Skip navigation

Recidivism Among Federal Offenders, USSC, 2016

Download original document:
Brief thumbnail
This text is machine-read, and may contain errors. Check the original document to verify accuracy.
Recidivism Among Federal Offenders:
A Comprehensive Overview

UNITED STATES SENTENCING COMMISSION

UNITED STATES SENTENCING COMMISSION
ONE COLUMBUS CIRCLE, N.E.
WASHINGTON, DC 20002
WWW.USSC.GOV

Patti B. Saris
Chair
Charles R. Breyer
Vice Chair
Dabney L. Friedrich
Commissioner
Rachel E. Barkow
Commissioner
William H. Pryor, Jr.
Commissioner
Michelle Morales
Ex Officio
J. Patricia Wilson Smoot
Ex Officio

Kenneth P. Cohen
Staff Director
Glenn R. Schmitt
Director
Office of Research and Data

March 2016

Kim Steven Hunt, Ph.D
Senior Research Associate
Robert Dumville
Research Associate

TABLE OF CONTENTS

PART I

INTRODUCTION

2

PART II

SUMMARY OF KEY FINDINGS

6

PART III

DESCRIPTION OF METHODOLOGY AND STUDY GROUP

8

PART IV

PART V

Defining and Measuring Recidivism

12

Methodology

13

The Study Group

14

DETAILED RECIDIVISM FINDINGS

18

General Recidivism Rates

20

Most Serious Recidivism Offense

22

Recidivism and Criminal History

23

Recidivism and an Offender’s Federal Offense

25

Recidivism and Sentences Imposed

27

Recidivism and Offender Characteristics

28

CONCLUSION

30

ENDNOTES

34

APPENDIX

36

i

PART I

Introduction

This report provides a broad overview of key findings from the United States Sentencing Commission’s study of recidivism of federal offenders. The Commission
studied offenders who were either released from federal prison after serving a sentence of imprisonment or placed on a term of probation in 2005. Nearly half (49.3%)
of such offenders were rearrested within eight years for either a new crime or for some other violation of the condition of their probation or release conditions. This
report discusses the Commission’s recidivism research project and provides many additional findings from that project. In the future, the Commission will release
additional publications discussing specific topics concerning recidivism of federal offenders.

The United States Sentencing Commission 1 began studying recidivism
shortly after the enactment of the Sentencing Reform Act of 1984 (SRA). 2 Past
studies, together with ongoing multi-year research on this subject, advance the
Commission’s mission to conduct research on sentencing issues related to the
purposes of sentencing set forth in the SRA. 3 Recidivism information is central
to three of the primary purposes of punishment as described in the SRA –
specific deterrence, incapacitation, and rehabilitation – purposes which focus on
prevention of future crimes through correctional intervention. 4 Information
about recidivism also is relevant to the Commission’s obligation to formulate
sentencing policy that “reflect[s], to the extent practicable, advancements in
knowledge of human behavior as it relates to the sentencing process.” 5

Considerations of recidivism by federal offenders were central to the
Commission’s initial work, as it chose to develop the Guidelines Manual’s
criminal history provisions in significant part on offenders’ risk of reoffending. 6
The Commission’s 2004 report, Measuring Recidivism, served as a “performance
review” of the predictive ability of these provisions, i.e., the predictive statistical
power of the criminal history measure to reflect subsequent recidivism among
federal offenders. That report concluded that these provisions largely
succeeded in predicting subsequent risk of reoffending. 7 Two additional
Commission reports in that same period further investigated federal offender
recidivism and reviewed features of the federal sentencing guidelines. 8 Two
later publications examined recidivism by federal offenders sentenced under
the guidelines for child pornography and crack cocaine. 9
Recent developments have refocused the Commission’s interest on the
recidivism of federal offenders, particularly the recent public attention on the

Page 3

size of the federal prison population and the cost of incarceration. 10 Well over
1.5 million offenders have been sentenced under the federal sentencing
guidelines since their inception in 1987. In recent years, approximately 80,000
offenders have been sentenced each year for federal felonies and Class A (nonpetty) misdemeanor offenses. 11 Nine out of ten of those federal offenders today
receive sentences of imprisonment, while one out of ten is sentenced to
probation. 12 The federal prison population today is slightly under two hundred
thousand inmates 13 (which is around one-eighth of the total prison and jail
population in the United States). 14 As these numbers have grown, courts and
correctional officials have sought greater information about reoffending.
Recidivism information has recently been used in decisions by the Commission
to reduce the periods of incarceration established for certain offenders through
retroactive application of sentence reductions in the guidelines. 15 And changes
in the sentencing guidelines at the policy level have a significant effect on the
imposition of individual sentences. 16 It is in this policy climate that the
Commission has begun its multi-year study of recidivism by federal offenders.

The Commission’s current recidivism research substantially expands
on the scope of previous Commission recidivism projects. In addition to a
different set of offenders – U.S. citizen federal offenders released in 2005 – the
current study group (25,431 offenders) is much larger than those in previous
Commission studies. A larger study group provides the opportunity to develop
statistically useful conclusions about many subgroups of federal offenders,
including those sentenced under different provisions in the guidelines. This is
the first report on the results of the Commission’s study of the recidivism of the
federal offenders released during this time period.

PART II

Summary of Key Findings

The key findings of the Commission’s study are:
•

•

•

•

•

Over an eight year follow-up period, almost one-half of federal offenders
released in 2005 (49.3%) were rearrested for a new crime or rearrested
for a violation of supervision conditions.

Almost one-third (31.7%) of the offenders were also reconvicted, and
one-quarter (24.6%) of the offenders were reincarcerated over the
same study period.

Offenders released from incarceration in 2005 had a rearrest rate of 52.5
percent, while offenders released directly to a probationary sentence
had a rearrest rate of 35.1 percent.

Of those offenders who recidivated, most did so within the first two years
of the eight year follow-up period. The median time to rearrest was 21
months.

About one-fourth of those rearrested had an assault rearrest as their
most serious charge over the study period. Other common most serious
offenses were drug trafficking, larceny, and public order offenses.

•

A federal offender’s criminal history was closely correlated with
recidivism rates. Rearrest rates range from 30.2 percent for offenders
with zero total criminal history points to 80.1 percent of offenders in the
highest Criminal History Category, VI. Each additional criminal history
point was generally associated with a greater likelihood of
recidivism.

•

A federal offender’s age at time of release into the community was also
closely associated with differences in recidivism rates. Offenders
released prior to age 21 had the highest rearrest rate, 67.6 percent, while
offenders over sixty years old at the time of release had a recidivism rate
of 16.0 percent.

•

Other factors, including offense type and educational level, were
associated with differing rates of recidivism but less so than age and
criminal history.
More detailed findings are contained below in Part IV.

Recidivism Study Offenders
The offenders studied in this project are 25,431 federal offenders who:
•
•
•
•
•

Page 5

are citizens;

re-entered the community during 2005 after discharging their sentence of incarceration or by commencing a term of probation in 2005;

have valid FBI numbers which could be located in criminal history repositories (in at least one state, the District of Columbia, or federal records);
are not reported dead, escaped, or detained, and

have a pre-sentence investigation report that was submitted to the Commission with a federal sentence that was not vacated.

PART III

Description of Methodology and Study Group

Defining and Measuring Recidivism
Recidivism “refers to a person’s relapse into criminal behavior, often
after the person receives sanctions or undergoes intervention for a previous
crime.” 17 Recidivism measures can provide policy makers with information
regarding the relative threat to public safety posed by various types of
offenders, and the effectiveness of public safety initiatives in (1) deterring crime
and (2) rehabilitating or incapacitating offenders. 18 Recidivism measures are
used by numerous public safety agencies to measure performance and inform
policy decisions and practices on issues such as pretrial detention, prisoner
classification and programming, and offender supervision in the community. 19
Recidivism is typically measured by criminal acts that resulted in the
rearrest, reconviction, and/or reincarceration of the offender over a specified
period of time. 20 These are the three recidivism measures used in this report.
Providing multiple measures of recidivism allows users to select the
performance measure best suited to their outcome of interest. 21

Rearrest classifies a person as a recidivist if he or she has been
arrested for a new crime after being released into the community directly on
probation or after serving a term of imprisonment. Rearrest also includes
arrests for alleged violations of supervised release, probation, or state parole. 22
The number of rearrests in the Commission’s analysis is based on the number of
unique arrest dates, regardless of the number of individual charges arising from
a single arrest event. Thus, if an offender was arrested on a single occasion for
both driving under the influence and possession of cocaine, that arrest date
would constitute a single rearrest event.
Reconviction classifies a person as a recidivist if an arrest resulted in a
subsequent court conviction. Violations and revocations of supervision are not
included in reconvictions since no formal prosecution occurred.
Reincarceration classifies a person as a recidivist if a conviction or
revocation resulted in a prison or jail sentence as punishment. The
reincarceration measure counts offenders who were reported returned to the
Federal Bureau of Prisons, state prison, or local jail for any term of
incarceration.

For example, many rearrests do not ultimately result in a reconviction
or reincarceration for reasons relating to procedural safeguards (e.g., the
suppression of evidence for an unconstitutional search or seizure), lack of

Page 7

sufficient evidence to convict or revoke, and prosecutorial or judicial resource
limitations. To the extent that the rearrest event is an accurate indicator of
relapse into criminal behavior, excluding events due to non-conviction or nonincarceration will result in underestimation of recidivism. Even using the least
restrictive measure, rearrest, does not count the full extent of offender
recidivism, as many crimes go unreported to police or, if reported, do not result
in an arrest. For these reasons, no measure is perfect, and reporting several
measures provides a more complete and nuanced picture of reoffending. 23 The
three measures overlap in some areas – meaning all offenders who were
reconvicted or reincarcerated also were necessarily rearrested, too. Some
offenders who were reconvicted, however, were not reincarcerated. Generally
speaking, however, the measure of rearrest is larger than the measure of
reconviction, which in turn is larger than the measure of reincarceration.

