of any classification system. Much of the work of a process evaluation can be
accomplished easily if the system is fully automated. An automated classification system facilitates the taking of periodic statistical “snapshots” of the prisoner population
that are needed to produce management reports and answer questions such as whether
prisoners are being classified in a timely manner and housed according to the classification system and whether overrides are being used properly.
A process evaluation of a classification system should assess the system’s reliability
and validity. These components of the evaluation are discussed in the following two
sections.

41

Chapter 5

Exhibit 7. Summary of Evaluation Standards
General Standards for the Evaluation of Objective Classification Systems
An objective classification system should be evaluated to determine if it:
❖ Is implemented properly.
❖ Meets its goals.
❖ Can be improved.
An evaluation should be:
❖ Based on accurate and comprehensive data.
❖ Fair, timely, and useful.
An evaluation report should be clearly written and understandable to users.
Standards for Evaluation Goals
A comprehensive evaluation of a classification system should include process, validation, and
impact goals.
An impact evaluation should examine the intended impacts of a system but should also explore
unintended, unanticipated, and latent impacts.
With rare exceptions, an impact evaluation should not be conducted until a process evaluation has
demonstrated that the classification system is functioning as designed.
If a process evaluation demonstrates that a classification system is functioning as intended and an
impact evaluation demonstrates that the intended impact has not been achieved, then a validation
study should be conducted.
Evaluation goals should be achievable with available resources and should be likely to have a
practical effect.
Standards for Evaluation Questions
Evaluation questions should be framed to be answered by analysis of observations and objective data.
Evaluation questions should relate to the stated evaluation goals.
Process questions should address how the classification system is operating.
Impact questions should consist of independent and dependent variables and seek to determine if the
classification system is having an effect on prisoners, staff, and/or the prison system in general.
Validation questions should specify the type of validity under consideration and the outcomes to be
validated.
Standards for Evaluation Designs and Methods
A process design should identify the major components of the objective classification system and
compare the plan to actual performance.
An impact design should be experimental, with random assignment of subjects into experimental
and control classification systems. If such a design is not feasible, a quasi-experimental design using
matched control groups should be used.
A time-series design should be used to measure the impact of a system on aggregate levels of prisoner
misconduct, escapes, employee attitudes, and costs.
42

Evaluations of Prison Classification Systems

Exhibit 7 continued
An impact evaluation design should identify possible confounding factors and design effects and show
how the evaluation will account for them.
Both qualitative and quantitative methods should be used in conducting process and impact evaluations.
Standards for Measures
Evaluations should use multiple measures of concepts.
Measures should be reliable, valid, sensitive, comparable, convincing, timely, and efficient.
Evaluations should identify obstacles to collecting reliable and valid measurements and should specify
strategies for overcoming these obstacles.
Standards for Sampling
Every time general conclusions are drawn from partial observations, the universe and population should
be specified, and the sample selection method should be specified.
If probability samples are used, evaluations should also report sample selection bias, sampling frame,
the confidence limit, and tolerated error.
The rationale for the sample strategy must state limits on generalizations from the sample to the
population and the universe.
Standards for Data Collection
Data collection instruments and raw data collected for an evaluation should be maintained permanently
and should be accessible to other professionals (within the limits of confidentiality).
Data collection procedures should be pilot tested.
Data should be cleaned: missing, inconsistent, and implausible data should be reviewed and rectified
if appropriate and possible.
Data collection methods should be assessed for accordance with the operational definitions of the
measures.
The evaluation should report precisely what data were collected and how issues of reliability and
validity were addressed.
Evaluations of prison classification systems should collect both qualitative and quantitative data on
multiple measures.
Standards for Statistics
In general, the evaluation team should include staff or advisors with specialized training in applied
statistics to guide decisions on the proper use and interpretation of statistics.
Evaluation reports should include a frequency distribution table that shows the mean, standard deviation,
and number of valid cases for each variable used in the analysis.
Variables with 10 percent or more of the information missing should be excluded from statistical
analysis and classification scoring criteria.
In conducting tests of association or correlation, researchers must ensure that the statistics being
applied are appropriate for the type of data collected.
In presenting findings, researchers should distinguish between substantive and statistical levels of
significance.
43

Chapter 5

Reliability Assessment

The system is reliable
if the scoring of each
custody rating item
agrees in 80 percent
of the cases tested
and the assignment
of custody level
agrees in 90 percent
of the cases tested.

Assessment of the reliability of the classification system—that is, the degree of consistency in the decisionmaking process—is a key component of a process evaluation. There are two types of reliability: interrater and intrarater. Interrater reliability
refers to consistency among raters in reaching similar classification decisions when
using the same criteria. Having different classification officers classify the same
prisoner can test interrater reliability. Intrarater reliability refers to an individual
rater’s consistency in using classification criteria over time. Having the same classification officer reclassify the same prisoner on several dates can test intrarater reliability. Both inter- and intrarater reliability are important to ensure that the system
is being implemented as designed. If the level of reliability in the classification decisionmaking process is low (i.e., less than 80 percent), then the classification system
will have little or no validity and will be unlikely to have a positive impact on prison
operations and safety.
Some of the complicated classification and risk assessment instruments have been
the subject of several evaluation studies. Notably, reliability studies of the Level of
Service Inventory–Revised (LSI–R), Level of Service Inventory–Revised: Screening Version (LSI–R:SV), Adult Internal Management System (AIMS), and Adult
Internal Classification System (AICS) have all shown that without a strong staff
training and monitoring component, these instruments will fail to perform as
designed (Andrews and Bonta, 1995; Andrews and Bonta, 1998; Quay, 1984;
Hardyman, Alexander, and Davies, 2002). Correctional agencies need to ensure that
the staff responsible for conducting these primarily psychometric tests are certified
to perform them. A simple test of interrater reliability is to draw random samples of
prisoners who have been classified and having another staff person recompute each
prisoner’s score. If the scoring of each item used for a custody rating agrees in at
least 80 percent of the cases tested and the assignment of custody level agrees in at
least 90 percent of the cases, then the system is reliable. Percentages below these
levels are unacceptable. Moreover, if a classification or risk instrument is unreliable,
by definition it cannot be a valid instrument.
In general, the more complicated the classification process, the less reliable it will
be. In a study that applied five different classification systems, including AIMS, to
a sample population at a Federal Bureau of Prisons penitentiary and camp, Van
Voorhis (1994) found that AIMS had an unacceptably low level of reliability. Van
Voorhis also tracked the classified prisoners for 6 months to determine how well
their classification status related to their disciplinary and psychological adjustments
to prison. She found that in contradiction to the AIMS prediction, Kappas, especially at the beginning of their terms, were more likely to be prey than Sigmas (Van
Voorhis, 1994).5
Austin and colleagues found that the LSI–R risk assessment system was not reliable
in its application to prisoners who appeared before the Pennsylvania Parole Board
(Austin, Dedel-Johnson, and Coleman, 2003). Of the 54 items used to predict future
criminal behavior on the LSI–R, only a handful met the 80 percent threshold criteria.

44

Evaluations of Prison Classification Systems

Interrater reliability tended to be highest for LSI–R items that measured the prisoner’s
criminal history and other factual items. Furthermore, the two LSI–R raters disagreed markedly regarding the prisoner’s risk level (high, medium, or low). The
interrater agreement rate in this study was only 71 percent.
More positive results have been noted for classification and risk instruments that
have fewer than 10 factors and can be scored from official documents, as opposed
to those that use a self-administered questionnaire or survey. However, these results
are also based on the assumption that the classification staff and those involved in
the scoring process are professionally trained and tested on their scoring skills
(Hardyman, Austin, and Tulloch, 2002). The bottom line is that reliability is an
essential feature of an objective prison classification system.

Validity Studies
By definition, a classification or risk instrument that is unreliable is not a valid instrument. Therefore, the validity of the system can be evaluated only after it has passed
the reliability tests. As explained in chapter 2, the concept of validity encompasses
face validity—whether the items used for classification make sense to those who are
using them, or have face value—and predictive validity—whether the items demonstrate the capacity to predict risk based on a statistical test of association. Validation
studies track the misconduct of a sample of prisoners (e.g., an admission, release, or
current population cohort) over a given period to determine whether the risk factors
scored by the classification system are associated with prisoner misconduct. Statistical tests are used in completing the analysis of the risk factors. Note that a risk factor may pass the face validity test but not the predictive test and vice versa.
The majority of prisoners never become disruptive or difficult to manage. The most
serious forms of disruptive behavior within a prison (homicide, escape, aggravated
assault on prisoners or staff resulting in serious injuries, and riots) are rare, and most
staff and prisoners never become the victims of such incidents. Moreover, because
such events are rare, it is difficult, if not impossible, to predict which prisoners are
likely to become involved in them and under what circumstances.

