Dept of Justice, the First Step Act of 2018 - Risk and Needs Assessment System UPDATED, 2020

• Locations: United States of America • Topics: First Step Act

Download original document:
Brief thumbnail

Document text

This text is machine-read, and may contain errors. Check the original document to verify accuracy.

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

The First Step Act of 2018:
Risk and Needs Assessment
System - UPDATE
January 2020

__________________________

Table of Contents
I.

Introduction ...................................................................................... 1

II.

Summary of DOJ’s Efforts Following Publication of the
July 2019 Risk and Needs Assessment Report ....................... 2

III.

Changes to PATTERN Based on Suggestions Received
After the July 2019 Report ............................................................. 7

IV.

Conclusion ...................................................................................... 26
Attachment A: Examples of Scoring Under PATTERN 1.2
(Using General Risk Score) ....................................................... 27
Appendix I: Technical Corrections ......................................... 35
Appendix II: Revised Points Assigned in the PATTERN
Risk Assessment Models ............................................................ 37

___________________________

__________________________

Introduction
The efficient and effective implementation of the First Step Act (FSA) continues to
be a priority for the Department of Justice (DOJ or the Department) and for this
Administration. In July 2019, the Department announced a new Risk and Needs
Assessment System (RNAS or the System). In developing the System, the Department
worked to make the benefits of the FSA as widely available as possible without
compromising predictive reliability. The initial System, which was based on the
information available at the time, was robust, and the Department committed to
improve it, with more time, consultation, data, and research.
In the months that followed, the Department has worked vigorously to fulfill this
commitment. Importantly, we continued to work with the Independent Review
Committee (IRC) and our experts, Drs. Grant Duwe and Zachary Hamilton, to
identify ways to improve the Department’s RNAS, the Prisoner Assessment Tool
Targeting Estimated Risk and Need (PATTERN), while maintaining its high level of
predictability. Moreover, we continued to engage a range of stakeholders and have
listened to their concerns and suggestions. This input has been invaluable as we strive
to ensure the equity and effectiveness of PATTERN.
In this follow-up report, the Department is highlighting the changes made to
PATTERN as a direct result of the input and suggestions on this crucial issue. The
Department has taken the feedback seriously and considered the various and, at times,
competing views presented. In all cases, DOJ strove to engage stakeholders and
implement the most equitable, effective, and predictive tool possible to ensure that the
goals of the FSA are achieved.

________________________
1

__________________________

II.

Summary of DOJ’s Efforts Following Publication of the
July 2019 Risk and Needs Assessment Report

a. Independent Review Committee (IRC) Meetings
The IRC’s input has been critical to the Department’s development and review related
to PATTERN. On October 7, 2019, staff from the Office of the Deputy Attorney
General (ODAG), the National Institute of Justice (NIJ), and the Federal Bureau of
Prisons (BOP) met with the IRC to discuss the IRC’s views and recommendations
regarding PATTERN and potential improvements and advancements for the
subsequent version of the risk tool. Following that meeting, BOP provided NIJ
contractors with additional programmatic and disciplinary infraction data to facilitate
further analysis of PATTERN. In the weeks that followed, NIJ’s consultants spent
more than 100 hours addressing questions and concerns raised by the IRC. While not
all of the IRC’s recommendations were accepted by DOJ, many of them, in addition
to the additional analyses completed by the consultants, helped confirm DOJ’s
confidence in the accuracy of PATTERN.
The Attorney General met with the IRC on November 19, 2019, to discuss proposed
changes to PATTERN, including several refinements suggested by the IRC. The IRC
recommended that the Attorney General implement the changes described herein and
agreed to continue providing advice to the Department on the implementation of
these changes and other issues identified by the IRC regarding the FSA. BOP and NIJ
concurred with this recommendation.

NIJ LISTENING SESSIONS. Antoine Prince Albert of The Leadership Conference on Civil and
Human Rights / Education Fund speaks to David Muhlhausen of NIJ, Antoinette Bacon of ODAG,
Kathleen Hawk Sawyer of BOP, and Hugh Hurwitz of BOP at the listening session on September 11,
2019.

________________________
2

__________________________

b. Stakeholder Listening Sessions and 45-Day Public Comment Period on
PATTERN
On September 10-11, 2019, NIJ hosted two listening sessions in response to the
release of PATTERN. Criminal justice stakeholders, advocates, and interested citizens
were invited to Washington, D.C. to provide comments 1 on DOJ’s implementation of
the FSA and, specifically, the development of PATTERN. Eight stakeholder
organizations provided public comment. An additional eight sets of comments were
provided to NIJ for consideration.

NIJ LISTENING SESSIONS. Laura Mate (left), Jeremiah Mosteller (center), and James Felman
(right) participate in the listening session on September 11, 2019.

Following the release of PATTERN in July, DOJ provided a specific email address
(FirstStepAct@usdoj.gov) for interested members of the public to provide comment
and feedback on the risk tool and FSA Report. NIJ actively monitored the feedback
submitted to the email address—nearly 175 comments and statements were received
during the 45-day public comment period. Additionally, twenty statements were
submitted by external stakeholders to NIJ as part of an invitation to participate in a
third round of listening sessions. Through both written comment and in oral
presentations, stakeholders raised a diverse set of issues and offered suggestions and
opportunities for improvement.

The full set of comments submitted to NIJ can be found here:
https://www.ncjrs.gov/pdffiles1/nij/grants/254142.pdf

________________________

__________________________

ENGAGEMENT. The DOJ delegation poses for a photo with CSC staff.

c. Correctional Service of Canada (CSC) National Headquarters Engagement
On September 17-18, 2019, staff from ODAG, NIJ, BOP, Dr. Patti Butterfield, a
member of the IRC, and Drs. Zachary Hamilton and Angela Hawken, two of NIJ’s
external expert consultants, traveled to the national headquarters of the CSC. CSC is
the Canadian federal government agency responsible for the incarceration and
rehabilitation of convicted criminal offenders sentenced to two years or more. CSC
explained that it has taken decades to achieve its current level of sophistication, and it
continues to be a leader in the advancement and development of risk and needs
assessments. Meeting with CSC provided an important opportunity for DOJ to learn
about the successes and challenges of developing and implementing an innovative risk
and needs assessment system in a large jurisdiction.
For CSC, treatment begins on day one. Importantly, CSC provides cognitivebehavioral therapy to all offenders in their custody starting at the term of their
confinement. CSC delivers such therapy to all offenders regardless of their specific
needs. In this way, offenders receive the opportunity to learn about criminal thinking
and other aspects of criminal behavior that may have affected their lives. Offenders
with specific needs receive appropriate treatment with regard to dosage.
To review CSC operations and discuss risk and needs assessments further, the
aforementioned group of staff visited the Joyceville Institution, an intake processing
facility located in Kingston, Ontario, where inmates undergo an extensive initial needs
assessment. Offenders spend hours completing assessment instruments to more fully
understand an individual’s complex needs. The group toured the facility, discussed
existing program offerings, and were provided briefings on future plans for additional
programs being developed and expanded at the facility. They also met with facility
management to compare U.S. and Canadian federal systems and spoke with inmates
as to their intake progress.

________________________
4

__________________________

d. Needs Assessment Symposium
On September 23-25, 2019, BOP conducted a Needs Assessment Symposium to
gather input and feedback from experts to assist BOP in addressing the needs
assessment requirements of the FSA. The group met for three days with the goal of
developing recommendations for enhancing BOP’s current needs assessment tools.
The event included subject-matter experts from state corrections and research
organizations and involved facilitated discussions of needs assessment best practices.
The symposium resulted in recommendations regarding which needs should be
assessed, how to assess those needs, and ways in which BOP could improve its overall
infrastructure and needs assessment processes. These recommendations will be
considered by BOP’s Executive Staff to determine which recommendations are
consistent with BOP’s strategic plans and can be implemented.
The following experts participated in the symposium:
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•

