U.S. Department of Justice
Office of the Attorney General

The First Step Act of 2018:
Risk and Needs Assessment
System

The First Step Act of 2018:
Risk and Needs Assessment
System

Table of Contents
Letters From Leadership .......................................................................................................iv
Introduction.................................................................................................................................1
Chapter 1
Developing the First Step Act of 2018
Risk and Needs Assessment System ..........................................................................................4
Chapter 2
Characteristics of an Effective
Risk and Needs Assessment System ........................................................................................25
Chapter 3
The First Step Act of 2018
Risk and Needs Assessment System ........................................................................................42
Chapter 4
Implementing the First Step Act of 2018
Risk and Needs Assessment System ........................................................................................70

-----~o~----iii

(@fffre of t~e 1\ttorttet,? ~enentl
B'as-4ingfon, m. <!!. 20,5.30
July 19, 2019
The efficient and effective implementation of the First Step Act is a priority for the
Department of Justice and this Administration. Today, the Department is announcing three
major achievements related to the implementation ofthis important criminal justice reform
measure.
First, the Bureau of Prisons will release approximately 3,100 from its custody, as the First
Step Act's provision increasing good conduct time takes effect. This action is in addition to the
approximately 1,691 individuals whose sentences have been reduced due to the retroactive
application of the Fair Sentencing Act.
Second, the Department will fully fund the $75 million authorized by the First Step Act
in FY2019. I visited a BOP facility and saw first-hand how effective programs can have a
positive impact on inmates. Therefore, I directed that existing resources be reallocated in
FY2019 to expand vocational training and job readiness programs, to increase the availability of
Medication Assisted Treatment, to support programs tailored to the needs of the female inmate
population, and to increase education opportunities for inmates.
Third, we have developed a new Risk and Needs Assessment System called the Prisoner
Assessment Tool Targeting Estimated Risk and Needs, or simply "PATTERN." In crafting the
System, the Department worked to make the benefits of the FSA as widely available as possible
without compromising predictive reliability. The System is a good initial step based on the
available information, and we will aim to improve it, with more time, consultation, data, and
research.
The Department drew on the expertise of a wide range of stakeholders in developing
PATTERN. I would like to thank the Director of the Bureau of Prisons, the Director of the
Administrative Office of the United States Courts, the Director of the Office of Probation and
Pretrial Services, the Director of the National Institute of Justice, the Director of the National
Institute of Corrections, and the members of the Independent Review Committee (IRC), with
whom I consulted in developing the System. I also want to thank the experts, practitioners, and
policy makers, the public, and other stakeholders who engaged in this process. Finally, thank
you to the dedicated staff at the Department of Justice, who have worked tirelessly to faithfully
implement the First Step Act.
While the launch of PATTERN is a big step, it is only the first step. The Department is
committed to making the benefits of the Act widely available while maintaining the System's
predictive reliability. To that end, the Department is holding a 45-day public study period, in
which the public will be able to review the System and consider ways in which it may be

improved. Following this study period, the Department in September will invite stakeholders,
public interest organizations, and the public to comment on the system. The experience of these
communities will aid the Department as it works to improve the System.
Our communities are safer when we do a better job of rehabilitating offenders in our
custody and preparing them for a successful transition to life after incarceration. The
Department remains committed to this important aim and will continue to work to make
America's communities safer.

William P. Barr
Attorney General

U.S. Department of Justice
Federal Bureau of Prisons
Washington D.C. 20534

July 19, 2019
I am extremely proud of the more than 35,000 staff of the Bureau of Prisons (BOP) and the
work we do every day keeping the American people and our Nation's communities safe. Throughout
our nearly 90-year history, providing programs and training to prepare federal inmates for a
successful return to the community has been central to our mission. The First Step Act (FSA) – the
first meaningful criminal justice reform in many years – will allow us to enhance and improve these
efforts.
The BOP, along with the Department of Justice, is fully committed to the successful
implementation of the FSA. A number of important changes have already taken place. Over 1,500
inmates received sentence reductions under the FSA’s retroactive application of the Fair Sentencing
Act of 2010. BOP expanded its compassionate release and Second Chance Act home confinement
programs. The BOP also continues to update existing policies to ensure they reflect the changes now
memorialized in federal law. The BOP has completed new projected release date calculations for
over 3,000 inmates due for release because of the FSA’s Good Conduct Time changes. We wish
these individuals every success in their transition back into our communities. For those still in
custody, the BOP is working to ensure inmates maintain close ties with loved ones by transferring
inmates closer to their release residences in accordance with BOP policy and practice, and consistent
with the FSA.
We are committed to helping inmates return to their communities to lead productive, crimefree lives. The new risk and needs assessment system will ensure we conduct individualized
assessments of each inmate’s specific needs. These assessments will help us to create plans with
programs and productive activities tailored to meet identified needs. These plans will provide
inmates the opportunity for improvement and growth and promote successful reentry into their
communities.
This is an exciting time for the Bureau of Prisons. The FSA will have a lasting impact on
federal corrections and may serve as a model for state and local corrections as well. The dedicated
corrections professionals throughout our 122 federal prisons nationwide remain mission-focused on
helping change lives and enhancing public safety.

Hugh J. Hurwitz
Acting Director, Federal Bureau of Prisons

U.S. Department of Justice
Office of Justice Programs
National Institute of Justice
_________________________________________________________________________
Office of the Director

Washington, D.C. 20531

July 19, 2019
An offender’s first steps out of prison are important steps toward a crime-free and productive life.
The First Step Act of 2018 aims to reduce recidivism and reform the federal prison system so that those
exiting the system do not return.
The National Institute of Justice (NIJ) is playing a key role in fulfilling those objectives. We are
working with our federal partners, specifically the Bureau of Prisons in several areas, including
prioritizing who should receive evidence-based recidivism reduction programming and determining who
may be eligible for early release into the community. We are assisting to develop recommendations
based, as always, on empirical evidence: Which recidivism reduction programs and activities work best?
How are these programs assessed, and can we strengthen the best programs, and improve or end the
worst?
At NIJ, we understand that research, development, and evaluation cannot happen in a vacuum.
You have to listen to those on the front lines and get input from all perspectives. The development of a
new risk and needs assessment system is no different. As an example, and in cooperation with the
Department of Justice, we have hosted three listening sessions to obtain a diverse set of perspectives
about the development of a risk and needs assessment system.
Beyond conducting research on recidivism and prison reform, NIJ has already taken critical steps
in the development of the new risk and needs assessment system. We have reviewed and continue to
review existing risk and needs assessment systems and the available empirical research to determine how
NIJ can contribute to this important line of inquiry and development. We will continue to consult with
external experts and stakeholders, including researchers, think tanks, prison reform advocates, corrections
and law enforcement groups, and victims and victim advocacy groups, to learn more about how NIJ can
support research to inform evidence-based practices.
The current recidivism statistics are not encouraging. Far too many offenders leave prison only
to return. Taking steps toward reducing recidivism and reforming the federal prison system will take a
sustained investment and commitment. NIJ is and remains dedicated to addressing these important
challenges.

David B. Muhlhausen, Ph.D., Director
National Institute of Justice

viii

Introduction
“America wins when citizens with a criminal record can contribute
to their communities as law-abiding members of our society.”
-President Donald J. Trump
June 13, 2019, from the East Room of the White House

O

Chapter 1

n December 21, 2018, President
Donald J. Trump signed the First
Step Act of 2018 into law. Title I of
the First Step Act of 2018 (FSA or the Act)
is focused on reforms to reduce recidivism
among the federal prison population. Many
of Title I’s reforms hinge on the creation of a
risk and needs assessment system.

Chapter 1, Developing the First Step
Act of 2018 Risk and Needs Assessment
System, details the requirements of the FSA
regarding the development of a risk and
needs assessment system, including the
responsibilities of the Attorney General and
the Independent Review Committee. Tis
chapter also summarizes the Department of
Justice’s work to fully implement the Act’s
requirements in creating the System.

Under the FSA, the Attorney General is
charged with developing and releasing a
risk and needs assessment system for use in
the federal prison system. With this report,
Attorney General William P. Barr releases
the First Step Act of 2018 Risk and Needs
Assessment System.

Chapter 2
In Chapter 2, Characteristics of an Efective
Risk and Needs Assessment System, this
report identifes those characteristics and
principles that are fundamental to developing
an efective risk and needs assessment system.
Tis chapter also describes the valuable data
and information that the Department of
Justice received from our federal and state
partners and experts in the feld on developing
a strong risk and needs assessment system.
Tese characteristics, principles, and data
informed the development of the System.

Tis report outlines the work of the
Department of Justice to develop and
implement the Risk and Needs Assessment
System (System). It also introduces the new
System that the Federal Bureau of Prisons
will deploy in its facilities. And the report
announces the Department of Justice’s
strategic plan to evaluate, validate, and
enhance the System over time.

-----~o~----1

Chapter 3

Chapter 4

Chapter 3, Te First Step Act of 2018 Risk
and Needs Assessment System, describes the
adopted System in detail, including the new
assessment tool that will be deployed in the
Federal Bureau of Prisons. Tis chapter then
provides an explanation of the strengths of
the tool and enhancements ofered by the
new System.

Chapter 4, Implementing the First Step Act
of 2018 Risk and Needs Assessment System,
presents the Department’s strategic plan to
fully and completely implement the System
in the feld. It also includes an agenda
for continued engagement with experts,
stakeholders, and the public on the System.
Te chapter concludes by describing the
signifcant resources that the Department
of Justice is expending and will expend to
implement the System.

-----~o~----2

3

Chapter 1

Developing the First Step Act of 2018
Risk and Needs Assessment System
BACKGROUND

P

resident Donald J. Trump signed
the First Step Act (FSA or the Act),1
into law on December 21, 2018. Te
Act passed both houses of Congress with
bipartisan support. As President Trump has
said, the FSA “will give countless current
and former prisoners a second chance at life
and a new opportunity to contribute to their
communities, their states, and their nation.”2
Te FSA makes signifcant reforms to the
criminal justice system.

As discussed below, several of the FSA’s
reforms turn on the development of a risk
and needs assessment system for use in
the Federal Bureau of Prisons (BOP). Tis
chapter discusses the FSA’s requirements
with respect to that system, including the
responsibilities of the Attorney General
and the Independent Review Committee
mandated by the Act. It also summarizes the
Department of Justice’s work to implement
the requirements related to the system.

President Trump signs the First Step Act on December 21, 2018, in the Oval Ofce at the White House.

________Jo~----4

I. Statutory Obligations
A. Description of the Risk and
Needs Assessment System
In passing the First Step Act, Congress
placed a particular emphasis on the
development and deployment of a risk
and needs assessment system in the BOP
Indeed, Section 101 of the FSA requires the
Attorney General, in consultation with an
Independent Review Committee (discussed
in further detail below), to “develop and
release publicly on the Department of Justice
website a risk and needs assessment system.”3
Te FSA provides that the Attorney General
“may use existing risk and needs assessment
tools, as appropriate.” 4
Pursuant to the FSA, the risk and needs
assessment system “shall be used” to:
(1) determine the recidivism risk of
each prisoner as part of the intake
process, and classify each prisoner as
having minimum, low, medium, or
high risk for recidivism;
(2) assess and determine, to the extent
practicable, the risk of violent or
serious misconduct of each prisoner;
(3) determine the type and amount of
evidence-based recidivism reduction
programming that is appropriate
for each prisoner and assign each
prisoner to such programming
accordingly, and based on the
prisoner’s specifc criminogenic
needs, and in accordance with
subsection (b);

(4) reassess the recidivism risk of
each prisoner periodically, based
on factors including indicators of
progress, and of regression, that are
dynamic and that can reasonably be
expected to change while in prison;
(5) reassign the prisoner to appropriate
evidence-based recidivism reduction
programs or productive activities
based on the revised determination
to ensure that (A) all prisoners at each risk level
have a meaningful opportunity to
reduce their classifcation during
the period of incarceration;
(B) to address the specifc
criminogenic needs of the prisoner;
and
(C) all prisoners are able to
successfully participate in such
programs;
(6) determine when to provide
incentives and rewards for successful
participation in evidence-based
recidivism reduction programs or
productive activities in accordance
with subsection (e);
(7) determine when a prisoner is
ready to transfer into prerelease
custody or supervised release in
accordance with section 3624; and
(8) determine the appropriate use of
audio technology for program course
materials with an understanding of
dyslexia.5

5
---------,0.-------------

Tus, the system is to be used to “determine
the recidivism risk of each prisoner as part
of the intake process, and classify each
prisoner as having minimum, low, medium,
or high risk for recidivism [and] assess and
determine, to the extent practicable, the risk
of violent or serious misconduct of each
prisoner.”6 Te system also must be used
periodically to reassess the recidivism risk
of each prisoner “based on factors including
indicators of progress, and of regression,
that are dynamic and that can reasonably be
expected to change while in prison.”7

activities that shall be assigned for each
prisoner.”12 For example, the FSA requires
the BOP to prioritize access to recidivism
reduction programs for those who are at a
higher risk of reofending.13 Results from
the system also will be used to reassign
prisoners to evidence-based recidivism
reduction programs or productive activities
based on the periodic reassessments of their
recidivism risk.14
Te system’s other requirements include
using results to provide “guidance on
program grouping and housing assignment
determinations” and, afer taking account
of the safety of prisoners and others at the
prison, “provid[ing] that prisoners with a
similar risk level be grouped together in
housing and assignment decisions to the
extent practicable;”15 providing incentives
and rewards for prisoners “to participate in
and complete evidence-based recidivism
reduction programs,” including phone and
visitation privileges, transfer to an institution
closer to a prisoner’s release residence
(taking account of availability, security, and
the recommendation of the warden where
the prisoner is currently held), and other
increased privileges and time credits;16
determining “when to provide incentives
and rewards for successful participation
in evidence-based recidivism reduction
programs or productive activities;”17 and
determining “the appropriate use of audio
technology for program course materials
with an understanding of dyslexia.”18

A prisoner’s risk classifcation level may
afect the prisoner’s ability to receive certain
rewards and incentives. For example, eligible8
prisoners are to earn 10 days of time credits
for every 30 days of successful participation
in evidence-based recidivism reduction
programming or productive activities.9
Eligible prisoners who are classifed in the
minimum or low risk categories and who
have not increased their risk of recidivism
over two assessments may receive fve
additional days of time credit for every 30
days of successful participation in evidencebased recidivism reduction programming or
productive activities.10 Tese time credits are
available for eligible inmates who successfully
complete evidence-based recidivism reduction
programming or productive activities. A
prisoner’s risk assessment over time also
afects that prisoner’s ability to use those
time credits to receive prerelease custody or
early supervised release.11
Te results from the system will be used
to “provide guidance on the type, amount,
and intensity of evidence-based recidivism
reduction programming and productive

Te system must be released no later than
210 days afer enactment of the FSA—July
20, 2019.19

-----~o~----6

B. Responsibilities of the
Attorney General
In carrying out his responsibilities with
respect to the system, the Attorney General
must consult with the Director of the Bureau
of Prisons, the Director of the Administrative
Ofce of the United States Courts, the
Director of the Ofce of Probation and

Pretrial Services, the Director of the
National Institute of Justice, the Director of
the National Institute of Corrections, and
the Independent Review Committee (IRC)
authorized under the FSA.20

Attorney General William P. Barr and Housing and Urban Development Secretary Benjamin S. Carson,
Sr., M.D., attend a First Step Act celebration in the East Room of the White House on April 1, 2019.
Photo: Chip Somodevilla (Getty Images).

Te FSA specifcally directs21 the Attorney
General, in consultation with the IRC, to:

that exist in State-operated prisons
throughout the United States;

(1) review the efectiveness of
evidence-based recidivism reduction
programs that exist as of the date
of enactment of this subchapter in
prisons operated by the Bureau of
Prisons;

(3) identify the most efective
evidence-based recidivism reduction
programs;

(2) review available information
regarding the efectiveness of
evidence-based recidivism reduction
programs and productive activities

(4) review the policies for entering
into evidence-based recidivism
reduction partnerships described in
section 3621(h)(5); and
(5) direct the Bureau of Prisons
regarding -

7
---------,0.-------------

recidivism

programming that most efectively
reduces the risk of recidivism; and

(B) the ability for faith-based
organizations to function as a
provider of educational evidencebased programs outside of the
religious classes and services
provided through the Chaplaincy;
and

(D) products purchased by Federal
agencies that are manufactured
overseas and could be manufactured
by prisoners participating in a prison
work program without reducing job
opportunities for other workers in
the United States.

(A) evidence-based
reduction programs;

C. Independent Review Committee

(C) the addition of any new
efective evidence-based recidivism
reduction programs that the
Attorney General fnds.

1. Responsibilities
Te IRC must assist the Attorney General
in carrying out his responsibilities under 18
U.S.C. §§ 3631(b), 3632, and 3633.25 Tat
includes26 assisting in:

Te Attorney General also must “develop and
implement training programs for Bureau of
Prisons ofcers and employees responsible
for administering” the system and “monitor
and assess” the use of the system, including
conducting annual audits of the BOP’s use of
the system.22

(1) conducting a review of the existing
prisoner risk and needs assessment
systems in operation on the date of
enactment of this Act;

Afer releasing the system, the Attorney
General must annually “review, validate, and
release publicly on the Department of Justice
website the risk and needs assessment system.”23
Consistent with this annual requirement, the
Attorney General must “conduct ongoing
research and data analysis”24 on:

(2) developing recommendations
regarding evidence-based recidivism
reduction programs and productive
activities;

(A) evidence-based recidivism
reduction programs relating to
the use of prisoner risk and needs
assessment tools;

(A) evidence-based recidivism
reduction programs relating to
the use of prisoner risk and needs
assessment tools;

(B) the most efective and efcient
uses of such programs;

(B) the most efective and efcient
uses of such programs; and

(C) which evidence-based recidivism
reduction programs are the most
efective at reducing recidivism, and
the type, amount, and intensity of

(C)
which
evidence-based
recidivism reduction programs
are the most efective at reducing
recidivism, and the type, amount,

(3) conducting research and data
analysis on—

-----~o~----8

and intensity of programming that
most efectively reduces the risk of
recidivism; and
(4) reviewing and validating the risk
and needs assessment system.
Te Director of the Bureau of Prisons is
required to assist the IRC in performing
its duties and in promptly responding to
requests from the IRC for access to BOP
facilities, personnel, and information.27
Within two years of the enactment of the FSA,
the IRC must produce a report to the Senate
and House Judiciary Committees and Senate
and House Appropriations Subcommittees
on Commerce, Justice, Science, and Related
Agencies.28
2. Composition of the IRC
Te members of the IRC “shall all have
expertise in risk and needs assessment
systems.”29 Te FSA requires the National
Institute of Justice (NIJ) to “select a
nonpartisan and nonproft organization with
expertise in the study and development of
risk and needs assessment tools to host the
Independent Review Committee.”30 Tat
outside organization then must appoint not
fewer than six members to the IRC.31

II.

First Step Act Implementation

A. Department Implementation
Eforts
Following the enactment of the FSA, the
Department of Justice began the process
of reviewing and implementing the Act’s
requirements.
Department leadership,
including the Ofce of the Deputy Attorney
General and the Ofce of Legal Policy, began

to coordinate implementation eforts of
relevant components, including the BOP, the
NIJ, the Executive Ofce for United States
Attorneys and the U.S. Attorney community,
the Criminal Division, and the Ofce of
Legislative Afairs. Tis coordinated efort
has ensured that the Department carefully,
diligently, and fully implements the FSA’s
requirements. For example, as discussed in
Chapter 2, the Department has consulted
with experts from state and federal criminal
justice systems to inform its implementation
eforts.
B. Appointing the Independent
Review Committee
On April 8, 2019,32 NIJ announced33 that
the nonpartisan and nonproft Hudson
Institute would host the IRC. In selecting
an organization, NIJ considered a variety
of factors, including the organization’s
legislatively mandated status as a nonproft
and nonpartisan organization, the ability to
build a qualifed team of experts, expertise
regarding risk and needs assessments,
stafng capabilities, and ability to complete
high-profle projects in a timely manner.
Te Host Committee’s role is akin to that of
a project manager, hosting, supporting, and
monitoring the activities of the IRC. Hudson,
as the Host Committee, administers the grant,
establishes a secure document-retention
system for archiving all ofcial records and
fnancial transactions, coordinates logistics,
and selects the IRC members.
Hudson is not the IRC and has no formal role
in developing the Risk and Needs Assessment
System. However, its selection is necessary

9
---------,0.-------------

to create the IRC, whose members would be
leaders in their feld and have expertise to design and implement a new system.
Te Hudson Institute subsequently identifed34 six people to serve as members of the
IRC. Tose members include experts in
criminology and prison systems, as well as
former senior justice sector policymakers.
Biographies of the IRC members are included below.

organization and membership were named.35
Subsequently, the IRC independently
organized its activities and gave advice to
Department subject matter experts, the
NIJ’s outside experts, and the Attorney
General. Tis advice occurred through
formal meetings with the Attorney General,
Deputy Attorney General, and Department
subject matter experts and through informal
consultation sessions between IRC members
and NIJ outside experts.

Within a week of funds being made available to select the IRC, the IRC host
Patti Butterfeld, Ph.D. Dr. Butterfeld is an adjunct faculty member
at Southern New Hampshire University and a former Senior Deputy
Assistant Director in the Federal Bureau of Prisons (BOP) Reentry
Services Division. In addition to serving in leadership in the
BOP’s reentry services programs, Dr. Butterfeld previously served
as a prison psychologist and coordinator for inmate treatment
programs. As a result, Dr. Butterfeld has an appreciation for the
needs of inmates within a prison and the needs of defendants as
they reenter society.
James M. Byrne, Ph.D. Dr. Byrne is a Professor and Associate Chair at the
School of Criminology and Justice Studies at University of Massachusetts
Lowell and Director of the Global Community Corrections Initiative. Dr.
Byrne received his undergraduate degree in Sociology (Summa
cum Laude) from the University of Massachusetts, Amherst (1977),
and his Masters (1980) and Doctoral degree (1983) in Criminal
Justice from Rutgers University. He is the author of several books,
monographs, journal articles, and research reports on a range of
criminal and juvenile justice policy and program evaluation issues.
Dr. Byrne is the Editor-in-Chief of the journal, Victims and Ofenders: An International
Journal of Evidence-based Research, Policy, and Practice. Dr. Byrne also serves on the
editorial boards of two other journals, Criminology and Public Policy, and the European
Journal of Probation, and on the National Advisory Committee for the journal, Federal
Probation, a publication of the Administrative Ofce of the U.S. Courts.