Two principal research choices can affect the relative magnitude of
recidivism as measured in any study: first, which events are being included as
evidence of reversion to criminal behavior; and second, over what time period
these events are counted. The period of time over which events are counted
following release into the community is the “follow-up period.” Recidivism
analysis begins with a starting event, such as release from prison into the
community, and may have one or more subsequent events, such as arrests,
recorded before the close of the follow-up period. In some studies, follow-up
periods may be quite short, for example six months. Other studies follow
offenders in the community for substantially longer periods, which may extend
to several years. The longer the follow-up period, the higher the reported
recidivism rate, because some offenders who did not recidivate within the first
year of release recidivate in the second year, and other offenders who did not
recidivate during the first two years recidivated thereafter. Longer follow-up
periods are in this sense less restrictive, and come closer to estimating true rate
of desistance from crime.
The Commission selected an eight year follow-up period for its
research. The Commission also considered all recidivism events (including
felonies, misdemeanors, and “technical” violations of the conditions of
supervision) except the Commission excluded minor traffic offenses. In this
report and accompanying appendices, the Commission will include findings
using all three measures: rearrest, reconviction, and reincarceration. The
measures of rearrest, reconviction, and reincarceration are reported as the
percentage of offenders who recidivated by that measure. As noted, those
measures overlap in some areas.

This report provides additional important information regarding the
recidivism of federal offenders. First, the Commission reports the time to the
first recidivism event for those offenders who recidivated. Studying the timing
of recidivism can help understand the process of desistance, as some offenders
may be able to remain in the community for a considerable time before
recidivating, while others recidivate very quickly. The Commission also reports
the median number of recidivism events for those who recidivated, again using
the three recidivism measures. With a follow-up period of eight years, many
offenders recidivate more than once. Knowing how often they recidivate and
for what crimes can provide important information to policymakers. Finally,
the Commission reports the most serious post-release event for those who
recidivated, as some crimes represent greater harm to society than others.

This report presents the findings of a much larger group of offenders
than previous Commission reports, due largely to recent technological advances
described below that make this possible. As with previous Commission reports,
the present multi-year study of recidivism of federal offenders seeks not only to
describe recidivism results generally, but also to relate these results to current
sentencing policy questions as may be appropriate in light of the information
obtained.

Two other recent recidivism reports are particularly relevant to the
present Commission study, although differing with respect to follow-up period
and in some other ways. First, the Bureau of Justice Statistics of the U.S.
Department of Justice began issuing reports on recidivism among prisoners
released from state prisons in the 1980s with Recidivism of Prisoners Released in
1983, and recently issued Recidivism of Prisoners Released in 30 States in 2005:
Patterns from 2005 to 2010. 24 Among state prisoners released in 2005, 67.8
percent of released prisoners were arrested for a new crime within three years,
and 76.6 percent were arrested within five years. Many prisoners (36.8%) who
were arrested during the five-year period were arrested within the first six
months, and 56.7 percent were arrested by the end of the first year.
Second, the Office of Probation and Pretrial Services of the
Administrative Office of the U.S. Courts (AO) sponsored a report in 2012 on the
recidivism of federal offenders prepared by Abt Associates. The AO study,
Recidivism of Federal Offenders on Federal Community Supervision, 25 also
followed federal offenders released into the community in fiscal year 2005. The
AO found that 38 percent of offenders recidivated within five years, including
24 percent who were re-arrested for a new crime and 13 percent who were
revoked from probation or supervised release. Unlike the Commission’s
present research, the AO study excluded most public order offenses from
criminal acts counted in its recidivism measure. The report also examined how

recidivism rates varied with the presence of various offender risk (e.g.,
substance abuse issues) and protective (e.g., positive social support) factors.

Methodology

Technological advances have allowed the Commission to examine all
U.S. citizen federal offenders released in 2005 who meet the study criteria
discussed below, greatly expanding the number of federal offender records
analyzed over past Commission analysis. Following a practice pioneered by the
Bureau of Justice Statistics, 26 the Commission entered into a data sharing
agreement with the FBI’s Criminal Justice Information Services (CJIS) Division
and the Administrative Office of the United States Courts to provide the
Commission with electronic access to criminal history records (i.e., RAP sheets)
through the CJIS’s Interstate Identification Index (III). The CJIS’s III allows
authorized agencies to determine whether any repository has criminal history
records on an individual. Results received over this system provide an
individual’s criminal history record maintained by all U.S. states, the District of
Columbia, and federal agencies. Once automated records were obtained, the
Commission went through an extensive process to standardize offense
descriptions across all reporting agencies, because states use a variety of statespecific text descriptions and other citations to report the type of offense.
Disposition information received from law enforcement agencies and courts
was also standardized.
Using an automated software program that analyzed and classified
offender criminal records, the Commission identified and processed the
criminal records of more than 35,000 offenders who had a valid FBI number in
either Commission or Bureau of Prisons records and were released during the
study period, which included primarily calendar year 2005 but was extended
before and after this year to expand the sample for certain groups. To be
included in the final cohort, study subjects must have been federal offenders:
•
•

•
•

who are citizens;
who re-entered the community after discharging their sentences of
incarceration or by commencing a term of probation in 2005; 27
whose pre-sentence investigation report was submitted to the
Commission;
who have valid FBI numbers which could be located in criminal history
repositories (in at least one of the 50 states, DC, or federal records);
and

Page 8

•
•

who were not reported dead, escaped, or detained; and
whose federal sentence was not vacated.

The Commission examined all offenses committed by the offenders in
the study group and ranked the offenses in order of seriousness. In general, the
Commission followed a widely accepted ranking scheme used by the Bureau of
Justice Statistics in prior recidivism research. 28 The ranking generally begins
with the most serious violent crimes, proceeds to less serious violent crimes,
next ranking property, drug, and public order crimes. The exception to this
general pattern is that the Commission ranked drug trafficking just below
violent crimes and above all property crimes.

The study expands on previous Commission practice in another sense.
This study cohort was tracked in the community for eight years, compared to
previous follow-up periods that ranged from two to five years. Because the risk
of re-offending may extend for long periods after release into the community,
longer follow-up periods may reveal recidivism that would otherwise be
undetected. The longer period may also provide insight into offenders with
multiple recidivism incidents, or periods of activity and inactivity.

Figure 1.
Demographic Characteristics of Recidivism Study Offenders

The Study Group
This report examines 25,431 offenders who were released into the
community (either from federal prison or on to probation) in calendar year
2005 and met all of the above criteria. The advantages of this large study group
are substantial. Having several thousand offenders allows more precise
estimates of recidivism rates across different subgroups. For example, there are
4,664 female offenders in this study, and 1,048 released offenders were older
than sixty years of age.

Offender Demographics

Male offenders constituted more than four of every five (81.7%)
offenders in this study. White offenders were the largest group (43.7%),
followed by Black (33.9%), Hispanic (17.8%), and Other Race (4.6%).

The median age at time of sentencing was 33 years for offenders in this
study. The median age at release was 36 years. However, over four percent of
offenders were released into the community after age sixty. Regarding the
highest education level attained by released offenders, about one-third (34.3%)
did not complete high school, while most (65.7%) completed at least high
school, including some offenders (7.5%) who were college graduates.

SOURCE: U.S. Sentencing Commission’s 2005 Recidivism Release Cohort Datafile, RECID05. Of the 25,431 cases in this study, the Commission excluded cases from this analysis that were missing information necessary to perform the analysis.

Page 9

Federal Offense Type and Criminal History
Over two in five (41.7%) federal offenders released in 2005 and
included in this study group were released after receiving a conviction for a
federal offense of drug trafficking. Other federal offenses which were common
to this study group are fraud (13.6%); unlawful receipt, possession, or
transportation of firearms (12.8%); robbery (4.3%); larceny (3.9%); and
immigration 29 (3.5%). All other offenses constituted the remaining 20. 3
percent.

An offender’s prior criminal record is an important consideration at
sentencing. Offenders sentenced in the federal system usually receive criminal
history points for each prior sentence based on the number and severity of past
criminal conduct. For example, each prior sentence of imprisonment exceeding

one year and one month generally receives three points. In addition to points
for prior sentences, if the defendant committed the federal offense while under
any criminal justice sentence, such as probation, two additional points are
added. Some prior sentences are not counted because of staleness, their minor
nature, or other reasons. 30 The total number of criminal history points
determine the Criminal History Category (I-VI) to which the offender is assigned
for the purpose of determine the sentencing guideline range.
Criminal History Category (CHC) I comprises 53.6 percent of the study
group, and is assigned to offenders with zero or one criminal history point, the
lowest or least serious level of past criminal conduct. In contrast, 7.6 percent of
offenders in this study were assigned to CHC VI (13 or more criminal history
points), indicating the highest or most serious level of past criminal conduct.

Figure 2.
Offense Types and Criminal History Scores of Recidivism Study Offenders

SOURCE: U.S. Sentencing Commission’s 2005 Recidivism Release Cohort Datafile, RECID05. Of the 25,431 cases in this study, the Commission excluded cases from this analysis that were missing information necessary to perform the analysis.

Page 10

Sentences Originally Imposed
The offenders in this study were sentenced for their instant federal
offenses between 1990 and 2005. About one-quarter of offenders were
sentenced before 2002 while about three-quarters of offenders were sentenced
before 2005. Of all the offenders in the study, over three-quarters (77.8%) were
sentenced before the Supreme Court’s January 12, 2005 decision in United
States v. Booker, 31 which held that the sentencing guidelines were advisory only.