By definition, a classification or risk
instrument that is
unreliable is not a
valid instrument.
Therefore, the validity of the system can
be evaluated only
after it has passed
the reliability tests.

Despite these limitations, over the past three decades a considerable amount of
research has been conducted on those factors that have been shown to be predictive
of prisoner behavior and institutional misconduct. Regardless of some of the difficulties associated with prediction, objective prison classification systems that use
reliable and valid scoring criteria have repeatedly been shown to be successful in
classifying prisoners according to their level of risk of becoming involved in prison
misconduct. The following factors have been shown to be the most predictive of
prisoner behavior:
Current age: Older prisoners are less involved in all forms of misconduct.
Gender: Women prisoners are less involved in violent incidents.
45

Chapter 5

History of violence: Prisoners with a recent history of violence are more likely to continue that behavior.
History of mental illness: Prisoners with a history of mental health problems
are more likely to be involved in all forms of misconduct.
Gang membership: Gang members are more likely to be involved in all forms
of misconduct.
Program participation: Prisoners who are not involved in programs and
have never completed a program are more likely to be involved in all forms of
misconduct.
Recent disciplinary actions: Prisoners who recently (within the past 12 months)
have been involved in misconduct are more likely to continue to be involved in
future disruptive behavior.
Perhaps more interesting is that many factors used for classification have little if any
predictive capability but exert a strong influence on the custody designation process.
Common nonpredictive factors include:
Severity of the current offense.
Sentence length.
History of escape.
Time left to serve.
Detainers.
Alcohol and drug use.
This is not to say that these factors should not be used. In many ways, they reflect
correctional policy that is often held accountable to a nonachievable zero tolerance
for escape and violence. Although very few persons serving long prison terms for
murder or a sex offense become management problems or escape, if and when one
does, the assault by the media and politicians on the correctional agency is simply
too unbearable to assume these prisoners present a low risk to institutional safety
and security. However, agencies need to review their classification policies regularly
to ensure they are not being overly restrictive.6
Women prisoners, who are far less likely to become involved in serious or potentially violent behavior while incarcerated than men, are as a class more likely than
men to be overclassified under a system that has been normed on a male prisoner
population. For this reason, the classification system for women prisoners should be
adjusted based on a separate study of the female population’s misconduct rates to
ensure that such overclassification does not occur.
46

Evaluations of Prison Classification Systems

Assess the Impact of the Classification System
In the terminology of conventional experimental design, the outcome variables
selected for an impact evaluation are the dependent variables (y), and the new classification system is the independent variable (x). The main goal of the impact study
is to examine whether there is any cause-effect relationship between the new system
and the dependent variables. The new classification system or policy may be regarded as the cause (i.e., the intervention or treatment), and the selected outcome variable as the effect.
Conventional experimental designs deal with cause-effect analysis by applying the
treatment to the experimental group while a matched control group is given an alternative or conventional treatment (e.g., the old classification system). Random
assignment of cases to each of these groups is used to achieve a high level of comparability between the groups and to rule out other causal factors. This experimental setup allows a clear inference regarding the impact of the intervention and its link
to the outcome.
Quasi-experimental designs offer another possibility but are vulnerable to confounding factors. A useful approach in prisons is the interrupted time-series design,
which can be used when a long stream of data are available for the prison before and
after the introduction of the classification system. Depending on the number of confounding factors (e.g., new staff, new policies), some tentative inferences may be
made regarding the impact of the newly implemented classification system.
Various confounding factors prevent a direct cause-effect inference in the context of
single case studies. These might include, for example, the hiring of new classification staff, additional training programs, the opening of a direct supervision wing, or
new policies. The presence of these additional changes would confound any simple
inference regarding the impact of the classification system.
Longitudinal data are amenable to a variety of statistical analyses that usually involve
the analysis of time series or trends and differences in performance levels before and
after the change. When examining such time series in graphical form, the analyst
looks for any change in the “slope” or level of the trend line after the change occurred
in comparison with the slope of the line prior to the implementation of the classification system. However, simply establishing a statistically significant change in the
trend line does not mean that the observed change is related to the new classification
procedures. Other rival hypotheses or explanations must be considered and ruled out.
These other explanations could include changes in reporting standards, changes in
prisoner population attributes, or changes in prisoner management.

47

six

Chapter

Classification of Women Prisoners
Between 1990 and 2002, the number of women under the jurisdiction of state and
federal prison authorities increased from 44,065 to 97,291 (Harrison and Beck,
2003). At year-end 2002, women in state and federal prisons constituted 6.8 percent
of the nation’s prisoners, compared with only 5.7 percent in 1990. Since 1990, the
number of male prisoners has grown by 77 percent, while the number of women
prisoners increased by 108 percent (Beck and Harrison, 2001). Although these statistics do not surprise correctional administrators, the nation’s penal systems remain
ill equipped to address the security, programming, and special needs presented by
women prisoners. Because U.S. correctional systems—both facilities and policies—
were originally designed to accommodate male prisoners, they are based on risk factors that have a tenuous relationship, at best, to the behavior of women prisoners.
Although the literature regarding the design and effectiveness of correctional programming for women offenders is somewhat limited and dated, the unique needs
and issues of this population have been well documented. There is widespread
agreement that incarcerated women differ from their male counterparts in terms of
their offenses, institutional behavior, and medical, substance abuse, mental health,
and family issues (Greenfield and Snell, 1999). The constellation of characteristics
and needs commonly seen in prisoners manifests itself differently in women than in
men. Whereas women generally pose little threat of institutional violence or escape,
their significant substance abuse and mental health needs can produce behaviors that
are difficult to predict. These differences are particularly important to institutional
classification systems, yet they are underresearched.

Whereas women
generally pose little
threat of institutional
violence or escape,
their significant substance abuse and
mental health needs
can produce behaviors that are difficult
to predict.

In 1994, the National Institute of Justice (NIJ) sponsored a study to assess programming needs and promising approaches for incarcerated women (Morash,
Bynum, and Koons, 1998). The identified management issues included problems
emerging from overcrowded facilities, a lack of programs, and shortcomings in classification procedures. More specifically, administrators reported that their systems
did not collect and compile adequate information on the risk factors and needs of
women offenders. Thus, the systems were not useful in matching the women to
appropriate custody levels or programming. Furthermore, classification and screening instruments were often unrelated to where the women were housed or to which
programs they had access. This was true even in larger states with a broader range
of programming and housing options. Lack of bedspace and constant movement of
large numbers of women were cited as specific operational barriers.
Dissatisfied with current classification systems, correctional administrators have been
faced with three basic options: 1) use the current instruments and override the scored

49

Chapter 6

Although objective

custody levels; 2) modify the current risk factors and/or scale cut points; or 3) discontinue use of the current instruments and classify women prisoners based on a subjective, intuitive process. Regardless of the option selected, the result is that women
are classified according to risk factors that are not relevant to their custody, housing,
or programming needs. Thus, although objective prison classification systems for
male prisoners are well established in virtually every state, objective classification for
women prisoners has been long neglected and consequently is poorly developed.

prison classification
systems for male
prisoners are well
established in virtually every state, objective classification for
women prisoners has
been long neglected
and consequently is
poorly developed.

Common Themes
During the past 5 years, several states have undertaken initiatives with NIC support
to learn how best to assess the risks posed by their female prison population. The
goal of these initiatives was to be able to place women prisoners in the least restrictive environment and assign them to appropriate housing and programming that best
addressed their risk factors and needs. The initiatives began with the assumption that
the institutional behavior of women prisoners differs from that of male prisoners and
that a different set of risk factors and/or a different classification process may be
required to manage this population efficiently and effectively.
As noted above, correctional administrators have employed three basic strategies to
make classification systems more responsive to the risk factors and needs of women
prisoners: using current instruments and overriding the scored custody levels; modifying the current risk factors and/or scale cut points; or discontinuing use of current instruments and classifying women based on a subjective, intuitive process. Our
work with these jurisdictions has provided some insight into the viability of these
three options.