10
Q
,----------

Faye S. Taxman, Ph.D. Dr. Taxman is an University Professor and
Director of the Center for Advancing Correctional Excellence!
at George Mason University. Dr. Taxman’s areas of research
include evidence-based courts and corrections programs,
including program design and interventions. She is currently
researching, among other things, the development of web-based
risk-needs-responsivity tools for federal, state, and local criminal
justice agencies; she has developed the RNR Simulation Tool to
determine whether the services provided are responsive to the risk and need factors of
ofenders in their correctional systems. Dr. Taxman is on the Editorial Boards of the
Journal of Experimental Criminology, Criminology and Public Policy, and Journal of
Ofender Rehabilitation. She is co-editor of Health & Justice. She received the University
of Cincinnati award from the American Probation and Parole Association in 2002 for her
contributions to the feld. She is a Fellow of the Academy of Experimental Criminology
and a member of the Correctional Services Accreditation Panel (CSAP) of England. In
2008, the American Society of Criminology’s Division of Sentencing and Corrections
recognized her twice as Distinguished Scholar. In 2017, she received the Joan McCord
Award from the Division of Experimental Criminology. In 2018, she was awarded a
Fellow of the American Society of Criminology.

George J. Terwilliger III. Mr. Terwilliger has served in numerous
leadership roles in the Department of Justice, including as
Deputy Attorney General, Acting Attorney General, and United
States Attorney for the District of Vermont. While serving as
Deputy Attorney General, Mr. Terwilliger oversaw the Justice
Department’s operations, including the operation of the Bureau
of Prisons. In total, Mr. Terwilliger served for 15 years with the
Justice Department. Mr. Terwilliger currently is a partner at the Washington, DC, ofce
of McGuireWoods.

11
---------,0.-------------

John P. Walters. Mr. Walters is chief operating ofcer of the Hudson
Institute and former director of the Ofce of National Drug Control
Policy (ONDCP) in the cabinet of President George W. Bush. Mr.
Walters served as the nation’s “Drug Czar” from 2001 until 2009. In
that position, Mr. Walters “guided all aspects of federal drug policy
and programs—supporting eforts that drove down teen drug use
25 percent, increased substance abuse treatment and screening in
the healthcare system and dramatically dropped the availability of
cocaine and methamphetamine” in the United States.36 Mr. Walters previously served as
Chief of Staf and Deputy Director of Supply Reduction at ONDCP. Mr. Walters also has
served as Assistant to the Secretary and Chief of Staf at the United States Department of
Education.
John E. Wetzel. Secretary Wetzel is Secretary of Corrections
for the Commonwealth of Pennsylvania and the Immediate Past
Chair and member of the Executive Committee of the Council of
State Governments Justice Center. Afer initially being appointed
Secretary of Corrections in 2011, Secretary Wetzel was reappointed
in 2015 and 2019. Secretary Wetzel has almost 30 years in the feld of
corrections. He is currently the President of the Association of State
Correctional Administrators and a member of Harvard’s Executive
Session on Community Corrections. Secretary Wetzel previously was appointed as the
corrections expert to the Charles Colson Task Force on Federal Corrections.37
C. Working with Outside Experts to
Develop the System

Drs. Duwe and Hamilton quickly began to
collect and review data from BOP. Tat data
included information on BOP prisoners,
prisoners released into the community, and
BOP’s risk prediction instrument, BRAVO
(Bureau Risk Assessment Verifcation and
Observation). BRAVO was designed to
predict serious misconduct in prison. Drs.
Duwe and Hamilton also learned about
BOP’s revised tool BRAVO-R, which is
designed to address recidivism in the
community. One of the experts completed
a tour of BOP’s Designation and Sentence
Computation Center, located in Grand
Prairie, Texas. Visiting the center enabled

In addition to receiving advice from the
IRC, the Department went beyond the
requirements of the FSA by working with
outside experts and researchers throughout
the country to develop the Risk and Needs
Assessment System. On April 8, the NIJ
announced that it began contracting with
three experts, two of whom (Dr. Grant
Duwe and Dr. Zachary Hamilton) focused
on the development of the Risk and Needs
Assessment System. Biographies of Drs.
Duwe and Hamilton are included below.

-----~o~----12

the expert to learn about current risk and
need tools that in are place and how those
tools ft into the BOP’s operations. Tese

experts likewise met regularly with members
of the IRC and consulted personally with the
Deputy Attorney General.

Grant Duwe, Ph.D. Dr. Duwe is the Director of Research for the
Minnesota Department of Corrections and a nationally recognized
expert on the development of recidivism risk assessment systems. Dr.
Duwe created the Minnesota Screening Tool Assessing Recidivism
Risk (MnSTARR), as well as three other separate risk assessment
instruments. Dr. Duwe was awarded the American Society of
Criminology’s Practitioner Research Award based on his development
of the MnSTARR assessment system. Dr. Duwe has written over 60
articles that were published in peer-reviewed academic journals, including Criminology,
Criminology and Public Policy, Criminal Justice and Behavior, and Sexual Abuse: A Journal
of Research and Treatment. Dr. Duwe is a non-resident scholar at the Institute for Studies
of Religion at Baylor University.

Zachary Hamilton, Ph.D. Dr. Hamilton is an Associate Professor
of Criminal Justice and Criminology and the Director of the
Washington State Institute for Criminal Justice. Dr. Hamilton has
developed nationally renowned risk-need assessment systems for
both juveniles and adults in Washington State, Nebraska, Tennessee,
Delaware, and Iowa. Dr. Hamilton is a co-editor of the Handbook
of Recidivism Risk/Needs Assessment Tools (2018). Dr. Hamilton
has authored or co-authored numerous peer-reviewed publications, reports, and
manuscripts on ofender risk assessments. Dr. Hamilton’s recent publications have
appeared in Experimental Criminology, Justice Quarterly, Criminology and Public Policy,
Criminal Justice & Behavior, and Ofender Rehabilitation.

13
---------,0.-------------

D. Consulting with Experts, Stakeholders,
and Other Interested Parties
As part of the System’s development, the
Department held three listening sessions
to enable experts, stakeholders—including
organizations representing crime victims—

and public interest organizations to comment
on the development of the Risk and Needs
Assessment System. Te sessions took place
on April 3, 2019, April 5, 2019, and May 3,
2019, at the Ofce of Justice Programs in
Washington, DC.

First Step Act – Risk and Needs Assessment System listening session held on April 3, 2019, at the Ofce of
Justice Programs.

First Step Act – Risk and Needs Assessment System listening session held on April 5, 2019, at the Ofce of
Justice Programs.

14
Q,-------------

First Step Act – Risk and Needs Assessment System listening session held on May 3, 2019, at the Ofce of
Justice Programs.

Senior Department leadership, including the
Director of NIJ and representatives from the
Ofce of Deputy Attorney General and the
BOP, welcomed those who ofered comments
and others who attended.
Tose who ofered oral and written
comments included:
Aamra Ahmad, Sentencing Resource
Counsel of the Federal Public and
Community Defenders;
Sarah Anderson, Federal Afairs Manager,
FreedomWorks;
Sakira Cook, Director of Justice Reform
Program, Te Leadership Conference on
Civil and Human Rights;
Monique Dixon, Deputy Director of Policy,
NAACP Legal Defense and Educational
Fund, Inc.;
Ellen Donnarumma, President, International
Community Corrections Association, and

Vice President of Justice Services, Community
Resources for Justice;
Jim Felman, American Bar Association;
Richard Fiano, Regional Vice President,
Association of Federal Narcotics Agents
(AFNA);
Mark Holden, Senior Vice President, Koch
Industries;
John Hollywood, Senior Operations
Researcher, RAND Corporation;
DeAnna Hoskins, President and CEO,
JustLeadershipUSA;
Susan Howley, Center for Victim Research,
Justice Research and Statistics Association;
Jessica Jackson, National Director and
Co-Founder, #cut50;
John Koufos, National Director of Reentry
Initiatives, Right on Crime;

15
---------,0.-------------

Paul Larkin, Senior Legal Research Fellow,
Te Heritage Foundation;

Te participants ofered constructive advice
and comments on the following topics,
among others: the contents of a useful risk
assessment; the process for constructing
a risk assessment; the need to avoid racial
bias in the drafing and administration of
risk assessments; the need to implement the
FSA fully and expeditiously; and the desired
contents of future criminal justice reform
legislation. Additionally, some participants
emphasized the need to take account of the
harms sufered by crime victims.

Nancy La Vigne, Vice President, Justice
Policy, Urban Institute;
Mary Graw Leary, Chair, United States
Sentencing Commission Victims Advisory
Group;
Tom Manger,
Association;

Major

Cities

Chiefs

Marc Mauer, Executive Director, Te
Sentencing Project;
Jesselyn McCurdy, Deputy
American Civil Liberties Union;

Te full written statements of the listening
session participants will be available on the
Department of Justice website.

Director,

E. Summary of Work to Develop the
Risk and Needs Assessment System

Pat O’Carroll, Executive Director, Federal
Law Enforcement Ofcers Association;

Te Department, its outside experts, and the
IRC have conscientiously and expeditiously
worked to meet the FSA’s requirements with
respect to developing the Risk and Needs
Assessment System. Tis section highlights
the Department’s eforts.

Sarah Picard, Research Director, Center for
Court Innovation;
Arthur Rizer, Director, Criminal Justice &
Civil Liberties, R Street Institute;
Christopher Scott, Director of Criminal
Justice, Education, and Policing Reform,
Open Society Policy Center;

1. Overview
As an initial matter, the Department and
NIJ’s experts researched and evaluated the
attributes of a successful risk and needs
assessment system. As discussed in Chapter
2, the Department requested and received
information and recommendations from
state prison systems and the BOP regarding
recidivism and risk assessment tools. Te
outside experts from NIJ learned about
BOP’s revised recidivism tool, BRAVO-R,
and analyzed it using relevant data from
BOP. Tose same experts looked for ways
to improve the predictive accuracy of

Rabbi Levi Shemtov, Director, American
Friends of Lubavith (Chabad);
Lindsay Silverberg, Te Network for Victim
Recovery of DC;
Faye Taxman, Professor, Criminology,
Law and Society Department, George
Mason University, and Director, Center for
Advancing Correctional Excellence!; and
Jonathan Tompson, Executive Director
and CEO, National Sherifs’ Association.

-----~o~----16

BRAVO-R and to build a strong risk and
needs assessment system.
While working on the Risk and Needs
Assessment System, NIJ’s experts regularly
collaborated with members of the IRC. As
part of this consultation process, the IRC
reviewed NIJ’s eforts, asked questions, and
provided input.
In addition to the work of the Department of
Justice’s experts, Attorney General William
P. Barr and Deputy Attorney General Jefrey
A. Rosen were personally involved in the
process of developing the Risk and Needs
Assessment System. Both the Attorney
General and the Deputy Attorney General
received information and input from
Department components. And both of the
Department’s top leaders met individually
with the members of the IRC to discuss the
development of the System.

Te Attorney General and Deputy Attorney
General both personally visited BOP facilities
to examine and understand the ways that a
new risk and needs assessment system would
impact inmates and the BOP system.
On July 8, 2019, the Attorney General toured
the Federal Correctional Institution (FCI)
in Edgefeld, South Carolina. Te Attorney
General met with and was accompanied
by U.S. Senators Lindsey Graham and Tim
Scott of South Carolina. Senators Graham
and Scott strongly supported passage of the
FSA, and Senator Scott served as an original
cosponsor of the bill. While at FCI Edgefeld,
Attorney General Barr spoke with prison
ofcials, including vocational training,
psychology, and education staf. Te Attorney
General also spoke with inmates who are
participating in model BOP programs
aimed at reducing recidivism and promoting

Attorney General William P. Barr, along with Senators Lindsey Graham and Tim Scott, receive a presentation
from Supervisor Stephanie Ewing, of the Culinary Arts Vocational Training program at FCI Edgefeld.

- - - - - - - - - O17r - - - - - - - - - -

successful reentry into the community.
Te Attorney General toured Edgefeld’s
UNICOR clothing and textile factory,
an occupational and vocational training
program, as well as other occupational and
vocational training programs ofered at the
facility. Inmates in UNICOR clothing and
textile factories cut, sew, weave, embroider,
and use silk screening processes, among
other things. In occupational and vocational
programs, inmates learn a variety of skillsbased, marketable job disciplines, such as
the culinary arts, computer applications,
and automobile service and maintenance.
Te Attorney General also learned about the
facility’s Residential Drug Abuse Program
(RDAP), Non-Residential Drug Abuse
Program (NR-DAP), faith-based programs,

and reentry services that help inmates
successfully reintegrate into society.
Te Deputy Attorney General visited the FCI
in Englewood, Colorado, on July 2, 2019.
During his visit, the Deputy Attorney General
focused on programming opportunities
for inmates. Tese opportunities included
the National Roofng and Paving Program
based at FCI Englewood, RDAP, NR-DAP,
Culinary Arts Vocational Training (VT),
UNICOR, and educational opportunities,
among others. Consistent with the FSA’s
focus on evidence-based programming, the
treatment philosophy in RDAP and NRDAP is evidence-based, with both using
Cognitive Behavioral Terapy.
Skilled
training, certifcates, and licenses such as the
Culinary Arts VT, UNICOR, and National

Inmate at FCI Englewood demonstrates Computer Aided Design (CAD) Drafing to Deputy Attorney General
Jefrey A. Rosen.

018

Roofng and Paving Program also are key to
success under the FSA because they provide
inmates with real-life employment skills and
a better appreciation for the importance of
developing responsible work habits.
2. FSA Statutory Obligations Met
Troughout the process of developing the
Risk and Needs Assessment System, the
Attorney General satisfed the obligations
listed in 18 U.S.C. §§ 3631(a) and 3633.
As required by section 3631(a), the Attorney
General consulted with the Director of
BOP, the Director of NIJ, the Director of
the Administrative Ofce of the United
States Courts, the Director of the Ofce of
Probation and Pretrial Services, the Director
of the National Institute of Corrections, and
the IRC in developing the System.
Te Directors of BOP and NIJ have
consistently supported and advanced the
Department’s eforts to develop the Risk and
Needs Assessment System. For example,
as noted previously, NIJ hired and worked
closely with outside experts on risk and needs
assessment systems to research, evaluate,
and make recommendations as part of the
process of developing the System. BOP has
supported those experts’ work and made its
own experts available to work with NIJ on
the risk and needs assessment tool. While
the Department was developing the System,
BOP also acted diligently to implement the
FSA’s initial requirements in its facilities
through new and updated Bureau-wide
policies. BOP and NIJ will remain closely
involved as the Risk and Needs Assessment
System is implemented and enhanced over
time.

Both the Director of the Administrative
Ofce of the United States Courts and
the Director of the Ofce of Probation
and Pretrial Services consulted with the
Department regarding the FSA.
Te Acting Director of the National Institute
of Corrections (NIC) consulted with the
Department during NIC’s Advisory Board
Meeting on Friday, June 21, 2019. During
this meeting, the Acting Director and
NIC’s Advisory Board members ofered
recommendations and input regarding
developing and implementing a new
risk and needs assessment system. In
addition to this in-person consultation,
the Acting Director subsequently sent
the Attorney General a detailed letter
containing recommendations regarding the
development and implementation of a risk
and needs assessment system. Te Attorney
General considered the NIC’s suggestions as
part of the process of developing, publishing,
and implementing the Risk and Needs
Assessment System.
As discussed previously, the Attorney General
and Deputy Attorney General personally
consulted with the IRC. Te IRC’s members
also regularly consulted with other members
of the Department, including the NIJ and its
experts. Te IRC provided valuable input
and raised probing questions to improve
the risk and needs assessment tool. Te
Department will continue to consult with
the IRC as the Department implements the
Risk and Needs Assessment System.
In addition to satisfying the FSA’s consultation
requirements in developing the Risk and
Needs Assessment System, the Attorney

19
---------,0.-------------

Attorney General William P. Barr and members of the Independent Review Committee discuss the First Step
Act of 2018 Risk and Needs Assessment System on July 11, 2019.

of the VOA, hear from individuals who have
“graduated” from the center, and observe
classes and a substance abuse support group.

General satisfed the fve requirements
outlined in 18 U.S.C. § 3633(a).38 First, as
discussed previously, the Attorney General,
through the eforts of BOP, NIJ, and other
Department components, reviewed the
efectiveness of evidence-based recidivism
reduction programs in the BOP. Te
Attorney General also received input and
recommendations regarding programming
from the IRC.

Tird, as a result of the above eforts, the
Attorney General identifed the most
efective
evidence-based
recidivism
reduction programs and has included aspects
of those programs in the Risk and Needs
Assessment System. Fourth, the Attorney
General reviewed the policies for entering
into evidence-based recidivism reduction
partnerships. Fifh, the Attorney General
directed the BOP regarding evidence-based
recidivism reduction programs, the addition
of any new evidence-based recidivism
reduction programs, and the ability of faithbased organizations to provide educational
evidence-based programs outside of
religious classes and services. In carrying
out these obligations, the Attorney General
“consider[ed] the prevalence and mitigation
of dyslexia in prisons.”39

Second, the Attorney General reviewed
available information regarding such
programs in state-operated prisons
throughout the country. For example,
Department representatives on June 27, 2019,
visited the Volunteers of America Reentry
Center (VOA) in Baltimore, Maryland. At
the state reentry center, men and women
receive counseling, job training, and
customized support to help ensure success
as they reenter their communities. Te
visitors had the chance to meet the leaders

-----~o~----20

Tus, the Department has taken informed and
proactive steps to study, develop, and publish
a robust risk and needs assessment system.
Tis report’s subsequent chapters discuss in
greater detail the information learned and
the steps taken to develop and implement the
Risk and Needs Assessment System. Chapter

2 addresses the characteristics of an efective
risk and needs assessment system. Chapter
3 discusses the Risk and Needs Assessment
System that the Department has developed.
To conclude, Chapter 4 addresses processes
to implement that System.

Attorney General William P. Barr and members of the Independent Review Committee. From lef to right:
George J. Terwilliger III, Dr. James M. Byrne, Dr. Patti Butterfeld, the Attorney General, Secretary John E.
Wetzel, and John P. Walters. Dr. Faye S. Taxman is not pictured.

21
------0

NOTES

14

Id. § 3632(a)(5).

15

Id. § 3632(c).

Remarks by President Trump at 2019
Prison Reform Summit and FIRST STEP
Act Celebration (April 1, 2019), https://
www.whitehouse.gov/briefngs-statements/
remarks-president-trump-2019-prisonreform-summit-first-step-act-celebration/
(last visited July 11, 2019).

16

Id. § 3632(d).

17

Id. § 3632(a)(6).

18

Id. § 3632(a)(8).

19

Id. § 3632(a).

20

Id. § 3631(a).

Pub. L. 115-391, § 101 (creating 18 U.S.C.
§ 3632(a)).

21

1

Pub. L. 115-391.

2

Id. § 3633(a); see also 18 U.S.C. § 3631(b)
(1), (b)(2) (requiring the Attorney General
to “develop recommendations regarding
evidence-based
recidivism
reduction
programs and productive activities in
accordance with section 3633”). In carrying
out his obligations under section 3633(a),
the Attorney General “shall consider the
prevalence and mitigation of dyslexia in
prisons.” 18 U.S.C. § 3633(b).

3

4

18 U.S.C. § 3632(a).

5

Id.

6

Id. § 3632(a)(1), (2).

7

Id. § 3632(a)(4).

See Id. § 3632(d)(4)(D), (E) (listing
categories of prisoners who are ineligible
to receive time credits). Eligible prisoners
may earn time credits only by “successfully
complet[ing] evidence-based recidivism
reduction programming or productive
activities.” Id. § 3632(d)(4)(A).
8

9

22

Id. § 3631(b)(4); see also id. § 3631(b)(5)
(requiring the Attorney General to “make
any revisions or updates to the risk and
needs assessment system that the Attorney
General determines appropriate pursuant to
the review under paragraph (4)”).

23

Id. § 3632(d)(4)(A)(i).

10

Id. § 3632(d)(4)(A)(ii).

Id. §§ 3624(g)(1), 3632(a)(7). Section
3624(g) takes efect “beginning on the
date that the Attorney General completes
and releases the risk and needs assessment
system.” First Step Act, Pub. L. 115-391,
102(b)(2).

24

Id. § 3631(b)(3).

25

First Step Act, Pub. L. 115-391, § 107(e).

26

Id.

27

Id. § 107(f).

11

Id. § 107(g). Te Attorney General also
must provide a report to Congress, as
discussed in 18 U.S.C. § 3634.
28

18 U.S.C. § 3632(b); see also 18 U.S.C. §
3632(a)(3).

12

13

18 U.S.C. § 3632(f), (g).

Id. § 3621(h)(6).

-----~o~----22

29

Id. § 107(d).

30

Id. § 107(b).

31

Id. § 107(c).