At their original sentencings for their instant federal offenses, 14.1
percent of the offenders in the study group were sentenced to probation only
(i.e., probation with no condition of confinement); 4.7 percent were sentenced
to probation with a condition of confinement; 3.9 percent were sentenced to a
“split” sentence of imprisonment to be followed by a non-prison alternative
confinement; and 77.4 percent were sentenced to a term of imprisonment
only. 32 Because the vast majority of offenders who received probation with a
condition of confinement served that period of confinement in home detention
or in a halfway house rather than in a jail or prison, 33 such offenders will be
analyzed together with offenders who received probation without a condition of
confinement. Furthermore, because offenders who received “split” sentences
received a median term of imprisonment of 5 months, those offenders will be
analyzed together with offenders who received sentences of imprisonment only.
Figure 3.
Sentences Imposed on Recidivism Study Offenders

Therefore, for ease of analysis and presentation, the discussion of recidivism
rates of the study group by sentence type that appear in Recidivism and
Sentences Imposed below will consider only the following two 34 larger groups of
offenders: (1) offenders who received a probationary sentence, regardless of
whether it had a condition of confinement (18.8 percent of the study group), or
(2) offenders who received a sentence of imprisonment, regardless of whether
it was a “split” sentence or a sentence of imprisonment only (81.2 percent of the
study group). 35
The median length of imprisonment for all released offenders was 37
months; however, one-tenth (12.3%) of the offenders received a sentence of ten
years or more.

A term of supervised release may be ordered to follow any term of
imprisonment, and must be ordered if required by statute. 36 The most common
length of supervised release ordered at sentencing (54.0%) was three years.
One-quarter (25.2%) of released offenders were ordered to serve five years or
more of supervised release, including 0.6 percent ordered to serve ten years or
more. Because the study follow-up period was eight years, and the typical
length of supervision was not more than three years, most offenders completed
their term of supervision, and were no longer on supervision for much of the
follow-up period.

SOURCE: U.S. Sentencing Commission’s 2005 Recidivism Release Cohort Datafile, RECID05. Of the 25,431 cases in this study, the Commission excluded cases from this analysis that were missing information necessary to perform the analysis.

Page 11

It is important to note that most sentenced offenders released during
the study period, with the exception of most probationers, were sentenced prior
to the Booker decision 37 that rendered the federal sentencing guidelines
“effectively advisory.” 38 The impact of this decision, as reported in the
Commission’s Report on the Continuing Impact of U.S. v. Booker on Federal
Sentencing, is that “the rates of non-government sponsored below range
sentences have increased in most districts.” 39 Three-quarters (77.8%) of
offenders discussed in this study were sentenced before Booker.

Although the study group was released several years ago, this group is
generally well-matched to citizen offenders sentenced in 2014. Both the study
group and citizen offenders sentenced in 2014 are predominantly male (81.7%
and 81.2%, respectively). White offenders are the most numerous
race/ethnicity group, 43.7 percent for the study group and 38.1 percent for the
2014 sentenced offenders. Black offenders (33.9% for the study group and
32.7% for the 2014 sentenced offenders, respectively) and Hispanic offenders
(17.8% and 23.4%, respectively) are sufficiently well-matched. The median age
at sentencing for the study group (33 years old) is close to the median for 2014
sentenced citizens (35 years old).

The Criminal History Categories are also well-matched. CHC I is the
most common category for both groups, 53.7 percent for the study group and
46.8 percent for 2014 sentenced offenders. Conviction for a federal offense of
drug trafficking is the most common offense, 41.7 percent of the study group
and 36.4 percent for the 2014 sentenced group. Offenses involving firearms
(12.8% and 17.0%, respectively) and fraud (13.6% and 14.2%, respectively) are
also well-matched. Given these similarities, it is reasonable to assume that the
recidivism findings available from the study group analysis are relevant to
currently sentenced offenders.

Page 12

Page 13

PART IV

Detailed Recidivism Findings

General Recidivism Rates
As mentioned earlier, this report uses three different measures of
recidivism—rearrest, reconviction, and reincarceration.

Rearrest

Almost one-half of offenders released in 2005 (49.3%) were rearrested
for a new crime or for an alleged violation of the conditions of their supervision
over the eight year follow-up period. The median time to rearrest was 21
months, meaning that for one-half of the offenders rearrest occurred in less
than two years following their initial release from prison or placement on
probation. Among those who recidivated, the median number of rearrest
events (events occurring on separate days) was two, but 21.2 percent of
recidivist offenders were rearrested five times or more. When considering only
the “most serious” offense committed by those who were rearrested, the most
serious event that was most prevalent was assault. About one-fourth (23.3%)
of those rearrested had an assault rearrest as their most serious charge over the
study period. Other common “most serious” offenses were other public order
(15.5%), drug trafficking (11.5%), and larceny (7.7%).
Table 1.
Overview of Recidivism Study Findings

Reconviction
Almost one-third (31.7%) of offenders were reconvicted over the study
period. Most offenders who were reconvicted were reconvicted once. The
median time to reconviction, as measured from the date of the arrest that led to
the reconviction, was 30 months. About one-fourth (23.9%) of those
reconvicted had an assault conviction as their most serious charge over the
study period. Other common most serious offenses were drug trafficking
(13.5%), other public order (8.7%), and larceny (8.5%).

Reincarceration

One-quarter (24.6%) of offenders were reincarcerated over the study
period. Most offenders who were reincarcerated were reincarcerated once.
The median time to arrest for the offense that led to reincarceration was 29
months. More offenders (23.8%) were reincarcerated for an assault crime as
their most serious offense than for any other offense. Other common most
serious offenses were drug trafficking (14.5%), other public order (8.4%), and
larceny (7.8%).

Comparison to State Prisoners

Compared to a cohort of state prisoners released into the community in
2005 and tracked by the Bureau of Justice Statistics, federal offenders had a
lower recidivism rate. BJS found that 76.6 percent of offenders released from
state prison were rearrested within five years. The Commission, using a
comparable five year follow-up period and including only federal offenders
released from prison (i.e., excluding those sentenced to probation or a fineonly), found the recidivism rate for these federal offenders was 44.9 percent. 40
Using reconviction as the measure of recidivism, BJS found that 55.4
percent of state offenders had an arrest within five years that led to a
conviction. The reconviction rate for federal offenders over the same length of
time was 26.0 percent. When recidivism is measured by reincarceration, BJS
found a 28.2 recidivism rate for state offenders within five years that led to an
incarceration, compared to 20.7 percent of the federal offenders in the
Commission’s study.

Page 15

Time to First Recidivism Event

Table 2.
Rearrest Rates for Recidivism Study Offenders

The measure of the time to first recidivism event can be useful in
distinguishing offenders who recidivate early from those who eventually
recidivate, but are apparently crime-free for a longer interval. Tracking the
length of time to failure can also help policymakers determine an appropriate
period of supervision after the release from prison, for example by extending
supervision through the peak crime-prone interval. 41 Time from release to first
arrest is shown for all rearrests, reconvictions, and reincarcerations.

During the first year following release into the community, 16.6 percent
of the offenders in the Commission’s study were rearrested for the first time.
Each subsequent year fewer people were rearrested for the first time than in
the previous year, going out to year seven. For example, 10.5 percent of the
total were rearrested in the second year, and 6.6 percent of the total were
rearrested in the third year. An additional 1.8 percent of offenders who were
not previously arrested were rearrested in the eighth year, as demonstrated
below.
Figure 4.
Time to First Rearrest of Recidivism Study Offenders

SOURCE: U.S. Sentencing Commission’s 2005 Recidivism Release Cohort Datafile, RECID05.

SOURCE: U.S. Sentencing Commission’s 2005 Recidivism Release Cohort Datafile, RECID05. The reconviction and reconfinement lines indicate time to first arrest that led to a conviction and time to first arrest that led to a confinement, respectively.

Page 16

Most Serious Recidivism Offense
The Commission ranked new offenses in order of seriousness for those
who reoffended. When considering only the “most serious” post-release offense
committed by those who were rearrested, assault was the most prevalent.
About one-fourth (23.3%) of those rearrested had an assault rearrest as their
most serious post-release event over the study period. Other common “most
serious” offenses were public order (15.5%), drug trafficking (11.5%), and
larceny (7.7%).
Figure 5.
Recidivism Study Offenders by Most Serious Recidivism Offense

SOURCE: U.S. Sentencing Commission’s 2005 Recidivism Release Cohort Datafile, RECID05. Of the 25,431 cases in this study, the Commission excluded cases from this analysis that were missing information necessary to perform the analysis.

Page 17

Recidivism and Criminal History
The relationship between prior criminal record and recidivism has
been recognized by the Commission since its inception in the mid-1980s, as
discussed in Chapter Four of the Guidelines Manual. 42 Empirical research
assessing correlates of recidivism and criminal career behavior was consulted
in formulating the criminal history scoring system 43 and recent research
confirms this relationship. 44 The previously referenced Commission study,
Measuring Recidivism (2004), confirmed that Chapter Four’s criminal history
provisions were working as designed, and recidivism rates rise as criminal
history points increase and as Criminal History Categories increase. The
present analysis confirms that these criminal history provisions continue to
work as designed.
Figure 6.
Rearrest Rates for Recidivism Study Offenders by Criminal History Points

An offender’s total criminal history points, which determines the CHC
to which the offender is assigned for the purpose of calculating the sentencing
range under the guidelines, is designed in part to reflect recidivism potential. In
order “to protect the public from further crimes of the particular defendant, the
likelihood of recidivism and future criminal behavior must be considered.” 45
Fully consistent with its previous recidivism studies, the Commission’s present
study found that recidivism rates are most closely correlated with total criminal
history points. For example, 30.2 percent of offenders with zero total criminal
history points were rearrested within eight years, compared to 81.5 percent of
offenders with more than 10 total criminal history points. In fact, each
additional criminal history point is generally associated with a greater
likelihood of recidivism. For example, the rearrest rate of offenders with three
total criminal history points is 52.7 percent, compared to 59.4 percent for four
point offenders. This pattern continues even at higher total points, with
rearrest rates ranging from 71.1 percent (seven point offenders), 74.0 percent
(eight point offenders), and 76.1 percent (nine point offenders). 46

SOURCE: U.S. Sentencing Commission’s 2005 Recidivism Release Cohort Datafile, RECID05. Of the 25,431 cases in this study, the Commission excluded cases from this analysis that were missing information necessary to perform the analysis.

Page 18

Because an offender’s criminal history score determines the CHC to
which he or she is assigned, recidivism rates are also correlated with the CHC.
That is, the higher the CHC (a result of more prior crimes and/or more serious
crimes), the higher the recidivism rate. Rearrest rates ranged from a low of 33.8
percent for those in CHC I to a high of 80.1 percent for those in CHC VI.