Holly Busby, National Institute of Corrections
Dr. Patti Butterfield, Independent Review Committee
Dr. Marie Garcia, National Institute of Justice
Wendy Goodman, Virginia Department of Corrections
Dr. Angela Hawken, New York University and NIJ Contractor
Dr. Matthew Makarios, University of Northern Iowa
Dr. Jon Mandracchia, Missouri Western State University
Gary Mohr, Former Director of the Ohio Department of Rehabilitation and
Correction
Dr. Jennifer Pankow, Texas Christian University
Dr. Emily Salisbury, University of Nevada, Las Vegas
Dr. Jennifer Skeem, University of California, Berkeley
Shirley Moore Smeal, Pennsylvania Department of Corrections (retired)
Dr. Solveig Spjeldnes, Ohio University
Dr. Jody Sundt, Indiana University—Purdue University Indianapolis
Donna Tebought, Georgia Department of Corrections
Dr. Stephen Tripodi, Florida State University
Dr. Wendy Williams, Alabama Department of Corrections

e. Use of Risk and Needs Assessment in Prisons Workshop
On August 28-29, 2019, NIJ, in collaboration with the RAND Corporation, hosted a
workshop on the Use of Risk and Needs Assessment in Prisons. The workshop
brought together staff from four state departments of corrections – Minnesota, Ohio,
Pennsylvania, and Tennessee – and staff from BOP’s Reentry Services Division, as
well as scholars with expertise in risk and needs assessments, including Drs. Zachary
Hamilton and Faye Taxman, a member of the IRC.
________________________
5

__________________________

Over the two-day meeting, participants discussed and identified the challenges prisons
face when implementing an RNAS. To illustrate the complexity and uses of risk and
needs assessment, each department of corrections discussed the risk and needs
processes in its department. This discussion was valuable in comparing and contrasting
the use of RNAS in the state corrections departments. A full report detailing identified
issues and needs, as well as recommendations from the workshop, will be available in
Spring of 2020.

________________________
6

__________________________

III.

Changes to PATTERN Based On Suggestions Received
After the July 2019 Report

1. In response to feedback that PATTERN should encourage inmates to take
advantage of programming and should incentivize change, DOJ considered
adding more dynamic factors and removing static factors.
In developing a new risk tool, the Department placed emphasis on a system that
accurately measures an inmate’s change during incarceration, and provides
opportunities for inmates to reduce their risk scores post-intake during periodic
reassessments. Accordingly, PATTERN includes a diverse set of factors to determine
the risk of recidivism for BOP inmates; many of these factors are dynamic, meaning
that an inmate’s risk of recidivism could change with appropriate programming and
services or could be affected by the inmate’s behavior. These factors, in conjunction
with an inmate’s commitment to reform, are crucial to the assessment process.
Importantly, adding dynamic factors to the risk tool and minimizing static factors
(those features of an offenders’ history that are not amenable to change) provides a
greater opportunity for offenders to reduce their risk scores over time during periodic
reassessments. The inclusion of dynamic factors in PATTERN will assist BOP in
identifying an offenders’ needs, determining the services and programming responsive
to the needs, and determining more accurately an offender's actual risk of recidivism.
The dynamic factors initially included in PATTERN were:
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.

Infraction convictions (any) 2
Infraction convictions (serious and violent) 3
Number of programs completed (any)
Number of Technical or Vocational Courses
Federal Prison Industry employment (only for women)
Drug treatment while incarcerated
Drug education while incarcerated
Non-Compliance with financial responsibility (only for women)
Education Score (only for women)
History of Violence
History of Escape

In response to suggestions from the IRC and others, the Department sought to
identify other dynamic factors that would be appropriate measures to add to
PATTERN. As part of that process, BOP provided additional programmatic data to
NIJ’s consultants, who performed further analyses to assess whether the addition of
new programmatic data and the augmentation and removal of variables from the
The “Infraction convictions (any)” measure includes a decay function. Any infraction convictions that
occurred ten or more years prior are removed from the count.
3 The “Infraction convictions (serious and violent)” measure includes a decay function. Serious or
violent infraction convictions that occurred ten or more years prior are removed from the count.
2

________________________

__________________________

model affected its ability to accurately predict an offender’s risk of recidivism. Based
on results from the analyses, and upon the recommendation of the IRC, DOJ intends
to augment PATTERN as follows:
•
•

Include an additional dynamic measure of offender’s “infraction free” period
during his or her current term of incarceration; 4 and
Modify programming measures by adding psychology treatment programs
(BRAVE, Challenge, Skills Program, Sex Offender Treatment (both residential
and non-residential), STAGES, and Step Down programs), the faith-based
Life Connections Program (LCP), and the BOP’s Drug Education program to
the “Number of programs completed (any)” measure and combine
technical/vocational and UNICOR into a new work programming measure.

WORK PROGRAMS. Inmates participate in a horticulture vocational program at FPC Bryan.

2. The Department removed certain variables in furtherance of its commitment
to ensuring that PATTERN is fair and accurate.
The goal of a risk assessment is to create a fair and accurate prediction of risk that an
inmate will commit a crime. The Department received correspondence stating that the
criminal justice system discriminates against minorities and advocating that
PATTERN must avoid perpetuating or exacerbating racial disparities. The
Department agrees that inmates and the public are not served by a risk assessment tool
that is racially biased or does not accurately reflect inmate risk.

This measure identifies the duration of time an individual was infraction free prior to their release. As
a dynamic measure of behavior, this measure is an indicator of an offender’s readiness for release.

________________________

__________________________

As previously stated, PATTERN used many factors that are scientifically-weighted
based on their predictability of reduced recidivism. The Department analyzed
hundreds of different iterations and variables in order to find those factors most
predictive of the risk of recidivism. The process of weighting the variables was based
on scientific research and analysis.
The Department measured PATTERN to ensure that it was predictive across all races
and genders. As shown in the chart below, PATTERN is a neutral assessment tool, as
evidenced by the nearly equal scores on the Area Under the Curve (AUC) analysis.
Recidivism
General
PATTERN
Violent
PATTERN

Male
All White
0.80 0.81
All White
0.77 0.80

AA 5
0.78
AA
0.75

Hispanic
0.78
Hispanic
0.78

Other
0.83
Other
0.82

Female
All White
0.79 0.79
All White
0.78 0.75

AA
0.80
AA
0.80

Hispanic
0.76
Hispanic
0.75

Other
0.81
Other
0.81

The risk assessment tool cannot correct for any outside biases that lead to higher
recidivism. DOJ agrees, however, that the risk assessment tool should avoid adding or
exacerbating any bias that may exist. Critically, it must predict recidivism fairly,
accurately, and objectively with the available information.
Although PATTERN uses scientifically-based factors that help predict an inmate’s risk
of recidivism, the Department removed or changed certain measures that might be
associated with bias, especially racial bias, in order to implement the most fair and
predictive tool possible. The following factors were removed from PATTERN:
•
•

Age of first arrest/conviction; and
Voluntary surrender.

These changes reduce PATTERN’s predictive accuracy by approximately one percent.
The IRC and DOJ both viewed this decrease as acceptable, if it prevents the actual or
perceived perpetuation of any bias.
One criticism relayed to DOJ was that considering “supervised release violations” as
a factor in the risk score would disproportionately affect African Americans and
Hispanics because it was alleged they are more likely to have their supervised release
revoked due to biases in the criminal justice system. DOJ researched whether this
assumption was true. After research and analysis of the issue, the Department found
that the opposite was actually true—removal of supervised release violations from the
data would actually increase the potential racial disparity and have a negative impact
on the predictability of the tool.
There are two categories of supervised release violations: (1) the commission of a new
crime, and (2) the violation of a term of supervised release. BOP consulted with the
5

African-American is abbreviated here as “AA.”

________________________

__________________________

Administrative Office of the United States Courts (AOUSC) and found that statistics
demonstrated that African American and Hispanic offenders were not returned to
prison for a supervised release violation more often than white offenders, as many
critics assumed. The following tables illustrate this finding.
BOP Inmates Released in FY 2009 – FY 2015
1. Inmates who recidivated only for a Supervised Release Violation, out of all releases
in FY 2009 through FY 2015.
Group
Number
%
Total Number

White
6,294
7.89%
79,740

AA
6,382
7.27%
87,838

Hispanic
3,213
7.51%
42,760

Other 6
1,746
13.82%
12,632

Total
17,635
7.91%
222,970

2. Inmates who recidivated only for a Supervised Release Violation and were
Returned to the BOP, out of all inmates who recidivated only for a Supervised
Release Violation.
Group
Number
%
Total Number

White
AA
Hispanic
3,417
3,336
1,595
54.29% 52.27% 49.64%
6,294
6,382
3,213

Other
1,120
64.15%
1,746

Total
9,468
53.69%
17,635

As the tables show, there is little difference in the rates of supervised release violations
between White, African American, and Hispanic inmates. Thus, there is no evidence
that racial disparity or racial bias is increased by including supervised release violations
in the recidivism outcome measure for PATTERN.
With the changes set forth in Item 1 above, and this item, the updated version of
PATTERN is a more streamlined version of the risk tool previously released. As
updated, PATTERN contains fifteen factors, eleven dynamic and four static, as
follows:
Dynamic Factors:
1.
2.
3.
4.
5.
6.
7.