Using Current Instruments and Overriding the Scored Custody
Levels
Using an agency’s existing classification instruments but overriding the scored custody levels these instruments indicate is the most popular strategy for making classification systems more responsive to women prisoners. This strategy is used as an
interim solution by many states until they can undertake a validation study and
incorporate the changes necessary to make their classification system gender specific. Although useful as a short-term means for addressing a state’s concerns about
overclassification, this strategy is problematic because the classification decisions
are based on subjective overrides rather than statistically validated risk factors.
One state that has employed this strategy is Wisconsin (Hardyman, 2001). As of
March 2000, the Wisconsin Department of Corrections (DOC) used the same instruments and basic assessment process for both its male and female prisoners.
Although the actual rates of discretionary decisions and overrides were not available, it was apparent that the risk level indicated by the risk assessment instruments
was not given much weight in decisions on custody, programming, and housing.
Because the Wisconsin DOC did not formally assess the reliability or validity of its
50

Classification of Women Prisoners

classification system, the predictive validity of the system (i.e., its ability to identify statistically distinct custody levels that were correlated with institutional adjustment) was unknown. However, the face validity of the system (i.e., the perception
among experienced staff that the system’s risk factors identified accurate custody
levels) was poor. During the reclassification process, for example, staff routinely
modified or overrode the risk rating. This disregard of the risk assessment instrument, particularly at reclassification, diminished its role and value to the custodial
and administrative staff. Therefore, the Wisconsin Female Offender Agency Plan,
which was developed by a cross-divisional team that examined issues critical to
managing women offenders, recommended the development of a gender-specific
classification system.

Modifying the Current Risk Factors and/or Scale Cut Points
Modifying current risk factors and/or scale cut points is the most common strategy
employed by systems that have undertaken a validation study and found a statistically significant difference in the predictive power of the risk factors for their male
and female prisoners. Delaware, Idaho, New York, Oklahoma, and the Federal
Bureau of Prisons are some of the jurisdictions that have used this strategy. Across
jurisdictions, the research findings have been somewhat inconsistent. However, the
most common risk factors modified to better assess the risk posed by women
offenders are age, criminal history, current offense, and stability factors. The preliminary findings regarding these factors are presented in the following sections.
Age as a risk factor for women prisoners. Most validation studies have found that
age is a statistically significant predictor of institutional adjustment for both male
and female prisoners; however, the behavior patterns vary by gender. The validation
studies conducted for several states (e.g., Delaware, Florida, Idaho, Tennessee, West
Virginia, and Wyoming) found that the relationship between age and institutional
adjustment differed for male and female prisoners. The most common pattern
observed was that the rate of institutional infractions decreased at an earlier age for
men than women (i.e., male prisoners committed fewer infractions as they reached
their mid to upper thirties, whereas women prisoners continued to receive institutional disciplinary reports into their mid to late forties). Thus, modifying risk factors by adding different age categories for male and female prisoners has enhanced
the predictive power of classification instruments.

The most common
risk factors modified
to better assess the
risk posed by women
offenders are age,
criminal history,
current offense,
and stability factors.

Criminal history as a risk factor for women prisoners. Several researchers have
observed differences in the pathways male and female offenders take to involvement
in the criminal justice system. For example, researchers have noted differences in
the number and types of crimes for which female and male offenders are convicted
and incarcerated (Owens, Bloom, and Covington, 2003). The primary question considered in the design and validation of a classification system is what difference, if
any, these patterns make in determining appropriate custody level for women prisoners. The data have been rather mixed in that some studies have shown criminal
history risk factors to have about the same predictive power for male and female
prisoners, while others have suggested that the severity of prior convictions is a
51

Chapter 6

stronger predictor for men than women. As a whole, criminal history factors are
poor predictors of institutional adjustment, particularly at reclassification. Therefore, some states have excluded criminal history risk factors or reduced their weight
or score on the reclassification instrument developed for the women prisoners.

Modifications of current offense as a risk
factor were among
the first genderspecific changes to
classification systems.

Few, if any, generalizations can be made from validation studies comparing the predictive power of criminal history for men and women because the predictive power
of the factor varies according to its operational definition. Generally, the predictive
power of risk factors that consider the number of prior convictions or incarcerations
is poor. On the other hand, in some states the severity of prior convictions has been
statistically correlated with institutional adjustment. Pilot testing of alternative operational definitions for criminal history has been the most useful strategy for developing a valid and reliable criminal history risk factor for the women prisoners.
Idaho, for example, completely revised the criminal history risk factor on its initial
classification instruments for women and deleted this risk factor from its reclassification instrument based on analyses of its women prisoners’ history and institutional
adjustment.
Current offense as a risk factor for women prisoners. Modifications of current
offense as a risk factor were among the first gender-specific changes to classification systems. New York, for example, has scored the severity of the current offense
differently for male and female prisoners since the 1980s. The common argument
for assigning different weights to the current offense for women offenders is that
violent crimes committed by women are often against family members or within the
context of personal relationships. Women offenders are seen as less predatory and
thus are believed to pose less risk to institutional security than male offenders.
Another argument is that because nonviolent crimes are motivated by substance
abuse or economic factors, a history of nonviolent crimes does not pose significant
risks to the safety and security of the institution.
Unfortunately, often either the data required to test these hypotheses are not available or the number of cases is insufficient for valid statistical analysis. Women prisoners convicted of violent crimes tend to have higher rates of disciplinary
infractions than those convicted of nonviolent crimes, although the differences do
not always achieve statistical significance. One exception to this pattern was
observed in Oklahoma. Data were collected on the relationship between the victim
and offender, the role of substance abuse in the offense, and the relationship
between the offender and her codefendant (Hardyman and Tulloch, 2000). In contrast to the hypothesis, the type of victim (child, familiar adult, acquaintance, or
stranger) was not statistically related to the rate of institutional infractions. The
women whose crimes involved a spouse, partner, or a child as a victim had slightly
higher rates of institutional infractions than women incarcerated for crimes against
strangers, but these differences were not statistically significant. As expected,
women incarcerated for victimless crimes (e.g., drug-related or property crimes)
had statistically fewer infractions.

52

Classification of Women Prisoners

In a second analysis of the circumstances of the crime, the same study examined
the role of the woman in the commission of the offense. Crimes were differentiated according to whether the woman had an accomplice and, if so, his or her identity. The data showed the highest rates of institutional infractions for women who
were involved with a male codefendant or family member. These findings suggested that women who are involved with negative peers in the community are likely to
be more aggressive and disruptive within the institution than those who did not have
a codefendant.
Unfortunately, the number of cases for which these data were available was small,
and the data were therefore inconclusive. The findings did not support modification
of the operational definitions for rating the severity of the current offense of women
prisoners. The only consistent observation across multiple states is that women
incarcerated for violent offenses tend to have higher rates of disciplinary infractions
than women incarcerated for nonviolent offenses. Differentiation among types of
violent crimes has not been particularly useful, although prisoners incarcerated for
some street crimes (e.g., robbery, aggravated assault, and weapons offenses) have
been found to have higher rates of institutional infractions than those incarcerated
for other violent offenses (e.g., murder, rape, and kidnapping).
Stability as a risk factor for women prisoners. Many state prison classification
systems include various indicators of offender stability on their initial classification
instruments and also as dynamic risk factors (i.e., factors that can change throughout the prisoner’s incarceration) on their reclassification instruments. The most
common stability factors considered at initial classification are employment at the
time of arrest, education, and substance abuse. Stability factors used as dynamic
reclassification risk factors often include institutional behavior, participation in
institutional programming and treatment, and institutional substance abuse. Age is
a stability factor frequently included on the initial classification instrument and the
reclassification instrument. Research findings related to some of the more commonly used stability factors are discussed in the following sections.
Institutional adjustment. Analyses of the relationship between dynamic stability
factors and institutional adjustment among women prisoners have been instructive
because the data provide some insight into how to make classification systems
gender-specific and account for differences between male and female prisoners in
terms of their institutional behavior, substance abuse, and medical, mental health,
and family issues. Although the results have been inconsistent across states, these
factors clearly require special consideration when attempting to refine classification
instruments.
As illustrated by findings from Colorado, Kentucky, and West Virginia, women
whose primary role was homemaker or caretaker of a child at the time of arrest had
rates of institutional infractions comparable to those of women with full-time
employment (Hardyman and Davies, 2001a, 2001c; Van Voorhis et al., 2001). These
finding suggests the need to expand the operational definition of employment to
include childcare/homemaker roles as indicators of community stability.
53

Chapter 6

Education. Kentucky and West Virginia also considered education as a stability factor. The findings from these states were interesting in that education appeared to be
an indicator of stability among male but not among female prisoners. More specifically, rates of institutional misconduct were lower for male prisoners who had at
least a high school or general equivalency diploma than for those who did not have
a diploma. In contrast, rates of institutional misconduct were higher for women prisoners with a high school or general equivalency diploma than for those who did not.