Te Department selected the nonproft and
nonpartisan Hudson Institute as the host
organization on April 8, 2019, afer the 30
day deadline included in the FSA. Tis delay
was unavoidable for two related reasons.
First, under the Anti-Defciency Act, the
Department was unable to fund the IRC
without an appropriation or congressionallyapproved reprogramming of existing funds.
Second, one day afer the FSA was signed,
many of the appropriations that fund the
Department lapsed and the Department
was partially shut down until Friday,
January 25, 2019. Most of the staf working
on FSA-related projects were furloughed
until that date, and nearly all FSA-related
work, including identifying funds to be
reprogrammed, did not qualify as excepted
work.
32

Despite these challenges, the Department
diligently worked to ensure the creation of
the IRC. When the Government opened on
Monday, January 28, 2019, the Department
began the process of identifying organizations
that might serve as the Host Committee
and worked to identify sources of funding
for the IRC. Te Department submitted a
reprogramming request to the House and
Senate Appropriations Committees in early
March, 2019.

the Hudson Institute. At the end of that
process, Hudson was selected as the most
qualifed organization.
Te House and Senate Appropriations
Committees’ approval for the reprogramming
request was completed on April 2, 2019.
Within days of this approval, the Department
obligated those reprogrammed funds and
selected the Host Organization.
Department of Justice Announces First
Step Act Implementation Progress (April
8, 2019), https://www.justice.gov/opa/pr/
department-justice-announces-first-stepact-implementation- (last visited July 12,
2019).
33

Hudson Institute To Host First Step
Act’s Independent Review Committee
(April 8, 2019), https://www.hudson.
org/research/14945-hudson-institute-tohost-first-step-act-s-independent-reviewcommittee (last visited July 12, 2019).

34

35

See Note 33.

John P. Walters Biography Page, https://
www.hudson.org/experts/559-john-pwalters (last visited July 12, 2019).

36

John Wetzel Biography Page, https://
www.cor.pa.gov/Pages/Secretary%20of%20
Corrections.aspx (last visited July 12, 2019).

37

38

See also 18 U.S.C. § 3631(b)(1), (2).

39

18 U.S.C. § 3633(b).

While awaiting Congress’s determination,
the Department engaged with those
organizations deemed the most qualifed to
serve in this administrative role, including
23
---------,0.-------------

24

Chapter 2

Characteristics of an Effective Risk and Needs
Assessment System

I

and post-supervision agencies in the United
States utilize tools to assess ofender risk.1
Some of those agencies leverage tools to
measure ofender needs.2 Tese tools are
used to determine an ofender’s risk of
carrying out misconduct while in custody,
what criminogenic needs (i.e., needs that
research shows are associated with an
ofender’s risk of recidivism) an ofender
may have and which programs may address
those needs, and an ofender’s risk for
reofending once the ofender returns to
the community.3 Although there are many
diferent assessment tools in use, the most
accurate assessment tools share a number of
common characteristics, including dynamic
individualized assessment, periodic revalidation and update, racial and ethnic
neutrality, and a proper assessment of
criminogenic needs.

n order to develop a robust risk and needs
assessment system that is efective for the
federal prison population, it is critical
to frst examine and defne the general
characteristics of an efective risk and needs
assessment system. Tis chapter discusses
the fundamental aspects of an efective
risk and needs assessment system and its
common characteristics. It also discusses
the necessity of validating a risk and needs
assessment tool against the population for
which it is being used and the importance of
using scales to assess risk and needs. Finally,
this chapter summarizes the information
that experts in the feld shared with the
Department regarding the development of a
robust risk and needs assessment system.

I. Common Elements of Efective
Risk and Needs Assessment Tools
Risk and needs assessments are cornerstones
of correctional systems and ofender
supervision management. Many correctional

-----~o~----25

Common Elements of Efective Risk and Needs Assessment Tools
• Dynamic Individualized Assessment
• Periodic Re-Validation and Update
• Racial and Ethnic Neutrality
• Assessment of Criminogenic Needs

A. Dynamic Individualized
Assessment
Te federal prison population changes
over time, both demographically and
culturally. Likewise, individual inmates also
typically change over the course of their
incarceration.4 An efective risk and needs
assessment (RNA) system should be able to
recognize changes in each inmate and update
that inmate’s assessment score accordingly.
Te First Step Act (FSA or Act) requires an
RNA system to “determine the recidivism
risk of each prisoner as part of the intake
process, and classify each prisoner as having
minimum, low, medium, or high risk for
recidivism.”5 Te Act also requires periodic
reassessment of prisoners.6 And the FSA
requires the system to provide “information
on the best ways that the Bureau of Prisons
can tailor . . . programs to the specifc
criminogenic needs of each prisoner so as
to most efectively lower each prisoner’s risk
of recidivism.”7 Providing individualized
assessments ensures the measurement of
individual needs and the assignment of

programming are based on an inmate’s
personal characteristics.
Nearly all efective RNA tools incorporate
both static and dynamic factors.8 Static
factors are characteristics of inmates that are
historical and therefore unchangeable, such
as ofense severity, age at frst arrest, and
criminal history at prison entry. By contrast,
dynamic factors are variables that may
change over time and may refect more recent
inmate behavior, such as prison misconduct
or substance abuse. Te use of dynamic
factors in a risk and needs assessment tool
helps agencies monitor individual changes in
an inmate’s risk or needs, and is a requirement
of the FSA.9 Dynamic factors are also more
responsive to treatment options, because
static items by defnition do not change,
even in response to an efective treatment.
In many cases, however, dynamic items are
only incrementally predictive of criminal
behavior, as compared to the static items in
various RNA tools.10

26
---------,0.-------------

Additionally, as one of its core functions, the
risk tool must be able to diferentiate inmates
at a higher risk for recidivism from inmates
at a lower risk for recidivism. Based on
the principles of Risk-Needs-Responsivity,
the focus of treatment and programming
is prioritized for high risk ofenders.11 Tis
approach is key in areas where resources are
limited. Consistent with these principles, the
FSA requires the Bureau of Prisons (BOP)
to ensure that “priority for participation in
recidivism reduction programs shall be given
to medium-risk and high-risk prisoners,
with access to productive activities given to
minimum-risk and low-risk prisoners.”12

B. Periodic Re-Validation
and Update
A robust risk and needs assessment tool
should be re-validated (reviewed to confrm
the tool is still accurate) and periodically
updated to ensure it continues to measure risk
and needs accurately. Te tool should adapt
as the inmate population changes. Te FSA
incorporates this best practice by requiring
the Attorney General, “on an annual basis,
[to] review, validate, and release publicly on
the Department of Justice website the risk
and needs assessment system.”15 An RNA
tool should be re-validated by analyzing the
most recent ofender information available,
including both custodial and post-release
conduct. An RNA tool should contain
as much current information as possible,
while accepting that there will be a time
lag to collect recidivism data for the system
re-validation process.

For purposes of needs assessment and
programming, correctional staf should
conduct a needs assessment for each
inmate at or near the beginning of service
of the sentence to determine the inmate’s
programming requirements. Te specifc
needs identifed by this assessment should
include areas empirically associated with
ofender recidivism, such as substance
abuse and educational and vocational
needs.13 Te needs assessment also may
identify additional critical targets for
intervention, such as reducing symptoms of
mental illness, improving ofender quality of
life, and improving institution adjustment.
A needs assessment should also be updated
periodically, through reassessments, to refect
an ofender’s progress toward addressing
programming needs or to identify the
emergence of any additional needs during
incarceration.14

Te needs assessment tool should
be validated to ensure that the needs
assessments accurately measure an inmate’s
actual needs. In addition, the needs measures
used in the tool should be relatively reliable,
in that scorings by diferent assessors of the
measure would yield the same approximate
score. Consistent evaluations of a system’s
needs assessment will ensure its continued
reliability and validity.
As noted previously, the FSA requires the
Attorney General to “review, validate, and
release publicly” the RNA system annually.16
To fulfll that requirement, an efective RNA
tool may collect data on new test items that
receive no weight or have a minor impact
on the inmate’s risk assessment. A test

-----~o~----27

item is a suspected risk factor that is not
currently included in the RNA tool and the
correctional organization does not currently
collect information about; for example, “age
at frst arrest” is a common risk factor.17
Without information or data, researchers
cannot test whether it is an efective
recidivism predictor that could enhance the
RNA tool. By collecting the necessary data,
researchers could determine if the test item
should be included in a future version of the
risk assessment. When the risk assessment
is later re-validated, the test item may be
shown as an efective predictor of recidivism,
in which case it may be fully scored and used
in future versions of the RNA tool.
C. Racial and Ethnic Neutrality
An efective RNA system includes risk tools
that are racially unbiased, where “racially
unbiased” is defned as having the tool
correctly calibrated or standardized within
racial and ethnic groups. To be racially
unbiased or neutral, the tool should ensure
race and ethnicity have no efect on the
tool’s outcomes, specifcally the prediction of
whether an individual will recidivate, once
the total score is controlled.18 In concrete
terms, this means that if an RNA tool applies
the same risk score to two ofenders of
diferent races, the two ofenders must still
be proven to have the same recidivism rate.19
Tis methodology was primarily how the
Administrative Ofce of the U.S. Courts’ Post
Conviction Risk Assessment (PCRA) tool
was previously tested for racial bias.20 It was
also how one study of Correctional Ofender
Management Profling for Alternative
Sanctions (COMPAS) validated that tool as
racially unbiased.21

Alternately, an efective RNA tool can be
proven to be racially or ethnically neutral
based on some other acceptable criterion for
fairness and equality. Tere is an ongoing
academic debate regarding how to develop
such a tool. For example, a ProPublica article
criticized that same COMPAS RNA tool for
having more false positives and fewer false
negatives among black ofenders than among
white ofenders,22 even as COMPAS was
properly calibrated within groups.23 Tere
are multiple defnitions for racial fairness,
and they are mutually exclusive.24 When
defning racial fairness as calibrated groups,
COMPAS was racially unbiased, but to meet
the standard of calibrated groups, it had to
be racially biased by the defnition of equal
false-positive rates. Unless the racial groups
have the exact same recidivism base rates, an
RNA tool cannot achieve racial fairness by
more than one defnition of racial fairness at
the same time.25
D. Assessment of Criminogenic Needs
Although RNA tools should assess
criminogenic needs, an efective RNA
system distinguishes needs from risks.26
Whereas risk refers to an inmate’s statistical
propensity for recidivism or some other
adverse outcome, needs represents the areas
of intervention that must be addressed
to mitigate the risk of recidivism.27 On
theoretical grounds, criminogenic needs are
directly or indirectly linked to an adverse
outcome, such as recidivism. However,
they may not be predictive of that outcome
and should not be included in the risk score
unless they are predictive.28

28
---------,0.-------------

Although research varies about how each
of these needs is linked to recidivism risk,
criminal thinking, antisocial peers, substance
abuse, and education and vocational
needs typically are identifed as important
criminogenic needs.29 “Antisocial peers”
means the extent to which an ofender’s
social network is comprised of other criminal
ofenders. Criminal thinking encompasses
the “attitudes, values, and/or behaviors of
ofenders that infuence their involvement in
criminal behavior.”30

be identifed and placed in the appropriate
programs, thus maximizing taxpayer value
and the total number of inmates who can
participate in appropriate services.
Finally, an efective RNA system is practical
about the needs assessment. It measures
needs that correspond to the evidence based
recidivism reducing programs and treatments
that are available in the correctional system,
currently or in the near future. Efective RNA
tools do not collect data on measures that
have no operational value.

An efective RNA system includes tools
to assess criminogenic needs for two key
reasons. First, to the extent that needs
assessments are conducted over time, inmate
progress can be dynamically monitored
consistent with that inmate’s reentry goals.
Certain needs areas, such as substance abuse
patterns, can be assessed repeatedly. Tat
repeated assessment allows correctional
staf to determine whether an inmate’s risk
is adequately mitigated in that area or if
the inmate would beneft from additional
treatment. Other needs areas are more
difcult to assess dynamically, but measures
do exist and may be deployed with some
degree of success.31

Efective RNA tools must be validated on
the population for which they will be used.
“Te intent of actuarial risk assessment is
to identify subgroups within an ofender
population who have signifcantly diferent
rates of recidivism.”32 One example of
subgroups is male and female. As diferent
correctional systems will have diferent
subgroups within them, an efective RNA
tool must be tailored to best identify and
diferentiate the subgroups that exist in each
correctional system.

Second, a needs assessment can assist in
programming decisions, as correctional
departments match inmates to programs
that address their specifc criminogenic
needs. For example, if an inmate has no
assessed need for substance abuse treatment,
then such treatment is likely to be of no
beneft to the inmate and likely to represent a
poor use of resources. Te treatment will be
more cost efective if high-need inmates can

When it comes to risk assessments, there is
no “one size fts all” tool. Tis is particularly
important since the BOP inmate populations
difer from the ofender populations in state
and local correctional systems, in numerous
ways. Based on data collected by the Bureau
of Justice Statistics (BJS),33 the following
table displays the diferences in demographic
and population characteristics between
the prison populations of federal and state
correctional jurisdictions.

II. Validating a Tool Against
the Appropriate Population

-----~o~----29

Comparison of Federal and State Inmates in 2017,
based on BJS National Prisoner Statistics Data
Category

Descriptor

Federal Inmate State Inmate

Gender

Female

6.8%

7.6%

Race/Ethnicity* White

27.6%

31.1%

Black

37.1%

32.6%

Hispanic

31.9%

21.6%

Other

3.4%

14.7%

Citizenship

Non-US Citizen

19.7%

5.8%

Offense Type

Drug Offense

47.3%

14.8%

Weapons

17.0%

4.2%

Immigration

6.7%

0.0%

Other Public Order (non-weapon,
non-immigration)

14.9%

7.7%

Homicide

1.6%

14.2%

Robbery

3.8%

13.1%

Other Violent Offense

2.4%

27.9%

Burglary

0.2%

9.4%

Other Property Offense

5.7%

8.1%

* “Hispanic” includes all races with Hispanic ethnicity; all other categories include non-Hispanics
of that race.
Te BJS table also indicates that most federal
inmates are incarcerated for drug trafcking,
weapons charges, and other (meaning nonweapon and non-immigration) public order
ofenses. Te majority of state inmates are in
prison for violent ofenses, such as murder
and robbery.34 State correctional populations
also include more inmates with convictions
for property ofenses (mainly burglary),
whereas the relatively few property ofenders

in the federal system are usually convicted
of fraud. In terms of race and ethnicity,
federal inmates are more ofen Hispanic or
non-US citizens than state inmates, but they
are slightly more likely to be black. Finally,
federal inmates are slightly less likely to be
female than state inmates.
Notably, these BJS statistics were drawn from
the sentenced federal prison population. As

30
---------,0.-------------

compared to the population of incoming or
outgoing inmates, the general population
will skew toward more severe crimes and
longer sentences. For example, 14% of state
inmates are in prison for murder, but far less
than 14% of state crimes are murder. Rather,
murderers tend to serve signifcant sentences
and take up a disproportionate share of the
prison beds. State ofenders convicted of
minor crimes may be numerous, but because
they move in and out of prison more quickly,
relatively few are using prison beds at any
given time.

the BOP imported a state-validated RNA tool
into the federal prison system and made no
adjustments, the imported tool would likely
overestimate the inmate’s history of violence
as an important risk predictor. Instead, a
good risk and needs assessment tool should
be validated for the population for which it
will operate.

III. Te Importance of Using Scales
to Assess Risk and Needs
Risk and needs assessment tools ofen include
scales to assess ofender risk and needs.
Scales are standardized measures that permit
an objective and reliable way of interpreting
a response. For example, afer the question
“how satisfed are you with our service” you
might fnd the following scale of responses:
1 – very unsatisfed; 2 – unsatisfed;
3 – neutral; 4 – somewhat satisfed; or
5 – very satisfed. For an inmate’s risk
and needs assessment, specifcally, data
are collected regarding the inmate, scored,
and compared with normative data to yield
a determination of specifc needs areas
requiring intervention.

Due to these diferences, state inmates may
have more variety in their criminal histories
than federal inmates, which would afect
the predictive nature of a risk tool designed
for a state prison population. A risk factor
can better predict recidivism not only if it is
more strongly related to recidivism, but also
if there is more variety in the risk factor’s
scores and more individuals with extreme
scores. State and federal prisons both have
many ofenders with extremely low violent
histories, but if state prisons have more
inmates with very violent histories, then
violent history may be a stronger predictor
of recidivism in state prisons. Accordingly, if

Example of Standardized Measures in a Scale
Question: How satisfied are you with our service?
1 – Very Unsatisfied

2 – Unsatisfied

3 – Neutral

4 – Somewhat Satisfied

5 – Very Satisfied

3 – Not Sure

4 – Agree

5 – Strongly Agree

3 – Not Sure

4 – Agree

5 – Strongly Agree

Statement: Sometimes I lose my temper.
1 – Strongly Disagree

2 – Disagree

Statement: I have friends that I trust.
1 – Strongly Disagree

2 – Disagree

-----~o~----31

As an example of a needs assessment
tool, the Tests of Adult Basic Education
(TABE) is a measure currently used in the
BOP and other adult education settings to
assess the needs of adult learners.35 Based
on an inmate’s results on the TABE, that
inmate may be referred to literacy or other
educational programs.
In addition to
standardized measures like the TABE, needs
assessments may also include empiricallybased questionnaires. For example, the
Adverse Childhood Experiences Survey has
been shown to predict suicide risk and other
critical outcomes, and is therefore a strong
predictor of the need for mental health
programming.36
Te use of scales in assessment tools helps
ensure that data are collected and used in
a more quantifable, consistent, reliable,
and accountable manner. Actuarial risk
assessments are generally more efective
than subjective professional judgment at
predicting recidivism and other adverse
outcomes.37 Additionally, it is more efcient
to train staf to use a scale than it is to teach
the years of experience required to leverage
subjective professional judgment. It is
also more efcient to validate, amend, and
improve the scales over time. By contrast,
professional judgment is harder to test
systematically and harder to improve in an
objective manner. Tus, scales are key to an
efective RNA system.

IV. Guidance from the Experts
Te Department has consulted with experts
from state and federal criminal justice
systems to inform its implementation eforts.

Specifcally, experts from the Administrative
Ofce of the United States Courts,
Pennsylvania Department of Corrections,
Minnesota Department of Corrections, and
the BOP provided informative briefngs on
the
research,
development,
and
implementation of risk and needs assessment
systems. Tese briefngs provided critical
information and evidence to assist in
developing an efective risk and needs
assessment system.
A. Administrative Ofce of the
United States Courts
Te Administrative Ofce of the United
States Courts, Probation and Pretrial
Services Ofce (Administrative Ofce),
briefed the Department on its pre-trial
risk assessment and post-conviction
risk assessment tools. Tis included a
discussion of the federal Post Conviction
Risk Assessment (PCRA), which is used to
help probation ofcers manage defendants
while on supervision. Te Administrative
Ofce explained the prolonged process that
it employed to evaluate existing commercial
risk assessment tools and ultimately to decide
to create its own risk assessment tool.38 For
example, the Administrative Ofce met with
the developers of major existing risk and
needs assessment tools, assembled a panel
of experts, and began a pilot project in fve
districts. Afer completing that process, the
Administrative Ofce decided “that creating
an instrument with data specifc to the
federal probation system was preferable.”39
Te Administrative Ofce also discussed the
need for high inter-rater reliability—that is,
the extent to which diferent raters (in this

32
---------,0.-------------

case, corrections and probation ofcials)
yield consistent results. Te Administrative
Ofce also noted that training staf to enter
data accurately and consistently was a key
requirement for an efective system.

bias. While a 2016 study found “that black
ofenders obtained higher average PCRA
scores than white ofenders, most of the
racial diferences in the PCRA scores (about
69%) were attributable to the criminal
history domain[.]”43 Te remainder of the
diference was attributable to employment
and education, as well as social networks and
attitudes.44 Additionally, PCRA “strongly
predicted rearrests for both black and white
ofenders across the instrument’s risk levels.
Stated diferently, a given PCRA score had
essentially the same meaning—i.e., the same
probability of recidivism—across the two
race groups[.]”45

Te Administrative Ofce also provided
guidance on best practices that contributed
to positive changes in an individual
probationer’s behavior: “1) intensive
correctional interventions should be
directed to higher risk rather than lower
risk persons under supervision (risk
principle), 2) dynamic risk factors should
be targeted for change (need principle), and
3) strategies such as cognitive behavioral
treatment should be delivered in a way that is
specifcally responsive to the characteristics
of the individual (responsivity principle).”40

B. Pennsylvania Department
of Corrections
Te Pennsylvania Department of Corrections
(PA DOC) briefed the Department of Justice
on its risk and needs system and the research
underlying that system. PA DOC explained
the process used to select its suite of risk
assessment tools and the situations in which
those tools are deployed.

Te Administrative Ofce shared research
regarding the predictive efect of PCRA.41In
one study, “changes in ofender risk were
associated with changes in re-arrest rates.
Specifcally, high-, moderate-, and low/
moderate-risk ofenders with decreases in
either their risk classifcations or overall
PCRA scores had lower recidivism rates
compared to their counterparts whose risk
levels or scores either remained unchanged
or increased, while increases in ofender
risk were associated with higher rates of
Te Administrative Ofce
rearrests.”42
emphasized that PCRA was designed to
assess behaviors for federal probationers
and was not intended to be transferable to
other jurisdictions or ofender populations
without appropriate modifcation.

PA DOC emphasized the need to connect
the risk assessment tool to the population
for which it is being used–for example,
by developing the tool based on that
population. PA DOC promoted the idea of
using computers to automate the process of
calculating risk assessment scores to improve
scoring consistency and to allow staf to
spend more time engaged in other benefcial
activities. PA DOC also emphasized the
need to train staf on how to administer risk
assessment tools properly.

Research studies also have evaluated
PCRA for forms of bias, including racial

-----~o~----33

Like other organizations that provided a
brief, PA DOC used the area under the curve
(AUC) to test the predictive validity of various
tools. AUC represents the probability that

a randomly selected recidivist would have
received a higher risk rating than a randomly
selected non-recidivist. Te closer the AUC
value is to 1, the more predictive it is.