Infraction convictions (any) current incarceration
Infraction convictions (serious and violent) current incarceration
Infraction-free (any) current incarceration
Infraction-free (serious and violent) current incarceration
Number of programs completed (any)
Work programming
Drug treatment while incarcerated

The “Other” category represents Asian and Native American inmates.

________________________

__________________________

8.
9.
10.
11.

Non-compliance with financial responsibility 7
History of violence
History of escapes
Education score 8

Static Factors:
12.
13.
14.
15.

Age at time of assessment
Instant violent offense
Sex offender (Walsh)
Criminal history score

With these changes, 9 PATTERN still retains a high level of predictability, across races
and sexes, as reflected in the chart below.
AUC Comparisons Across Race and Ethnicity, and Gender
for the Revised Version of PATTERN
Recidivism
General
PATTERN
Violent
PATTERN

Male
All White
0.79 0.80
All White
0.77 0.80

Female
AA Hispanic Other All White AA
0.78 0.78
0.82 0.78 0.78 0.79
AA Hispanic Other All White AA
0.75 0.78
0.82 0.77 0.75 0.79

Hispanic
0.76
Hispanic
0.75

Other
0.80
Other
0.80

3. In response to feedback that under a model found on the Internet, 10 PATTERN
does not appear to allow inmates to move from higher to lower risk categories
over the course of incarceration, the Department considered assigning more
weight to dynamic factors and provided illustrative examples of inmates’
abilities to move categories.
A crucial component of the FSA is the ability of inmates to show that they are
reformed through the use of training, programming, and other factors to lower their
risk score and reduce the time they serve in secure confinement. PATTERN was
validated using numerous factors that are scientifically-weighted based on their
predictability of reducing recidivism. The Department analyzed more than a hundred
different iterations and variables in order to find those factors most predictive of the
risk of recidivism. The process of weighting the variables was based on scientific
research and analysis. As noted above, the Department will be updating PATTERN

For females, this factor was included only in the general recidivism model. It was not included in the
violent recidivism model because it was not predictive of this outcome. Additionally, this factor was
predictive only for females, not males.
8 Education score was found to be predictive only for females, not males.
9 The appendix to this report contains technical clarifications and corrections to errata in the July 2019
Report. Those changes were non-substantive and do not affect the AUC analysis.
10 The IRC informed the Department that the IRC had identified some errors in that Internet model.
7

________________________

__________________________

to include additional programming measures and will take into account behaviors that
occur during the current period of incarceration.
The Department tested the updated version of PATTERN to ensure that individuals
could successfully move from a higher to a lower risk score. The examples in
Attachment A show how different types of inmates can use the dynamic factors in the
updated PATTERN to lower their risk score and how a motived inmate’s risk score
will improve compared to an unmotivated inmate who does not thoroughly complete
evidence-based recidivism reduction programs.

WORK PROGRAMS. Inmates learn how to properly perform a four-wheel vehicle wheel alignment
as part of an automotive chassis vocational program at FCI Bennettsville.

4. The Department has clarified the terms “general” and “violent” recidivism to
further promote transparency.
The Department received criticism about the lack of clarity regarding “general” and
“violent” recidivism and understands this confusion. There are many different
definitions of violent and general recidivism. Risk assessment tools and crime statistics
are not uniform in the definition employed. In an attempt to clarify this confusion, as
well as improve overall transparency regarding the tool and its functions, the
Department has tried to be as consistent as possible in the terms it uses across the risk
assessment tool and in the crime statistics it has collected.
PATTERN calculates two risk scores: one for general recidivism risk and one
specifically for violent recidivism risk. The definitions of general and violent recidivism
mirror those used by the Bureau of Justice Statistics (BJS). General recidivism is
defined as a return to BOP custody or a re-arrest within three years of release from
BOP custody, excluding all traffic offenses except driving under the influence (DUI)
and driving while intoxicated (DWI). Violent recidivism is defined as a re-arrest for a
________________________
12

__________________________

suspected act of violence within three years of release from BOP custody. Examples
of the violent offenses captured in this definition include, but are not limited to,
firearms violations, homicide, child abuse, robbery, sex trafficking, and sexual assault. 11
5. In response to suggestions to narrow the definition of “recidivism” to include
only convictions, not arrests, the Department considered whether changing
from the longstanding definition of “recidivism” would be beneficial and
appropriate.
There is no uniform definition of “recidivism” in criminal justice systems across the
United States. As discussed previously, “recidivism” is currently defined in PATTERN
to include a return to BOP custody or a re-arrest within three years of release from
BOP custody, as well as DUI and DWI offenses. This definition is consistent with the
definition used by BJS. Some have suggested that it would be fairer to narrow the
definition of recidivism so that it only includes convictions and does not count those
who were arrested but either never charged with a crime or were acquitted. We
explored the possibility of changing the definition to exclude acquittals and arrests not
resulting in charges over a five-year period 12 and found that, unfortunately, the data
for such a review is not available.
BOP does not have access to accurate and complete data for case dispositions (i.e.,
convictions or acquittals). BOP receives state recidivism data as keyed by the states,
which each have different reporting processes. Some jurisdictions report only arrests,
while others report arrests and convictions. BOP lacks the authority to compel the
states to enhance their reporting and to dedicate the resources needed to include
disposition. BOP’s review of criminal history data for purposes of Second Chance Act
studies indicates that an estimated fifty percent of reconviction data is missing. This
missing data, if used to determine recidivism rates, would significantly bias the analysis
of PATTERN and have a profound impact on its predictability.
DOJ also considered adopting one of the definitions of recidivism used by different
states. Two states define recidivism as returns to custody in only their state, and DOJ
considered defining recidivism as only a return to BOP custody. That definition is
problematic, because if someone is released from BOP custody and then is convicted
in state court of murder, that person would not count as a recidivist, even though that
person committed a very serious offense. Another state defines recidivism as the
percentage of people who are convicted of a felony, released from incarceration in a
A detailed list of violent offense codes can be provided upon request.
The change from a three-year to five-year period is necessary to account for the lengthy period
between arrest and judgment. In the federal system, there is a time gap between arrest, trial, verdict, and
judgment entry that could run well over one year in many cases. Consider the case of a person arrested
for murder two-and-a-half years after his/her release from prison, who was convicted at trial eight
months later, and the judgment filed and sentence imposed three months beyond. If recidivism was
measured by conviction only, that person would be counted as a “non-recidivist,” despite a conviction
for a violent offense. Omissions such as these types of cases would have a profound impact on
PATTERN’s predictability and would undermine confidence in the system.

11
12

________________________

__________________________

local jail or the state’s Department of Correction facility, and are reincarcerated within
three years of release. BOP found that definition also lacking, as a defendant charged
with RICO two years after release from prison, who was convicted after trial and
sentenced to prison 18 months later, would not count as a recidivist, and neither would
a repeat con artist sentenced to prison for stealing senior citizens’ life savings,
sentenced to prison but given bond pending appeal.
Having considered a variety of definitions, the Department sees benefits associated
with using the long-standing definition of arrests over a three-year period. For
example, keeping the same definition helps the BOP assess and study the effectiveness
of its programming. The FSA’s goal is to identify programs that reduce recidivism,
offer those programs to inmates, and allow inmates to reduce their risk of recidivism.
Over forty-four percent of inmates released from BOP custody to U.S. communities
during FY 2013 recidivated, as defined previously (44.2%). By keeping the same
definition of recidivism, BOP can more accurately measure how well a program is
working. No one benefits from continuing programs that do not work. Therefore,
accurately measuring changes in recidivism year over year is critical to the success of
the FSA. By changing the definition, the ability to conduct this type of accurate
comparison is lost.
Additionally, because the definition of recidivism is the most widely used in the federal
system, keeping that definition would make comparisons across risk assessment and
criminal justice systems easier. The U.S. Sentencing Commission, BJS, and AOUSC
all rely on post-release arrest data for more accurate and complete analyses. By using
the same definition, BOP and the public have more meaningful data with which to
assess progress.
The Department intends to continue exploring other ways to readily obtain complete
and accurate disposition data. In the meantime, to avoid any delay associated with
deploying PATTERN and the corresponding FSA benefits, the Department intends
to continue using the standard definition of recidivism.
6. DOJ has begun the process of identifying a fourth group to independently test
and validate PATTERN in response to requests to publish the underlying
PATTERN data to allow stakeholders to independently test its validity.
The Department understands the desire for stakeholders and advocates to
independently review and validate PATTERN. The Department has encouraged
thoughtful criticism and input on how to improve PATTERN. Unfortunately, the
distribution of the data underlying PATTERN is restricted because the data includes
arrest and conviction information provided directly to DOJ by state and local
jurisdictions, who have a significant interest in protecting their data. The retrieval,
disclosure, and redistribution of that criminal history data is prohibited by the sharing
agreements used to acquire the underlying data.
With that said, and in order to ensure independence and integrity in the development
of the risk tool, DOJ arranged for external experts to evaluate and validate PATTERN.
________________________
14