Relationships and
mental health are two
of the most common
institutional risk factors identified by correctional system staff
working with women
prisoners.

The assessment of Florida’s internal classification system suggested that although
academic achievement was not statistically correlated with institutional adjustment
among women prisoners, the relationship between academic achievement and institutional adjustment appeared to be different for male and female prisoners (Hardyman, 2000). Although the data from the Florida DOC do not statistically replicate the
findings from West Virginia and Kentucky, the Florida data do support the hypothesis that academic achievement is a gender-specific risk factor. Thus, if academic
achievement is included on the classification instrument, the operational definition
should be tailored to accurately reflect the behavior of male versus female prisoners.
History of substance abuse. The offender’s history of substance abuse has been
problematic as a stability factor because of concerns about data reliability and validity. The reliability of this factor is often questionable because the operational definition allows for subjective bias and interpretation as to what constitutes substance
abuse. Some staff consider any use of an illicit substance an indication of substance
abuse because it is a criminal offense, whereas other staff define substance abuse
either as the involvement of illicit substances and/or alcohol in the current offense
or as daily use of these substances. Data to score the factor are dependent on the
biases of the presentence report writer and the offender’s self-report.
The demographic and need data compiled as part of the NIC classification validation initiatives suggest that 75–80 percent of the women prisoners had substance
abuse problems. Therefore, even if the data are reliable, the pervasiveness of the
problem among women offenders will often render the factor useless for classification purposes. Under these conditions, it is not surprising that the various validation
studies have yielded mixed results.
Relationships and mental health. Relationships (both institutional and community)
and mental health are two of the most common institutional risk factors identified
by correctional system staff working with women prisoners. Unfortunately, few data
are available to guide the development of reliable, objective risk factors to assess
prisoner relationships. Because these relationships vary throughout the term of
incarceration, reliability of the item is a concern. Florida, for example, developed
scales for rating these relationships as positive or negative as a part of its internal
classification system. Preliminary reliability and validity analyses of the risk factors—
child welfare, intimate relationships, and family relationships—indicated these factors
were unreliable and were not correlated consistently with institutional adjustment
(Hardyman, 2000; Hardyman and Davies, 2001b).

54

Classification of Women Prisoners

Data from West Virginia, on the other hand, indicated that institutional relationships
were a valid predictor—specifically, that women for whom institutional relationships were a stress factor had higher rates of institutional infractions. Neither having children nor being involved in legal issues were directly correlated with
institutional adjustment. However, the presence of multiple stress factors was highly correlated with institutional adjustment. These data suggest that a woman’s experiences both inside and outside prison affect her institutional adjustment.
The Oklahoma DOC developed an institutional stability classification item that considered a prisoner’s need for medical, mental health, emotional stability, and substance abuse services. This item was based on the need areas most frequently cited
by correctional staff as critical to women prisoners’ adjustment to institutional life.7
The data suggested that stability was an important factor for a woman’s initial
adjustment to prison but was not statistically correlated with long-term institutional
adjustment. This finding contradicted the observations of correctional staff. The low
correlation at the custody reclassification review has several possible explanations.
For example, after a woman’s stability needs are identified and addressed by institutional services and programs, stability either is obtained or the woman’s needs
continue to change throughout the reassessment period, making the item unreliable
and, therefore, invalid.
Although the inconsistencies in the relation between all these stability factors and
women prisoners’ institutional adjustment throughout incarceration were contrary to
observations by correctional staff, the finding that stability factors are more important at initial classification is logical. An alternative explanation is that as women
prisoners become more institutionalized, their behaviors are more affected by the
day-to-day relationships and activities in the institution than by noninstitutional
influences, relationships, and concerns. However, as observed by institutional staff,
these factors are very dynamic. Therefore, their ability to predict behavior over a 6to 12-month period may be diminished. Thus, their value as a classification or custody reclassification risk factor needs further research.

Discontinuing Current Instruments and Classifying Women
Based on a Subjective, Intuitive Process
This third option is rarely chosen. The most common variant of this option is to classify women prisoners using the agency’s standard instruments and process. However, the woman’s custody level has little impact on the facility, housing unit,
program(s), or institutional jobs to which she is assigned. Regardless of their custody levels, all women prisoners have access to the same housing units, programs,
institutional work assignments, recreation activities, visitation privileges, etc. In
effect, the classification system only determines a woman’s eligibility for work
assignments outside the security perimeter and the supervision requirements if she
leaves the facility grounds for court hearings or medical appointments.

55

Chapter 6

This failure to use the objective classification system when making decisions for
managing women prisoners is often a function of the physical structure of the correctional facility, overcrowding, and limited programming options. Thus, the classification instrument becomes just another seemingly meaningless form to be
completed. The form is completed and then filed in the prisoner’s case file. Decisions regarding housing, working, programming, and so forth are based on subjective factors such as bed or program availability, the relationships between the
women and/or staff members, and/or efficiency.

Modifying current
risk factors and/or

Given the extremely low level of violence within correctional facilities for women,
the short sentences served by most women prisoners, and the homogeneity of the
population with respect to criminal history, the option of disregarding the formal
classification system has offered some correctional systems a short-term solution
for managing women prisoners. However, this benefit does not eliminate concerns
voiced by correctional staff that the classification system is not responsive to the risk
and needs posed by women prisoners. In fact, disregarding the formal classification
system is often counterproductive because the information required for managing
the prisoner population, determining appropriate staffing levels for the facility/unit,
projecting bedspace needs, or planning programming and services is no longer
available. Development of gender-specific community risk instruments, internal
classification systems, and/or minimum community security screening processes is
required to bring correctional agencies up to the industry standard of objective, reliable systems that place prisoners within the least restrictive environment.

scale cut points is the
best option for making classification systems more responsive
to the risk and needs
of women prisoners.

56

Implications and Future Steps
NIC has long advocated for the validation of any classification system within the
population to which it is to be applied. The analyses highlighted in this chapter
should be replicated in other jurisdictions before final conclusions are drawn. Nevertheless, modifying current risk factors and/or scale cut points is the best option for
making classification systems more responsive to the risk and needs of women prisoners because it refines risk factors and tests them in the population to which they
will be applied. This strategy also provides the opportunity to develop and test new
factors to assess the risk posed by women prisoners. The other two strategies, which
do not rely on objective, reliable assessments, are at best short-term options for
managing women prisoners during the development and pilot testing of a more
gender-responsive system.

Classification of Women Prisoners

The struggle by state and local correctional systems to assess and fine-tune their
classification systems for women prisoners speaks to the need to continue research
on this topic. The inconsistencies observed thus far in the risk factors for this population suggest much remains to be learned about the classification of women
offenders. It is an urgent problem because the number of women prisoners under
correctional supervision continues to grow while resources decline. The need to
develop reliable and valid classification systems for women prisoners that will
enable correctional agencies to manage and provide services for this population
with fewer resources becomes more critical each year. Future research and validation efforts should focus on developing systems that are both practical and feasible.
Scarce resources should be used to provide maximum returns, and future initiatives
should therefore concentrate on models that require reasonable efforts in terms of
staff training, validation, and implementation.

57

seven

Chapter

Other Special Topics and Issues
in Classification
Two issues have been identified as of special concern with regard to the management of prison populations: the impact of environmental factors and prison management on prisoner and staff behavior and the need to link prison classification and
risk assessment with release decisions. These two topics are discussed in more detail
below.

Very little is understood about how the

The Impact of Environmental Factors and Prison
Management on Prisoner Behavior

physical environment

Very little is understood or appreciated about how the physical environment of the
prison and style of prison management influence prisoner and staff behavior. It
would be difficult to find a correctional official, warden, superintendent, or line officer who does not agree that a facility’s architectural design influences prisoner
behavior. Facilities that rely on open views of housing, dining, and recreation areas
tend to produce fewer episodes of disruptive and potentially dangerous behavior
than those with numerous “blind” spots. Unfortunately, few if any studies have
assessed the impact of architecture on suppressing or controlling prisoner behavior,
and it is unlikely in today’s fiscal environment that many of the antiquated prison
facilities still in use will be replaced in the near future. Nonetheless, there are potentially many lessons to be learned about the impact of architectural design on suppressing or controlling prisoner behavior.

style of prison man-

of the prison and

agement influence
prisoner and staff
behavior.