Figure 7B.
Rearrest Rates for Recidivism Study
Offenders by Career Offender/Armed
Career Criminal Status

Other guideline provisions that account for an offender’s prior crimes
also serve as good predictors of future recidivism. For example, offenders
designated under the guidelines as career offenders 47 and armed career
criminals 48 receive substantially increased sentences because
they are repeat offenders with serious criminal backgrounds. These offenders
have substantially higher recidivism rates than other offenders in the study
group. In fact, those two groups of offenders have the highest recidivism rates
of any group in this report. Taken together, these offenders had a rearrest rate
of 69.5 percent, compared to a rearrest rate of 48.7 percent for all other
offenders.

Figure 7A.
Rearrest Rates for Recidivism Study Offenders by Criminal History Category

SOURCE: U.S. Sentencing Commission’s 2005 Recidivism Release Cohort Datafile, RECID05. Of the 25,431 cases in this study, the Commission excluded cases from this analysis that were missing information necessary to perform the analysis.

Page 19

Recidivism and an Offender’s Federal Offense
The Commission did not find a strong correlation between the severity
of the offender’s federal offense conduct, as determined under the sentencing
guidelines, and future recidivism. Under the guidelines, the seriousness of an
offender’s federal crime is measured by a final offense level score ranging from
one to 43. There is not a strong correspondence between final offense level and
recidivism. For example, offenders whose federal offense was assigned to the
lowest final offense levels (one through eight) had a rearrest rate of 45.2
percent, almost the same rearrest rate as those assigned the highest
final offense levels of 31 through 43 (45.7%). It should be noted, however, that
the offense levels in the federal sentencing guidelines were intended to reflect
multiple purposes of punishment, including just punishment and general
deterrence (which are unrelated to offender recidivism). 49

Although the specific numerical offense severity determined under the
sentencing guidelines appears to not be correlated with recidivism, the type of
crime the offender committed does have some correlation with the risk of
future crime. Offenders whose federal offense involved firearms were most
likely to be rearrested (68.3%), followed by those arrested for robbery (67.3%),
immigration (55.7%), drug trafficking (49.9%), larceny (44.4%), other (42.0%),
and fraud (34.2%).
Offenders who received an enhanced sentence for a weapon, either
through a conviction under 18 U.S.C. § 924(c) or by application of a specific
offense characteristic in the guidelines for having a weapon present during
commission of a crime, 50 had higher recidivism rates than other offenders.
Offenders with such a weapon enhancement had a rearrest rate of 55.4 percent,
compared to 48.6 percent for all other offenders.

Figure 8.
Rearrest Rates for Recidivism Study Offenders by Final Offense Level and Federal Offense Type

SOURCE: U.S. Sentencing Commission’s 2005 Recidivism Release Cohort Datafile, RECID05. Of the 25,431 cases in this study, the Commission excluded cases from this analysis that were missing information necessary to perform the analysis.

Page 20

Offenders who were organizers, leaders, managers, or supervisors of an
offense received an aggravating role adjustment of 2, 3, or 4 levels 51 under the
guidelines, increasing their total offense level. Those with such an aggravating
role adjustment were less likely to recidivate than other offenders. Those
receiving no adjustment (93.6 percent of all offenders in the study) had a
rearrest rate of 50.2 percent. However, those receiving aggravating role
adjustments had recidivism rates of: 37.3 percent (two level increase); 35.6
percent (three level increase); and 33.7 percent (four level increase).
Offenders who were found to have a minor or minimal role in an
offense received a mitigating role adjustment of 2, 3, or 4 levels 52 under the
guidelines, decreasing their total offense level. Offenders in the study with such
a mitigating role adjustment (10.6% of all offenders) were somewhat less likely
to recidivate than other offenders. Those receiving no adjustment had a
rearrest rate of 49.7 percent. However, those receiving mitigating role
adjustments had recidivism rates of: 47.2 percent (two level decrease); 37.5
percent (three level decrease); and 43.6 percent (four level decrease).

An offender who demonstrates acceptance of responsibility for his
offense can receive a reduction in offense level of 2 or 3 levels. 53 The vast
majority of all offenders in the Commission’s study (90.7%) received a
reduction in their guideline range for acceptance of responsibility. Such an
adjustment was not associated with lower recidivism rates. Offenders with no
adjustment under §3E1.1 had a rearrest rate of 46.3 percent. Offenders who
received a two-level decrease had a rearrest rate of 46.6 percent, and those with
a three level decrease were higher still (50.9%).

Figure 9.
Rearrest Rates for Recidivism Study Offenders by Selected Guideline Characteristics

SOURCE: U.S. Sentencing Commission’s 2005 Recidivism Release Cohort Datafile, RECID05. Of the 25,431 cases in this study, the Commission excluded cases from this analysis that were missing information necessary to perform the analysis.

Page 21

Recidivism and Sentences Imposed
Recidivism rates differ according to the type of sentence imposed. As
previously noted, 81.2 percent of the study group were sentenced to some
amount of imprisonment. These offenders had the highest rate of rearrest, 52.5
percent. Conversely, offenders sentenced to probationary sentences (18.8
percent of the study group) had a rearrest rate of 35.1 percent. 54

Offenders with shorter lengths of imprisonment generally had lower
recidivism rates. For instance, offenders with sentences of imprisonment of
fewer than six months had the lowest rearrest rate at 37.5 percent, followed by
offenders with sentences from six to 11 months (50.8 percent), and 12 to fewer
than 24 months (50.8%). Conversely, the highest recidivism rates are generally
found among offenders with longer sentences. Those with sentences from 60
months to fewer than 120 months had the highest rate (55.5%), followed
closely by those with 24 to fewer than 60 months (54.0%), and 120 months or
more (51.8%). 55

The correlation between sentence type and length and recidivism is
not, of course, entirely a coincidence. The guidelines are intended, in part, to
incapacitate offenders whose criminal records indicate a greater risk of future
criminality. Those with prison sentences are incapacitated in a manner that
those receiving no term of incarceration are not, and those receiving longer
terms of incarceration as a result of their higher CHCs are also at greater risk of
recidivism than those receiving no incarceration or a shorter period of
incarceration who generally had lower CHCs.

The most common length of supervised release imposed for those
offenders who received terms of supervised release to follow their terms of
imprisonment was 36 months. Offenders ordered to serve a term of supervised
release of three years were the majority of offenders, and this group had the
highest rearrest rate, 55.4 percent. The lowest rearrest rate was found among
offenders serving a term of supervised release of ten years or more (42.7%),
and the second lowest rate occurred with offenders serving two year terms
(46.8%).

Figure 10.
Rearrest Rates for Recidivism Study Offenders by Federal Sentences Originally Imposed

SOURCE: U.S. Sentencing Commission’s 2005 Recidivism Release Cohort Datafile, RECID05. Of the 25,431 cases in this study, the Commission excluded cases from this analysis that were missing information necessary to perform the analysis.

Page 22

Recidivism and Offender Characteristics
Studies have repeatedly shown that older offenders at sentencing are at
lower risk for reoffending, and the Commission’s research confirms these
findings. 56 Offenders sentenced when younger than twenty-one had a 71.1
percent rearrest rate, compared to 14.0 percent of offenders who are sentenced
after age sixty.

Age at release also is associated with different rates of recidivism.
Those released into the community who were below age twenty-one had the
highest rearrest rate, 67.6 percent. Conversely, those oldest at age of release,
over sixty years old, had the lowest recidivism rate, 16.0 percent. For each age
grouping shown below, the older the age group, the lower the rearrest rate. The
same pattern holds for reconviction and reincarceration.

Figure 11.
Rearrest Rates for Recidivism Study Offenders by Age at Sentencing and Release

SOURCE: U.S. Sentencing Commission’s 2005 Recidivism Release Cohort Datafile, RECID05. Of the 25,431 cases in this study, the Commission excluded cases from this analysis that were missing information necessary to perform the analysis.

Page 23

Male offenders (52.2%) were rearrested at higher rates than females
(36.4%).

Eight years after release into the community, Black offenders had been
arrested at the highest rates (59.1%), followed by Other Race (49.4%), Hispanic
(49.1%), and White (41.7%) offenders. However, the apparent relationship
between race/ethnicity and recidivism is much less pronounced if the prior
criminal history of these offenders is also examined. For offenders assigned to
higher CHCs, recidivism results are similar, regardless of race or ethnicity.
Rearrest rates for White, Black, and Hispanic offenders in CHC IV are 69.7
percent, 77.6 percent, and 75.4 percent, respectively. At CHC V for the same
three groups, the rearrest rates are 75.6, 78.5, and 79.9 percent, respectively. At
CHC VI for the same three groups, the rearrest rates are 77.1, 82.4, and 79.3
percent, respectively. Much of the difference in overall recidivism rates appears
to be the result of the differing proportion of Black offenders in CHC I (38.9%),
compared to White offenders (59.7%) and Hispanic offenders (62.2%).

Education levels are also associated with different rates of recidivism.
Offenders with less than a high school diploma had the highest recidivism rates
(60.4%), followed by high school graduates (50.7%) and those with some
college (39.3%). College graduates had the lowest rates (19.1%).

Figure 12.
Rearrest Rates for Recidivism Study Offenders by Selected Demographic Characteristics

SOURCE: U.S. Sentencing Commission’s 2005 Recidivism Release Cohort Datafile, RECID05. Of the 25,431 cases in this study, the Commission excluded cases from this analysis that were missing information necessary to perform the analysis.