__________________________

Initially, DOJ contracted with two external experts (Drs. Grant Duwe and Zachary
Hamilton) to statistically analyze an anonymized version of the data and use it to
develop PATTERN. The IRC also received access to subsequent results from the
analysis and reanalysis of PATTERN to help verify the assumptions, analysis, and
conclusions incorporated. Researchers were only able to access information about
PATTERN after completing required background investigations. Since background
checks are expensive and lengthy, in the absence of a contractual relationship, sharing
the data widely is thus not a feasible solution to increasing transparency with outside
stakeholders.
To demonstrate the Department’s commitment to addressing concerns about peer
review and transparency, and in order to further promote independence and to
enhance transparency of the process, NIJ will release a funding opportunity to solicit
proposals for a five-year project to review and revalidate the new version of
PATTERN. Upon completion of appropriate background investigations, and
permission from relevant justice agencies to access administrative data, these experts
will access and analyze data to validate PATTERN’s reliability. NIJ expects to fund
this research in 2020.
7. To promote further transparency, the Department is reaffirming its
commitment to publishing the annual validation data, as well as annual
recidivism data.
DOJ intends to publish results from the annual review and validation of the risk
assessment system and recidivism data in its annual report required by 18 U.S.C. §
3634, which will include the programs completed by inmates. The FSA specifically
requires the Department to submit a report to congressional committees that includes,
among other items:
•
•

•
•

A summary of the activities and accomplishments in carrying out the FSA.
A summary and assessment of the types and effectiveness of the evidencebased recidivism reduction programs and productive activities in prisons
operated by BOP, including—
o Evidence about which programs have been shown to reduce
recidivism;
o The capacity of each program and activity at each prison, including the
number of prisoners along with the recidivism risk of each prisoner
enrolled in each program; and
o Identification of any gaps or shortages in capacity of such programs
and activities.
Rates of recidivism among individuals who have been released from Federal
prison; and
The status of prison work programs at facilities operated by BOP.

________________________
15

__________________________

This report is due to Congress on December 18, 2020, two years from the date the
FSA was passed into law. DOJ has already begun collecting this information and fully
intends to meet the deadline.
The IRC and other interested stakeholders recommended that DOJ publish recidivism
data in real-time. DOJ sees benefits in releasing this data as we continue to study and
to refine PATTERN and to robustly implement the FSA, however, there is no feasible
mechanism to retrieve real-time arrest data for a group of individuals. Instead, DOJ is
capable of retrieving criminal history data for groups of individuals on a monthly
basis. 13 The Department notes that publishing monthly recidivism data might lead to
confusion, as seasonal trends might skew perceptions or lead to hasty or improper
conclusions about the efficacy of the FSA based on a time period that is not
traditionally used for recidivism analysis (i.e. monthly vs. three-year). 14 Additionally,
random irregularities in the data could cause the recidivism rate to spike or dip, leading
to inaccurate conclusions about the risk of the relevant group of releasees in relation
to others.
Considering the many concerns and interests, the Department has concluded that it
will begin a pilot program to publish recidivism data on a quarterly basis, beginning in
or around April 2020, and will publish annual data in its report. The Department will
continue to assess the efficacy and practicality of periodic reporting and will be poised
to make adjustments as appropriate.
8. In response to a claim that PATTERN’s separate modelling for men and
women raises constitutional concerns, the Department further considered
whether to collapse PATTERN into one model for both men and women.
PATTERN employs separate modeling for men and women to account for their
gender-specific pathways to crime and risk factors. Evidence-based research shows
that gender shapes the pathways to crime differently for men and women. 15 Moreover,
13

Arrest data is retrieved from Nlets, a private non-profit corporation owned by the States, which was
created to facilitate the cooperative exchange of criminal justice information among law enforcement
agencies. As part of an agreement with the FBI's Criminal Justice Information Service (CJIS) and
Nlets, BOP is able to retrieve information on post-prison arrests occurring in all 50 states and by
federal agencies in order to fulfill statutorily-authorized (e.g. the Second Chance Act) reporting
requirements. For typical recidivism analysis, and consistent with other DOJ reports (e.g. the Bureau
of Justice Assistance), recidivism is defined as (a) a new arrest in the U.S. by federal, state, or local
authorities within 3-years of release or (b) a return to Federal prison within the 3-years of release.
Reporting of criminal history data is not uniform across jurisdictions.
14
DOJ anticipates that the post-FSA recidivism rate for initial releasees will differ from its traditional
reported rate for all releases because not all inmates recidivate in the first month or year. The quarterly
rate will steadily rise until a post-FSA 3 year recidivism rate is established and able to be used as a
benchmark.
15 See Salisbury, E. J., & Van Voorhis, P. (2009). Gendered pathways: A qualitative investigation of
women probationers’ paths to incarceration. Criminal Justice & Behavior, 36(6), 541-566; Bloom, B.,
Owen, B., & Covington, S. (2003). Gender responsive strategies: Research, practice, and guiding
principles for women offenders. Washington, DC: U.S. Department of Justice, National Institute of
Corrections.

________________________

__________________________

some stakeholders echoed this evidence and encouraged DOJ to employ genderspecific models. Because the connection to crime varies for men and women, it was
imperative that PATTERN model the risk of recidivism for these groups separately.
Furthermore, in PATTERN’s dataset, women are outnumbered by men almost 6 to
1. If men and women were scored using a single model, women would be scored
according to a model that centered on men. When other risk assessments have scored
women’s risk using a male model, women are more likely to receive higher risk scores
than their relatively-low recidivism rate would merit. This approach has resulted in
women being over-classified and over-supervised, which results in unfair treatment of
women.
By modeling the risk of men and women separately, BOP will be better positioned to
address the specific risks and needs of offenders in their custody. Addressing the needs
of offenders with appropriate evidence-based recidivism reduction programs and
productive activities will improve the lives of offenders returning to their communities
and enhance public safety, key goals of the FSA and the Department.
9. The Department is committed to ensuring that victims of crime are treated
fairly throughout the criminal justice process. In response to feedback that
PATTERN did not meet victims’ needs, the Department confirmed that the
Victim Notification System will alert victims to changes in release date, should
an inmate receive early release due to the FSA. The Department further
reinforced that victims’ views should be included as part of FSA
implementation.
Every act of recidivism results in victimization of an individual or harm to the
community. That is why DOJ is committed to implementing the FSA—to reduce
recidivism, which in turn, reduces the number of victims and makes our communities
safer.
DOJ and other federal agencies are committed to ensuring that victims of federal crime
are treated fairly as their case moves through the criminal justice system. DOJ will
continue to engage and communicate with organizations that represent crime victims
as it works to fully implement PATTERN, the forthcoming needs assessment, and
other provisions of the FSA.
In order to provide victims with information on case events, as required by the
Victims' Rights and Restitution Act and the Crime Victims’ Rights Act, DOJ has
developed the Victim Notification System (VNS), which is funded by the Crime
Victims Fund of the Office for Victims of Crime. The Fund was established by the
Victims of Crime Act of 1984. The Fund is financed by fines and penalties paid by
convicted federal offenders, not from tax dollars. As part of the VNS notification
service, BOP provides notifications to the VNS when inmates move to pre-release
custody (such as a residential reentry center or home confinement) or release. VNS
then provides the information to registered individuals. We encourage all victims who
are interested in learning of these and other events to register with VNS. Victims may
learn more about the automated system and register by browsing the VNS website.
________________________
17

__________________________

•

EDUCATION. BOP inmates participate in a GED program.