Area Under the Curve (AUC) Scores

II

0.10

I

0.20

0.30

AUC=0.00
(Perfectly wrong
answers every time)

I

0.40

II
0.60

AUC= 0.50
(Random guessing, like
a 50-50 coin toss)

C. Minnesota Department
of Corrections
Te Minnesota Department of Corrections
(MNDOC) briefed the Department of Justice
on its risk and needs system, Minnesota
Screening Tool Assessing Recidivism Risk
(MnSTARR assessment system) and the
research that Minnesota found to be most
relevant. MNDOC summarized the history
of risk assessment tools and how current
tools assess a defendant’s risk and treatment
needs. MNDOC also summarized the work
of Daniel Kahneman and Amos Tversky,46
who researched how cognitive biases afect
decision-making. An example of such
biases includes anchoring (giving too much
weight to unimportant variables). Efective
risk assessment tools minimize or address
cognitive biases by relying on empirically
validated factors.

0.70

I I

0.80

0.90

I

AUC= 1.00
(Perfect Accuracy)

MNDOC also noted how current risk
assessment tools include both static and
dynamic risk factors. Examples of dynamic
risk factors include education, substance
abuse, and criminal thinking. Current tools
also more completely integrate protective
factors (factors that reduce the risk of
recidivism).
MNDOC discussed the benefts of using
automated risk assessment tools, which
leverage technology to allow staf simply to
answer questions and input information.
Such tools are more efcient, requiring less
staf time manually calculating scores.
MNDOC emphasized the need for
risk assessments to be designed for the
population for which they are being used.
For example, a facility should not assume
that a risk assessment based on one prison

----------~Q
34 r - - - - - - - - - - - -

BOP noted its risk assessment tool has
included dynamic factors for several
decades. Tose dynamic factors include
history of escape or attempts to escape;
history of violence in prison; age; drug use
while incarcerated; education level; program
participation; contact with family and
community; and criminal behavior while in
prison.

population will be an accurate predictor of
recidivism for another prison population.
Tis is because risk factors may vary among
prison populations (such as males and
females). MNDOC (like PA DOC) stated
that it is essential that an assessment tool be
used on the same population that was used
to develop the tool.
D. Federal Bureau of Prisons

Additionally, BOP demonstrated how a
dynamic risk needs tool called BRAVO-R,
customized to predict recidivism, was
superior to BRAVO at accurately predicting
recidivism for both male (AUC of 0.79) and
female (AUC of 0.77) ofenders and superior
to other relevant instruments commonly
used in predicting recidivism for both males
and females.47 As shown in the following
Figure, BOP compared the AUC statistic for
BRAVO-R to the AUCs for other well-known
tools (including PCRA) and found higher
AUCs for BRAVO-R than those reported for
the other state and federal risk instruments.

In addition to the information received from
other experts, researchers from the Bureau of
Prisons provided a briefng on its recidivism
risk prediction instrument, BRAVO
(Bureau Risk Assessment Verifcation and
Observation). BRAVO is specifcally used to
predict serious misconduct in prison.
Consistent with the Administrative Ofce,
PA DOC, and MNDOC, BOP discussed the
AUC and the implications of the AUC values
for its instruments. Using AUC, BOP found
that BRAVO is a more dynamic predictor of
recidivism for male ofenders, with an AUC
of 0.77.

-----~o~----35

AUCs for Commonly-Used Recidivism Predictors and the BRAVO-R Risk Assessment Tool

1
0.95
0.9
0.85

0.79

0.8
0.73

0.75
0.7

0.67

0.74
0.66

0.64

0.65

0.77

0.73

0.6
0.55
0.5

COMPAS

LSI-R

SPIN-W

STRONG

WRN-R

PCRA

BRAVO-R
(Male)

BRAVO-R
(Female)

Commonly Used Recidivism Predictor Tools
Acronym
COMPAS
LSI-R
SPIN-W
STRONG
WRN-R
PCRA

Full Name
Correctional Offender Management Profile
for Alternative Sanctions
Level of Service Inventory – Revised

I
I

Service Planning Instrument – Women
Static Risk and Offender Needs Guide
Wisconsin Risk and Needs – Revised
Federal Post Conviction Risk Assessment

r

Examples of Locations Where
the Tool is Being Used
FL, MI, NM, NY, and WI
CA, CO, DE, HI, IA, IL, NE,
PA, and WV
CT
WA
WI
Administrative Office of the
U.S. Courts for Federal
Probation

The six recidivism predictor tools shown on the above graph were selected for comparison based on their
level of use and the existence of a published AUC, which facilitates direct comparison.

E. Common Principles

BOP also emphasized that a risk assessment
tool must be validated on each population
for which it is used. For example, federal
ofenders are diferent from state ofenders in
material ways and those diferences should
be refected in the RNA.

Several common principles arose from the
expert briefngs. For example, multiple
briefers emphasized that risk assessment
tools should be developed using the

--------,0.-------36

populations for which they ultimately will be
used. Likewise, risk assessment tools should
be statistically validated on the populations
for which they will be used.

makers should continually improve risk
assessment tools based on ongoing research
regarding the predictive value of certain
risk factors. Te AUC is widely used to test
the predictive validity of risk assessment
tools. Tese principles were instructive in
developing an efective and robust risk and
needs assessment system.

Te experts also agreed that risk assessment
tools should include a mixture of static
and dynamic risk factors. Finally, decision-

Common Principles from Federal and State Authority Briefngs
• Develop Tools Using Appropriate Inmate Population
• Validate Tools on Appropriate Inmate Population
• Deploy Static and Dynamic Factors
• Continually Improve Tools Trough Ongoing Research

-----~o~----37

NOTES
Handbook of Recidivism Risk/Needs
Assessment Tools 3-6 (Jay P. Singh et al.
eds., 2018).
1

2

Id. at 7-8.

Tomas H. Cohen, et al., Te Federal PostConviction Risk Assessment Instrument: A
Tool for Predicting Recidivism for Ofenders
on Federal Supervision, in Handbook of
Recidivism Risk/Needs Assessment
Tools 77-100 (Jay P. Singh et al. eds., 2018).
3

Craig Haney, Te Psychological Impact of
Incarceration: Implications for Post-Prison
Adjustment, From Prison to Home: the
Effects of Incarceration and Reentry
on Children, Family and Communities
Project (2001).
4

5

18 U.S.C. § 3632(a)(1).

6

18 U.S.C. § 3632(a)(4).

7

18 U.S.C. § 3632(b)(2).

8

Supra note 1.

9

18 U.S.C. § 3631(b)(4)(C).

Faye S. Taxman, Risk assessment: Where
do we go from here?, in Handbook of
Recidivism Risk/Needs Assessment
Tools 271-284 (Jay P. Singh et al. eds., 2018).
10

Risk-need-responsivity model for ofender
assessment and rehabilitation 2007-06,
Public Safety Canada, https://www.
publicsafety.gc.ca/cnt/rsrcs/pblctns/rsk-ndrspnsvty/index-en.aspx (last modifed Jan.
31, 2018).

11

12

18 U.S.C. § 3621(h)(6).

James Bonta & Douglas A. Andrews, The
Psychology of Criminal Conduct 180181 (6th ed. 2017).
13

14

See 18 U.S.C. § 3632(d)(5).

15

18 U.S.C. § 3631(b)(4).

16

Id.

Scott D. Camp, Federal Bureau of
Prisons, A Risk Assessment Tool for
Recidivism by Federal Inmates (2018).

17

American Educational Research Association
et al., Standards for Educational and
Psychological Testing (1985).

18

While the FSA does not explicitly require
the Attorney General to create a race-neutral
RNA system, 18 U.S.C. § 3631(b)(4)(E)
requires the Attorney General to periodically
conduct “an evaluation of the rates of
recidivism among similarly classifed prisoners
to identify any unwarranted disparities,
including disparities among similarly classifed
prisoners of diferent demographic groups, in
such rates.”
19

Jennifer L. Skeem & Christopher T.
Lowenkamp, Risk, Race, and Recidivism:
Predictive Bias and Disparate Impact. 54
Criminology 680-712 (2016).
20

Anthony W. Flores et al., False Positives,
False Negatives, and False Analyses: A
Rejoinder to Machine Bias: Tere’s sofware
used across the country to predict future
criminals. And it’s biased against blacks. 80
Federal Probation 38 (2016).

21

Julia Angwin et al., Machine bias: Tere’s
sofware used across the country to predict
22

38
---------,0.-------------

Tests of Adult Basic Education, Data
Recognition Corporation (2019), https://
tabetest.com.

future criminals. And it’s biased against
Blacks. ProPublica (2016).
23

35

Supra note 21.

Tifany M. Jones et al., Modeling life course
pathways from adverse childhood experiences
to adult mental health. 80 Child Abuse &
Neglect 32–40 (2018).

36

Richard Berk et al., Fairness in Criminal
Justice Risk Assessments: Te State of the Art,
Sociological Methods & Research 1
(2018). Found at https://journals.sagepub.
com/doi/full/10.1177/0049124118782533.

24

25

John A. Swets et al., Better Decisions through
Science. 283(4) Scientific American 82-87
(2000).
37

Id. at 17-24.

Christopher Baird, A Question of Evidence:
A Critique of Risk Assessment Models Used in
the Justice System, Nat’l Council on Crime
and Delinquency (2009).
26

See An Overview of the Federal Post
Conviction Risk Assessment, Administrative
Ofce of the United States Courts Probation
and Pretrial Services Ofce (June 2018),
https://www.uscourts.gov/sites/default/fles/
overview_of_the_post_conviction_risk_
assessment_0.pdf (last visited May 13, 2019).
38

Tony Ward & Shadd Maruna,
Rehabilitation: Beyond the Risk
Paradigm, 48-49 (2007).
27

28

Supra note 26.

39

Id. at 9.

29

Supra note 13.

40

Id. at 17.

See Federal Post Conviction Risk Assessment
Risk Compendium, Administrative Ofce
of the United States Courts Probation and
Pretrial Services Ofce (March 2018).
https://www.uscourts.gov/sites/default/fles/
federal_post_conviction_risk_assessment_
research_compendium_0.pdf (last visited
May 13, 2019).

41

Faye S. Taxman et al., Construct and
Predictive Validity of Criminal Tinking
Scales. 38(2) Criminal Justice and
Behavior 174–87 (2011).

30

Alese Wooditch et al., Which Criminogenic
Need Changes Are Most Important in
Promoting Desistance From Crime and
Substance Use? 41(3) Criminal Justice and
Behavior, 276-99 (2014).

31

32

Supra 26 at page 3.

Supra note 3.

43

Id.

Jennifer L. Skeem and Christopher T.
Lowenkamp, Risk, Race, and Recidivism:
Predictive Bias and Disparate Impact
(2016), https://onlinelibrary.wiley.com/doi/
full/10.1111/1745-9125.12123. (last visited
May 24, 2019).
44

U.S. Dep’t of Justice, Bureau of
Justice Statistics, Prisoners in 2017
1 – 44 (2019), https://www.bjs.gov/index.
cfm?ty=pbdetail&iid=6546.

33

34

42

Id. at 22.

-----~o~----39

45

Supra note 41 at 3.

See, e.g., Daniel Kahneman and Amos
Tversky, Thinking, Fast and Slow (2011).

46

47

Supra note 1.

40
---------,0.-------------

41

Chapter 3:
The First Step Act of 2018
Risk and Needs Assessment System
Introduction

instrument the BOP currently uses to predict
misconduct for custody-level classifcation
purposes.

I

n late March 2019, the National Institute
of Justice (NIJ) contracted with Dr.
Zachary Hamilton and Dr. Grant Duwe1
to serve as consultants for the development
and validation of a risk and needs assessment
system to be implemented by the Federal
Bureau of Prisons (BOP) pursuant to the
First Step Act (FSA or the Act). Hamilton
and Duwe began their work by attending a
series of listening sessions organized by NIJ.
Moreover, they participated in numerous
conference calls with staf from the BOP, NIJ,
and the Independent Review Committee in
which they discussed data availability and
background information on BOP policies
and practices.

In May, the NIJ experts and BOP staf held
a series of conference calls to exchange
contextual information regarding quantitative
data available for the development of risk
and needs tools. Te NIJ experts and BOP
ORE and Reentry Services Division staf
attended the weekly calls. Additional
conference calls were also completed with
BOP reentry staf to assess the availability
of needs assessment data. On May 21, 2019,
Hamilton visited the BOP’s Designation
and Sentence Computation Center (DSCC)
in Grand Prairie, Texas. At the DSCC,
Hamilton met with custody designation staf
responsible for gathering many of the data
collected about inmates admitted to BOP
facilities. Hamilton observed the assessment
process, learned how the BOP uses its
classifcation assessment, and identifed
other routinely collected data that could be
used to develop and/or improve the risk and
needs assessment system.

Hamilton and Duwe began reviewing
research material that BOP Ofce of
Research and Evaluation (ORE) staf had
prepared on Bureau Risk and Verifcation
Observation (BRAVO), the BOP’s current
classifcation system,2 and the BRAVORecidivism (BRAVO-R) tool. BRAVO-R is
a risk assessment instrument developed by
ORE and designed to predict recidivism—a
new arrest or return to federal prison within
three years afer release from prison—for the
BOP inmate population. BRAVO-R, which
contains both static and dynamic items, is a
modifed version of the BRAVO assessment

In early May, Hamilton and Duwe received
access to de-identifed data, including threeyear rearrest data, on BOP inmates. Te
dataset contained 278,940 BOP inmates
released from BOP facilities between 2009
and 2015. Te dataset included only those

-----~o~----42

inmates released to the community. It
excluded released inmates who died during
the three-year follow-up period and inmates
scheduled for deportation. In addition to
three-year rearrest data, the dataset contained
a large number of variables relevant for the
assessment of risk and needs. In particular,
the dataset contained measures commonly
associated with recidivism risk, such as
criminal history, demographic characteristics
such as age, prison misconduct, and
participation in programming. Te dataset
also included measures used in both BRAVO
and the recently-developed BRAVO-R.
As described below in more detail, Hamilton
and Duwe used the BOP datasets to develop
and validate a new tool, the Prisoner
Assessment Tool Targeting Estimated Risk
and Needs (PATTERN).3 Te PATTERN
instrument is an assessment designed
to predict the likelihood of general and
violent recidivism for all BOP inmates over
a three-year follow-up period. As required
by the FSA,4 the PATTERN assessment
instrument contains static risk factors as well
as dynamic items that are associated with
either an increase or a reduction in risk. Like
BRAVO-R, PATTERN is a gender-specifc
assessment providing predictive models, or
scales, developed and validated for males
and females separately. Tese eforts make
the tool more gender responsive, as prior
fndings have indicated the importance of
gender-specifc modeling.5
Operating under an extremely short timeline,
the development of PATTERN is a signifcant
advancement in the U.S. Department of
Justice’s (the Department’s) implementation

of the FSA for the following reasons:
• Te PATTERN assessment tool achieves
a high level of predictive performance
and surpasses what is commonly found
for risk assessment tools for correctional
populations in the United States;
• Te PATTERN instrument makes great
use of dynamic factors, enabling inmates
to work to make behavioral changes while
in prison;
• PATTERN’s predictive performance
reduces bias and aims to improve parity
across race and ethnic classifcations; and
• Te PATTERN instrument is a single
assessment, incorporating a design that
seeks to encourage risk reduction behavior
over time by inmates.
Te following section briefy reviews the
BRAVO-R assessment tool from which
PATTERN builds. Te next section then
discusses the methods used to develop and
validate PATTERN. And the following section
details the enhanced predictive performance
results for PATTERN. Finally, the chapter
discusses the needs assessment component
of the Risk and Needs Assessment System.

I. BRAVO-R
Te BRAVO-R instrument is, as noted
above, a modifed version of the existing
classifcation assessment the BOP created
in the 1970s and has updated periodically
as part of regular reviews. Te current

43
---------,0.-------------

classifcation assessment, the BRAVO
instrument, enables staf to perform an initial
assessment at the DSCC once they have
received the sentencing documentation from
the U.S. Probation Ofce.6 A reclassifcation
assessment is completed afer an inmate has
been in prison for seven months and every
12 months thereafer. Te initial assessment
contains items that measure whether the
inmate has a detainer (a separate pending
sentence of proceeding), ofense severity,
criminal history, history of escape/attempts,
history of violence, voluntary surrender/precommitment,7 age, drug use, and education
level. Te reclassifcation assessment
contains these items along with the following
additional measures: percentage of time
served, program participation, living skills,
prison misconduct type and frequency,
and family/community ties. Terefore,
the BRAVO instrument contains not only
static factors, but also dynamic items that
measure whether risk has increased or
decreased while an inmate is confned. Te
same version of both assessments—initial
and reclassifcation—are given to males and
females, however the groups are measured
using diferent cut points to account for
gender diferences in risk.

items than BRAVO. More specifcally, the
detainer, ofense severity, and percentage of
time served items were removed because the
BOP found them to be inconsistent measures
in predictive accuracy.
Analyses performed by BOP ORE show that
BRAVO performs very well in predicting
misconduct for the BOP population.
Independent analyses conducted by the NIJ
consultants reveal that BRAVO-R performs
similarly well in predicting recidivism.

II. Methodology for Developing
PATTERN
In an efort to expand, enhance, and
provide an assessment that meets the FSA’s
requirements, the NIJ experts were tasked
with the development and creation of a risk
and needs assessment tool. Trough calls
and site visits, every attempt was made to
identify all available data elements that could
be used in the development of the risk and
needs tools. Using available, quantifable
data, the NIJ experts, in consultation with
the Department and considering the advice
of the Independent Review Committee,
developed PATTERN, the risk and needs
assessment tool that will be deployed in
BOP facilities. PATTERN builds from
the BRAVO-R tool, which was already
quite good at predicting recidivism. Te
analyses used to develop PATTERN refect
recent advancements in risk assessment
tool construction adding both gender and
outcome (i.e., general and violent recidivism)
specifcity. Recognizing the FSA’s emphasis
on using dynamic factors, additional items

Taking advantage of the overlap between
misconduct and recidivism, the BOP
recalibrated its classifcation tool to predict
recidivism once an inmate is released into
the community.8 Similar to the BRAVO
instrument, the BRAVO-R instrument
contains
initial
and
reclassifcation
assessments, which are the same for males
and females with diferent cut points.
BRAVO-R, however, contains three fewer

-----~o~----44

relating to an inmate’s program participation
and performance were added to the possible
predictors. PATTERN’s prediction models
were then developed and assessed for
predictive validity and bias. Te details of the
analysis plan are provided below.
A. Data and Sample
As indicated, several factors were selected
for testing from BRAVO/BRAVO-R, while
others were created based on BOP-provided
indicators. BRAVOBRAVO-R indicators
selected for the PATTERN models included
criminal history score, history of violence,
history of escapes, voluntary surrender,
and education score.9 Additional indicators,
including several dynamic factors, were
also developed using routinely collected
BOP data elements. Te new or redesigned
dynamic predictors include:10
• Infraction convictions (any), a count
measure11 identifying the number of
infractions resulting in a guilty fnding
that the inmate has incurred during the
current incarceration.12
• Infraction convictions (serious and
violent), a count measure identifying the
number of serious infractions (identifed
as being at the 100 or 200 level)13 resulting
in a conviction that the inmate has
incurred during the current incarceration.
• An overall measure of the number
of benefcial programs completed by
the inmate (for example, education,
parenting, drug treatment, and technical
and vocational programs), converted into
ordinal categories.

• Te number of technical and vocational
courses created as a count metric.
• Federal industry employment, also known
as UNICOR, was indicated “yes” if the
individual worked in UNICOR during his
or her current incarceration.
• Drug treatment was assessed as a
categorical
measure,
identifying
programming need based on the BRAVO
drug/alcohol abuse indicator and further
assessing residential, non-residential,
or no treatment completed during the
current incarceration.
• Drug education completed during the
current incarceration.
• And an ofender’s willingness to use
income earned during incarceration
for payment toward victim restitution
and dependents was indicated with the
created “non-compliance with fscal
responsibility” measure.14
Te additional or redesigned static factors
include:
• Age at frst conviction,15 an ordinal
measure identifying the number of years
between the individual’s date of birth and
his or her frst recorded conviction by
BOP assessment staf.
• Age at the time of assessment, calculated
as the diference between the ofender’s
current age when the BRAVO assessment
is conducted and an individual’s date of
birth.