Correctional directors also have long known that similarly designed facilities with
similarly situated prison populations can produce very different rates of prisoner misconduct, both within and across state prison systems. Each major system with multiple facilities has wardens who are able to handle problem prisoners who cannot be
handled elsewhere. The field is also filled with stories of prisoners who were transferred to another state correctional system and suddenly started behaving differently.
Such variations in misconduct rates for prisons that are equivalent in design and
prison population are likely related to differences in the management style adopted
by each prison administrator. Again, no studies have substantiated this observation,
except for a few recent evaluations of the use of internal classification systems
(although there are books about great wardens of the last century who ruled with an
iron fist and a velvet glove). However, correctional administrators have reported that
59

Chapter 7

a “back-to-basics” management style that includes new methods in risk assessment
offers the best promise of reducing and controlling prison violence.
Also needed are formal assessments of the often advocated, but still highly controversial, super maximum-security facilities. Specifically, basic research is needed to
determine how best to identify prisoners who require this level of segregation, how
long they should remain segregated from the general population, what interventions
should be used to control their high-risk behavior, how to safely reintegrate them
into the general population, and how they behave after release from these units.
Without such basic research, it will be difficult to propose new methods for identifying high-risk prisoners and to apply interventions that will help control and manage them.

The Need To Link Prison Classification and Risk
Assessment With Release Decisions
The recent developments in prison classification and risk assessment systems offer
an interesting opportunity to use well-established correctional management and risk
assessment tools to assist state correctional agencies facing budgetary and other
emerging issues. The past three decades have witnessed an unparalleled increase in
the nation’s prison population. In 1970, state and federal prisons held only 196,429
prisoners (Blumstein and Cohen, 1973). Today, the number has reached 1.4 million,
which does not include another 690,000 inmates held in local jails and nearly
134,000 youth in juvenile facilities (Harrison and Karberg, 2004; Sickmund, 2004).
Despite this dramatic increase in the use of incarceration, some states are showing
signs of beginning to reduce their prison populations. The U.S. Department of Justice’s Bureau of Justice Statistics reported that the nation’s prison population grew
by 2.6 percent in 2002, which was less than the average growth of 3.6 percent since
yearend 1995 (Harrison and Beck, 2003). During 2001, the nation’s prison population grew by only 1.1 percent, which also was less than the average growth since
yearend 1995. Additionally, in 2001, the nation’s prison population increased at the
lowest rate since 1972 and had the smallest absolute increase since 1979 (Harrison
and Beck, 2002).
Perhaps more interesting is the trend toward a decline in the prison populations of
several states. At midyear 2003, nine states experienced either a decline or no
growth in their prison populations (Harrison and Karberg, 2004). In 2001, 10 states
reported prison population decreases, led by New Jersey (–5.5 percent), followed by
Utah, (–5.2 percent), New York (–3.8 percent), and Texas (–2.8 percent) (Harrison
and Beck, 2002). Between 1999 and 2000, 15 states reported either a decrease or no
growth in their prison populations. These included states with some of the largest
prison populations: California (–0.1 percent), New York (–3.7 percent), Ohio (–2.2
percent), and Texas (–3.1 percent) (Beck and Harrison, 2001).

60

Other Special Topics and Issues in Classification

Further reductions are expected in several states as a result of declining crime rates
and administrative and legislative actions to either divert offenders or reduce their
period of incarceration. Some states with indeterminate sentencing structures and
discretionary release policies are implementing new parole guidelines that increase
the rate of parole (personal communication, state correctional administrators from
Kentucky, Pennsylvania, and Texas). Other states are restricting the readmission
of parolees who have violated their terms of parole for technical reasons or for
misdemeanor-level arrests (personal communication, state correctional administrators from Arkansas, Connecticut, Louisiana, Michigan, and Ohio).
Interest in reentry—the transition from incarceration to the community—is also
growing. As the number of released prisoners increases by nearly 600,000 per year,8
commitment is growing to reduce the number of offenders who are rearrested and/or
reincarcerated. This interest is further fueled by the trend toward releasing prisoners with no form of parole or community supervision. Moreover, concerns are being
voiced about the lack of programming and services for incarcerated prisoners as
well as for those who have been released to the community.
As pressures to control or reduce prison populations continue to increase, there will
be a related need to use classification and risk assessment instruments to inform the
following key decisions regarding prisoners:
What custody level and type of programs are appropriate for the prisoner during incarceration?
When should the prisoner be released and under what forms of supervision and
services?

Increasing pressures
to control or reduce
prison populations
will increase the need
to use classification
and risk assessment
instruments to inform key decisions
regarding release as
well as incarceration.

Answering these two basic questions will require well-coordinated and virtually
seamless classification and risk assessment processes from the time the prisoner is
admitted to the prison system through the prisoner’s eventual release from parole or
other forms of postincarceration supervision. Improving the ability to assess and
manage the level of risk posed by the millions of persons who pass through the
nation’s probation, prison, and parole systems each year is a goal correctional agencies can no longer afford to ignore or neglect.

61

Notes
1. These cases include Austin et al. v. Wilkinson et al. (Ohio); Cain v. Michigan;
Calvin R. v. Illinois; Busey et al. v. Corrections Corporation of America (District of
Columbia/Ohio); Gartrell et al. v. Ashcroft et al. (DC/Virginia); Holloway et al. v.
King County (Washington); Montoya v. Gunter et al. (California); Ruben Henriquez
v. Camden County et al. (New Jersey); Ruiz v. Lynaugh (Texas); United States v.
The Parish of Orleans Criminal Sheriff’s Office (Louisiana); United States v. State
of Florida and DOC; and USA v. Michigan.
2. Chapter 5 offers a more detailed discussion of the concepts of reliability and
validity.
3. See California Department of Corrections, Data Analysis Unit, Inmate Incidents
in Institutions: Calendar Year 2002, Sacramento: California Department of Corrections, 2003. Available at the California Department of Corrections Web site:
http://www.corr.ca.gov/OffenderInfoServices/Reports/Annual/BEH1/BEH1d2002.
pdf.
4. This issue is discussed in more detail in chapter 5.
5. Kappas and Sigmas are two of the five types of inmates identified by AIMS. A
third type, Alphas, are characterized as being likely to be assaultive and/or manipulative predators. In contrast, Kappas are likely to be neither predators nor prey,
whereas Sigmas are sluggish, inept, tense, anxious, and likely to be prey. An underlying assumption of AIMS is that the five types it identifies are unchanging personality types (Quay, 1984).
6. An example of an overly restrictive policy would be one that required all prisoners convicted of homicide or a violent sex offense to be housed in maximum security for an extended period, even when it is clear that many such prisoners can safely
be housed in a medium-security setting.
7. One variation of the stability factor included substance abuse, emotional stability,
mental health, sex offender, and reintegration needs. This factor was not statistically
correlated with institutional adjustment.
8. See Bureau of Justice Statistics, “Reentry Trends in the United States: Releases
From State Prison.” Accessed December 4, 2003, on the Web at www.ojp.usdoj.
gov/bjs/reentry/releases.htm/.

63

References
Andrews, D., and J. Bonta. 1995. The Level of Supervision Inventory–Revised
(LSI–R). North Tonawanda, NY: Multi-Health Systems.
———. 1998. Level of Service Inventory–Revised: Screening Version: User’s
Manual. North Tonawanda, NY: Multi-Health Systems.
Austin, J., K. Dedel-Johnson, and D. Coleman. 2003. Reliability and Validity of the
LSI-R for the Pennsylvania Board of Probation and Parole. Washington, DC: The
George Washington University.
Austin, J., and K. McGinnis. 2004. Classification of High-Risk and Special Management Prisoners: A National Assessment of Current Practices. Washington, DC:
U.S. Department of Justice, National Institute of Corrections.
Beck, A.J., and P.M. Harrison. 2001. Prisoners in 2000. Washington, DC: U.S.
Department of Justice, Office of Justice Programs, Bureau of Justice Statistics.
Bloom, B., B. Owen, and S. Covington. 2003. Gender-Responsive Strategies:
Research, Practice, and Guiding Principles for Women Offenders. Washington, DC:
U.S. Department of Justice, National Institute of Corrections.
Blumstein, A., and J. Cohen. 1973. “A Theory of the Stability of Punishment.” The
Journal of Criminal Law and Criminology 64 (2): 198–207.
Greenfield, L.A., and T.L. Snell. 1999. Women Offenders. Washington, DC: U.S.
Department of Justice, Office of Justice Programs, Bureau of Justice Statistics.
Hardyman, P.L. 2000. Assessment of the Florida Department of Corrections Risk
and Needs System: An Ambitious Internal Classification Design and Implementation Effort. Washington, DC: The Institute on Crime, Justice and Corrections.
———. 2001. Wisconsin Department of Corrections Objective Classification System: Observations and Recommendations for Identifying and Addressing the GenderSpecific Needs of Female Prisoners: Final Report. Washington, DC: The Institute
on Crime, Justice and Corrections.
Hardyman, P.L., J. Alexander, and G. Davies. 2002. “Assessment of the Reliability
and Validity of the Missouri Department of Corrections Adult Internal Classification
System.” Washington, DC: The Institute on Crime, Justice and Corrections.
Hardyman, P.L., J. Austin, J. Alexander, K.D. Johnson, and O.C. Tulloch. 2002.
Internal Prison Classification Systems: Case Studies in Their Development and
Implementation. Washington, DC: U.S. Department of Justice, National Institute of
Corrections.
Hardyman, P.L., J. Austin, and O.C. Tulloch. 2002. Revalidating External Prison
Classification Systems: The Experience of Ten States and Model for Classification
65