10. The Department received feedback expressing concerns about BOP’s
programming capacity, as according to the July 2019 Report, very few inmates
appear to participate in programming. In response to that feedback, the
Department is clarifying the data in the July Report and is providing an update
on some of the efforts to expand programming.
BOP believes that reentry begins at day one and encourages inmates to participate in
its robust program offerings. The vast majority of Bureau programs do not have
capacity-related constraints. The term “waitlist” is colloquially used to refer to a staging
mechanism that ensures inmates are enrolled in needed courses at the appropriate
times in their incarceration. Our case management and programming staff monitor
these lists based on inmate need and release date/plans, to ensure relevant programs
are completed in appropriate timeframes.
Federal Prison Industries (FPI) does have a waitlist. BOP is grateful to Congress for
the expansions to FPI in the FSA, and anticipates these expansions will over time allow
it to increase business and re-open factories, thereby increasing capacity and decreasing
waitlists. However, this expansion will take time, as growing the businesses, hiring on
additional staff to manage the business lines, and re-opening factories is not a quick
process.
The Resolve Program, which provides trauma treatment, also has a true capacitydriven waitlist, but BOP has used some of the funding in its FY 2019 base budget to
increase staffing and materials, and BOP believes it will improve program availability.
Note that prior to the FSA, inmates could volunteer to take programs regardless of
need. As part of the FSA implementation, the BOP is assigning codes to approved
evidence-based recidivism reduction programs and productive activities to enable
tracking and monitoring of their capacity and use. BOP will also begin assigning
inmates to specific programs to address identified needs, which will allow it to further
________________________
18

__________________________

examine inmate interest and program capacity. Based upon these changes, BOP can
expand or contract capacity consistent with the inmate needs and interests.
BOP does not compel inmates to complete recidivism-reduction programs and, as
such, BOP historically has not tracked all inmate program participation prior to the
FSA. Consequently, the data concerning program participation and availability set
forth in the July 2019 report might have caused confusion and could benefit from
further explanation.
At the outset, the chart on page 47 of the report did not reflect every inmate’s
participation in every BOP program. It only listed participation in a small subset of
programs. BOP has hundreds of programs in operation across the country, many of
which were not previously tracked using standardized assignments demonstrating
completion. Therefore, the report data was under-inclusive.
Because programming is optional, low participation rates do not necessarily equate
with lack of program capacity. BOP has not historically tracked the number of inmates
who declined to participate in programs. As part of the FSA implementation, BOP is
working diligently to improve tracking of its program participation. Such improved
tracking will allow reporting of participation rates for the full complement of BOP
programs.
Moreover, the report data included a large number of short-term inmates (less than
one year), who, by virtue of their short sentence, would have had relatively little time
to complete programs. The chart below has removed those inmates serving sentences
of less than one year and provides the corresponding risk reduction rates under
PATTERN. It is worth repeating that these charts are limited to the select number of
programs that BOP tracked and is not indicative of all programs.
Program Completion Rates for Inmates Confined for One Year or More
That Were Released Between FY 2009 and FY 2015

■

o Completions

■

1 Completion

■

2-9 Completions

■

1 O+ Completions

= 174,323 inmates ,vithout retainers released

________________________
19

__________________________

PATTERN Risk Scoring
Number of Programs Completed (any)

Male Female

Male

General

Female

Violent

No (0) Programs Completed

One (1) Program Completed
More than One (> 1), Less than or Equal to
Three (<= 3) Programs Completed
More than Three ( > 3), Less than or Equal
to Ten (<= 10) Programs Completed
More than Ten (> 10) Programs Completed

-2

-1

-4

-2

-6

-3

-8

-4

As the chart above reflects, close to 50 percent of inmates confined for one year or
more in BOP custody and released between 2009 and 2015 completed three or more
programs. As the second chart shows, when applying the scoring for PATTERN to
this population, inmates that participate in programming could earn a reduction of
three to twelve points in their risk score depending on the amount of programming
taken and type of recidivism predicted. For example, males who completed more than
ten programs would earn a twelve-point reduction in General Recidivism. Similarly,
females who completed ten programs would receive an eight-point reduction in their
General Recidivism score.
BOP shares the stakeholders’ concerns about program availability and is preparing to
meet an anticipated increased demand as inmates will be eligible to receive incentives
for completing programs.

WORK PROGRAMS. A vocational student grinds a test plate that was welded in the Gas Metal Arc
Welding (MIG) Process at FCI Morgantown.

________________________
20

__________________________

WORK PROGRAMS. Students apply the concepts of horticulture through hands-on learning at FCC
Coleman in Sumter County, FL.

BOP has already begun expanding programs and hiring staff to deliver the programs.
While the FSA did not come with appropriated funds in FY 2019, the Deputy Attorney
General announced in July 2019 that BOP had taken the initiative to adjust funding
within its budget to cover a variety of targeted FSA activities, including:
•
•
•
•
•
•
•
•

Increasing Volunteers/Partnership Opportunities
Enhancing Medication Assisted Treatment
Providing English as a Second Language Workbooks and Textbooks
Expanding Education Programs and Programs for Women
Providing Certifications for Vocational Training
Increasing Vocational Training Opportunities
Performing Evaluations for Evidence-Based Programs
Developing a Needs Assessment System

With respect to FY 2020, 16 the Administration requested approximately $116 million
for the following:
•

•
16

Expansion of Evidence-Based Reentry Programs ($52.3 million, of which $30
million is for buildings and facilities): Expands physical space and program
capacity for existing mental health, life skills, special needs, and educational
and vocational programs; also adds new programs as they are identified and
evaluated.
Expansion of Capacity for Prerelease Custody ($10.0 million): Expands BOP’s
capacity to place inmates into halfway houses (also known as Residential

FY2020 runs from October 1, 2019 through September 31, 2020.

________________________

__________________________

•

Reentry Centers) prior to the expiration of their prison sentence to help them
transition back into society.
Expansion of Medication-Assisted Treatment (MAT) Nationwide ($42.6
million): Expands the number of inmates who can participate in treatment
designed to prevent relapse to opioid use upon reintegration in the community;
this treatment may also reduce an inmate’s risk for recidivism. Vivitrol will
continue to be available at all BOP facilities, and Buprenorphine and
Methadone will be available at half of BOP institutions.
Increase of Information Technology Services for Inmates ($6.3 million):
Increases BOP’s capacity to support the I-Connect inmate education network
and expand recidivism program offerings, as well as provide inmate video relay
services for deaf and hard-of-hearing inmates, and connect inmates with their
families potentially through video communication services.
Evaluation of Programs and Services ($5.0 million): Supports a multi-year
effort to evaluate the programs and services offered by BOP in order to
develop recommendations regarding evidence-based recidivism reduction
programs and productive activities. The risk tool developed for BOP will also
be revalidated and assessed as required by FSA.

The approved FY 2020 budget for DOJ from both the House and the Senate provides
seventy-five million dollars ($75 million) for FSA implementation activities. 17

WORK PROGRAMS. Inmates participate in a cosmetology vocational program at FPC Bryan.

17 From October 1, 2019 through December 20, 2019, BOP was operating under a continuing resolution
and had not received any FSA funding.

________________________

__________________________

11. The Department is continuing to work toward updating and improving its
needs assessment tool, as comments rightly noted that the current tool could
benefit from improvement and PATTERN, in its current form, does not
contain a new needs tool.
The Department agrees that the current needs assessment system could be enhanced
and has been implementing the plan set forth in the July 2019 report to identify and
to make those enhancements. In the meantime, DOJ staff has been using BOP’s
current needs assessment system to ensure assessment and programming are not
delayed while we endeavor to improve the system.
Since the publication of the July 2019 report, the BOP has begun updating its needs
assessment process. For example, BOP identified 13 needs areas 18 and began assessing
them for all inmates. Additionally, BOP adopted methods to assess each inmate in
every one of these need areas. A list of programs that address each of the needs will
be used to ensure inmates enroll in the proper programs to address their individual
needs, with the goal of eventually lowering their recidivism risk.
BOP also took several steps toward further developing the needs assessment system.
Those efforts include:
•

•
•

•

BOP engaged the MITRE Corporation 19 to assist with various FSA-related
tasks, specifically, developing a framework to independently evaluate programs
submitted by external entities; conducting market research on needs
assessment best practices; and developing a design and prototype of the new
needs assessment system.
BOP implemented a new trauma screening tool as part of its intake processing.
In the July 2019 report, BOP noted inclusion of a dyslexia screening as part of
the needs assessment tool. BOP completed negotiation with the National
Union and published its Management of Inmates with Disabilities policy. This
update will enable the implementation of dyslexia screening, as required by the
FSA, for inmates.
BOP trained all staff on the updated needs assessment process and the use of
additional screening tools to identify specific criminogenic needs. This training
will enable case management staff to assign relevant programs. This process
will be used while DOJ, BOP, and MITRE formalize a final needs assessment
tool. When the final needs assessment tool is in place, BOP will be able to