45
---------,0.-------------

• If the crime of conviction that resulted
in the current BOP incarceration was
violent,16 a “yes” response was identifed
in the instant ofense measure.

location in the United States and had received
a BRAVO assessment. Tis provided an
eligible sample size of 222,970 individuals.
Consistent with prior risk assessment
development, continuous measures were
converted to ordinal, categorical indicators.
Category ranges were constructed based on
the prevalence of responses and the impact
on recidivism. Response prevalence for the
development sample is provided in Table 1.
While a large pool of items was considered,
to provide a concise review of model
descriptive statistics, only items selected via
model prediction algorithms are displayed.18

• Finally, a “yes” response was indicated
for an individual identifed as a “sex
ofender” based on that individual’s initial
assessment under the Walsh criteria.17
To evaluate the PATTERN development
sample, the NIJ experts computed basic
statistics describing the sample. Eligible
members of the sample included those
who were released from a BOP facility to a

Table 1. Sample Descriptive Statistics (N = 222,970)
MEASURE
Age at frst arrest

PERCENT OF SAMPLE
>= 35
< 35, >= 25
< 25, >= 18
< 18

14
15
52
18

> 60
< 50, <= 60
> 40, <= 50
> 29, <= 40
> 25, <= 29
>= 18, <= 25

4
11
22
38
14
11

0
> 0, <= 1
> 1, <= 2
>2

63
17
8
12

Age at time of assessment

Infraction convictions (any)

-----~o~----46

MEASURE
Infraction convictions (serious and violent)

PERCENT OF SAMPLE
0
> 0, <= 1
> 1, <= 2
>2

82
12
3
3

0
1
> 1, <= 3
> 3, <= 10
> 10

49
15
16
15
5

>1
> 0, <= 1
0

8
10
82

Yes
No

8
92

No need indicated
Completed residential drug treatment during incarceration
Completed drug treatment during incarceration
Need indicated but no treatment during incarceration
Drug education while incarcerated
No
Yes
Non-Compliance with fnancial responsibility
No
Yes
Instant ofense violent
No
Yes

22
8
13
57

Number of programs completed (any)

Number of technical or vocational courses

Federal industry employment (UNICOR)

Drug treatment while incarcerated

60
40
97
3
71
29

47
---------,0.-------------

MEASURE
Sex ofender (Walsh)

PERCENT OF SAMPLE
No
Yes

90
10

0-1 points
2-3 points
4-6 points
7-9 points
10-12 points
13+ points

33
14
17
13
9
14

None
>10 years minor
>15 years serious
5-10 years minor
10-15 years serious
<5 years minor
5-10 years serious
<5 years serious

58
4
6
5
5
6
8
6

None
>10 years minor
5-10 years minor
<5 years minor or any serious
BRAVO-R Initial: Voluntary Surrender
Yes
No
BRAVO-R Initial: Education Score
HS degree or GED - Verifed
Enrolled and progressing in GED
No verifed degree and not participating in GED program
Sample
Training
Test

74
4
3
5

BRAVO-R Initial: Criminal History Score19

BRAVO-R Initial: History of Violence

BRAVO-R Initial: History of Escapes

25
75
67
15
18
67
33

-----~o~----48

MEASURE
Gender

PERCENT OF SAMPLE
Male
Female

85
15

General
Violent

47
15

Recidivism

B. Assessment Construction and
Validation
Te NIJ consultants used a dataset of all
BOP releases from 2009 to 2015 to develop
and validate PATTERN. Te vast majority
of items within the dataset were drawn
from SENTRY, the BOP’s centralized inmate
management system. Te BOP also provided
data on recidivism, which was measured as a
new arrest or return to BOP custody within
three years of release.
Te NIJ experts used both split-sample and
K-fold procedures to develop and validate the
predictive models. Te classic split-sample
procedure is a simpler approach in which
data are partitioned into two equal halves,
and one subset is used for training while the
other is used to test how well the tool predicts
recidivism. Tis method is limited in that it is
unable to use all available data for each step,
potentially resulting in prediction instability.
Split-sample procedures are also limited in
that the training set is smaller than what is
otherwise achieved via the K-fold method,
resulting in a less accurate fnal model.
Despite these limitations, the split-sample
procedure also makes it possible to carry out

an objective, “apples-to-apples” comparison
with BRAVO-R on a test (or validation) set.
To create the PATTERN instrument, the
NIJ experts placed all BOP inmates released
between 2009 and 2013 in the training set
(i.e., the data used to develop the predictive
models). As shown in Table 1 above, roughly
two-thirds (66 percent) of the development
sample were included in the training set, and
the remaining one-third (33 percent) was
reserved for the testing sample.
Gender is also relevant when trying to create
an ideal assessment tool. Gender responsivity
has been regarded as an essential component
of risk instrument development.20 Genderresponsive risk metrics are developed by
separating males and females into individual
samples to produce gender-specifc
prediction models, which improves both
the context and accuracy of prediction.21 To
accommodate gender-specifc modeling, two
samples were also subdivided to create male(85 percent) and female-only (15 percent)
samples.
Most risk assessments utilize Burgess, or
unweighted, scoring schematics.22 Such

49
---------,0.-------------

D. Evaluating Predictive Performance
on the Test Set

schemes treat each item equally in predicting
recidivism, and the overall risk score is then
summed. Based on fndings of previous
studies,23 analytically weighting assessment
items improves predictive accuracy, and this
was the method adopted for PATTERN.

Relying on the AUC as the primary metric
for evaluating predictive validity, the
research team analyzed how the PATTERN
instruments performed in predicting
recidivism on the test set, which consisted
of inmates released between 2014 and
2015. More specifcally, risk points for each
instrument (that previously were created in
the training set) were calculated for subjects
in the test set. AUC results showed how each
assessment performed in predicting general
and violent recidivism for males and females.
Further analyses were completed to assess
potential issues related to racial and ethnic
neutrality.

Related to gender-responsive modeling
techniques, outcome specifcity is also
a concern. Instruments used to predict
general recidivism may not be accurate if
they are used only to identify a specifc type
of reofending.24 To accommodate outcome
specifcity, two PATTERN models were
computed to predict general recidivism (any
arrest or return to BOP custody following
release) and violent recidivism (violent
arrests following release) within three years.
As indicated in Table 1, the base recidivism
rate for all ofenders is roughly 47 percent for
general and 15 percent for violent recidivism.

E. Development of Risk Level Cut
Points
Consistent with the FSA’s requirement to
“classify each prisoner as having minimum,
low, medium, or high risk for recidivism,”
18 U.S.C. § 3632(a)(1), the NIJ consultants
identifed risk level categories (RLCs) via cut
points, which are risk score thresholds that
place individuals into four categories: high,
medium, low, or minimum. While there are
several known methods for creating RLC
cut points, PATTERN’s cut points were set
around the recidivism base rate. Briefy,
each item response (see Table 2) results in
a score, and an individual’s response scores
are summed to compute a total score. Te
total scores for individuals are analyzed, and
their RLC is ascertained based on the score’s
reference to all other released inmates in
the sample. An “aggregate,” or collection of

Te NIJ experts used this same process to
develop predictive models for general and
violent recidivism for males and females
separately. More specifcally, the team created
four distinct predictive models: 1) general
recidivism for males; 2) general recidivism
for females; 3) violent recidivism for males;
and 4) violent recidivism for females.
C. Predictive Performance Metrics
Model predictive validity, also known as
discrimination, measures the degree to
which an assessment separates the recidivists
from the non-recidivists. One of the most
widely used metrics for evaluating predictive
performance is the area under the curve
(AUC).25

-----~o~----50

scores, can be identifed by determining the
distribution of BOP releasees’ total scores.
Typically, this aggregation approximates a
normal distribution in which most of the
scores are near the average.
Cut points were next established to
determine an individual’s RLC. To identify
which individuals are high, medium, low,
and minimum risk, PATTERN uses the base
rate, which is roughly the “average” rate of
recidivism for the entire BOP development
data set. Te cut point is then set at a value
above or below the base rate. As an example,
let us assume a risk instrument had scores
ranging from -50 to 100, where the average
risk score was 25 with an accompanying 40
percent recidivism base rate (see Figure 1).
To provide an objective understanding of
what low risk means for this population, one
could then identify the risk score associated
with a 20 percent probability of reofending
(or half the base rate) and set that as the
mark for low risk. Using this example, if the
average risk score is 25, and the risk score
associated with 20 percent probability is 5,
then any individual scoring 5 or less would
be classifed as low risk (see Figure 2).
Similarly, the BOP base rates for general
and violent recidivism were used to help
determine how inmates could be assigned to
the four risk level categories for PATTERN.
Specifcally, the low risk threshold was set at
roughly half the base rate (or 24 percent) for
general recidivism and just over two-thirds
the base rate for violent recidivism (or 12
percent). Te minimum risk category was set
at just under one-quarter of the base rate (or

10 percent) for general recidivism and onethird of the base rate for violent recidivism
(or 5 percent). Te high risk category was
set at roughly two-thirds above the base rate
(or 80 percent) for general recidivism, and
just over twice the base rate (or 33 percent)
for violent recidivism. Tose individuals
not identifed as minimum, low, or high
risk were classifed as medium risk. Tis
specifc set of cut points was one of nearly
a dozen examined and tested. Te current
cut points endeavor to set the appropriate
balance between maximizing the number
of inmates eligible to earn early release
time credits and to participate in evidencebased programming that would reduce
their recidivism risk to a low or minimum
category, while also considering public safety
and the risk of recidivism upon release.
Te four categories were created for both
general and violent recidivism risk scores. A
fnal set of categories was created where an
individual must be identifed as minimum
risk of both general and violent recidivism to
be classifed as minimum in the fnal RLCs.
An individual that was identifed as lower
than medium risk in both the general and
violent models was labeled as low risk in
the fnal RLC. Tose individuals identifed
as high risk in either the general or violent
models were classifed as high risk in the
fnal RLCs. Finally, those not classifed as
minimum, low, or high risk were identifed
as medium risk in the fnal RLCs.
To prevent over-classifcation of female
ofenders, this process was completed
separately for male and female samples, using
the gender-neutral base rates for general

51
---------,0.-------------

Figure 1. Example Risk Score Distribution
Average Risk Score

Higher Risk Scores

Lowest Risk Scores

-50
Lowest Risk Score

25
Average Risk Score

100
Highest Risk Score

Figure 2. Illustration of Aggregate Risk Score Distribution and Cut Point Creation

Base Rate
½ ˜e
Base Rate

Low
Risk
20%
5
Low
Cut Point

40%
25
Average
Risk Score

52
Q,-------------

and violent recidivism (47 and 15 percent,
respectively). Once again, the NIJ experts
analyzed the impact of the RLCs on race/
ethnicity, including diagnostic odds ratios
to assess risk level category discrimination
between FSA-eligible and non-eligible RLCs.
In order to identify potential sources of bias,
a relative rate index (RRI) was computed
to assess the magnitude of disparity across
risk categories. Te RRI compares the rate,
or proportion, of individuals across race/
ethnicity groups in the RLCs of PATTERN.26

III. PATTERN Results
Tis section provides the fndings from the
developed PATTERN general and violent
risk models, as well as the created RLCs.
Results from the prediction models are
displayed frst, with comparisons between
the PATTERN and BRAVO-R models,
showing the achievements of PATTERN.
Ten, the RLC proportions and category

recidivism rates are presented. Finally,
tests and comparisons for racial and ethnic
neutrality are provided.
A. PATTERN Risk Prediction Models
Four risk models were created for the BOP
sample of individuals released.27 Table 2
provides the response scoring for each of the
four models created.
Each column represents a diferent genderand outcome-specifc prediction model.
Cell values represent response scores, where
blank cells indicate items not identifed as
predictive for a given model. Larger values
indicate greater prediction strength for
a given model. Of note is the overlap in
predictors; they are more common than
distinct across models. However, where some
items distinctly predict for certain models,
fndings are consistent with expectations of
recidivism specifcity.

Table 2. Points Assigned in the PATTERN Risk Assessment Models
General
Male
Female

MEASURE
Age at frst conviction
>= 35
< 35, >= 25
< 25, >= 18
< 18

0
4
8
12

0
5
10
15

Violent
Male
Female
0
2
4
6

0
1
2
3

53
---------,0.-------------

General
Male
Female

MEASURE
Age at time of assessment

Violent
Male
Female

> 60
> 50, <= 60
> 40, <= 50
> 29, <= 40
> 25, <= 29
>= 18, <= 25

0
6
12
18
24
30

0
3
6
9
12
15

0
3
6
9
12
15

0
1
2
3
4
5

0
> 0, <= 1
> 1, <= 2
>2
Infraction convictions (serious and violent)
0
> 0, <= 1
> 1, <= 2
>2
Number of programs completed (any)
0
1
> 1, <= 3
> 3, <= 10
> 10
Number of technical or vocational courses
0
> 0, <= 1
>1
Federal industry employment (UNICOR)
Yes
No

0
3
6
9

0
2
4
6

0
2
4
6

0
1
2
3

0
2
4
6

0
2
4
6

0
2
4
6

0
2
4
6

0
-3
-6
-9
-12

0
-2
-4
-6
-8

0
-1
-2
-3
-4

0
-1
-2
-3
-4

-2
-1
0

-4
-2
0

Infraction convictions (any)

-1
0

-1
0

-----~o~----54

MEASURE
Drug treatment while incarcerated
No need indicated
Completed residential drug treatment during
incarceration
Completed drug treatment during incarceration
Need indicated but no treatment during
incarceration
Drug education while incarcerated
No
Yes
Non-compliance with fnancial responsibility
No
Yes
Instant ofense violent
No
Yes
Sex ofender (Walsh)
No
Yes
BRAVO Initial: Criminal History Score
0-1 points
2-3 points
4-6 points
7-9 points
10-12 points
13+ points
BRAVO Initial: History of Violence
None
>10 years minor
>15 years serious
5-10 years minor
10-15 years serious
<5 years minor

General
Male
Female

Violent
Male
Female

-6
-4

-9
-6

-3
-2

-3
-2

-2
0

-3
0

-1
0

-1
0

0
-1

0
-1

0
4

0
5

0
3

0
1

0
1

0
-1
0
3

0
6
12
18
24
30

0
6
12
18
24
30

0
2
4
6
8
10

0
2
4
6
8
10

0
1
2
3
4
5

0
1
2
3
4
5

0
1
2
3
4
5

0
1
2
3
4
5

55
---------,0.-------------

MEASURE
5-10 years serious
<5 years serious
BRAVO Initial: History of Escapes
None
>10 years minor
5-10 years minor
<5 years minor or any serious
BRAVO Initial: Voluntary Surrender
Yes
No
BRAVO Initial: Education Score
HS degree or GED - Verifed
Enrolled and progressing in GED
No verifed degree and not participating in
GED program
0 – Lowest
1 – Low/Moderate
3 - Moderate
5 – High
7 – Greatest
B. Predictive Validity

General
Male
Female
6
6
7
7

Violent
Male
Female
6
6
7
7

0
2
4
6

0
2
4
6

0
1
2
3

0
2
4
6

-12
0

-9
0

-1
0

-2
0

-2
-1
0

-2
-1
0
0
1
3
5
7

0
1
3
5
7

1.0 means there is a 100 percent probability
in the result, which is essentially perfection.
Many risk assessments in the United States
have an AUC in the .6 to .75 range. Afer
computing the test sets’ total scores using
the PATTERN and BRAVO-R models, the
predictive validity was compared. Using the
two recidivism types (general and violent),
AUC values were computed for males and
females. Given that BRAVO-R has two tools,
one at intake and one for reclassifcation,
scores were combined so that if an individual
possessed a reclassifcation score, it was

Te strength of the predictive validity is of
key importance. As described previously,
predictive validity is measured through AUC
values. As shown in Figure 3, PATTERN
achieves a higher level of predictability and
surpasses common risk assessment tools for
correctional population in the U.S.
Simply put, an AUC of .5 means there is a 50
percent probability that the result is accurate,
which is the same as a coin toss. An AUC of

-----~o~----56

Figure 3. AUC Values for Tools Used in the United States
1.00
0.95
0.90
0.85
0.80

0.8

0.75

0.73

0.70

˜e Average
AUC of Other
Tools is .684

0.65 0.67
0.60

0.73

0.67

0.64

0.67

0.66

WRN

WRN_R

0.79

0.57

0.55
0.50

0.74

0.76

COMPAS

LSI-R

LSI:SV

RMS

SPIN-W STRONG

PCRA

PCRA PATTERN PATTERN
Violent
Male
Female

PATTERN is
15% More
Predictive
˜an Other
Tools

Table 3. Te PATTERN and BRAVO-R AUC Comparisons

Model
PATTERN AUC Testing Dataset
BRAVO-R AUC Testing Dataset
AUC Enhancement
used. By contrast, an initial score was used
only if a reclassifcation score was not
available for a given ofender. Findings from
these analyses are presented in Table 3.
When examining the models’ AUC values,
PATTERN demonstrates a consistent,
modest improvement when compared to
BRAVO-R, which was already a high-scoring
tool. Specifcally, there was a 1 to 3 percent
improvement in AUC using the PATTERN
model scoring.
Next, cut points were established for
PATTERN based on the analytic placement

General
Male
Female
0.80
0.79
0.77
0.77
0.03
0.02

Violent
Male
Female
0.78
0.77
0.75
0.76
0.03
0.01

method previously described.28 Test set
fndings are presented in Table 4. As
indicated, cut points were developed for the
general and violent models and for females
and males separately. Te presented fndings
below show the model cut points, percentage
of the population, and recidivism rates for
each RLC.
Importantly, given the scoring for all four
assessment models (see Table 2) and the cut
point system developed to assign risk levels
(see Table 4), 99 percent of ofenders have
the ability to become eligible for early release

57
---------,0.-------------

Table 4. Te PATTERN RLC Cut Point Scores, Population Distribution and Recidivism Rates
RLC
Any
Minimum
Low
Medium
High
RLC
Violent
Minimum
Low
Medium
High

Cut Point
Male
10
33
45
Cut Point
Male
21
35
40

Population
Female
9
29
45

Recid.
20
28
17
35

Population
Female
22
30
35

10
32
51
73
Recid.

49
35
9
8

4
19
32
44

Note: Te base rate is 47 percent for general recidivism and 15 percent for violent recidivism.

percent. Odds ratios comparing minimum
and low risk ofenders to those ofenders
who are not FSA eligible, or to those in
the medium and high risk categories, were
identifed. With regard to general recidivism,
minimum and low risk ofenders possess 86
percent reduced odds of recidivating, and
88 percent reduced odds of recidivating
violently, as compared to medium and high
risk ofenders. Collectively, these fndings
show that a substantial proportion of the
population is deemed eligible for early
release through the accumulation of FSA
earned time credits, and the odds ratio
statistics identify a substantial demarcation
in recidivism rates between early-releaseeligible (minimum and low) and noneligible (medium and high) risk categories.
Furthermore, the RLC levels identify their
achieved goal, balancing public safety
concerns (i.e., a low rate of recidivism) while

through the accumulation of earned time
credits even though they may not be eligible
immediately upon admission to prison. Tat
is, as ofenders participate in evidence-based
programming, carry out work assignments,
and complete courses, nearly all have the
ability to reduce their risk score to the low
risk category.
Next, the RLC proportions and each
category’s associated rate of recidivism
were examined (see Table 5). Nearly half of
the population (48 percent) identifed as
low or minimum risk and thus eligible for
early release under the FSA through earned
time credits. When examining recidivism
proportions, the proportion of recidivism
increases as risk increases from minimum
to high. General recidivism rates increase,
on average, 22 percent with each successive
RLC category and an average increase of 10

-----~o~----58

Table 5. Te PATTERN RLC Sample Proportions and Recidivism
PATTERN
RLC
Minimum
Low
Medium
High

Population
20
28
17
35

Violent
Recid.

Recid.
9
31
51
73

Recid OR
1
6
13
30

0.14

Violent OR
0.12

Table 6. Te PATTERN RLC Gender Comparisons
PATTERN
RLC
Minimum
Low
Medium
High

Recidivism
Violent
Population
Recidivism
OR
Recidivism
Violent OR
Male Female Male Female Male Female Male Female Male Female
17
38
10
9
1
1
25
42
30
35 0.14 0.16
7
4 0.13 0.19
18
14
50
60
14
11
40
6
73
74
31
17

simultaneously maximizing the proportion
of individuals eligible for early release
through the accumulation of earned time
credits.
Next, gender diferences were examined
using the PATTERN RLCs. Te population
proportions and recidivism fndings are
presented in Table 6. In contrast to the
overall RLC breakdown (see Table 5) there
is a substantial distinction in population
proportions when comparing males and
females. Specifcally, 42 percent of males
versus approximately 79 percent of females
are identifed to be eligible for early release
under the FSA through the accumulation
of earned time credits. However, the rates
of both general and violent recidivism are

similar across all four RLCs. Furthermore,
the odds ratios identity similar prediction
strength with minimum and low risk males
identify 86 percent reduced odds of general
and 87 percent reduced odds of violent
recidivism; where FSA eligible females
identify 84 percent reduced odds of general
and 81 percent reduced odds of violent
recidivism.
Collectively, these fndings
indicate gender parity when examining
recidivism by RLC category.
C. Race/Ethnicity
Te fnal set of tests examined potential issues
related to racial and ethnic neutrality. While
nearly all indicators likely to be used within
a risk assessment model have the potential

59
---------,0.-------------

individuals being identifed as FSA-eligible,
or the likelihood that minimum and low
risk ofenders will be eligible for the time
credits. ORs are then provided, examining
the diferences between PATTERN’s general
and violent recidivism rates for both white
and non-white FSA-eligible individuals.

to be correlated with socio-economic
status (SES), race, and/or ethnicity, recent
fnding29 have indicated that bias is reduced
when assessments include dynamic needs
indicators. Dynamic programmatic needs
measures tend to have lower associations
with SES/race/ethnicity30 and, in turn, tend
to reduce the impact of criminal history
measures that are more highly correlated
with SES/race/ethnicity. Terefore, when
developing PATTERN, there was an
attempt to include many predictive dynamic
indicators. Tis efort served two purposes:
to reduce potential sources of racial disparity
and to develop an assessment that would
allow ofenders the opportunity to reduce
their risk scores over time via good behavior
(i.e., a lack of infraction behavior) and
successful completion of programming.

To provide a detailed examination, race/
ethnicity fndings are broken down by gender.
As displayed in Table 7, AUC fndings are
consistent across these groups. In particular,
the PATTERN AUC values are consistently
large. Specifcally, AUC values for general
recidivism range from 0.78 to 0.83 for males
and from 0.76 to 0.81 for females. With regard
to violent recidivism, AUC values range for
0.75 to 0.82 for males and from 0.75 to 0.81
for females. Overall, these fndings indicate
strong and comparable prediction strength
for PATTERN models across all race/
ethnicity categories, suggesting minimal
racial/ethnic disparity for PATTERN’s
prediction strength.

Below, AUC comparisons broken down by
race/ethnicity are provided for PATTERN.
Next, RLC population distribution variations
are examined across the four primary BOP
race/ethnicity categories—white, AfricanAmerican (AA), Hispanic, and Other. Te
Relative Rate Index (RRI) is then computed
to assess the likelihood of non-white

In performing their analysis, the NIJ experts
also examined gender and race/ethnicity
variations across population distributions

Table 7. Te PATTERN Race/Ethnicity AUC Comparisons
Recidivism
General
PATTERN
Violent
PATTERN

All
0.80
All
0.77

Male
White
AA Hispanic
0.81 0.78 0.78
White
AA Hispanic
0.80 0.75 0.78

Other
0.83
Other
0.82

All
0.79
All
0.78

White
0.79
White
0.75

Female
AA Hispanic
0.80 0.76
AA Hispanic
0.80 0.75

Other
0.81
Other
0.81

-----~o~----60

for PATTERN. Tose results are shown in
Tables 8 -10.