Page 24

Page 25

PART V

Conclusion

Conclusion
The study of recidivism by federal offenders directly relates to multiple
statutory purposes of punishment as set forth in the Sentencing Reform Act of
1984. Using automated criminal history data, the Commission tracked 25,431
federal offenders who either discharged a federal prison sentence or
commenced a term of probation in 2005 during an eight-year follow-up period.
The Commission found that almost one-half of these federal offenders who
reentered the community in 2005 (49.3%) were rearrested for a new crime or
rearrested for a violation of supervision conditions during the follow-up period.
The Commission found that, consistent with existing research, 57 two
factors – offenders’ criminal histories and their ages at the time of release into
the community – were most closely associated with differences in recidivism
rates. Younger offenders recidivated at significantly higher rates than older
offenders, and offenders with more extensive criminal histories recidivated at
significantly higher rates than offenders with lesser criminal histories.
Regarding criminal history in particular, the Commission found that an
offenders’ total criminal history points, as determined under Chapter Four of
the Commission’s Guidelines Manual, were closely correlated with recidivism
rates.

Page 27

Other factors, including educational achievement and offense type, also
were associated with differences in recidivism rates, but less so than age and
criminal history. Certain factors related to provisions in Chapters Two and
Three of the Guidelines Manual – including adjustments in an offender’s offense
level as well as the total offense level – were not associated with differences in
rates of recidivism. However, unlike Chapter Four’s provisions – which were
based in large part on recidivism data – the guideline’s offense level
computations are based on several purposes of punishment, some of which are
unrelated to recidivism, including general deterrence and retributivist
concerns.
In the coming months, the Commission will issue additional reports
based on its recidivism study.

ENDNOTES

1

The United States Sentencing Commission (“Commission”) is an independent agency in the judicial branch of government. Established by the Sentencing Reform Act of 1984,
its principal purposes are (1) to establish sentencing policies and practices for the federal courts, including guidelines regarding the appropriate form and severity of punishment for
offenders convicted of federal crimes; (2) to advise and assist Congress, the federal judiciary, and the executive branch in the development of effective and efficient crime policy;
and (3) to collect, analyze, research, and distribute a broad array of information on federal crime and sentencing issues. See 28 U.S.C. §§ 995(a)(14), (15), (20).

2

See U.S. SENTENCING COMM’N, SUPPLEMENTARY REPORT ON THE INITIAL SENTENCING GUIDELINES AND POLICY STATEMENTS 41-44 (1987),
http://www.ussc.gov/sites/default/files/pdf/guidelines-manual/1987/manual-pdf/1987_Supplementary_Report_Initial_Sentencing_Guidelines.pdf (hereinafter SUPPLEMENTARY
REPORT); see also U.S. SENTENCING COMM’N, RECIDIVISM AMONG OFFENDERS RECEIVING RETROACTIVE SENTENCE REDUCTIONS: THE 2007 CRACK COCAINE AMENDMENT
(2014), http://www.ussc.gov/sites/default/files/pdf/research-and-publications/research-projects-andsurveys/miscellaneous/20140527_Recidivism_2007_Crack_Cocaine_Amendment.pdf (hereinafter CRACK COCAINE RECIDIVISM REPORT); U.S. SENTENCING COMM’N, REPORT TO
CONGRESS: FEDERAL CHILD PORNOGRAPHY OFFENSES 293-310 (2012), HTTP://WWW.USSC.GOV/SITES/DEFAULT/FILES/PDF/NEWS/CONGRESSIONAL-TESTIMONY-AND-REPORTS/SEXOFFENSE-TOPICS/201212-FEDERAL-CHILD-PORNOGRAPHY-OFFENSES/FULL_REPORT_TO_CONGRESS.PDF (hereinafter CHILD PORNOGRAPHY REPORT); U.S. SENTENCING COMM’N, A
COMPARISON OF THE FEDERAL SENTENCING GUIDELINES CRIMINAL HISTORY CATEGORY AND THE U.S. PAROLE COMMISSION SALIENT FACTOR SCORE (2005),
http://www.ussc.gov/sites/default/files/pdf/research-and-publications/research-publications/2005/20050104_Recidivism_Salient_Factor_Computation.pdf (hereinafter SALIENT
FACTOR SCORE); U.S. SENTENCING COMM’N, RECIDIVISM AND THE “FIRST OFFENDER” (2004), http://www.ussc.gov/sites/default/files/pdf/research-and-publications/researchpublications/2004/200405_Recidivism_First_Offender.pdf (hereinafter FIRST OFFENDER); U.S. SENTENCING COMM’N, MEASURING RECIDIVISM: THE CRIMINAL HISTORY
COMPUTATION OF THE FEDERAL SENTENCING GUIDELINES (2004), http://www.ussc.gov/sites/default/files/pdf/research-and-publications/researchpublications/2004/200405_Recidivism_Criminal_History.pdf (hereinafter MEASURING RECIDIVISM).
3

18 U.S.C. § 3553(a)(2)(C).

4

The SRA provides that:
The court, in determining the particular sentence to be imposed, shall consider. . . the need for the sentence imposed—
(A) to reflect the seriousness of the offense, to promote respect for the law, and to provide just punishment for the offense;
(B) to afford adequate deterrence to criminal conduct;
(C) to protect the public from further crimes of the defendant; and
(D) to provide the defendant with needed educational or vocational training, medical care, or other correctional treatment in the most effective manner.

Id.; see also 28 U.S.C. § 991(b) (“[T]he United States Sentencing Commission [shall] establish sentencing policies and practices for the Federal criminal justice system that . . .
assure the meeting of the purposes of sentencing as set forth in section 3553(a)(2) of title 18, United States Code.”). The SRA requires courts and the Commission to consider both
general and specific deterrence. See United States v. Martin, 455 F.3d 1227, 1240 (11th Cir. 2006).
5

28 U.S.C. § 991(b)(2).

6

See SUPPLEMENTARY REPORT, supra note 2, at 41-44.

Page 29

7

See MEASURING RECIDIVISM, supra note 2.

8

See generally SALIENT FACTOR SCORE, supra note 2; FIRST OFFENDER, supra note 2.

9

See CRACK COCAINE RECIDIVISM REPORT, supra note 2; CHILD PORNOGRAPHY REPORT, supra note 2, at 293-310.

10

See, e.g., Final Priorities for Amendment Cycle, 79 FED. REG. 49378, 49379 (Aug. 20, 2014) (“Pursuant to 28 U.S.C. § 994(g), the Commission intends to consider the issue of
reducing costs of incarceration and overcapacity of prisons, to the extent it is relevant to any identified priority.”).

11

See, e.g., U.S. SENTENCING COMM’N, ANNUAL REPORT: FISCAL YEAR 2014, A-5 (2014).

12

Courtney Semisch, U.S. SENTENCING COMM’N, ALTERNATIVE SENTENCING IN THE FEDERAL CRIMINAL JUSTICE SYSTEM (2015),
http://www.ussc.gov/sites/default/files/pdf/research-and-publications/research-projects-and-surveys/alternatives/20150617_Alternatives.pdf.
13

FEDERAL BUREAU OF PRISONS, Statistics, https://web.archive.org/web/20160211120327/https://www.bop.gov/about/statistics/population_statistics.jsp (Feb. 11, 2016) (reporting
195,730 inmates, including 44,362 non-citizens).
14

BUREAU OF JUSTICE STATISTICS, PRISONERS IN 2014, http://www.bjs.gov/content/pub/pdf/p14.pdf (1,561,500 state prisoners as of 2014).

15

See USSG, App’x C, amend. 782 (“Reason for Amendment”) (in voting to apply an amendment to the drug-trafficking guidelines that reduced offense levels for many
offenders, the Commission noted that it “was informed by its studies that compared recidivism rates for offenders who were released early as a result of retroactive application of
the Commission’s 2007 crack cocaine amendment with a control group of offenders who served their full terms of imprisonment”).

16

Judges are required to apply the federal sentencing guidelines, now “advisory” guidelines following United States v. Booker, 543 U.S. 220 (2005), before deciding what
sentence to impose and these guidelines continue to have a significant policy effect on federal sentencing. See, e.g., Peugh v. United States, 133 S. Ct. 2072, 2084 (2013)
(“[C]onsiderable empirical evidence indicat[es] that the Sentencing Guidelines have the intended effect of influencing the sentences imposed by judges.”).
17

See NAT’L INST. OF JUSTICE, Recidivism, https://web.archive.org/web/20160120175242/http://www.nij.gov/topics/corrections/recidivism/pages/welcome.aspx (Jan. 20, 2016).

18
See MICHAEL D. MALTZ, RECIDIVISM 7-20 (2001); see also Ryan King & Brian Elderboom, Improving Recidivism as a Performance Measure, URBAN INSTITUTE (2014),
https://www.bja.gov/Publications/UI-ImprovingRecidivism.pdf.
19

See, e.g., Christopher T. Lowenkamp, Marie VanNostrand, and Alexander Holsinger, INVESTIGATING THE IMPACT OF PRETRIAL DETENTION ON SENTENCING OUTCOMES (2013),
http://pretrialnola.org/wp-content/uploads/2015/09/Investigating-the-Impact-of-Pretrial-Detention-on-Sentencing-Outcomes-2013.pdf; Elizabeth

Drake, Steve Aos, and Marna Miller, Evidence Based Public Policy Options to Reduce Crime and Criminal Justice Costs: Implications in Washington State, in 4 VICTIMS AND
OFFENDERS 170-96 (2009); NAT’L RESEARCH COUNCIL, PAROLE, DESISTANCE FROM CRIME, AND COMMUNITY INTEGRATION, 19-31 (2008).
20

Recidivism, supra note 18 (“[R]ecidivism is measured by rearrest, reconviction or return to prison.”); See also, NAT’L INST. OF JUSTICE, Measuring Recidivism,
https://web.archive.org/web/20160129195540/http://www.nij.gov/topics/corrections/recidivism/pages/measuring.aspx (Jan. 29, 2016) (“The timing of recidivism is key not only to
its measurement but also to understanding the processes underlying the effects of sanctions and interventions with respect to the propensity of the individual to commit crime.”)
21

See MALTZ, supra note 18, at 20-21, 28-29, 55-58; Improving Recidivism as a Performance Measure, supra note 19.

Page 30

22

Arrests for alleged violations of probation, supervised release, or state parole (as well as actual revocations of any these types of supervision) were counted as rearrests.
Revocations were not counted as reconvictions because the offenders were not convicted of a new offense (even if the basis for revocation was a “new law violation”). Offenders
whose terms of supervision were revoked and who were sentenced to imprisonment were treated as reincarcerations.