References

Reform. Washington, DC: U.S. Department of Justice, National Institute of
Corrections.
Hardyman, P.L., and G. Davies. 2001a. Revalidation of the Kentucky Department of
Correction Objective Classification System: Final Report. Washington, DC: The
Institute on Crime, Justice and Corrections.
———. 2001b. Validation of the Florida Department of Correction Objective Classification System: Draft Report. Washington, DC: The Institute on Crime, Justice
and Corrections.
———. 2001c. Validation of the West Virginia Department of Correction Objective
Classification System for the Female Prisoner Population: Final Report. Washington, DC: The Institute on Crime, Justice and Corrections.
Hardyman, P.L., and O.C. Tulloch. 2000. Validation of the Oklahoma Department
of Correction Objective Classification System for the Female Prisoner Population:
Final Report. Washington, DC: The Institute on Crime, Justice and Corrections.
Harrison, P.M., and A.J. Beck. 2002. Prisoners in 2001. Washington, DC: U.S.
Department of Justice, Office of Justice Programs, Bureau of Justice Statistics.
———. 2003. Prisoners in 2002. Washington, DC: U.S. Department of Justice,
Office of Justice Programs, Bureau of Justice Statistics.
Harrison, P.M., and J.C. Karberg. 2004. Prison and Jail Inmates at Midyear 2003.
Washington, DC: U.S. Department of Justice, Office of Justice Programs, Bureau of
Justice Statistics.
Morash, M., T.S. Bynum, and B.A. Koons. 1998. Women Offenders: Programming
Needs and Promising Approaches. Research in Brief. Washington, DC: U.S. Department of Justice, Office of Justice Programs, National Institute of Justice.
Quay, H. 1984. Managing Adult Inmates: Classification for Housing and Program
Assignments. College Park, MD: American Correctional Association.
Sickmund, M. 2004. Juveniles in Corrections. Bulletin. Washington, DC: U.S.
Department of Justice, Office of Justice Programs, Office of Juvenile Justice and
Delinquency Prevention.
Van Voorhis, P. 1994. Psychological Classification of the Adult Male Prison Inmate.
Albany, NY: State University of New York Press.
Van Voorhis, P., J. Peiler, L. Presser, G. Spiropoulis, and J. Sutherland. 2001. Classification of Women Offenders: A National Assessment of Current Practices and the
Experiences of Three States. Washington, DC: U.S. Department of Justice, National Institute of Corrections.
66

Suggested Readings
Alexander, J., and J. Austin. 1992. Handbook for Evaluating Objective Prison Classification Systems. Washington, DC: U.S. Department of Justice, National Institute
of Corrections.
American Correctional Association. 1993. Classification: A Tool for Managing
Today’s Offenders. Laurel, MD: American Correctional Association.
Anno, J., C. Graham, J.E. Lawrence, and R. Shansky. 2004. Correctional Health
Care: Addressing the Needs of Elderly, Chronically Ill, and Terminally Ill Inmates.
Washington, DC: U.S. Department of Justice, National Institute of Corrections.
Austin, J., ed. 1986. Crime & Delinquency. Special Issue: Prison Classification
Systems. 32 (3).
———. 2001. “Prisoner Reentry: Current Trends, Practices, and Issues.” Crime &
Delinquency 47 (3): 314–334.
Austin, J., J. Clark, P.L. Hardyman, and D.A. Henry. 1999. “The Impact of ‘Three
Strikes and You’re Out’.” Punishment & Society 1 (2): 131–162.
Austin, J., and G. Coventry. 2001. Emerging Issues on Privatized Prisons. Washington, DC: U.S. Department of Justice, Office of Justice Programs, Bureau of Justice Assistance.
Austin, J., S. Cuvelier, and A. McVey. 1992. “Projecting the Future of Corrections:
The State of the Art.” Crime & Delinquency 38 (3): 285–308.
Austin, J., and P.L. Hardyman. 2004. “The Risks and Needs of the Returning Prisoner Population.” Review of Policy Research 21 (1): 13–29.
Austin, J., P. Hardyman, and J. Irwin. 2002. “Exploring the Needs and Risks of the
Returning Prisoner Population.” Washington, DC: U.S. Department of Health and
Human Services and the Urban Institute, “From Prison to Home” Conference, January 30–31, 2002.
Austin, J., D. Holien, L. Chan, and C. Baird. 1989. Reducing Prison Violence by
More Effective Inmate Management: An Experimental Field Test of the Prisoner
Management Classification (PMC) System. San Francisco, CA: National Council on
Crime and Delinquency.
Austin, J., and J. Irwin. 2001. It’s About Time: America’s Imprisonment Binge. 3d
ed. Belmont, CA: Thomson/Wadsworth.
Austin, J., K.D. Johnson, and M. Gregoriou. 2000. Juveniles in Adult Prisons and
Jails: A National Assessment. Washington, DC: U.S. Department of Justice, Office
of Justice Programs, Bureau of Justice Assistance.
67

Suggested Readings

Bench, L., and T. Allen. 2003. “Investigating the Stigma of Prison Classification: An
Experimental Design.” The Prison Journal 83 (4): 367–382.
Bloom, B.E., ed. 2003. Gendered Justice: Addressing Female Offenders. Durham,
NC: Carolina Academic Press.
Brennan, T., J. Alexander, and D. Wells. 2004. Enhancing Prison Classification Systems: The Emerging Role of Management Information Systems. Washington, DC:
U.S. Department of Justice, National Institute of Corrections.
Briggs, C., J. Sundt, and T. Castellano. 2003. “The Effect of Supermaximum Security Prisons on Aggregate Levels of Institutional Violence.” Criminology 41 (4):
1341–1373.
Bureau of Justice Assistance. 1996. National Assessment of Structured Sentencing.
Washington, DC: U.S. Department of Justice, Office of Justice Programs, Bureau of
Justice Assistance.
Carlson, P.M., and J.S. Garrett. 1999. Prison and Jail Administration: Practice and
Theory. Gaithersburg, MD: Aspen Publishers, Inc.
Clements, C. 1982. “The Relationship of Offender Classification to the Problems of
Prison Overcrowding.” Crime & Delinquency 28 (1): 72–81.
Council of State Governments. 2002. “Criminal Justice/Mental Health Consensus
Project.” Washington, DC: Office of Justice Programs.
Cowden, J. 1984. Development and Validation of Risk Prediction and Needs Assessment Scales for Juvenile Offenders in Wisconsin. Summary. Madison, WI: Wisconsin Department of Corrections.
Cowden, J., and G. Durkin. 1984. Development and Validation of Differential Risk
Prediction Scales: An Enhancement of the Original Wisconsin Risk/Needs Classification System. Summary. Madison, WI: Wisconsin Department of Corrections.
Deemer, W., and A. Conine. 1984. “An Institutional Programming Model for the
Adult Developmentally Disabled Offender.” Corrections Today June: 30–31.
Evatt, P., S. Brown, and L. Fowler. 1989. “Offender Classification: Don’t Overlook
This Important Security Strategy.” Corrections Today July: 34–37.
Fagan, T.J. 2003. Negotiating Correctional Incidents: A Practical Guide. Lanham,
MD: American Correctional Association.
Farr, K.A. 2000. “Classification for Female Inmates: Moving Forward.” Crime &
Delinquency 46 (1): 3–17.