The need areas to be assessed are anger/hostility, antisocial peers, cognitions, dyslexia, education,
family/parenting, work, finance/poverty, medical, mental health, recreation/leisure/fitness, substance
abuse, and trauma.
19 The MITRE Corporation (MITRE) is a non-profit organization that operates federally funded
research and development centers (FFRDCs). FFRDCs are unique organizations that assist the United
States government with scientific research and analysis; development and acquisition; and systems
engineering and integration. MITRE also has an independent research program that explores new and
expanded uses of technologies to solve customer problems. For more information:
https://www.mitre.org/.
18

________________________

__________________________

•

track the amount of need and the proportion of inmates who receive
programming.
The BOP published a Request for Information to solicit feedback from
subject-matter experts and the public to further inform its work to improve
needs assessments. 20

12. The DOJ received feedback identifying significant barriers to successful
implementation, including adequate staffing, appropriate training on the tool,
and funding for expanding programs. The Department is providing an update
on efforts to overcome those challenges and to successfully implement the FSA.
The Department agrees that appropriate staffing, training, and funding are three keys
to successful implementation of the FSA, especially considering the expedited
statutory timelines.
In terms of funding, BOP leadership has repeatedly stated that adequate funding is
necessary to ensure successful implementation of the FSA. The Department is
committed to implementing the FSA at whatever level of funding it is given, but agrees
that more funding is crucial to the successful implementation of the Act. As stated
above, the BOP took the initiative to redirect $75 million in July 2019 to begin
expanding inmate programming. BOP has used some of that money to increase
staffing at female institutions and to enhance male and female trauma treatment and
vocational training offerings. 21 BOP has requested funds in the FY 2020 and FY 2021
budgets to continue program expansion.
Maintaining appropriate staffing levels is a second key to successful implementation.
BOP has been actively focused on hiring new employees, especially program staff and
case management staff, and as reflected above, has requested FY 2020 funding to
enhance resources and staffing for program delivery. As part of this enhancement, the
BOP is also seeking to amend the responsibilities and duties of BOP counselors, who
are part of the Unit Team, to provide greater support for inmate program participation.
Recently, BOP inmate-to-staff ratios were updated and approved by the BOP
Executive Staff for Unit Management departments to ensure the successful
implementation of the FSA.

See https://beta.sam.gov/search?keywords=15BNAS20RFI0005&sort=relevance&index=opp&is_active=true&page=1.
21 BOP is using a variety of strategies to very aggressively address its hiring challenges. These strategies
include hiring seven contractors to supplement BOP’s staffing unit in an effort to expedite hiring and
hiring twenty temporary positions to help staffing and background investigation units to expedite hiring.
BOP is also using incentives, including pay setting above the minimum rates, student loan repayment,
annual leave credit, recruitment incentives, special salary rates, retention incentives, and relocation
incentives to on-board more staff. Since March, BOP hiring has exceeded separations each month and
that is tremendous progress. Staffing up fully is an ongoing process, and the timeline will vary at BOP
institutions nationwide. Some sites can staff up quickly due to favorable applicant market conditions,
but for others it will take more time.
20

________________________

__________________________

First Step Act of 2018

TRAINING. BOP Unit Managers receive training at the FCC Victorville Complex in Victorville, CA.

Training is a third key to successful implementation. BOP has already deployed a
comprehensive training program for relevant program areas responsible for portions
of the risk assessment process.
Such training has included:
•
•
•

All Wardens, Associate Wardens, Central Office Administrators, Public
Information Officers, and Case Management Coordinators;
BOP Regional Office and Privatization staff; and
BOP Case Managers and Unit Managers in all 122 BOP institutions.

All case management staff have been trained and relevant programs staff received
initial training in November 2019. Additionally, BOP is developing a National Training
Conference for Correctional Counselors.

________________________
25

__________________________

IV.

Conclusion
Thanks to the helpful input from the IRC, NIJ’s expert consultants, and stakeholders,
the Department has launched an enhanced version of PATTERN. BOP has begun
the process of assessing the risk levels and needs for every inmate and anticipates
completing all risk assessments by the January 2020 statutory deadline. This tool, while
enhanced, is not static. DOJ intends to monitor its use, study the data, and in
consultation with the IRC, consider any improvements and adjustments that should
be made for future assessments.
We appreciate the feedback and input and look forward to continuing to implement
the FSA.

________________________
26

__________________________

Attachment A

Examples of Scoring Under PATTERN 1.2 (Using General Risk Score)
1. Young Drug Offender Taking Multiple Programs
Male. 25-year-old male with a 5-year sentence for drug trafficking. Previously
convicted for a felony aggravated assault at age 16, and a non-violent misdemeanor at
age 21. No degree. Has a drug abuse problem.
•

Score at entry = 57 (High risk)
o 35 points for age, 16 for criminal history, 6 for history of violence
• Model inmate score at release = 21 (Low risk)
o Completes maximum programs and no infractions.
o -14 for aging, -2 for staying clean on history of violence, -4 for earning
GED, -6 for completing RDAP, -8 for completing 11+ programs, and
-2 for completing 2+ work programs.
• Mixed prison record score at release = 32 (Medium risk, almost low)
o Completes 3 programs and 1 work program. Completes RDAP.
Skipped GED class. Committed two minor infractions, but early in the
prison term. Financially compliant.
o -14 for aging, -2 for staying clean on history of violence, +2 for
infractions, -6 for completing RDAP, -4 for programs, -1 for work
program.
• Non-compliant inmate score at release = 63 (High risk)
o Completes no programs. Commits 3 serious infractions in prison, of
which one was a serious violent incident within 3 months of release.
Financially non-compliant. Did everything wrong except try to escape
(extremely rare, except minor escapes at a minimum-security camp).
o -14 for aging, +1 for history of violence, +3 for infractions, +6 for
serious infractions, +6 for not being infraction-free, +3 for not being
serious-infraction-free, +1 for financial non-compliance.
Hypothetical: Young Male Drug Offender with Good Opportunities in PATTERN
Score at Entry
Aging
History of Violence
Earned GED
Drug Treatment
Completed Programs
Completed Work Programs
Infractions
Serious Infractions
Clean Infraction History
Clean Serious Infraction History
Financial Compliance
Score at Release

Model
57
-14
-2
-4
-6
-8
-2
0
0
0
0
0
21

________________________
27

Mixed
57
-14
-2
0
-6
-4
-1
+2
0
0
0
0
32

Non-compliant
57
-14
+1
0
0
0
0
+3
+6
+6
+3
+1
63

__________________________

Female. 25-year-old female with a 5-year sentence for drug trafficking. Previously
convicted for a felony aggravated assault at age 16, and a non-violent misdemeanor at
age 21. No degree. Has a drug abuse problem.
•
•

•

Score at entry = 47 (High risk)
o 25 points for age, 16 for criminal history, 6 for history of violence
Model inmate score = 15 (Low risk)
o Completes maximum programs and no infractions.
o -10 for aging, -2 for staying clean on history of violence, -6 for earning
GED, -8 for completing RDAP, -4 for completing 11+ programs, and
-2 for completing 2+ work programs.
Mixed prison record score = 26 (Low risk)
o Completes 3 programs and 1 work program. Completes RDAP.
Skipped GED class. Committed two minor infractions, but early in the
prison term. Financially compliant.
o -10 for aging, -2 for staying clean on history of violence, +2 for
infractions, -8 for completing RDAP, -2 for programs, -1 for work
program.
Non-compliant inmate score = 56 (High risk)
o Completes no programs. Commits 3 serious infractions in prison, of
which one was a serious violent incident within 3 months of release.
Financially non-compliant. Did everything wrong except try to escape.
o -10 for aging, +1 for history of violence, +3 for infractions, +3 for
serious infractions, +6 for not being infraction-free, +3 for not being
serious-infraction-free, +3 for financial non-compliance.