IV. Needs Assessment Process
Under the FSA, the risk and needs
assessment system must include both a risk
component (PATTERN, described above)
and a needs component. Specifcally with
respect to inmate needs, the system must
be used to determine “the type and amount
of evidence-based recidivism reduction
programming that is appropriate for each
prisoner and assign each prisoner to such
programming … based on the prisoner’s
criminogenic needs.”31 As discussed in
detail in Chapter 4, the BOP has a strong
needs assessment process in place to match
inmates with programs to address their
criminogenic needs. Trough the FSA
implementation process, and in consultation
with the IRC and others, the Department
identifed an opportunity to expand upon
the present system and to create an updated,
more comprehensive needs assessment. Te
details surrounding the project timeline are
set forth in Chapter 4.
Dyslexia Screening
Under the FSA, the BOP is required to
“incorporate a dyslexia screening program
into the System,”32 screen inmates for
dyslexia, and “incorporate programs
designed to treat dyslexia into the evidencebased recidivism reduction programs
or productive activities.”33 Te BOP
currently provides a variety of educational
opportunities to inmates housed within its
122 institutions.

Specifcally, Program Statement 5200.05,
Management of Inmates with Disabilities,
outlines the BOP’s strategy for managing
inmates living with disabilities, including
dyslexia. According to the policy and
consistent with the FSA’s requirements,
inmates are pre-screened at the time
designation takes place and assessed
for disabilities once they arrive at an
institution. Individualized accommodation
procedures require institutions to convene
interdisciplinary committees to ensure
inmates have appropriate access to treatment,
programs, and services.
As required by the FSA and to ensure a
strong screening process specifc to dyslexia,
the BOP expanded current requirements for
assessing and accommodating disabilities as
part of the work of developing the Risk and
Needs Assessment System. Psychologists and
education experts within the agency consulted
external experts and reviewed studies on the
prevalence of dyslexia in prison populations
within the United States and worldwide.34
Te FSA requires that tools used in dyslexia
assessment be psychometrically valid, efcient,
low cost, and readily available.35 Terefore,
the BOP dyslexia screening process consists
of two steps that ensure a quality assessment
allowing for a diagnosis. Te frst step is the
inmate screening. Te FSA defnes dyslexia
as “an unexpected difculty in reading for an
individual who has the intelligence to be a
much better reader, most commonly caused
by a difculty in the phonological processing
(the appreciation of the individual sounds
of spoken language), which afect the ability
of an individual to speak, read, and spell.”36

61
---------,0.-------------

-----~o~-----

62

White
30
27
15
29

AA
7
20
20
53

Hispanic
16
31
20
33

Other
19
20
16
45

NonWhite

0.63

White

0.41

White
9
31
51
73

AA
9
29
49
73

Hispanic
10
26
46
70

Other
8
38
65
81
0.13

White

PATTERN
Minimum
Low
Medium
High

White
1
6
12
26

AA Hispanic
2
2
8
5
16
10
34
26

Other
1
6
15
33

Male

Dif

1.54

RRI

White
39
43
14
5

NonWhite

0.15

White

0.13

-0.02

Dif

White
1
4
8
13

White
9
39
0.16 -0.03 62
75

NonWhite

Table 10. PATTERN Violent Recidivism Rates by Gender

PATTERN
Minimum
Low
Medium
High

Male

Table 9. PATTERN General Recidivism Rates by Gender

PATTERN
Minimum
Low
Medium
High

Male

Table 8. PATTERN Population Distribution

AA
1
5
14
23

AA
8
29
55
75

AA
36
39
15
10

Hispanic
1
5
9
13

Other
9
49
69
72

Female

Other
37
40
17
7

Other
1
4
12
26

Female

Hispanic
9
28
56
70

Hispanic
37
47
12
4

Female

0.25

White

0.17

White

0.21

White

0.17

NonWhite

0.17

NonWhite

0.22

NonWhite

+0.08

Dif

0.00

Dif

1.08

RRI

Because experts in diferent disciplines may
use other defnitions, the BOP considered
three additional diagnostic criteria to
inform their process: the Specifc Learning
Disorder,37 the Specifc Learning Disability,38
and the International Dyslexia Association’s
defnition of dyslexia.39
Based on these requirements, BOP staf
developed the Screening Checklist for
Dyslexia. Tis checklist consists of three
sections with questions related to historical
information (e.g., past diagnosis of a learning
disorder), processing skills, and perceptual
problems. Tis 30-question checklist will be
read aloud to the inmate at the time of intake
processing by education staf. A threshold
score is set that indicates the possible presence
of dyslexia. If this threshold is reached or if
additional information suggesting dyslexia
is obtained at any time during the screening,
the inmate proceeds to the second step of the
assessment process.
For inmates determined to possibly
experience symptoms of dyslexia, step two
of the process requires the BOP Special
Learning Needs teacher to administer both
batteries of the Woodcock-Johnson IV Tests
of Achievement and Oral Language, following
the guidance provided by its authors.40 Tis
information is then used to determine
whether criteria for dyslexia are met, and
if so, the inmate will receive appropriate,
individually determined accommodations.
Tis may include assistance (e.g., tutors,
extended time) with existing programming

or specialized programming designed to
treat dyslexia, such as courses on phonics.

Conclusion
Te Risk and Needs Assessment System
is strong. Utilizing available data metrics,
the expert team created an optimal risk
instrument, PATTERN, that outlines a single
assessment which incorporates a design that
encourages risk reduction behavior (i.e.,
lack of infractions and increases in program
participation) over time. PATTERN also
provides scales for general and violent
recidivism that are the same across all
assessments.
Te development of PATTERN is a
signifcant advancement in the Department’s
implementation of the FSA. First, PATTERN
achieves a high level of predictive
performance and surpasses what commonly
is found for risk assessment tools for
correctional populations in the United States.
Second, the PATTERN instrument more
appropriately aligns with the goals of the
FSA because it makes greater use of dynamic
factors. Tird, the PATTERN instrument’s
predictive performance is unbiased
across racial and ethnic classifcations.
Fourth, PATTERN is a single assessment,
incorporating a design that may encourage
risk reducing behavior by inmates over time.
Furthermore, BOP has needs assessment
and dyslexia screening processes to identify
and address the needs of inmates.

63
---------,0.-------------

Notes

either the U.S. Marshals Ofce or the place of
confnement. Additionally, this item applies
only to post-sentencing voluntary surrender,
and does not include cases where the inmate
surrendered to the U.S. Marshals on the
same day as sentencing. Voluntary Surrender
Credit may only be applied to the initial term
of confnement; it may not be applied to any
subsequent Supervised Release, Mandatory
Release or Parole Violation return to custody.
See BOP Program Statement 5100.08,
Inmate Security Designation and Custody
Classifcation: https://www.bop.gov/policy/
progstat/5100_008.pdf.

Dr. Alex Kigerl, assistant research professor
of criminal justice and criminology at
Washington State University, provided
valuable assistance to Duwe and Hamilton
in the development of PATTERN.
1

BRAVO, the BOP’s classifcation system, is
used to predict serious misconduct while an
inmate is incarcerated in BOP facilities. It is
also used to assign inmates to an appropriate
security level. PATTERN, on the other
hand, is used to predict general and violent
recidivism afer an inmate leaves BOP
custody.
2

Harer, M., Langan, N., & Gwinn, J. (2019).
Te Federal Bureau of Prisons Inmate
Classifcation Instrument as a BehavioralChange Predictor of Serious Prison Misconduct
and Post Release Recidivism. Unpublished
manuscript.
8

As presently developed, PATTERN does
not currently include a needs assessment
component. As discussed in Chapter 4, the
BOP will continue to use its current needs
assessment system, which is being modifed
and enhanced at the time of this writing.
Te BOP’s needs assessment system will
continue to be developed and enhanced over
the coming months.
3

4

For greater detail on selected BRAVO items
and their operationalization, see Harer, M.,
Langan, N., & Gwinn, J. (2019). “Te Federal
Bureau of Prisons Inmate Classifcation
Instrument as a Behavioral-Change
Predictor of Serious Prison Misconduct and
Post Release Recidivism.” Washington, DC:
Federal Bureau of Prisons, Ofce of Research
and Evaluation.
9

18 U.S.C. §3632(a)(4).

Hamilton, Z., Kigerl, A., Campagna,
M., Barnoski, R., Lee, S., van Wormer,
J., & Block, L. (2016). “Designed to Fit:
Te Development and Validation of the
STRONG-R Recidivism Risk Assessment.”
Criminal Justice and Behavior, 43, 230-263.
5

During the 45-day public study period
discussed in Chapter 4, the Department
welcomes input on what changes to these
variables may increase predictability.
10

See Chapter 4 of this report for a more
detailed discussion of how inmates are
processed into federal custody.

6

A count measure refers to data in which
the values are non-negative whole numbers
such as 0, 1, 2, 3, etc.
11

Voluntary surrender means the inmate was
not escorted by a law enforcement ofcer to

7

-----~o~----64

BOP Program Statement 5100.08,
Inmate Security Designation and Custody
Classifcation: https://www.bop.gov/policy/
progstat/5100_008.pdf.

12

Per the BOP’s Program Statement 5270.09,
Inmate Discipline Program, 100 security
level ofenses are categorized as the greatest
severity level prohibited acts. Tey include
but are not limited to “killing (100) and
“assaulting“[a]ssaulting any person, or an
armed assault on the institution’s secure
perimeter (a charge for assaulting any
person at this level is to be used only when
serious physical injury has been attempted or
accomplished) (101).” Also, 200 level ofenses
are categorized as high security level, for
example, “escape“[e]scape from a work
detail, non-secure institution, or other nonsecure confnement, including community
confnement, with subsequent voluntary
return to Bureau of Prisons custody within
four hours (200),” and “fghting with another
person (201).” See https://www.bop.gov/
policy/progstat/5270_009.pdf.
13

Age at frst conviction measures the frst
conviction for any ofense.
14

Violent recidivism is any arrest that falls into
the category “homicide/aggravated assault,”
“weapons/explosives,”
“sexual
assault,”
“assault,” “robbery,” or “other violent.”
15

Te Walsh criteria refers to whether the
inmate is a sex ofender as defned in the Sex
Ofender Registration and Notifcation Act
(SORNA), Title I of the Adam Walsh Act.
See https://www.justice.gov/archive/olp/pdf/
adam_walsh_act.pdf.
16

Diagnostic examinations were also
completed (but not displayed), including
bivariate assessments of item prediction
strength. All fndings indicated substantial
prediction strength of selected risk items,
and computed risk models were identifed to
meet model assumptions.
17

Te measure is the United States Sentencing
Guidelines Criminal History item, used as
part of the inmate’s sentence computation.
18

Van Voorhis, P., Wright, E.M., Salisbury, E.,
& Bauman, A. (2010). “Women’s Risk Factors
and Teir Contributions to Existing Risk/
Needs Assessment: Te Current Status of a
Gender-Responsive Supplement.” Criminal
Justice and Behavior, 37, 261-288.
19

Hamilton, Z., Campagna, M., Tollefsbol,
E., van Wormer, J., & Barnoski, R. (2017).
“A More Consistent Application of the RNR
Model: Te STRONG-R Needs Assessment.”
Criminal Justice and Behavior, 44, 261-292.
20

Hamilton, Kigerl, Campagna, Barnoski,
Lee, van Wormer, & Block (2016). “Designed
to Fit.”

21

Duwe, G. (2019). “Better Practices in the
Development and Validation of Recidivism
Risk Assessments: Te Minnesota Sex
Ofender Screening Tool–4.” Criminal Justice
Policy Review, 30, 538-564; Hamilton, Kigerl,
Campagna, Barnoski, Lee, van Wormer, &
Block (2016). “Designed to Fit”; and Duwe,
G. (2014). “Te Development, Validity, and
Reliability of the Minnesota Screening Tool
Assessing Recidivism Risk (MnSTARR).”
Criminal Justice Policy Review, 25, 579-613.
22

65
---------,0.-------------

Vincent, G.M., Laura, S.G., & Grisso, T.
(2012). “Risk Assessment in Juvenile Justice:
A Guidebook for Implementation.” Chicago,
IL: John D. and Catherine T. MacArthur
Foundation.

Note that cut points were created using the
training set and then applied on the test set.

23

28

See Hamilton, Z., Kowalski, M.A., Kigerl,
A., & Routh, D. (2019). “Optimizing
Youth Risk Assessment Performance:
Development of the Modifed Positive
Achievement Change Tool in Washington
State.” Criminal Justice and Behavior.
DOI: 10.1177/0093854819857108; and
Hamilton, Z., Kowalski, M.A., Schaefer,
R., & Kigerl, A. (2019). “Recrafing
Youth Risk Assessment: Developing the
Modifed Positive Achievement Change
Tool for Iowa.” Deviant Behavior. DOI:
10.1080/01639625.2019.1609302.

29

As Davis and Goadrich (2006) point out,
the AUC can provide an overly optimistic
estimate of predictive discrimination for
imbalanced datasets. Nevertheless, the AUC
is relatively robust across diferent recidivism
base rates and selection ratios. See Smith, W.
(1996). “Te Efects of Base Rate and Cutof
Point Choice on Commonly Used Measures
of Association and Accuracy in Recidivism
Research.”
Journal
of
Quantitative
Criminology, 12(1), 83-111; and Davis, J.
& Goadrich, M. (2006). “Te Relationship
Between Precision-Recall and ROC Curves,”
in Proceedings of the 23rd International
Conference on Machine Learning. Canada:
Banf, 1-12.
24

Puzzanchera, C., & Hockenberry, S.
(2013). An Interpretation of the National
DMC Relative Rate Indices for Juvenile
Justice System Processing in 2010 (National
Disproportionate
Minority
Contact
Databook). Washington, DC: National
Center for Juvenile Justice for the Ofce of
Juvenile Justice and Delinquency Prevention.
25

26

30

Ibid.

31

18 U.S.C. § 3632(a)(3).

32

18 U.S.C. § 3632(h)(1).

33

18 U.S.C. § 3632(h)(2).

External experts consulted and studies
reviewed on the prevalence of dyslexia in
prison populations within the United States
and worldwide include:
34

American
Psychiatric
Association
(2013). Defnition of Specifc Learning
Disorder, in Diagnostic and Statistical
Manual of Mental Disorders (5th ed.).

See detailed AUC discussion in Chapter 2.

Breaux, K., & Eichstadt, T. (2018).
Pearson Clinical Assessment Solutions: A
Dyslexia Toolkit.

As discussed, boosted regression models
were computed for each of the four models,
with selected items and weights created from
model coefcients. Coefcient values for
each model were converted (multiplied by
100) to whole numbers to improve ease of
risk scoring.
27

Fabelo, T., et al. (2004). “Te Impact
of Ignoring Dyslexia and Reading
Disabilities in the Criminal Justice
System: What We Know and Need to

-----~o~----66

Know.” In Report to the Dyslexia Research
Foundation, Inc.

Learning Difculties and Learning
Disabilities. London: Prison Reform
Trust.
http://www.prisonreformtrust.
org.uk/uploads/documents/NOKNL.
pdf.

Gilbert, J.K., et al. (2012). “Early
Screening for Risk of Reading Disabilities:
Recommendations for a Four-Step
Screening System.” Assessment for Efective
Intervention: Ofcial Journal of the Council
for Educational Diagnostic Services, 38(1),
6-14. https://journals.sagepub.com/doi/
pdf/10.1177/1534508412451491.

Mather, N., et al. (2016) “Use of the
Woodcock-Johnson IV for Dyslexia
Evaluations.” WJ Perspectives.
One Hundred Fifeenth Congress of the
United States (2018). S. 756 – 14, ss 3635.
Defnitions (1) Dyslexia. https://www.
congress.gov/115/bills/s756/BILLS115s756enr.pdf

Indiana Department of Education
(2018). Criteria for Dyslexia Screeners.
https://www.doe.in.gov/sites/default/
files/literacy/criteria-used-for-vettingand-approving-dyslexia- screeners.pdf.

Proctor, C.M., et al. (2015). “Use of
the Woodcock-Johnson IV for the
Assessment of Dyslexia.” WoodcockJohnson IV Assessment Service Bulletin, 6.

International Dyslexia Association
(2018). Dyslexia Screener for Adults.
https://dyslexiaida.org/screening-fordyslexia/dyslexia-screener-for-adults/.

Rose, J. (2009). “Identifying and
Teaching Children and Young People
with Dyslexia and Literacy Difculties.”
British Dyslexia Association. https://
www.bdadyslexia.org.uk/about.

International Dyslexia Association
(2002). Defnition of Dyslexia (2002).
http://www.dyslexiaida.org.
Jameson, M. (2014). “Screening
Checklists for Adults with Dyslexia,”
Releasing the Potential of Ofenders with
Dyslexia and Related Specifc Learning
Difculties: A Practical Guide for Staf
Working with Ofenders. http://www.
dyslexia-malvern.co.uk/docs/justice/
Releasing%20Potential.pdf.
Kooij, J. (2013). “Comorbidity and
Diferential Diagnosis: ADHD and
Dyslexia.” Adult ADHD Diagnostic
Assessment and Treatment, 83 (3rd ed.).
Loucks, N. (2006). Te Prevalence and
Associated Needs of Ofenders with

U.S. Department of Education. Defnition
of Specifc Learning Disability. Section
300.8: Child with a Disability. IDEA:
Individuals with Disabilities Education
Act.
https://sites.ed.gov/idea/regs/
b/a/300.8.
35

18 U.S.C. § 3635(2).

36

18 U.S.C. § 3635(1).

American Psychiatric Association (2013).
“Defnition of Specifc Learning Disorder.” In
Diagnostic and Statistical Manual of Mental
Disorders (5th ed.).
37

67
---------,0.-------------

U.S. Department of Education. Defnition
of Specifc Learning Disability. Section 300.8:
Child with a Disability. IDEA: Individuals
with Disabilities Education Act. https://sites.
ed.gov/idea/regs/b/a/300.8.
38

International Dyslexia Association (2002).
Defnition
of
Dyslexia.
http://www.
dyslexiaida.org.
39

Proctor et al. (2015). “Use of the WoodcockJohnson IV.”
40

-----~o~----68

Chapter 4

W

Implementing the First Step Act of 2018
Risk and Needs Assessment System
the Department seriously considered the
available data at the time of the System’s
release, the Department will consider
potential new sources of data, new research,
and new metrics on an ongoing basis. Te
Department also is committed to continuing
to work and to consult with the Independent
Review Committee (IRC) and rely on its
expertise to improve the System.

ith this report, the Attorney
General is publishing the First
Step Act of 2018 Risk and Needs
Assessment System (System). Te publication
of the System is only the frst step in an
ongoing, collaborative, and dynamic process
to enhance prison programming, improve
inmate outcomes, and ultimately reduce
recidivism and make every community safer.
Te First Step Act (FSA or the Act) requires
signifcant work to be completed in a short
period of time. Te Department met the FSA’s
ambitious deadline to publish the System
within 210 days of the Act’s enactment.1 In
crafing the System, the Department worked
to make the benefts of the FSA as widely
available as possible without compromising
the predictive reliability of the risk and needs
tool. In developing and validating the System,
the Department necessarily was limited by
the data available. Although the System is
a good initial step based on the available
information, it can be improved with more
time, consultation, data, and research.

During a 45-day public study period
following the publication of the System, the
public will be able to review the System and
consider ways in which it may be improved.
Further work can and will be done during
the 45 days to look for ways to improve
the System. Following this study period,
the Department in September will invite
stakeholders, public interest organizations,
and the public to comment on the System.
In seeking ways to make the System better,
the Department will welcome comment on
ways to make the benefts of the FSA widely
available while maintaining the System’s
predictive reliability. Te Director of the
Federal Bureau of Prisons (BOP) and other
senior Department leadership will be present
in September to receive input and feedback.
Afer reviewing the input, we will reassess
the tool, make any appropriate changes, and
begin the process of assessing each individual
inmate.

Te Department is committed to improving
the System’s operation and implementation.
To that end, the Department will engage in
regular and ongoing dialogue with a wide
range of stakeholders, experts, and the
public. Te experience and expertise of
these communities will aid the Department
as it works to improve the System. Just as

-----~o~----70

Angela Hawken, Ph.D. Dr. Hawken is a Professor of Public Policy at New
York University and the founder and director of NYU’s Litmus/BetaGov
program. Her team’s Segregation Solutions project works with corrections
agencies to reduce the use of segregated housing, reform practices to reduce
violent behavioral infractions, provide incentives for pro-social behaviors, and
reduce stress. Te project supports corrections agencies with strategic planning,
policy guidance, data analysis, and evaluation. She is the principal investigator
of Graduated Reintegration. Graduated Reintegration explores the feasibility
and efectiveness of a resource- reallocation program that allows prisoners to release early into supported
scattered-site housing and structured programming to ease the transition from prison to the community.
Dr. Hawken primarily will assist the Department in implementing the risk and needs assessment system.

In addition, the National Institute of
Justice (NIJ) has contracted an outside
expert, Dr. Angela Hawken, to assist with
implementation of the System. Dr. Hawken’s
biography is included above.
Tis chapter outlines the Department’s plan
to implement the Risk and Needs Assessment
System in BOP facilities. First, it outlines
the Department of Justice’s robust plan for
implementing the new System. Second, it
provides estimated deadlines for completion
of the steps in this work plan. Tird, it
describes how the System will be re-validated
through further research. Fourth, this
chapter chronicles the training that BOP will
provide to its staf members on the Risk and
Needs Assessment System, which is critical
to the implementation process. Fifh, this
chapter outlines the process for assessing the
implementation of the System to ensure an
efective and strong system. Sixth, the report
explains how the risk and needs assessment
tool will be used to determine the evidencebased recidivism reduction programming
and productive activities that are appropriate
for each inmate and assign those programs

and activities. Finally, the chapter outlines
how BOP will communicate efectively with
inmates and stakeholders throughout the
implementation process.