68

Objective Prison Classification: A Guide for Correctional Agencies

Ferguson, J.L. 2002. “Putting the ‘What Works’ Research Into Practice: An Organizational Perspective.” Criminal Justice and Behavior 29 (4): 472–492.
Fliter, J.A. 2001. Prisoner’s Rights: The Supreme Court and Evolving Standards of
Decency. Westport, CT: Greenwood Press.
Gaes, G., S. Wallace, E. Gilman, J. Klein-Saffran, and S. Suppa. 2002. “The Influence of Prison Gang Affiliation on Violence and Other Prison Misconduct.” The
Prison Journal 82 (3): 359–385.
Hagan, J., and J.P. Coleman. 2001. “Returning Captives of the American War on
Drugs: Issues of Community and Family Reentry.” Crime & Delinquency 47 (3):
352–367.
Haney, C. 2003. “Mental Health Issues in Long-Term Solitary and ‘Supermax’ Confinement.” Crime & Delinquency 49 (1): 124–156.
Hardyman, P.L., J. Austin, and J. Peyton. 2004. Prisoner Intake Systems: Assessing
Needs and Classifying Prisoners. Washington, DC: U.S. Department of Justice,
National Institute of Corrections.
Hardyman, P.L., and P. Van Voorhis. 2004. Developing Gender-Specific Classification Systems for Women Offenders. Washington, DC: U.S. Department of Justice,
National Institute of Corrections.
Herrick, E. 1989. “The Surprising Direction of Violence in Prison.” Corrections
Compendium 14 (6): 1–9.
Hills, H., C. Siegfried, and A. Ickowitz. 2004. Effective Prison Mental Health Services: Guidelines To Expand and Improve Treatment. Washington, DC: U.S. Department of Justice, National Institute of Corrections.
Hochstetler, A., D.S. Murphy, and R.L. Simons. 2004. “Damaged Goods: Exploring
Predictors of Distress in Prison Inmates.” Crime & Delinquency 50 (3): 436–457.
Jacobsen, V. 1983. “Industrial Unit Management.” Corrections Today August:
88–91, 108.
Jiang, S., and M. Fisher-Giorlando. 2002. “Inmate Misconduct: A Test of the Deprivation, Importation, and Situational Models.” The Prison Journal 82 (3): 335–358.
King, R. 1999. “The Rise and Rise of the Supermax: An American Solution in
Search of a Problem?” Punishment & Society 1 (2): 163–186.
Kouzes, J.M., and B.Z. Posner. 1990. The Leadership Challenge: How to Get Extraordinary Things Done in Organizations. San Francisco: Jossey-Bass Publishers.

69

Suggested Readings

———. 1993. Credibility: How Leaders Gain and Lose It, Why People Demand It.
San Francisco: Jossey-Bass Publishers.
Krebs, C. 2002. “High-Risk HIV Transmission in Prison and the Prison Subculture.”
The Prison Journal 82 (1): 19–49.
Kroner, D.G., and J.F. Mills. 2001. “The Accuracy of Five Risk Appraisal Instruments in Predicting Institutional Misconduct and New Convictions.” Criminal Justice and Behavior 28 (4): 471–489.
Lansing, D., J. Bogan, and L. Karacki. 1977. “Unit Management: Implementing a
Different Correctional Approach.” Federal Probation 41 (1): 43–49.
Leeke, W., and H. Mohn. 1986. “Violent Offenders.” Corrections Today May:
22–24.
Lerner, K., G. Arling, and S.C. Baird. 1986. “Client Management Classification
Strategies for Case Supervision.” Crime & Delinquency 32 (3): 254–271.
Levinson, R. 1988. “Developments in the Classification Process: Quay’s AIMS
Approach.” Criminal Justice and Behavior 15 (1): 24–38.
Levinson, R., and J.D. Williams. 1980. “Inmate Classification: Security/Custody
Considerations.” Federal Probation 43 (1): 37–43.
Liebling, A., D. Price, and C. Elliott. 1999. “Appreciative Inquiry and Relationships
in Prison.” Punishment & Society 1 (1): 71–98.
LIS, Inc. 2002. Staffing Analysis for Women’s Prisons and Special Prison Populations. Washington, DC: U.S. Department of Justice, National Institute of Corrections.
———. 2003. Corrections Agency Collaborations with Public Health. (Special
Issues in Corrections). Washington, DC: U.S. Department of Justice, National Institute of Corrections.
Lowenkamp, C.T., A.M. Holsinger, and E.J. Latessa. 2001. “Risk/Need Assessment,
Offender Classification, and the Role of Childhood Abuse.” Criminal Justice and
Behavior 28 (5): 543–563.
MacKenzie, D.L. 2001. “Corrections and Sentencing in the 21st Century: EvidenceBased Corrections and Sentencing.” The Prison Journal 81 (3): 299–312.
Marvell, T.B. 1995. “Sentencing Guidelines and Prison Population Growth.” The
Journal of Criminal Law and Criminology 85 (3): 696–709.
Mauer, M. 2001. “The Causes and Consequences of Prison Growth in the United
States.” Punishment & Society 3 (1): 9–20.
70

Objective Prison Classification: A Guide for Correctional Agencies

McShane, M.D., F.P. Williams III, and H.M. Dolny. 2002. “Do Standard Risk Prediction Instruments Apply to Female Parolees?” Women & Criminal Justice 13
(2/3): 163–182.
Megargee, E.I., and M.J. Bohn, Jr., with J. Meyer, Jr., and F. Sink. 1979. Classifying Criminal Offenders: A New System Based on the MMPI. Beverly Hills, CA:
Sage Publications.
Moffatt, G.K. 2002. A Violent Heart: Understanding Aggressive Individuals. Westport, CT: Praeger Publishers.
Morris, N., and D. Rothman, eds. 1995. Oxford History of the Prison: The Practice
of Punishment in Western Society. New York, NY: Oxford University Press.
National Council on Crime and Delinquency. 1992. Criminal Justice Sentencing
Policy Statement. San Francisco, CA: National Council on Crime and Delinquency.
Nelson, M., and J. Trone. 2000. Issues in Brief: Why Planning for Release Matters.
Washington, DC: Vera Institute of Justice, State Sentencing and Corrections
Program.
Phillips, R.L., and C.R. McConnell. 1996. The Effective Corrections Manager:
Maximizing Staff Performance in Demanding Times. Gaithersburg, MD: Aspen
Publishers, Inc.
Quay, H. 1981. An Empirical Approach to the Differential Behavioral Assessment of
the Adult Offender. Washington, DC: U.S. Department of Justice, Office of Justice
Programs, National Institute of Justice.
Quay, H., and C. Love. 1989. Behavioral Classification for Female Offenders in the
South Carolina Department of Corrections. Washington, DC: U.S. Department of
Justice, National Institute of Corrections.
Rivera, B.D., E.L. Cowles, and L.G. Dorman. 2003. “An Exploratory Study of Institutional Change: Personal Control and Environmental Satisfaction in a Gang-Free
Prison.” The Prison Journal 83 (2): 149–170.
Roberts, J.W., ed. 1994. Escaping Prison Myths: Selected Topics in the History of
Federal Corrections. Washington, DC: The American University Press.
Schram, P.J., B.A. Koons-Witt, and M. Morash. 2004. “Management Strategies
When Working with Female Prisoners.” Women & Criminal Justice 15 (2): 25–50.
Shutt, R. 1983. Behavioral Classification System of Adult Offenders in the South
Carolina Department of Corrections. Washington, DC: U.S. Department of Justice,
National Institute of Corrections.

71

Suggested Readings

Simourd, D.J. 2004. “Use of Dynamic Risk/Need Assessment Instruments Among
Long-Term Incarcerated Offenders.” Criminal Justice and Behavior 31 (3):
306–323.
Smith, W.A., and C.E. Fenton. 1978. “Unit Management in a Penitentiary: A Practical Experience.” Federal Probation 42 (3): 40–46.
Solomon, L. 1980. “Developing an Empirically Based Model for Classification
Decision-Making.” Prison Law Monitor 2 (9): 234–237.
Sorensen, J., and D. Stemen. 2002. “The Effect of State Sentencing Policies on
Incarceration Rates.” Crime & Delinquency 48 (3): 456–475.
Stojkovic, S., and M.A. Farkas. 2003. Correctional Leadership: A Cultural Perspective. Belmont, CA: Thomson/Wadsworth.
Thompson, M.D., M. Reuland, and D. Souweine. 2003. “Criminal Justice/Mental
Health Consensus: Improving Responses to People With Mental Illness.” Crime &
Delinquency 49 (1): 30–51.
Toch, H. 1992. Living in Prison: The Ecology of Survival. Revised Edition. Washington, DC: American Psychological Association.
Walters, G.D. 2003. “Changes in Criminal Thinking and Identity in Novice and
Experienced Inmates: Prisonization Revisited.” Criminal Justice and Behavior
30 (4): 399–421.
Waters, J.E., and W.L. Megathlin. 2002. “Evaluating Change in Social Climate in a
Close Security State Correctional Facility.” Journal of Offender Rehabilitation
34 (4): 71–84.
Wool, J., and D. Stemen. 2004. Issues in Brief: Changing Fortunes or Changing
Attitudes? Sentencing and Correction Reforms in 2003. Washington, DC: Vera Institute of Justice, State Sentencing and Corrections Program.
Wright, K.N. 1991. “A Study of Individual, Environmental, and Interactive Effects
in Explaining Adjustment to Prison.” Justice Quarterly 8 (2): 217–242.
———. 1994. Effective Prison Leadership. Binghamton, NY: William Neil
Publishing.