Hypothetical: Young Female Drug Offender with Good Opportunities in PATTERN
Score at Entry
Aging
History of Violence
Earned GED
Drug Treatment
Completed Programs
Completed Work Programs
Infractions
Serious Infractions
Clean Infraction History
Clean Serious Infraction History
Financial Compliance
Score at Release

Model
47
-10
-2
-6
-8
-4
-2
0
0
0
0
0
15

________________________
28

Mixed
47
-10
-2
0
-8
-2
-1
+2
0
0
0
0
26

Non-compliant
47
-10
+1
0
0
0
0
+3
+3
+6
+3
+3
56

__________________________

2. “Average” Released Inmate (based on PATTERN dataset, which is biased
toward short-term inmates)
Male. 36-year-old male with a drug charge for 2 years. Has two prior convictions, one
of which was for a serious violent offense at least 15 years ago. Has a drug abuse
problem and a degree (high school/GED).
•
•

•

Score at entry = 35 (Medium risk)
o 21 points for age, 16 for criminal history, 2 for history of violence, -4
for degree
Model inmate score = 19 (Low risk)
o Completes maximum programs (may be tough in 2 years) and no
infractions.
o -6 for completing RDAP, -8 for completing 11+ programs, and -2 for
completing 2+ work programs.
Average prison record score = 30 (Low risk, almost medium)
o Completed 1 program. Completes RDAP. Committed one minor
(almost serious) infraction, about 11 months before release. Financially
compliant.
o +1 for infraction, +2 for only 11 months infraction-free time, -6 for
completing RDAP, -2 for program.
Non-compliant inmate score = 59 (High risk)
o Completes no programs. Commits 3 serious infractions in prison, of
which one was a serious violent incident within 3 months of release.
Financially non-compliant. Did everything wrong except try to escape.
o +5 for history of violence, +3 for infractions, +6 for serious
infractions, +6 for not being infraction-free, +3 for not being seriousinfraction-free, +1 for financial non-compliance.
Hypothetical: Male “Average” Released Inmate
Model
35
0
0
0
-6
-8
-2
0
0
0
0
0
19

Score at Entry
Aging
History of Violence
Earned GED
Drug Treatment
Completed Programs
Completed Work Programs
Infractions
Serious Infractions
Clean Infraction History
Clean Serious Infraction History
Financial Compliance
Score at Release

________________________
29

Mixed
35
0
0
0
-6
-2
0
+1
0
+2
0
0
30

Non-compliant
35
0
+5
0
0
0
0
+3
+6
+6
+3
+1
59

__________________________

Female. 36-year-old female with a drug charge for 2 years. Has one prior conviction
for a minor violent offense at least 15 years ago. Has a drug abuse problem and a
degree (high school/GED).
•
•

•

Score at entry = 18 (Low risk)
o 15 points for age, 8 for criminal history, 1 for history of violence, -6
for degree
Model inmate score = 4 (Minimum risk, almost low)
o Completes maximum programs (may be tough in 2 years) and no
infractions.
o -8 for completing RDAP, -4 for completing 11+ programs, and -2 for
completing 2+ work programs.
Average prison record score = 10 (Low risk)
o Completed 1 program. Completes RDAP. Committed one minor
infraction early in the prison term. Financially compliant.
o +1 for infraction, -8 for completing RDAP, -1 for program.
Non-compliant inmate score = 42 (Medium risk)
o Completes no programs. Commits 3 serious infractions in prison, of
which one was a serious violent incident within 3 months of release.
Financially non-compliant. Did everything wrong except try to escape.
o +6 for history of violence, +3 for infractions, +3 for serious
infractions, +6 for not being infraction-free, +3 for not being seriousinfraction-free, +3 for financial non-compliance.
Hypothetical: Female “Average” Released Inmate
Model
18
0
0
0
-8
-4
-2
0
0
0
0
0
4

________________________
30

Mixed
18
0
0
0
-8
-1
0
+1
0
+2
0
0
10

Non-compliant
18
0
+6
0
0
0
0
+3
+3
+6
+3
+3
42

__________________________

3. An Aging Career Criminal
Male. A 42-year-old male with a 20-year sentence for aggravated assault. Has 3 prior
felonies and 7 misdemeanors. Earned a GED during a prior prison term. Has a drug
abuse problem.
•
•

•

Score at entry = 62 (High risk)
o 14 points for age, 40 for criminal history, 7 for history of violence, 5
for violent instant offense, -4 for degree
Model inmate score = 27 (Low risk)
o Completes maximum programs and no infractions. Plenty of time for
the programs, but tough to stay clean for 20 years.
o -14 for aging, -5 for staying clean on history of violence, -6 for
completing RDAP, -8 for completing 11+ programs, and -2 for
completing 2+ work programs.
Mixed prison record score = 40 (Medium risk)
o Completes 9 programs and 2 work programs. Completes RDAP.
Committed four minor infractions and two serious infractions. One
infraction was a fight 8 years ago, but no infractions in the past year.
Financially compliant.
o -14 for aging, -4 for staying clean on history of violence, +6 for
infractions, +4 for serious infractions, -6 for completing RDAP, -6 for
programs, -2 for work program.
Non-compliant inmate score = 67 (High risk)
o Completes no programs. Commits 3 serious infractions in prison, of
which one was a serious violent incident within 3 months of release.
Financially non-compliant. Did everything wrong except try to escape.
o -14 for aging, +3 for infractions, +6 for serious infractions, +6 for not
being infraction-free, +3 for not being serious-infraction-free, +1 for
financial non-compliance.
Hypothetical: Male Aging Career Criminal
Model
62
-14
-5
0
-6
-8
-2
0
0
0
0
0
27

________________________
31

Mixed
62
-14
-4
0
-6
-6
-2
+6
+4
0
0
0
40

Non-compliant
62
-14
0
0
0
0
0
+3
+6
+6
+3
+1
67

__________________________

Female. A 42-year-old female with a 20-year sentence for a major drug trafficking
offense. Has 3 prior felonies and 7 misdemeanors, but nothing violent. Earned a GED
during a prior prison term. Has a drug abuse problem.
•
•

•

Score at entry = 44 (Medium risk)
o 10 points for age, 40 for criminal history, -6 for degree
Model inmate score = 20 (Low risk)
o Completes maximum programs and no infractions.
o -10 for aging, -8 for completing RDAP, -4 for completing 11+
programs, and -2 for completing 2+ work programs.
Mixed prison record score = 28 (Low risk)
o Completes 9 programs and 2 work programs. Completes RDAP.
Committed four minor infractions and two serious infractions. One
infraction was a fight 8 years ago, but no infractions in the past year.
Financially compliant.
o -10 for aging, +3 for history of violence, +3 for infractions, +2 for
serious infractions, -8 for completing RDAP, -4 for programs, -2 for
work program.
Non-compliant inmate score = 59 (High risk)
o Completes no programs. Commits 3 serious infractions in prison, of
which one was a serious violent incident within 3 months of release.
Financially non-compliant. Did everything wrong except try to escape.
o -10 for aging, +7 for history of violence, +3 for infractions, +3 for
serious infractions, +6 for not being infraction-free, +3 for not being
serious-infraction-free, +3 for financial non-compliance.
Hypothetical: Female Aging Career Criminal
Model
44
-10
0
0
-8
-4
-2
0
0
0
0
0
20

________________________
32

Mixed
44
-10
+3
0
-8
-4
-2
+3
+2
0
0
0
28

Non-compliant
44
-10
+7
0
0
0
0
+3
+3
+6
+3
+3
59

__________________________

4. A White Collar Offender
Male. A 45-year-old male with a 3-year sentence for fraud. No prior record. Has a HS
degree and no drug problems.
•
•

•

Score at entry = 1 (Minimum risk)
o 14 points for age, -4 for degree, -9 for no drug need
Model inmate score = -9 (Minimum risk)
o Completes maximum programs and no infractions.
o -8 for completing 11+ programs, and -2 for completing 2+ work
programs.
Mixed prison record score = -2 (Minimum risk)
o Completes 3 programs and 1 work program. Committed two minor
infractions, but early in the prison term. Financially compliant.
o +2 for infractions, -4 for programs, -1 for work program.
Non-compliant inmate score = 36 (Medium risk)
o Completes no programs. Commits 3 serious infractions in prison, of
which one was a serious violent incident within 3 months of release.
Another serious infraction was drug use, leading to a diagnosis that the
inmate needs drug abuse treatment. Inmate refused treatment.
Financially non-compliant. Did everything wrong except try to escape.
o +7 for history of violence, +9 for untreated drug need, +3 for
infractions, +6 for serious infractions, +6 for not being infraction-free,
+3 for not being serious-infraction-free, +1 for financial noncompliance.
Hypothetical: Male White Collar Offender
Model
1
0
0
0
0
-8
-2
0
0
0
0
0
-9