23

See MALTZ, supra note 18; Improving Recidivism as a Performance Measure, supra note 18.

24

Matthew Durose, Alexia Cooper, and Howard Snyder, BUREAU OF JUSTICE STATISTICS, Recidivism of Prisoners Released in 30 States in 2005: Patterns from 2005 to 2010
(2014) (hereinafter Prisoners Released in 30 States in 2005), http://www.bjs.gov/content/pub/pdf/rprts05p0510.pdf.
25

William Rhodes, et al., ABT ASSOCIATES, Recidivism of Offenders on Federal Community Supervision (2012), https://www.ncjrs.gov/pdffiles1/bjs/grants/241018.pdf.

26
Recidivism of Prisoners Released in 30 States in 2005, supra note 24 (discussing the methodology employed by BJS for collecting and processing criminal records for
recidivism research). The Bureau of Justice Statistics is the United States' primary source for criminal justice statistics, including statistics on crime, criminal offenders, victims of
crime, and the operation of justice systems at all levels of government. See http://www.bjs.gov/index.cfm?ty=abu. Appendix B of this report discusses the procedures employed
by the Commission.
27

All of the federal offenders who re-entered the community after discharging their sentences of incarceration had served those sentences continuously from the time of the
imposition of the sentences until the time of the offenders’ re-entry in 2005. None of the offenders released from their sentences of incarceration in 2005 had previously been
returned to prison upon revocation of probation or supervised release terms imposed in connection with their instant federal offenses of conviction.
28

See Prisoners Released in 30 States in 2005, supra note 24.

29

Immigration crimes committed by citizens are most commonly sentenced under USSG §2L1.1 (alien smuggling) rather than USSG §2L1.2 (illegal reentry).

30

See generally USSG §§4A1.1; 4A1.2.

31

543 U.S. 220 (2005).

32

These four different sentence types correspond to the four “Zones” (A-D) in the Sentencing Table in the Guidelines Manual. See USSG, Ch. 5, Pt. A (Sentencing Table); see
also USSG §§5B1.1 & 5C1.1 (setting forth the sentencing options for Zones A-D). Zone A authorizes probation only; Zone B authorizes probation with a condition of
confinement; Zone C authorizes a “split” sentence of imprisonment and community confinement (e.g., home detention or a halfway house); and Zone D authorizes sentences of
imprisonment only. See USSG §§5B1.1 & 5C1.1.

33

Of the offenders who received a Zone B sentence of probation with a condition of confinement, only 0.9 percent served the confinement in a jail or prison. The remainder
served their confinement in home detention or in community confinement such as a halfway house.
34

The small percentage of offenders who received a fine only (65 members of the study group) are excluded from the analysis.

35

Imprisonment sentences are sentences to a term of confinement in the Federal Bureau of Prisons (BOP) with or without a sentence to community confinement or probation to
follow (as a condition of supervised release pursuant to USSG §5C1.1(c)(2) or (d)(2)). Probation sentences may include either a period of community confinement. The term
alternative confinement refers to a period of confinement in a location other than a BOP facility, usually home detention or a halfway house.
Typically, federal offenders sentenced to terms of imprisonment of less than one year serve their sentences in a local jail or detention center rather than a federal prison. See
FED. BUREAU OF PRISONS, LEGAL RESOURCE GUIDE TO THE FEDERAL BUREAU OF PRISONS, 2014, 10 (2014),

Page 31

https://www.bop.gov/resources/pdfs/legal_guide.pdf (“The BOP will often contract with a local jail or detention center to house an inmate sentenced to a term of one year or
less.”). Regardless of whether a federal offender served his or her sentence in a jail or prison, that sentence will be deemed a “sentence of imprisonment” for the analyses that
follow.
36

U.S. SENTENCING COMM’N, FEDERAL OFFENDERS SENTENCED TO SUPERVISED RELEASE 4-9 (2010).

37

Booker, 543 U.S. 220.

38

Id. at 245.

39

U.S. SENTENCING COMM’N, REPORT ON THE CONTINUING IMPACT OF U.S. V. BOOKER ON FEDERAL SENTENCING 7 (2012).

40

See Prisoners Released in 30 States in 2005, supra note 24. The inmates released from 30 state prisons in 2005 included in the BJS study differ from the population released
from federal prison that same year in several respects. A quarter (25.7 percent) of released state inmates had a violent commitment offense compared to only 6.8 percent of
inmates released from federal prison. State offenders were more likely to be under 40 (68.5 percent) and male (89.3 percent) than the federal offenders (60.1 percent under age 40
and 85.9 percent male). The BJS study also included non-U.S. citizens, a category of offender excluded from the Commission’s study.

41

The National Advisory Commission on Criminal Justice Standards and Goals recommends a three year follow-up period, but recognizes this as “an arbitrary figure.” NATIONAL
ADVISORY COMMISSION ON CRIMINAL JUSTICE STANDARDS AND GOALS, CORRECTIONS 529 (1973)
42

See USSG, Ch.4, Pt.A, intro. comment.

43

See SUPPLEMENTARY REPORT, supra note 2, at 41-44.

44

See RHODES, supra note 25 at 12; Alfred Blumstein, David P. Farrington, and Soumyo Moitra, CRIME AND JUSTICE: A REVIEW OF RESEARCH 216 (1985),
Delinquency Careers: Innocents, Desisters, and Persisters. In two major cohort studies, the recidivism probability of youthful offenders rose with successive involvements with
law enforcement; Paul Gendreau, Tracy Little, and Claire Goggin, Criminology, (2008), A Meta-Analysis Of The Predictors Of Adult Offender Recidivism: What Works! at 575. A
meta-analysis of 131 studies found the strongest predictors of recidivism included criminogenic needs, criminal history, social achievement, age/gender/race, and family factors.

45

USSG, Ch.4, Pt.A, intro. comment.

46

This group falls within CHC IV.

47

Career Offenders are persons who commit a crime of violence or drug trafficking crime after being convicted of two prior felony drug trafficking or crime of violence offenses.
See USSG §4B1.1.
48

Armed Career Criminals are persons subject to an enhanced sentence under the provisions of 18 U.S.C. § 924(e). USSG §4B1.4.

49

See, e.g., USSG §2H4.1, comment (backg’d) (1995) (“For purposes of deterrence and just punishment, the minimum base offense level is 15.”).

50

See, e.g., USSG §2D1.1(b)(1).

51

USSG §3B1.1.

Page 32

52

USSG §3B1.2.

53

USSG §3E1.1.

54
A more detailed analysis of offenders based on the type of sentence imposed – see supra note 32 (discussing the four types of sentences that correspond to Zones A-D of the
Sentencing Table) – is as follows: Offenders who received probation only had a rearrest rate of 34.5 percent; offenders who received probation with a condition of confinement
had a rearrest rate of 36.9 percent; offenders who received a “split” sentence had a rearrest rate of 37.2 percent; and offenders who received a sentence of imprisonment only had a
rearrest rate of 53.3 percent.
55

4,870 cases with zero months of imprisonment ordered were excluded.

56

See GENDREAU, Supra note 44 at 575; David P. Farrington, Age and Crime, in Crime and Justice: An Annual Review of Research Vol. 7 (Michael Tonry and Norval Morris,
eds., 1986); Jeffery T. Ulmer and Darrell Steffensmeier, The Age and Crime Relationship: Social Variation, Social Explanations in The Nurture versus Biosocial Debate in
Criminology 378 (K. Beaver, B. Boutwell, and J.C. Barnes, eds., 2014). “It is now a truism that age is one of the strongest factors associated with criminal behavior.”

57

See GENDREAU, supra note 44 at 575.

Page 33

APPENDIX A-1

Rearrest Rates Across Selected Variables

Rearrest Rates Across Selected Variables
Criminal History

Appendix A-1

Total

N

%

Total

25,431

12,527

49.3%

Criminal History Category
CHC I
CHC II
CHC III
CHC IV
CHC V
CHC VI

13,581
3,084
3,616
1,996
1,121
1,923

4,594
1,674
2,288
1,490
872
1,541

33.8%
54.3%
63.3%
74.7%
77.8%
80.1%

Criminal History Points
0
1
2
3
4
5
6
7
8
9
10
More than 10

10,600
2,973
1,251
1,838
1,361
1,041
1,277
693
741
708
463
2,400

3,196
1,394
701
969
809
659
853
493
548
539
358
1,956

30.2%
46.9%
56.0%
52.7%
59.4%
63.3%
66.8%
71.1%
74.0%
76.1%
77.3%
81.5%

Career Offender/Armed Career Criminal Status
No Career Offender/Armed Career Criminal
Career Offender/Armed Career Criminal

24,798
633

12,087
440

48.7%
69.5%

Rearrest Rates Across Selected Variables
Offense Level and Offense Type
Final Offense Level
1 to 8
9 to 10
11 to 12
13 to 16
17 to 21
22 to 25
26 to 30
31 to 43
Federal Offense Type
Drug Trafficking
Firearms
Fraud
Robbery
Larceny
Immigration
All Other

Total

N

%

3,020
1,598
1,972
3,981
5,283
3,676
3,214
2,580

1,364
679
1,023
1,949
2,783
1,945
1,538
1,180

45.2%
42.5%
51.9%
49.0%
52.7%
52.9%
47.9%
45.7%

10,591
3,244
3,450
1,100
994
891
5,159

5,288
2,216
1,181
740
441
496
2,165

49.9%
68.3%
34.2%
67.3%
44.4%
55.7%
42.0%

Appendix A-1

Rearrest Rates Across Selected Variables
Offense Characteristics

Appendix A-1

Total

N

%

Weapon Enhancement
No Weapon Enhancement
Weapon Enhancement

22,932
2,499

11,143
1,384

48.6%
55.4%

Aggravating Role
No Adjustment
+2 Levels
+3 Levels
+4 Levels

23,789
872
315
439

11,931
325
112
148

50.2%
37.3%
35.6%
33.7%

Mitigating Role
No Adjustment
-2 Levels
-3 Levels
-4 Levels

22,726
2,010
232
447

11,287
948
87
195

49.7%
47.2%
37.5%
43.6%

Acceptance of Responsibility
No Adjustment
-2 Levels
-3 Levels

2,372
7,297
15,723

1,099
3,403
8,003

46.3%
46.6%
50.9%

Rearrest Rates Across Selected Variables
Sentences Imposed

Total

N

%

4,754
20,575

1,670
10,810

35.1%
52.5%

1,048
762
3,655
8,023
4,552
2,521

393
387
1,857
4,334
2,525
1,306

37.5%
50.8%
50.8%
54.0%
55.5%
51.8%

Sentence Relative to the Guideline Range
Within Range
Above Range
5K1.1 Departure
Other Government Sponsored Below Range
Non-Government Sponsored Below Range