I. Implementing the Risk and Needs
Assessment System
Following the publication of the System,
the FSA requires BOP to complete an
initial intake risk and needs assessment for
each inmate within 180 days.2 Te BOP
is committed to completing this initial
assessment for each inmate within this
time frame. Te Act also requires BOP to
assign inmates to appropriate evidencebased recidivism reduction programs
and productive activities based on that
determination. Te BOP, in consultation
with the Department, is reviewing all
programs currently ofered at BOP facilities.
When that review is complete, no later than
January 2020, a full list of the approved
programs will be posted on the BOP’s
website. BOP is committed to completing
each of the tasks outlined below to meet the
FSA’s requirements. Te implementation of
the System, and other FSA requirements,

71
---------,0.-------------

will involve a signifcant commitment
of BOP resources. Te Department has
identifed funds it received in Fiscal Year
2019 that may be used to support First Step
Act priorities. Te Department is working
with the Administration and Congress to
ensure that necessary funds will be available
in Fiscal Year 2020 and beyond.

Sentence Computation Center (DSCC)
in Grand Prairie, Texas perform an initial
assessment of the inmate for purposes of
determining where an inmate will serve
his/her sentence (called “designation”).
Te DSCC staf review information from
the U.S. Probation Ofce and the U.S.
Marshals Service, including the Judgement
and Commitment Order, the Statement of
Reasons,3 and the Presentence Investigation
Report. Te DSCC staf may also receive
information from the U.S. Attorney’s Ofce
or Department component that prosecuted
the inmate.4 DSCC staf use this information
to enter the inmate into the BOP’s inmate
management system and conduct the initial
security scoring (via the Inmate Load and
Security Designation Form). Subject-matter
experts at the DSCC (including medical staf
and clinical psychologists) also perform a
preliminary review of the inmate’s needs
with regard to medical, mental health, and
criminogenic programming, which may
infuence the designation decision. As part
of the FSA implementation and in addition
to the designation steps listed above, DSCC
staf, with assistance from BOP Legal staf,
will use the Risk and Needs Assessment
System to assign each incoming inmate an
initial recidivism risk level of Minimum,

A. Assigning a Recidivism Risk Level
and Conducting a Needs Assessment
As discussed above, BOP is responsible
for assigning a recidivism risk level and
conducting a needs assessment for all
inmates, including newly committed
inmates and current inmates. Once the
Risk and Need Assessment System is fully
operational and BOP staf fully trained, newly
committed inmates will be assessed as they
are processed into the BOP. Current inmates
will have their needs assessed and their risk
level communicated to them during regular
Unit Management Team meetings, which are
explained further below.
1. Newly-Committed Inmates
When an inmate is sentenced (or the BOP
accepts custody of a sentenced individual),
trained staf at the Designations and

Te BOP Grand Prairie Ofce Complex, located in Texas.

-----~o~----72

Low, Medium or High.5 BOP will require
additional staf resources added at the DSCC
to support this work.
Once the inmate arrives at his or her
designated facility, he or she is assessed
again as part of intake in the Receiving and
Discharge area. Within the frst twenty-eight
days of incarceration, the inmate has an initial
interview by his or her Unit Management
Team (comprised of the Unit Manager, Case
Manager, and Correctional Counselor).
Additionally, the inmate is seen by medical
staf (which may include an exam) and mental
health staf to determine initial treatment
needs. Every inmate also participates in the
Admissions and Orientation program, which
is designed to familiarize inmates with the
facility, including the rules and procedures,
typical schedules for operations and visiting,
and available programs and services ofered.
Unit Management Team staf will interview
the inmate to determine and assign any
required programs (e.g., the GED program6).
Te automated workfow for the collection
and documentation of this assessment
uses a BOP application called Insight. Te
Unit Management Team will also make
recommendations for additional programs
as necessary. Te inmate will also receive
a work assignment, provided that he or she
is physically able to work.7 As part of this
assessment, the inmate will be formally
advised of his or her risk level under the FSA.
Inmates may grieve their FSA eligibility using
existing administrative remedy procedures.
Based on the interviews with the inmate,
as well as input from mental health staf
and information in the inmate’s sentencing

documentation, the Unit Management
Team will provide the inmate with an
initial Needs Plan. Te inmate will review
this individualized plan to ensure he or
she understands what is either required
or is being recommended. BOP’s Insight
application is used to create the plan.
Te inmate may also undergo additional tests
and screening. Educational staf will screen
inmates for disabilities and will use tests to
determine recommended courses for those
inmates that need literacy programming.
BOP provides a variety of educational
opportunities to the inmates within its 122
institutions.
Under the FSA, BOP is required to
“incorporate a dyslexia screening program
into the System,”8 screen inmates for dyslexia,
and “incorporate programs designed to treat
dyslexia into the evidence-based recidivism
reduction programs or productive activities.”9
As described earlier in this report, BOP has
expanded current requirements for assessing
and accommodating disabilities to ensure a
strong screening process for dyslexia.
Te BOP also performs additional assessment
screenings. A clinical psychologist interviews
inmates requiring mental health treatment.
Te psychologist will then develop an
appropriate treatment plan and communicate
this plan to the inmate’s case manager. A
Drug Treatment Specialist screens inmates
requiring drug treatment. Each inmate
is then interviewed by a Drug Treatment
Coordinator (who is a clinical psychologist)
to determine if the inmate should receive
residential drug treatment. Sex ofenders are
referred for appropriate treatment.10

73
---------,0.-------------

Inmates’ needs are reassessed every six
months (or every three months once the
inmate is within a year of release) and
documented using the Insight application.
Staf will meet individually with the inmate
and review the prior Needs Plan, document
the inmate’s progress, and update the Needs
Plan. Tis assessment includes feedback
solicited directly from relevant program
areas including Health Services, Education,
Religious Services, and Psychology. Unit
Management Team staf determine if an

inmate is successfully participating in
assigned programs, has good work reviews,
and is otherwise satisfying the terms of his or
her sentence (such as payment of fnes and
restitution). Staf will also assess the inmate’s
institution adjustment (e.g., whether he or she
has engaged in misconduct) and any personal
issues (e.g., lack of family contacts, etc.). If
unit management staf determine additional
programs are required or recommended,
they will describe the programs(s) to the
inmate and, in some cases, incentivize
participation in them. Inmates may choose

Attorney General William P. Barr, U.S. Senators Lindsey Graham and Tim Scott of South Carolina, BOP
Acting Director Hugh J. Hurwitz, and others observe a UNICOR clothing and textile facility at Federal
Correctional Institution (FCI) Edgefeld.

74
Q
,----------

to enroll in these programs or may decline
to do so.11
As inmates successfully complete programs
and succeed in addressing their assigned
programming needs, their progress is
captured in the BOP’s inmate management
system and in Insight as part of an educational
transcript. If an inmate has no remaining
needs to address, he or she can continue to
participate in productive activities in order
to continue improving his or her reentry
outcomes. A collaborative process with the
inmate throughout the needs assessment
process is optimal. Te inmate’s Needs Plan
is modifed as the inmate addresses his or her
needs by successfully completing programs
or demonstrates new needs areas.
2. Currently-Admitted Inmates
Under the FSA, inmates currently serving
a term of imprisonment must also have
a recidivism risk level assigned and their
needs assessed using the Risk and Needs
Assessment System. BOP subject-matter
experts from DSCC, the Correctional
Programs Division case management staf,
and the Ofce of Information Technology
(OIT) are conducting an initial review of
all current inmates to determine whether
any are ineligible under the FSA’s statutory
bar based on the inmate’s conviction.12 If
the exclusion process can be automated,
the BOP’s DSCC will work collaboratively
with the BOP’s Ofce of General Counsel
staf to conduct the review of all inmates,13
prioritizing assignment based on projected
release dates. Once those inmates are
assigned a recidivism risk level, the inmate’s
Unit Management Team will schedule them

for an initial needs assessment, as required
by the FSA, to determine what additional
programming, including evidence-based
recidivism reduction programs and
productive activities, would beneft the
inmate.
BOP will reassess the needs of existing
inmates on an ongoing basis in the same
manner as described for newly-committed
inmates.14 BOP ordinarily reviews and
assesses all inmates every six months. Te
inmate’s Unit Management Team reviews
the inmate’s programming and needs,
as well as conduct, by relying on several
sources of information, including a personal
interview with the inmate and input from
various departments in the facility, including
Education, Psychology, Health Services, and
Religious Services. Te Unit Management
Team performs the reassessment of an
inmate’s risk level.
B. Enhancing the Needs
Assessment Process
BOP is committed to enhancing its needs
assessment process to incorporate additional
objective tools and data to better capture
the inmate’s individualized needs and refne
program and productive activity assignment
to address those needs. For example, BOP has
held preliminary discussions with the United
States Probation Ofce about supplementing
the Presentence Investigation Report15 with
more detailed education information about
learning needs. Similarly, and as described
above, BOP is implementing an enhanced
screening process for dyslexia.
BOP is engaging an outside research
contracting partner to assist with FSA

75
---------,0.-------------

implementation in two areas: 1) program
evaluation and 2) design of an enhanced
needs assessment tool. Te FSA requires
BOP to expand available evidencebased recidivism reduction program and
productive activities, including those
ofered by external organizations.16 Te
IRC suggested the Department consider
implementing a process for analyzing new
programs that might qualify as evidencedbased, and cited a model used in the United
Kingdom as one example of such process.
Following that recommendation, the BOP is
forming a working group composed of NIJ
researchers, BOP researchers and program
staf, and the outside research contracting
partner to develop and publish a standardized
process and framework to review and analyze
requests from external entities for program
submission. Te working group will then
serve as a clearinghouse by providing
research expertise to determine whether
a proposed program satisfes the FSA
defnition for an evidence-based recidivism
reduction program, including that it “has
been shown by empirical evidence to
reduce recidivism or is based on research
indicating that it is likely to be efective in
reducing recidivism.”17 Any program or
activity approved for consideration under
the FSA will be assessed in light of security
procedures, BOP protocols, BOP mission
requirements, practicality, and resource
availability.

will include NIJ working closely with the
BOP to perform independent validations
of PATTERN and scientifcally rigorous
evaluations of BOP programs, including
programs designated under the FSA as
“evidence-based
recidivism
reduction
programs.”
Research and evaluations
will be conducted consistent with ethical,
legal, security, and operational concerns.
In addition, BOP, in close consultation with
the Independent Review Committee (IRC)
will conduct research on prevailing best
practices and tools in the area of needs
assessment, particularly as it relates to
corrections and recidivism. Te research
will include consideration of existing
assessment tools that may be available for
implementation by BOP. BOP will also
coordinate meetings and briefngs with
various state and international correctional
systems, including meetings with staf of
large correctional systems in the United
States. Briefngs will include discussions
with program staf, researchers, and
correctional management regarding the
available tools in use for needs assessment
and the efectiveness of the tools.
Consistent
with
the
Department’s
commitment to engage in continuous
dialogue to improve the Risk and Needs
Assessment System, BOP will partner
with NIJ to host a meeting in August 2019
that will focus on needs assessments and
how they are used in correctional settings.
Invitees include research directors from
Departments of Corrections, reentry and
programming staf, researchers, and other
relevant partners. Tis meeting will serve
as a platform for BOP’s continued work

Te Department, specifcally the National
Institute of Justice (NIJ), is committed
to sponsoring a long-term research and
evaluation agenda that will support the
full and sound implementation of the
FSA in the years to come. Tis agenda

-----~o~----76

in this area. Additionally, BOP will host a
three-day symposium on needs assessment
in October 2019. Invitees will include
1) practitioners with expertise on needs
assessment from state correctional systems
that conduct assessments at all prisons in
their system; 2) academic researchers with
demonstrated experience working on needs
assessment systems and an understanding
of diversity issues in implementation; and
3) staf representatives from federal partners
including NIJ and the National Institute of
Corrections. Tis symposium, moderated by
BOP program staf, will include presentations
from the diferent corrections systems and
discussions with researchers on how to apply
the material gleaned from the corrections
presentations.
Following this meeting, and in furtherance
of its research, the BOP will issue a request
for information (RFI) regarding potential
inclusions or sources of data for the needs
assessment process. Te purpose of this
RFI is to provide an opportunity for
advocacy groups, outside researchers, and
any individuals to share input and suggest
additions to the Risk and Needs Assessment
System process, in regards to needs
assessment. Te RFI will be published for 45
days to allow for a robust comment period.
Te information gleaned from the overall
research and the RFI solicitation will be used
to design an enhanced needs assessment tool
and process. BOP anticipates this overall
engagement and collection process to take
six months (culminating by March 2020).
Te BOP will screen all inmates for risk
using the PATTERN tool described in
Chapter 3.18 In addition, the BOP will screen
all inmates for needs using BOP’s current

needs assessment system. At the same time,
as described herein, the Department will
begin immediate work building a more
comprehensive needs assessment system.
While the research is being conducted, BOP
will work with NIJ to explore how additional
existing data can be incorporated into the
needs process and evaluate the impact of this
additional data on the accuracy of the riskscoring tool. In September, Department
staf, members of the IRC, and an NIJ expert
will travel to Ottawa, Canada to meet with
the Correctional Service of Canada to learn
about the development of its automated
needs assessment system. Meanwhile, the
BOP will continue to leverage its existing
processes to assess, review, and address
inmate needs to ensure that other aspects
of the FSA are being carried out. As the
assessment process is enhanced, changes will
be incorporated into the current assessment
and all inmates will continue to be reassessed
using the updated system.
In the next several months, the BOP will add
several additional data items into the current
needs assessment process. Te BOP is aware
of the incredibly high rates of trauma and
the associated impact of this trauma on the
inmate population. By September 30, 2019,
the BOP will add the Adverse Childhood
Experience questionnaire to its existing
screening process to better incorporate this
need into the Risk and Needs Assessment
System. Also by September 30, 2019, the
BOP Reentry Services and Correctional
Programs subject matter experts will meet
with the IRC to present on the current
needs assessment process and discuss ideas
for expanding upon the existing system.
Additionally, the Correctional Programs
Division will review the current Insight
system to determine if more discrete (non-

77
---------,0.-------------

narrative) data regarding job performance
and completion of specifc programs can be
captured to enhance evaluation of inmate
risk and needs.

the corrections cycle; 4) the evidence-based
recidivism reduction programs (both those
currently used and those that may be used)
that will best minimize the risk that the
prisoner will recidivate; 5) what measures
can be used to indicate progress and
regression; and 6) if and how to automate the
assignment of programs based on the results
of the assessment while allowing for sound
judgment and decision-making.

Te BOP’s overall research results will
then be used to drive the design of a needs
assessment tool that best addresses the
processes and composition of the federal
inmate population. Te BOP will also engage
the BOP’s National Union in the design and
development of the system to ensure that
the ideas and concerns of staf, particularly
case management and program staf, are
incorporated into the design.

Te prototype system addressing the items
above is expected to be available for testing
by the second quarter of 2020. Te testing
of the system will involve training BOP
staf in the use of the tools or instruments,
assessing the system’s efectiveness by
evaluating available metrics such as prison
misconduct or recidivism, and evaluating
whether the tools or instruments are being
applied in a consistent manner. Depending
on the complexity of the tools or needs
instruments, the BOP would either use
electronic questionnaires or develop sofware

Te BOP anticipates the design of the system
to encompass several key requirements
including, but not limited to, determination
of: 1) the static and dynamic factors which
identify the inmate’s needs; 2) those tools
and instruments which can best identify
an inmate’s specifc criminogenic and noncriminogenic needs; 3) when those tools
and instruments are best administered in

Enhancing the Needs Assessment Process
• Implement Enhanced Dyslexia Screening
• Engage Outside Research Partner
• Conduct Research and Issue RFI
• Hold Needs Assessment Meeting
• Convene BOP Symposium
• Review Current Insight System
• Develop New Needs Tool
• Pilot New Needs Tool
• Execute New Tool/Process Across BOP
• Train all BOP Staff
• Engage with BOP’s National Union

-----~o~----78

to integrate the tools and instruments into
Insight.
To collect a sufcient amount of data for
evaluation purposes, identify issues and
trends, and conduct a preliminary assessment
of efectiveness, BOP plans to test the system
using eight facilities, across four security
levels and six regions, for twelve months.
Afer testing has been conducted and the
results validated, BOP will implement the
fnal system and begin rolling it out to the
remaining 114 BOP sites.
During this time, any required modifcations
would be made to the system and additional
training provided to staf.
All case
management staf, programs staf, and
correctional management will be trained in
the new instruments and tools. Te BOP
will develop training materials, conduct inperson training, and leverage webinars and
a train-the-trainer approach as part of the
deployment strategy.
C. Implementing the Earned Time
Credit System
Under the FSA, an eligible inmate may earn
time credits for successfully completing approved evidence-based recidivism reduction
programs or productive activities.19 No later than January 2020, a list of approved evidence-based recidivism reduction programs
and productive activities will be posted on
the BOP’s website. As required by the Act,
an inmate cannot earn time credits if he or
she is serving a sentence for a disqualifying
ofense or has a disqualifying prior conviction.20
An eligible inmate may “earn 10 days of
time credits for every 30 days of successful

participation in evidence-based recidivism
reduction programming.”21 Additionally,
an inmate at a minimum or low risk for
recidivating may “earn an additional 5 days
of time credits for every 30 days of successful
participation in evidence-based recidivism
reduction programming or productive
activities.”22
BOP will track and award time credits
based on actual work and program activities
completed. BOP will ensure that every
evidence-based
recidivism
reduction
program has an approved standardized
curriculum, including duration and hours.
To track participation and completion,
assignment codes for each approved
program will be created in the BOP’s inmate
management system. BOP will automate the
calculation of the amount of programming
activities completed. Unit Management
Team staf will ensure that the inmate did
in fact successfully complete the program.23
BOP will also track inmate participation in
productive activities.
Tere are clear instances when inmates
would not be able to participate in evidencebased recidivism reduction programs or
productive activities. Such circumstances
include when inmates are: placed in a Special
Housing Unit; in designation status outside
the institution (e.g. for an outside medical
trip, an escorted trip, etc.); in the custody of
another jurisdiction (e.g. on state or federal
writ; transfer to state custody for service
of sentence; transfer to ICE, etc.); under
medical/mental health/psychiatric holds; in
detention as a material witness or for civil
contempt; under Adam Walsh or other civil
commitment; and refusing to participate in
programs.

79
---------,0.-------------

Tere are additional factors that may warrant
limitations on the earning or application of
time credits, such as being investigated for
ongoing criminal conduct while in custody
or while in the community in pre-release
custody. Additionally, an inmate’s recidivism
risk score could increase if he or she engages
in misconduct - for example, if the inmate
assaults a corrections ofcer or other inmate,
uses a contraband cell phone, or attempts to
escape.

Tese individuals provided feedback that
inmates ofen spend the majority of their
time in recreation, rather than on activities
that will prepare them for a more successful
transition. Inmates who participated in
activities, such as fnancial responsibility and
parenting, were ofen ridiculed or heckled.
Tose at the RRC encouraged the Department
to incentivize and promote participation in
productive activities centered on parenting,
education, and fnancial responsibility and to
decrease emphasis on recreation. Tis would
help change the culture within prisons and
remove pressure from those seeking to better
themselves. Te Department recognizes the
value that individuals in RRCs and those
who were recently released and achieved a
successful transition may contribute to this
process and intend to seek input from this
group in developing the list of productive
activities.

As discussed previously, case management
staf will be trained on which programs
and productive activities are approved for
earning of time credits and on how to use the
programs as part of their assessments using
the BOP’s Insight application.
Department staf toured an RRC and met
with recently-released individuals who had
participated in BOP programs and activities.

Inmates participating in the UNICOR program.