________________________
33

Mixed
1
0
0
0
0
-4
-2
+2
0
0
0
0
-2

Non-compliant
1
0
+7
0
+9
0
0
+3
+6
+6
+3
+1
36

__________________________

Female. A 45-year-old female with a 3-year sentence for fraud. No prior record. Has
a HS degree and no drug problems.
•
•

•

Score at entry = -8 (Minimum risk)
o 10 points for age, -6 for degree, -12 for no drug need
Model inmate score = -14 (Minimum risk)
o Completes maximum programs and no infractions.
o -4 for completing 11+ programs, and -2 for completing 2+ work
programs.
Mixed prison record score = -9 (Minimum risk)
o Completes 3 programs and 1 work program. Committed two minor
infractions, but early in the prison term. Financially compliant.
o +2 for infractions, -2 for programs, -1 for work program.
Non-compliant inmate score = 29 (Low risk)
o Completes no programs. Commits 3 serious infractions in prison, of
which one was a serious violent incident within 3 months of release.
Another serious infraction was drug use, leading to a diagnosis that the
inmate needs drug abuse treatment. Inmate refused treatment.
Financially non-compliant. Did everything wrong except try to escape.
o +7 for history of violence, +12 for untreated drug need, +3 for
infractions, +3 for serious infractions, +6 for not being infraction-free,
+3 for not being serious-infraction-free, +3 for financial noncompliance.
Hypothetical: Female White Collar Offender
Model
-8
0
0
0
0
-4
-2
0
0
0
0
0
-14

________________________
34

Mixed
-8
0
0
0
0
-2
-1
+2
0
0
0
0
-9

Non-compliant
-8
0
+7
0
+12
0
0
+3
+3
+6
+3
+3
29

__________________________

Appendix I:
Technical Corrections
The following errata were identified in the July report after publication:
1. On page 45, the report states: “Federal industry employment, also known as UNICOR,
was indicated ‘yes’ if the individual worked in UNICOR during his or her (emphasis
added) current incarceration.” The original version of PATTERN only scored this
item for women. For operational reasons, BOP already collects this information for
both genders, which also addresses any research needs.
2. On page 45, both infraction items are described as including all guilty findings “that
the inmate has incurred during the current incarceration.” In fact, under the original
PATTERN, this element should have included all guilty findings from the past 10
years, including past BOP incarcerations. However, this error, has been rendered moot
because the revised version of PATTERN will only count misconduct during the
current incarceration period.
3. On page 45, the report describes the Number of Programs Completed item as
including drug treatment, which was incorrect. Other items already covered drug
treatment. The Number of Programs Completed item was modeled with Adult
Continuing Education, parenting, technical, and vocational courses, and it was only
intended to count those courses. However, this issue is rendered moot in the revised
version of PATTERN.
4. The Offense Severity item was not included in the descriptive statistics (page 48) and
had its header deleted at the end of the scoring table (page 56). Offense Severity was
part of the PATTERN violence scores, so it should have been fully included in the
report. However, it has been dropped for the revised version of PATTERN.
5. On page 48, the values of “BRAVO-R Initial: History of Escapes” do not add to 100%.
The correction is that the “none” category should have been 87% instead of 74%.
6. Several items in PATTERN derive from BRAVO, but the descriptive statistics table
(page 48) describes those items as coming from BRAVO-R; technically, all PATTERN
items came from BRAVO, not BRAVO-R. BRAVO-R uses the same items as
BRAVO, with different scores for the items. PATTERN was developed using
BRAVO’s items and scores, not BRAVO-R’s scores.
7. The scoring for the male general recidivism scores was switched between the two
infraction items on page 54. Serious infractions are instead worth 0, 3, 6, or 9 points
and all infractions are worth 0, 2, 4, or 6 points.
8. The Education item should be scored as 0, -1, and -2. On page 56, the report indicated
it was 1, 0, and -1.
9. The Voluntary Surrender item incorrectly reports the violence scores as -1 and -2 for
men and women on page 56. Those are the weights, not the scores. The actual scores
should be -3 and -6, respectively.

________________________
35

__________________________

Two other items related to the discussion of PATTERN in the RNAS report warrant
clarification:
1. On page 72 and 73, the report states the Designation, Sentencing and Computation
Center (DSCC) completes the first PATTERN assessment for each inmate. The BOP
has determined that the inmate’s Unit Team, which meets within the inmate’s first
month of arrival at the institution, will complete the first PATTERN assessment.
2. On page 47, the “Drug treatment while incarcerated” item indicated 57% of the
sampled inmates had a drug treatment need but received no treatment. This statistic is
misleading and has generated significant confusion. According to a recent evaluation
of RDAP, 0.6% of inmates who needed and signed up for RDAP drug treatment did
not get a chance to participate in RDAP before release, not 57% of all inmates.
PATTERN’s drug treatment item measured “drug treatment need” by using the
“drug/alcohol abuse” item from the Bureau’s Risk and Verification Observation
(BRAVO) risk instrument. This BRAVO item was originally intended to be used
internally by BOP to measure substance abuse within the past five years. However, in
accordance with BOP’s classification policy, this item also includes any inmates with
“a conviction of a drug or alcohol-related offense.” Consequently, this item
accidentally counts drug dealers and traffickers as having a drug need, even if they do
not use drugs themselves. It would be more accurate to say that 57% of sampled
inmates have an untreated “potential drug need,” with a very loose definition of
potential drug need. It does not indicate a confirmed, untreated drug need. Within
BOP, inmates are identified as eligible for RDAP after a Psychology Services
evaluation where treatment staff examine conduct for the past year. More importantly,
participation in programs is voluntary and inmates may refuse engagement, which also
reduces the number of inmates with a drug need that actually receive treatment. In the
future, PATTERN will switch to a more appropriate, accurate measure of drug need
as soon as it becomes available in a centralized database.

________________________
36

__________________________

Appendix II:
Revised Points Assigned in the PATTERN
Risk Assessment Models
(Previously published on pages 53-56 of the July report.)
The table below displays the scores assigned to the items used to develop PATTERN.
Each column represents a different gender and outcome-specific prediction model,
i.e., general or violent recidivism. Cell values represent response scores, where blank
cells indicate items not identified as predictive for a given model. For example, the
score for a male over the age of 60 in the general recidivism category for the item ‘age
at time of assessment’ is 0. Larger values indicate greater prediction strength for a given
model. Each individual’s response scores are summed to compute a total score.
Item
Age at time of assessment
> 60
< 50, <= 60
> 40, <= 50
> 29, <= 40
> 25, <= 29
>= 18, <= 25
Infraction convictions
(any) current incarceration
0
> 0, <= 1
> 1, <= 2
>2
Infraction convictions
(serious and violent)
current incarceration
0
> 0, <= 1
> 1, <= 2
>2
Infraction-Free (any)
None/12 months+
7-12months
3-6months
<3 months
Infraction-Free (serious
and violent)
None/12 months+
7-12months
3-6months
<3 months

General Male

Violent Male

General Female

Violent Female

0
7
14
21
28
35

0
4
8
12
16
20

0
5
10
15
20
25

0
1
2
3
4
5

0
1
2
3

0
2
4
6

0
1
2
3

0
2
4
6

0
1
2
3

0
2
4
6

0
1
2
3

0
2
4
6

0
1
2
3

________________________
37

__________________________
Item
Number of programs
completed (any)

General Male

Violent Male

General Female

Violent Female

0
-2
-4
-6
-8

0
-1
-2
-3
-4

0
-1
-2

-9

-3

-12

-3

-6

-2

-8

-2

-3

-1

-4

-1

0
1

0
3

0
5

0
1
> 1, <= 3
> 3, <= 10
> 10
Work Programming
0
> 0, <= 1
>1
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
Non-Compliance with
financial responsibility
No
Yes
Instant offense violent
No
Yes
Sex offender (Walsh)
No
Yes
Criminal History Score
0-1 points
2-3 points
4-6 points
7-9 points
10-12 points
13+ points
History of Violence
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

0
3

0
1
0
8
16
24
32
40

0
4
8
12
16
20

0
8
16
24
32
40

0
4
8
12
16
20

0
1
2
3
4
5
6
7

________________________
38

__________________________
Item
History of Escapes

General Male

Violent Male

General Female

Violent Female

0
2
4
6

0
1
2
3

0
3
6
9

0
2
4
6

-4

-2

-6

-2

-1

-3

-1

None
>10 years minor
5-10 years minor
<5 years minor or any serious
Education Score
HS degree or GED Verified
Enrolled and progressing in
GED
No verified degree and not
participating in GED
program

________________________
39