15,680
197
5,112
520
2,134

7,881
117
2,430
261
964

50.3%
59.4%
47.5%
50.2%
45.2%

Length of Supervised Release
No Supervised Release
Less than 2 Years
2 Years
3 Years
4 Years
5 to 9 Years
10 or More Years

121
627
1,736
11,097
1,793
5,068
124

60
307
813
6,149
864
2,559
53

49.6%
49.0%
46.8%
55.4%
48.2%
50.5%
42.7%

Sentence Type
Probation Only
Prison Only
Length of Sentence
Up to 6 Months
6 to 11 Months
12 to 23 Months
24 to 59 Months
60 to 119 Months
120 Months or More

Appendix A-1

Rearrest Rates Across Selected Variables
Offender Characteristics

Total

N

%

Age at Sentencing
Younger than 21
21 to 25
26 to 30
31 to 35
36 to 40
41 to 50
51 to 60
Older than 60

1,226
4,737
4,746
3,895
3,347
4,569
2,125
741

872
3,105
2,755
1,984
1,587
1,639
462
104

71.1%
65.6%
58.1%
50.9%
47.4%
35.9%
21.7%
14.0%

Age at Release
Younger than 21
21 to 25
26 to 30
31 to 35
36 to 40
41 to 50
51 to 60
Older than 60

398
2,986
4,325
4,584
3,762
5,551
2,732
1,048

269
1,984
2,692
2,533
1,834
2,353
675
168

67.6%
66.4%
62.2%
55.3%
48.8%
42.4%
24.7%
16.0%

20,758
4,664

10,825
1,696

52.2%
36.4%

11,099
8,617
4,512
1,179

4,628
5,089
2,213
582

41.7%
59.1%
49.1%
49.4%

8,656
9,324
5,409
1,884

5,230
4,724
2,126
359

60.4%
50.7%
39.3%
19.1%

Gender
Male
Female
Race/Ethnicity
White
Black
Hispanic
Other
Level of Education
Less than High School
High School Graduate
Some College
College Graduate

Appendix A-1

APPENDIX A-2

Reconviction Rates Across Selected Variables

Reconviction Rates Across Selected Variables
Criminal History

Total

Appendix A-2

N

%

Total

25,431

8,062

31.7%

Criminal History Category
CHC I
CHC II
CHC III
CHC IV
CHC V
CHC VI

13,581
3,084
3,616
1,996
1,121
1,923

2,701
1,019
1,494
1,030
634
1,140

19.9%
33.0%
41.3%
51.6%
56.6%
59.3%

Criminal History Points
0
1
2
3
4
5
6
7
8
9
10
More than 10

10,600
2,973
1,251
1,838
1,361
1,041
1,277
693
741
708
463
2,400

1,844
856
432
587
530
435
543
320
385
377
246
1,473

17.4%
28.8%
34.5%
31.9%
38.9%
41.8%
42.5%
46.2%
52.0%
53.3%
53.1%
61.4%

Career Offender/Armed Career Criminal Status
No Career Offender/Armed Career Criminal
Career Offender/Armed Career Criminal

24,798
633

7,761
301

31.3%
47.6%

Reconviction Rates Across Selected Variables
Offense Level and Offense Type
Final Offense Level
1 to 8
9 to 10
11 to 12
13 to 16
17 to 21
22 to 25
26 to 30
31 to 43
Federal Offense Type
Drug Trafficking
Firearms
Fraud
Robbery
Larceny
Immigration
All Other

Total

N

%

3,020
1,598
1,972
3,981
5,283
3,676
3,214
2,580

898
437
688
1,291
1,843
1,217
943
704

29.7%
27.4%
34.9%
32.4%
34.9%
33.1%
29.3%
27.3%

10,591
3,244
3,450
1,100
994
891
5,159

3,259
1,541
731
501
296
336
1,398

30.8%
47.5%
21.2%
45.6%
29.8%
37.7%
27.1%

Appendix A-2

Reconviction Rates Across Selected Variables
Offense Characteristics

Appendix A-2

Total

N

%

Weapon Enhancement
No Weapon Enhancement
Weapon Enhancement

22,932
2,499

7,177
885

31.3%
35.4%

Aggravating Role
No Adjustment
+2 Levels
+3 Levels
+4 Levels

23,789
872
315
439

7,704
204
68
81

32.4%
23.4%
21.6%
18.5%

Mitigating Role
No Adjustment
-2 Levels
-3 Levels
-4 Levels

22,276
2,010
232
447

7,298
587
47
125

32.1%
29.2%
20.3%
28.0%

Acceptance of Responsibility
No Adjustment
-2 Levels
-3 Levels

2,372
7,297
15,723

661
2,244
5,143

27.9%
30.8%
32.7%

Reconviction Rates Across Selected Variables
Sentences Imposed

Total

N

%

4,754
20,575

1,028
7,007

21.6%
34.1%

1,048
762
3,655
8,023
4,552
2,521

246
243
1,239
2,840
1,613
819

23.5%
31.9%
33.9%
35.4%
35.4%
32.5%

Sentence Relative to the Guideline Range
Within Range
Above Range
5K1.1 Departure
Other Government Sponsored Below Range
Non-Government Sponsored Below Range

15,680
197
5,112
520
2,134

5,115
88
1,526
168
602

32.6%
44.7%
29.9%
32.3%
28.2%

Length of Supervised Release
No Supervised Release
Less than 2 Years
2 Years
3 Years
4 Years
5 to 9 Years
10 or More Years

121
627
1,736
11,097
1,793
5,068
124

41
188
533
4,130
524
1,558
30

33.9%
30.0%
30.7%
37.2%
29.2%
30.7%
24.2%

Sentence Type
Probation Only
Prison Only
Length of Sentence
Up to 6 Months
6 to 11 Months
12 to 23 Months
24 to 59 Months
60 to 119 Months
120 Months or More

Appendix A-2

Reconviction Rates Across Selected Variables
Offender Characteristics

Total

N

%

Age at Sentencing
Younger than 21
21 to 25
26 to 30
31 to 35
36 to 40
41 to 50
51 to 60
Older than 60

1,226
4,737
4,746
3,895
3,347
4,569
2,125
741

635
2,128
1,764
1,290
989
931
253
59

51.8%
44.9%
37.2%
33.1%
29.6%
20.4%
11.9%
8.0%

Age at Release
Younger than 21
21 to 25
26 to 30
31 to 35
36 to 40
41 to 50
51 to 60
Older than 60

398
2,986
4,325
4,584
3,762
5,551
2,732
1,048

193
1,424
1,779
1,614
1,151
1,428
366
94

48.5%
47.7%
41.1%
35.2%
30.6%
25.7%
13.4%
9.0%

20,758
4,664

7,022
1,035

33.8%
22.2%

11,099
8,617
4,512
1,179

2,999
3,237
1,409
408

27.0%
37.6%
31.2%
34.6%

8,656
9,324
5,409
1,884

3,484
3,041
1,279
196

40.3%
32.6%
23.7%
10.4%

Gender
Male
Female
Race/Ethnicity
White
Black
Hispanic
Other
Level of Education
Less than High School
High School Graduate
Some College
College Graduate

Appendix A-2

APPENDIX A-3

Reincarceration Rates Across Selected Variables

Reincarceration Rates Across Selected Variables
Criminal History

Appendix A-3

Total

N

%

Total

25,431

6,266

24.6%

Criminal History Category
CHC I
CHC II
CHC III
CHC IV
CHC V
CHC V

13,581
3,084
3,616
1,996
1,121
1,923

1,897
737
1,189
868
554
986

14.0%
23.9%
32.9%
43.5%
49.4%
51.3%

Criminal History Points
0
1
2
3
4
5
6
7
8
9
10
More than 10

10,600
2,973
1,251
1,838
1,361
1,041
1,277
693
741
708
463
2,400

1,296
600
303
434
401
355
442
263
322
326
215
1,281

12.2%
20.2%
24.2%
23.6%
29.5%
34.1%
34.6%
38.0%
43.5%
46.1%
46.4%
53.4%

Career Offender/Armed Career Criminal Status
No Career Offender/Armed Career Criminal
Career Offender/Armed Career Criminal

24,798
633

6,012
254

24.2%
40.1%

Reincarceration Rates Across Selected Variables
Offense Level and Offense Type

Final Offense Level
1 to 8
9 to 10
11 to 12
13 to 16
17 to 21
22 to 25
26 to 30
31 to 43
Federal Offense Type
Drug Trafficking
Firearms
Fraud
Robbery
Larceny
Immigration
All Other