-----~o~----80

To ensure that both inmates and staf
understand the award and suspension of
time credits and to ensure that stakeholders
have an opportunity to provide feedback on
this important issue, BOP will update federal
regulations and BOP policy to describe
the time credit methodology. To promote
transparency in the process, the BOP will
update and publish Program Statement
5100.08, Inmate Security and Custody
Classifcation Manual.
D. Automation of Workfow
Currently, the BOP staf assign an
inmate’s security designation and custody
classifcation scores using an automated
workfow, where using information
technology replaces manual and paperbased processes. Tis automation expedites
the processing of tasks and eliminates errors.
For these reasons, the BOP also plans to
calculate, track, and report the recidivism risk
level using an integrated automated process.
Accordingly, BOP must make changes to the
inmate management system and additional
sofware applications to perform the risk
level calculation, present the score in other
system displays, and calculate an inmate’s
time credits. BOP anticipates that it will take
approximately eight months to make the
required changes. While the scoring process
is being enhanced through automation, BOP
will manually score recidivism risk levels so
there is no delay in fully implementing the
System. Te BOP’s OIT will coordinate
with the Ofce of Research and Evaluation
(ORE) and the DSCC to determine how
to integrate and automate the data points
required to be included in the PATTERN
risk calculus, as well as how to score and

classify the inmate with the Risk and Needs
Assessment tool. OIT staf will modify
existing programs by frst, designing test
programs based on the system requirements,
validating their performance, seeking user
input, verifying data inputs and outputs
are as expected, and deploying the sofware
changes to production.
OIT estimates that approximately 30
programs will need to be modifed to capture
and display the required data for the risk tool
and ongoing workfow management. For
example, BOP’s discipline module tracks the
incidents, sanctions, and history of inmate
misconduct. With the addition of the FSA,
the BOP will need to modify this module to
account for sanctions related to time credits,
as well as the FSA requirement to restore time
credits. Discipline sanctions likewise impact
an inmate’s projected release date and the
new projected pre-release custody eligibility
date; sanctions may also impact the inmate’s
risk score. Tus, the BOP sofware programs
that are afected by the FSA must be updated
and integrated with each other.
OIT will focus frst on implementing the
programs associated with the inmates’
risk and needs classifcation, followed by
implementation of the time credit system,
and then application integration.24 OIT
will require additional personnel and
contractor resources to assist in this efort.
Application integration is the practice of
sharing processes and data among various
information technology applications. It
will take approximately 24 months to add
appropriate sofware logic to each of these
programs. Te frst eight months will focus
on launching programs related to the risk
tool PATTERN. Te other program updates

81
---------,0.-------------

and changes will be addressed at the same
time. Ultimately, these changes will allow
BOP staf to more efciently and accurately
compute, assign, and track the recidivism
risk level for each inmate.

provide comments. Afer the comment
period closes, the issuing agency reviews
and considers the submitted comments
on the proposed rule. Te next step
in the process is to draf a fnal rule. If
appropriate, the issuing agency may
modify the proposed rule to address
public feedback and comments in the fnal
rule. Te fnal rule is prepared and once
again reviewed by OIRA if it is deemed to
have signifcant economic impact or is a
signifcant policy matter. Te fnal rule
is published a second time in the Federal
Register and may go into efect no earlier
than 30 days afer publication.26

E. Policy Implementation
Te FSA also requires BOP to develop new
policies to address a number of requirements
in the Act. Specifcally, the Director of the
Bureau of Prisons is required to “develop
additional policies to provide appropriate
incentives for successful participation and
completion of evidence-based recidivism
reduction programming.”25 Additionally,
the BOP has a signifcant number of federal
regulations and policies that require update
or modifcation to implement the FSA and
to provide clear and timely communication
to all persons afected by the changes.

ii. Policies: To develop new policy or
update existing policy, the applicable
BOP program area reviews and modifes
existing policy or develops new policy
language. A program area is the specifc
ofce or Division responsible for oversight
and enforcement of the function, service,
or program within the BOP. Te BOP’s
National Policy and Information
Management (NPIM) staf review, format,
and distribute draf policies internally to
all Divisions and feld sites, as well as to the
BOP National Union, for comment. If the
program area concurs with the comments,
NPIM will incorporate the recommended
changes and prepare a fnal draf. Te
NPIM staf prepare a fnal version for
the BOP Director’s signature and provide
a copy of the signed policy to the BOP
Labor Management Relation Ofce for
review by the BOP National Union. If the
National Union does not formally invoke
negotiation27 on the policy, the policy
is published both internally and, if not

Tere are many steps involved in the
formulation of extensive and complete
federal regulations and BOP policy. A highlevel outline of those steps follows.
i. Regulations: Te issuance of federal
regulations frst requires the development
of draf language known as the “proposed
rule.” Te proposed rule is reviewed for
legal compliance by the component and
the Department of Justice. It is then sent
for review to the Ofce of Information
and Regulatory Afairs (OIRA) within
the Ofce of Management and Budget.
Once the proposed rule is approved, it
is published in the Federal Register for
public comment for 30-60 days as a Notice
of Proposed Rulemaking. Members of
the public and stakeholders will be able
to review the proposed rule in full and

-----~o~----82

law enforcement sensitive, on the BOP’s
public website. Any applicable Program
Review (BOP’s internal compliance and
audit process) guidelines are updated
to ensure staf are following the new
or updated policy beginning 30 days
afer publication. If the National Union
formally invokes negotiation, the policy
is put on hold until Management and the
Union agree on the fnal policy.
BOP estimates that approximately three
regulations and 11 policies require revision
to comply with the FSA’s risk and needs
assessment system requirements. Tere
also may be a need to develop new policies
to address the FSA. In general, policy
formation and implementation that afect
bargaining unit staf working conditions take
approximately six months, but the process
may be longer on key issues. Bargaining
unit staf are members of and represented
by a labor organization (“union”), and
coordination with the BOP’s National
Union is required per the BOP’s collective
bargaining agreement.
Tis work has
already begun and will continue until all of
the appropriate policies and regulations are
updated to refect the changes needed to
implement the Risk and Needs Assessment
System and to obtain the necessary approvals
to publish these documents.

II. Timeline for Initial Implementation
of the Risk and Needs Assessment
System
Tere are many tasks that need to be
accomplished before the new Risk and Needs
Assessment System is fully operational. Te
keys to implementing the FSA successfully
are meshing both new and current work
processes, automating the information
technology systems used to track and report
inmate data, establishing the regulations
and policies needed to convey the updated
organizational requirements and practices,
and, fnally, training all BOP staf.
BOP has already begun to implement
the FSA. Many BOP policies and some
regulations have been updated to include
language referencing the Act.28 Information
technology staf are mapping out a strategy
that will incorporate the changes needed in
SENTRY, Insight, and other BOP systems. In
addition, BOP senior leadership are sharing
information about the FSA and its impacts
on BOP.
However, crucial work remains to implement
the Risk and Needs Assessment. Te image
below tracks the major tasks that are to
be accomplished with estimated dates for
delivery of the product between July 19,
2019 and January 2020.29 Importantly, the
Department will meet the FSA’s requirement
to assess all BOP prisoners by January 2020.

83
---------,0.-------------

INITIAL IMPLEMENTATION OF THE
RISK AND NEEDS ASSESSMENT SYSTEM
JULY 2019 - JANUARY 2020
DEVELOP POLICIES

90 DAYS

POLICY NEGOTIATION

180 DAYS

DRAFT REGULATIONS

90 DAYS

FINAL RULE
PUBLICATION

120 DAYS

AUTOMATION OF WORK
FLOW

150 DAYS

DEVELOP AND BEGIN
TRAINING

120 DAYS

INITIAL RISK
ASSIGNMENT

180 DAYS

0

60

120

III. Re-Validating the Risk and
Needs Assessment System

180

as an efective predictor of recidivism over
the inmates’ subsequent three-year period
in the community. In a research setting, a
researcher would normally gather data to
re-validate a risk and needs assessment tool
three years afer a large group of inmates
were released from custody. Following this
schedule, PATTERN and any new items
would not be formally re-validated until
2024. To meet the annual FSA re-validation
requirement, one-year recidivism data,
instead of three-year, will be used to obtain
results. Tis will cause an incongruence in the
validation and re-validation processes, but

As mandated by the FSA, the RNA tool must
be reviewed and re-validated on an annual
basis.30 Re-validation will allow for any
needed adjustment of the tool’s algorithm
to ensure that the tool is still reliable given
any changes in the prison population or
other variables over time. An annual revalidation timeline, however, presents a
potential research and theoretical challenge.
PATTERN was developed using a seven-year
dataset of BOP releases, and it was validated

-----~o~----84

the results can serve as an interim estimate
of PATTERN’s predictive performance.
BOP’s ORE and the NIJ will work together
to re-validate PATTERN on an annual basis.
Until enough time passes that PATTERN
can be validated on three years of postrelease data, ORE and NIJ will annually rerun the analyses that developed PATTERN,
adding one additional year of recidivism
data. Consistent with the FSA, the Attorney
General will review and consider any
modifcations. Over time, the researchers
will continue to assess PATTERN, gather
external advice and recommendations, and
use the time to consider further adjustments
to PATTERN.
In addition to re-validating the System
as described previously, the Department
will look for opportunities to improve the
System through an ongoing dialogue with
key stakeholders, by working closely with the
IRC, and by considering new research and
potential sources of data. As the System is
updated and improved, subsequent versions
will be re-validated, as discussed previously.

IV. Training to Implement the Risk
and Needs Assessment System
Following the publication of the System,
all relevant BOP staf members in every
institution, Regional Ofce, and Central
Ofce will receive training on implementing
the new System. Te training of BOP
staf will be informed by careful review of
implementation research and use of best
practices in training and quality assurance.
Te level and type of training will vary based
on the role of the BOP staf member.

Foundational training will be sufcient for
many staf in the institutions, as well as in the
Regional and Central Ofces. Tis training
will provide an overview of the program,
as well as how to score inmates in the new
system and use the assessment results to
identify appropriate programs, activities,
and services for inmates.
Advanced training will be required for
the Unit Management Teams and other
appropriate staf at institutions, as they will
have primary responsibility for assessing
inmates and entering information into BOP
automated systems relevant to the System.
Additionally, on-the-job training will be
required to ensure that the information
entered is accurate and consistent. Unit
Management Teams and program staf
will receive training on the administration
and scoring of the needs assessment of the
System. Education and Psychology staf will
receive training on the needs assessment
procedures that will be implemented. In
addition, the DSCC staf and BOP case
management staf will receive training on
how to score the risk assessment. Finally,
Regional and Central Ofce program stafs
will be trained on how to oversee and review
institutions’ implementation of the System
and processes. Tis oversight and review is
key to avoiding the misuse of the tools.
To facilitate this required training, the
BOP will utilize a team of subject matter
experts from the Reentry Services
Division; Correctional Programs Division;
Information, Policy and Public Afairs
Division; and Ofce of General Counsel.
Tis team will be primarily responsible

85
---------,0.-------------

BOP staf members at a correctional facility.

for developing training and instructional
materials for the System and processes. It will
take approximately two months to develop
the foundational initial orientation training
and an additional two months to develop the
more intensive, advanced training. Training
will initially focus on staf conducting
recidivism risk assessments and performing
initial assessments for all inmates. BOP has
already provided broad briefngs on the
provisions of the FSA to staf who perform
these tasks. And the additional training will
focus on more detailed requirements of the
Risk and Needs Assessment System.

interaction in small group settings with
computer access. Using a “train-the-trainer”
model, BOP subject-matter experts will train
selected employees about the System, who
will in turn serve as local experts to train
others to use the new system. It is anticipated
the all impacted staf will receive training
during calendar year 2020.
As the System is improved, as discussed
earlier, training will be adjusted to account
for changes to the System. As capabilities
and processes are enhanced and to ensure
refresher training is consistent, the System
training will become part of institution
orientation for new staf. Additionally, BOP
will incorporate Risk and Needs Assessment
System training as a core module of Annual
Refresher Training, which is required for all
staf. Additionally, Central Ofce subject
matter experts and Regional Ofce staf will
maintain oversight of the implementation and
can provide additional training as needed.
In order to provide continuing support
to these training eforts, BOP may need

Education about the FSA for all staf will be
delivered via a variety of platforms: online
via the BOP’s Learning Management System;
using collaboration tools such as Adobe
Connect; using videoconferencing; via
communication materials published on the
agency intranet; and in-person instruction.
Te training will be delivered by the end of
2019. Te intensive training will be provided
through in-person training and will include

-----0
86

additional resources to develop, coordinate,
and provide the required training.

V. Assessing the Implementation of
the Risk and Needs Assessment
System
Te Director of the BOP has 180 days to
1) “implement and complete the initial intake
risk and needs assessment for each prisoner”
and “begin to assign prisoners to appropriate
evidence-based
recidivism
reduction
programs;” 2) “begin to expand the efective
evidence-based
recidivism
reduction
programs and productive activities;” and 3)
“begin to implement the other risk and needs
assessment tools necessary to implement
the risk and needs assessment system over
time.”31 To ensure the implementation
of the System is completed in an efcient
and efective manner, the BOP will use
established internal processes, described
below, to provide appropriate checks on
implementation.
Te BOP’s Program Review Division
provides oversight, guidance, and direction
to BOP facilities and staf to ensure that
policies are being followed and that
programs are performing as intended.
Tis is accomplished through monitoring
of program areas, facilitating guideline
assessments, and analyzing trends and
statistical data.
To test the competency and strength of
its programs, the BOP subjects each one
to a thorough examination by trained
reviewers who are subject matter experts
in the program area being reviewed. Tese
independent reviews examine compliance
with BOP policy, laws, rules, and regulations.

In addition, they examine the area’s efciency
of operations and efectiveness in achieving
program results. Tis process helps ensure
that BOP programs are operating within
policy and are free of fraud, waste, abuse,
mismanagement, and illegal activities.
Te Inmate Classifcation Workgroup
is responsible for assessing the overall
efectiveness of the BOP’s inmate
classifcation system.
Te Workgroup
reports its fndings to the Director and
the agency’s senior leadership team when
changes to the BOP’s system are warranted.
Te Inmate Classifcation Workgroup will
propose changes to the Inmate Security
Designation and Custody Classifcation
Policy to implement the changes to the
System. In addition, the Workgroup will
outline steps for a feld test of the policy draf.
Following the feld test of the draf policy,
the Workgroup will seek fnal approval
from senior leadership for a revised policy,
5100.08, the Inmate Security and Custody
Classifcation Manual. Tis Workgroup will
additionally take the lead in reviewing and
assessing the implementation of the System.

VI. Connecting the Risk and Needs
Assessment System to Programming
As described above, each inmate will
be assessed under the Risk and Needs
Assessment System. Tis assessment will
be used to determine the type and amount
of evidence-based recidivism reduction
programming and productive activities that
are appropriate for the inmate. Te outcome
of the needs assessment component of the
system will be an individualized “Needs
Plan” for each inmate designed to reduce his
or her risk of recidivism. Te Needs Plan

87
---------,0.-------------

will identify areas the inmate must address
and identify the evidence-based recidivism
reduction programs available to address
those needs. As the inmate successfully
addresses a need, or new needs are identifed,
the inmate’s Needs Plan will be modifed,
and the inmate’s risk score will be adjusted
as appropriate.

In accordance with the FSA, medium and
high risk inmates will receive priority
for evidence-based recidivism reduction
programs. Inmates at medium and high
risk levels may volunteer to participate
in productive activities. Inmates who are
assigned minimum or low risk with no
remaining needs on their Needs Plan will
be assigned to participate in productive
activities to remain productive. Assignment
to evidence-based recidivism reduction
programs will receive priority over
assignment to productive activities.

All evidence-based recidivism reduction
programs, including current BOP programs,
will be vetted, approved, and cataloged.
Indeed, BOP’s current programs have
evolved and improved over the years and
additional data is available for consideration.
All programs will be assessed on the
quality of their curriculum, data regarding
improvement outcomes, and the ability
of the BOP to ensure the programs can be
provided as intended.32

No later than January 2020, a list of the approved evidence-based recidivism reduction
programs and productive activities will be
posted on the BOP’s website.

-----~o~----88

VII. Communication to Inmates and
Stakeholders

Engage Inmates
• Town Halls
• Electronic Bulletin Boards
• Unit Management Team Meetings

Engage Stakeholders
• Risk and Needs Assessment System Listening Sessions
• Request for Information
• Publication of Policies and Regulations

Engage Staff
• Webinars and Intranet Articles
• In-person and Web-based Training
• Meetings with National Union

Because the requirements of the FSA are
varied and complex, it is important for
inmates and staf to be fully briefed on
the numerous changes required under
the law and being implemented at BOP
facilities. Accordingly, BOP will publish
key announcements on the main page of
its public website and provide information
and updates on an ongoing basis on its new
FSA resource page. Internally, in addition
to the training described above, BOP staf
will be briefed via video conferences, web
collaboration, and the BOP’s intranet site to
remain current on all FSA-related changes
and improvements.

Inmates’ Unit Management Teams will be
available for inmates to ask questions and
ofer feedback regarding the FSA. Inmates
will be advised of changes and requirements
via meetings with their Unit Management
Teams; by the Warden and local leadership
via institution Town Halls; and directly via
announcements posted to inmate bulletin
boards (including through TRULINCS, an
application that provides inmates with the
ability to view electronic notices). Members
of the public may visit the BOP’s website33 to
receive additional information.

89
---------,0.-------------

Attorney General William P. Barr, U.S. Senator Lindsey Graham, and BOP Acting Director Hugh J. Hurwitz
meet with an inmate at FCI Edgefeld.

As mentioned previously, BOP will hold listening sessions to gather additional feedback
from stakeholders and the public about the
implementation of the Risk and Needs Assessment System. Tese sessions will pro-

vide BOP with direct input from interested
and impacted parties about the strengths
and challenges of the System as BOP fully
implements the System in its facilities.

-----~o~----90

NOTES
1

18 U.S.C. § 3632(a).

2

18 U.S.C. § 3631(h)(1).

Te Statement of Reasons includes explanatory information about the sentence including any disagreements with the Presentence
Investigation Report and information about
how the sentence was determined (e.g. sentencing guideline departures). Some of this
information, such as mental capacity, drug
dependency, etc. informs the designation
process. See https://www.uscourts.gov/sites/
default/fles/ao245sor.pdf .

3

Te applicable United States Attorney’s Offce provides guidance to BOP if it believes
security risks exist with housing defendants
with specifc individuals (e.g. co-defendants
or inmate witnesses).

4

Tis determination is separate and distinct
from BOP’s security level classifcations,
which also include Minimum, Low, Medium and High. An inmate may be classifed
as Low security and designated to a commensurate facility because he or she poses
little risk of fight or institutional misconduct, while at the same time have a high risk
of recidivism. For example, an inmate with
an extensive criminal history for defrauding
senior citizens might have a high risk of recidivism, but a low security level because of
a clear disciplinary record during prior periods of confnement.
5

One exception to voluntary programs is the
General Educational Development (GED),
which is required by statute to be provided
to any inmate without a prior high school diploma or GED. See 18 USC § 3624(f).
6

Tis requirement confers several benefts to inmates and institutional operations,
including development of vocational and
pro-social skills, the earning of money to repay criminal fnes and victim restitution and
purchase desired commissary items, and the
reduction of inmate idleness.

7

8

18 U.S.C. § 3632(h)(1).

9

18 U.S.C. § 3632(h)(2).

Such persons are usually designated to one
of the nine facilities that ofer the BOP’s Sex
Ofender Management Program.
10

Note that failure to participate in a mandatory literacy program will result in an inmate failing to earn the maximum amount of
good conduct time under 18 USC § 3624(b).

11

12

See 18 U.S.C. § 3632(d)(4)(D).

If the exclusion process cannot be automated, rosters of inmates will need to be sent
from the BOP Central Ofce to each feld
site for case management staf to review and
discuss at each inmate’s Unit Management
Team meeting.

13

14

18 U.S.C § 3621(h).

Prior to sentencing, the probation ofcer
will interview the defendant and conduct an
investigation to provide the sentencing judge
with pertinent information relevant for sentencing. Tis information is captured in a
Presentence Investigation Report and includes details of the defendant’s family history, community ties, education background,
employment history and physical and mental health. See https://www.txs.uscourts.gov/
presentence-investigation.
15

91
---------,0.-------------

16

18 U.S.C. § 3621(h)(5).

17

18 U.S.C. § 3635(3)(A).

Te BOP’s case management application,
Insight, is used at all BOP institutions. Te
BOP uses a second case management application called the R3M system where Residential Reentry Management staf to communicate with Residential Reentry Center
contract staf to manage inmates in pre-release custody. Insight currently includes a
referral module that will be more tightly integrated into R3M.
24

Tis includes any modifcations that might
occur following the study period.
18

19

18 U.S.C. § 3632(d)(4).

For example, 18 U.S.C. § 3632(d)(4)(D)
(li) excludes inmates convicted of “(li) An
ofense described in section 3559(c)(2)(F),
for which the ofender was sentenced to a
term of imprisonment of more than 1 year,
if the ofender has a previous conviction, for
which the ofender served a term of imprisonment of more than 1 year, for a Federal or
State ofense, by whatever designation and
wherever committed, consisting of murder
(as described in section 1111), voluntary
manslaughter (as described in section 1112),
assault with intent to commit murder (as described in section 113(a)), aggravated sexual
abuse and sexual abuse (as described in sections 2241 and 2242), abusive sexual contact
(as described in sections 2244(a)(1) and (a)
(2)), kidnapping (as described in chapter 55),
carjacking (as described in section 2119), arson (as described in section 844(f)(3), (h),
or (i)), or terrorism (as described in chapter
113B).”
20

21

18 U.S.C. § 3632(d)(4)(i).

22

18 U.S.C. § 3632(d)(4)(ii).

25

18 U.S.C. § 3632(d)(3).

For a detailed discussion of the rulemaking process, see “A Guide to the Rulemaking
Process,” https://www.federalregister.gov/
uploads/2011/01/the_rulemaking_process.
pdf (last visited July 3, 2019).

26

Master Agreement, Federal Bureau of Prisons and Council of Prison Locals, American Federation of Government Employees,
July 21, 2014 – July 20, 2017, https://www.
afgelocal1034.org/ewExternalFiles/2014%20
New%20Master%20Agreement.pdf (last visited July 3, 2019).

27

BOP has already updated policies or memoranda that address: Recidivism-Based Partnerships, Secure Firearms Storage, Location
Within 500 Driving Miles of Inmate’s Residence, Compassionate Release/Reduction in
Sentence, Home Confnement, Inmate Identifcation, De-Escalation Training, Juvenile
Solitary Confnement, Dyslexia, and UnitBased Programs. Some of these documents
are currently under review by the Union. See
https://www.bop.gov/PublicInfo/execute/
policysearch.

28

BOP case managers use the Insight application to schedule, develop, and store inmate
assessments. Te Insight application also
integrates with SENTRY (the BOP’s inmate
management system) and the Residential
Reentry Referral Management (R3M) system, used by RRM staf to manage pre-release processing and case management.

23

Te timeframes listed are estimates and
may be afected by funding, resources, or legal requirements.
29

-----~o~----92

30

18 U.S.C. § 3631(b)(4).

31

18 U.S.C. § 3621(h).

Te BOP will use federal partner resources to screen external programs. For example, existing evaluation resources which can
be leveraged include: MITRE (https://www.
mitre.org/research/overview); NIJ’s Crimesolutions.gov (https://www.crimesolutions.
gov/about_starttofnish.aspx); NIJ’s betagov (http://www.betagov.org/index.html);
and GSA’s Ofce of Evaluation Sciences
(https://www.gsa.gov/about-us/organization/office-of-governmentwide-policy/offce-of-evaluation-sciences).

32

33

See www.bop.gov.

93
---------,0.-------------