72

Appendix
Glossary of Key Terms

Glossary of Key Terms
Classification
Administrative segregation: A portion of the prison population that is segregated
from the general population for security reasons. Assignment to administrative segregation is for an indefinite period, but the agency must have explicit criteria that
define the circumstances under which a prisoner will be allowed to return to the general population in a timely manner. Usually no more than 5 percent of the entire
prison population is to be in administrative segregation.
Custody level: The level of risk posed by a prisoner to the safety and security of the
institution, other prisoners, and the state, as determined by the external classification system. The traditional custody levels are minimum, medium, and maximum.
(See “Security level.”)
Disciplinary segregation: A portion of the prison population that is segregated
from the general population for violation of the agency’s rules and regulations.
Unlike administrative segregation, prisoners are placed in disciplinary segregation
for a specific period and must be released at the end of that period unless they have
committed additional violations. Usually no more than 2 percent of the entire prison
population is to be in disciplinary segregation.
Discretionary override: An override of a scored custody level, based on the professional judgment of trained classification staff. Discretionary overrides should
account for 5–15 percent of all custody level decisions and should be equally balanced between increases and decreases in custody levels.
External classification: A classification system designed to determine a prisoner’s
custody level assignment and the facility that best suits the prisoner.
General population: That portion of the prison population with no special securityrelated restrictions on access to basic programs and services. In general, approximately 80 percent of prisoners are in the general population; these prisoners are
classified as minimum, medium, close, or maximum custody.
Housing plan: The plan that determines the security and custody features of each
housing unit at the facility level. The housing plan is an essential part of the internal classification system.
Internal classification: A classification system designed to determine a prisoner’s
housing, program, and work assignments within a correctional facility.
Nondiscretionary override: An override of a scored custody level, based on the
agency’s policy of restricting the custody level of particular prisoners. Classification
staff have no discretionary power to disregard such overrides.

75

Appendix

Objective classification system: A classification system that determines prisoner
custody level and program needs using a set of criteria that have been tested to
demonstrate acceptable reliability and validity. Objective classification systems typically consist of initial classification and reclassification forms that use quantifiable
scoring items and scales to determine custody level.
Override: A feature of an objective classification system that allows classification
staff to depart from the scored custody level. (See “Discretionary override” and
“Nondiscretionary override.”)
Protective custody: A portion of the prison population that is segregated from the
general population for their own protection from other prisoners. Assignment to
protective custody is for an indefinite period, but the agency must have explicit criteria that define the circumstances under which a prisoner will be allowed to return
to the general population in a timely manner. In general, no more than 2 percent of
the entire prison population is to be in protective custody.
Security level: The degree of security afforded by the architectural and staffing
attributes of the prison facility and housing units within the prison. Security levels
are related to custody levels, and the traditional security levels are minimum, medium, and maximum. (See “Custody level.”)
Subjective classification: A classification system that relies on the subjective professional judgment of staff to determine prisoner custody levels and program needs.

Reliability
Internal reliability: Consistency among answers to multiple instrument items that
measure the same concept. For example, objective risk classification instruments
often have redundant items that measure the same characteristic. If the answers to
these items are consistent, the instrument has good internal reliability.
Interrater reliability: Consistency among raters, i.e., whether different classification staff use the classification criteria in a similar manner. Evaluated by having different staff members assess the same prisoner.
Intrarater reliability: The same rater’s consistency over time, i.e., whether a classification officer uses the classification criteria consistently over time. Evaluated by
having an officer rescore the same prisoner on several dates.

Sampling
Probability sampling: Procedure in which each case in the population has a known
probability (other than zero) of being selected for the sample.
Sample: One or more cases selected from the population.
76

Objective Prison Classification: A Guide for Correctional Agencies

Simple random sampling: Procedure in which cases in the sample frame are numbered sequentially and then numbers are selected randomly.
Stratified sampling: Procedure in which the population is broken into subgroups
and then each subgroup is randomly sampled.
Systematic sampling: Procedure in which a first case is randomly selected in the
sample frame and then every “nth” subsequent case is selected.

Statistical Terms
Analysis of variance: Method for determining the significance of the difference
between any number of sample means simultaneously. Referred to as “ANOVA.”
2
Chi-square: A nonparametric statistic (χ ) used to determine whether observed differences between two nominal-level variables are statistically significant.

Correlation: A measure (r) of the degree to which two or more variables are associated. It ranges from –1.00 to +1.00.
Dependent variable: A variable (y) the analysis wants to predict. Its value is directly related to, or depends on, the value of the independent variable.
Independent variable: A variable (x) believed to be associated or correlated with
the dependent variable, i.e., to produce an effect on the variable the analysis wants
to predict.
Linear regression: A statistic used to determine how changes in one factor (y)
affect another factor (x). Linear regression is closely related to correlation. It uses
the coefficient of correlation (r) and percent of variance explained (r2).
Logistic regression: A type of regression analysis often used to determine the independent effect of each of several explanatory variables by controlling for several
factors simultaneously.
Mean: The average number in a distribution.
Median: The midpoint or middle number of a distribution.
Mode: The number that occurs more frequently than any other in a distribution.
Range: The distance between the highest and lowest values in a distribution.
Standard deviation: The square root of the variance. (See “Variance.”)
Statistical significance: The chance of incorrectly rejecting the null hypothesis
(i.e., of finding no difference between the tested variables).
77

Appendix

Variance: The mean sum of all squared deviations from the mean of any distribution of values. It summarizes the amount of dispersion, or variation, of the scores
around the mean.

Validity
Content validity: Whether an instrument covers the variety of topics encompassed
by the subject being assessed. For example, a risk assessment instrument that
addresses discipline but ignores escape has weak content validity.
External validity: An instrument’s effectiveness when used for other prison populations. For example, if an instrument is designed for a particular sample of prisoners but that sample does not represent the composition of the entire prison
population (which may have changed over time), then the instrument has external
validity problems.
Face validity: The plausibility of an instrument, i.e., whether it appears to be valid.
For example, face validity might be determined by asking staff whether they think
the instrument has the right factors and whether the factors are weighted properly.
Face validity is the weakest sort of validity because what is plausible is not necessarily valid.
Internal validity: The adequacy of an instrument’s design and testing. For example, an instrument has internal validity problems if its designers based it on carelessly collected data or a biased sample.
Predictive validity: The ability of an instrument to predict prisoner behavior.
For example, predictive validity might be determined by relating scores on a risk
assessment instrument at classification to actual disciplinary adjustment in general
confinement.

78

User Feedback Form
Please complete and return this form to assist the National Institute of Corrections in assessing the value
and utility of its publications. Detach from the document and mail to:
Publications Feedback
National Institute of Corrections
320 First Street N.W.
Washington, DC 20534
1. What is your general reaction to this document?
______Excellent ______Good ______Average ______Poor ______Useless
2. To what extent do you see the document as being useful in terms of:
Useful

Of some use

Not useful

Providing new or important information
Developing or implementing new programs
Modifying existing programs
Administering ongoing programs
Providing appropriate liaisons
3. Do you believe that more should be done in this subject area? If so, please specify the types of
assistance needed.____________________________________________________________________
4. In what ways could this document be improved? ________________________________________________
5. How did this document come to your attention? ____________________________________________
6. How are you planning to use the information contained in this document? __________________________
7. Please check one item that best describes your affiliation with corrections or criminal justice.
If a governmental program, please also indicate the level of government.
_____ Citizen group
_____ College/university
_____ Community corrections
_____ Court
_____ Department of corrections or prison
_____ Jail
_____ Juvenile justice

_____ Legislative body
_____ Parole
_____ Police
_____ Probation
_____ Professional organization
_____ Other government agency
_____ Other (please specify)

8. Optional:
Name: ____________________________________________________________________________
Agency: ____________________________________________________________________________
Address: __________________________________________________________________________
Telephone:__________________________________________________________________________
Objective Prison Classification:
A Guide for Correctional Agencies