Total

N

%

3,020
1,598
1,972
3,981
5,283
3,676
3,214
2,580

631
320
536
1,034
1,488
934
744
546

20.9%
20.0%
27.2%
26.0%
28.2%
25.4%
23.2%
21.2%

10,591
3,244
3,450
1,100
994
891
5,159

2,475
1,258
514
443
225
272
1,079

23.4%
38.8%
14.9%
40.3%
22.6%
30.5%
20.9%

Appendix A-3

Reincarceration Rates Across Selected Variables
Offense Characteristics

Appendix A-3

Total

N

%

Weapon Enhancement
No Weapon Enhancement
Weapon Enhancement

22,932
2,499

5,550
716

24.2%
28.7%

Aggravating Role
No Adjustment
+2 Levels
+3 Levels
+4 Levels

23,789
872
315
439

5,994
154
46
67

25.2%
17.7%
14.6%
15.3%

Mitigating Role
No Adjustment
-2 Levels
-3 Levels
-4 Levels

22,276
2,010
232
447

5,693
442
34
92

25.1%
22.0%
14.7%
20.6%

Acceptance of Responsibility
No Adjustment
-2 Levels
-3 Levels

2,372
7,297
15,723

524
1,664
4,066

22.1%
22.8%
25.9%

Reincarceration Rates Across Selected Variables
Sentences Imposed
Total

N

%

4,754
20,575

624
5,620

13.1%
27.3%

1,048
762
3,655
8,023
4,552
2,521

169
180
1,015
2,293
1,283
676

16.1%
23.6%
27.8%
28.6%
28.2%
26.8%

Sentence Relative to the Guideline Range
Within Range
Above Range
5K1.1 Departure
Other Government Sponsored Below Range
Non-Government Sponsored Below Range

15,680
197
5,112
520
2,134

4,021
83
1,132
137
460

25.6%
42.1%
22.1%
26.4%
21.6%

Length of Supervised Release
No Supervised Release
Less than 2 Years
2 Years
3 Years
4 Years
5 to 9 Years
10 or More Years

121
627
1,736
11,097
1,793
5,068
124

35
144
421
3,361
396
1,242
19

28.9%
23.0%
24.3%
30.3%
22.1%
24.5%
15.3%

Sentence Type
Probation
Prison
Length of Sentence
Up to 6 Months
6 to 11 Months
12 to 23 Months
24 to 59 Months
60 to 119 Months
120 Months or More

Appendix A-3

Reincarceration Rates Across Selected Variables
Offender Characteristics
Total

N

%

Age at Sentencing
Younger than 21
21 to 25
26 to 30
31 to 35
36 to 40
41 to 50
51 to 60
Older than 60

1,226
4,737
4,746
3,895
3,347
4,569
2,125
741

528
1,679
1,382
994
738
707
187
42

43.1%
35.4%
29.1%
25.5%
22.1%
15.5%
8.8%
5.7%

Age at Release
Younger than 21
21 to 25
26 to 30
31 to 35
36 to 40
41 to 50
51 to 60
Older than 60

398
2,986
4,325
4,584
3,762
5,551
2,732
1,048

142
1,137
1,392
1,246
882
1,108
281
69

35.7%
38.1%
32.2%
27.2%
23.4%
20.0%
10.3%
6.6%

20,758
4,664

5,560
703

26.8%
15.1%

11,099
8,617
4,512
1,179

2,266
2,531
1,111
351

20.4%
29.4%
24.6%
29.8%

8,656
9,324
5,409
1,884

2,809
2,344
929
134

32.5%
25.1%
17.2%
7.1%

Gender
Male
Female
Race/Ethnicity
White
Black
Hispanic
Other
Level of Education
Less than High School
High School Graduate
Some College
College Graduate

Appendix A-3

APPENDIX B

Data Collection and Analysis Process

Appendix B: Overview
Previous Commission recidivism studies required manual coding of
criminal records from state and federal criminal history records (i.e., RAP
sheets). This project has used technological advances to greatly expand the
number of federal offender records analyzed as compared to previous
Commission studies by collecting RAP sheets electronically. This appendix
describes this data collection and analysis process.

The Commission entered into a data sharing agreement with the FBI’s
Criminal Justice Information Services (CJIS) Division and the Administrative
Office of the United States Courts (AO) to provide the Commission with secure
electronic access to criminal history records through the CJIS’s Interstate
Identification Index (III) and the International Justice and Public Safety Network
(NLETS). Results received using this system provide an individual’s criminal
history record maintained by all U.S. states, the District of Columbia, and federal
agencies. 1 Once automated records were obtained, the Commission went
through an extensive process to organize and standardize offense and court
disposition descriptions across all reporting agencies with the assistance of
NORC at the University of Chicago (NORC). The resulting database contained
30,182 offenders who had valid identifying information and were released
during the study period, which included primarily calendar year 2005 (25,431
offenders) but was extended before and after this year to expand the sample for
certain subgroups of offenders (4,751 offenders). This process, described
below, began in 2014 and the database was completed in August 2015.

Collecting the Study Group from Commission and Other Records
Identifying information, including FBI numbers required by the NLETS
system, was collected on 36,007 offenders from Commission, Federal Bureau of
Prisons (BOP), and AO sources. The primary study cohort includes all federal
offenders who were either released from federal prison after serving a sentence
of imprisonment or were placed on probation in 2005. For offenders released
from prison, the BOP provided release dates, identifying information, some
reincarceration information, and other relevant information which allowed the
Commission to identify offenders who could not be reliably studied (e.g., those
who died while incarcerated). For offenders placed on probation, the AO

Appendix B

provided identifying information, some revocation information, and other
relevant information (e.g., death while under supervision).

Criminal history records were also drawn for two secondary offender
groups for inclusion in possible future studies. The first group selected for
additional collection was citizen offenders sentenced under USSG §4B1.1
(Career Offenders) and §4B1.4 (Armed Career Criminals) re-entering the
community in calendar years 2004 and 2006 (in addition to those already
included in the 2005 cohort). The second group included additional citizen
offenders whose guideline calculations placed them in Zones B or C of the
Sentencing Table. 2 In addition to those offenders in Zones B and C who
reentered during the 2005 study period, these additional offenders expanded
the reentry period to include those reentering the community from May 1, 2003
(beginning of the PROTECT Act period) 3 through calendar year 2006. The
purpose of expanding the initial study group in this way was to increase the
number of offenders eligible for probation or a shorter incarcerative sentence.

Processing the Criminal Records
Following a practice pioneered by the U.S. Department of Justice’s
Bureau of Justice Statistics (BJS), 4 the Commission entered into a data sharing
agreement with the FBI’s CJIS Division and the Administrative Office of the
United States Courts to provide the Commission with electronic access to
criminal history records through the CJIS’s Interstate Identification Index (III)
using secure data exchange, which protected these confidential records from
inadvertent disclosure. The CJIS’s III allows authorized agencies to determine
whether any federal or state repository has criminal history records on an
individual. When an offender’s records were found, each central criminal
records repository, responding to III requests over the NLETS network,
provided the records via the Administrative Office’s Access to Law Enforcement
System (ATLAS) secure interface to the Commission. As a result, offender
criminal history records were collected from all state and federal agencies in
which those records resided.
The ATLAS system returns the literal text in the RAP sheets in the
format in which the original records appear: dates of criminal justice system
actions (e.g., arrests); offense categories which indicate the charges in the
terminology used by that agency (e.g., text strings or numeric categories);
subsequent action tied to arrest charges (e.g., charges filed by prosecutors,
court findings of guilt, etc.); and sentencing and corrections information. All of
these records are subject to availability from the originating source.

The ATLAS system also “parses” records from RAP sheets received
from all the states, the District of Columbia, and federal agencies. Parsing
records involves organizing key data elements into logical components: sorting
into separate criminal justice system stages (arrest, court, and correctional
events); organizing all records by date; and linking related elements (such as
court action taken in response to an arrest) into a single cycle. Key data
elements include offender identifiers, dates of key actions (such as arrests and
convictions), the criminal charges, and outcomes such as convictions and
sentencing information. The parsing process collates the multi-state records
into a uniform structure, regardless of the state, and produces a database for all
individuals with a valid FBI number who were found in one or more
repositories across the country.

The initial extraction process began in September 2014 and was
completed in April 2015. Commission staff examined the raw RAP sheet
information and compared it to the parsed records, looking for missing or
erroneous translation of information from its raw form to the parsed record.
This step is important because criminal history record repositories are
continually updating and improving their information systems in ways that
change the location or format of key text strings, making software written for an
earlier computer platform obsolete for purposes of accurately parsing the data
in its intended format and detail. Commission staff worked with ATLAS staff to
resolve possible issues in about one-half of the repositories, and as a result
records for several state repositories had to be reprocessed after changes to the
parsing procedures. In April 2015, this process was complete.

Standardizing the Criminal Records
Following the collection of the RAP sheet records on study group
offenders, the Commission contracted with NORC to consolidate all records on
each offender, organize the records chronologically, and remove duplicative or
extraneous material. 5 Minor traffic offenses (e.g., speeding) were removed, but
serious traffic-related offenses (e.g., driving while intoxicated) were not
removed from the analysis. Following these preliminary process steps, NORC
researched state and federal criminal codes and repository data definitions to
assign each unique state offense and disposition description to a single uniform
description. This step was needed because criminal records repositories are
primarily designed to store their records in ways that accurately reflect the
requirements of each state or federal repository, such as the criminal code for

that jurisdiction. As a result, any two repositories are likely to use many unique
text strings to indicate the nature of the criminal charges and actions taken in
response to those charges. Each jurisdiction’s information was standardized by
NORC for national-level analysis that reflected common definitions.

Using research on each repository, NORC created separate offense
“crosswalks” for every state and the District of Columbia, as well as a separate
crosswalk for federal repository records. These crosswalks translate any given
jurisdiction’s arrest and court records into standardized arrest and court codes.
Through this standardization, all records, regardless of originating source, were
brought into a common framework.

Within each arrest and court cycle, arrest and disposition of charges
were categorized using standardized coding which generally follows the BJS
model. 6 A charge severity index was created which incorporates both criminal
law classification (e.g., felony or misdemeanor) and offense severity. Offense
severity was first separated into four broad categories (violent, property, drug,
and public order), and then into 16 major arrest charge codes 7 and 98 detailed
arrest charge codes used by BJS. 8 The charge severity index ranks murder of a
public officer as the most serious charge, followed by 26 other detailed violent
charge codes. These detailed violent charge codes are followed in order of
severity by five drug trafficking, 16 property, 10 other drug, and 40 public order
detailed charge codes. Additionally, court events are coded for charge
disposition. After ordering, each charge is assigned a sequence number, and the
top three for each event are retained.
Finally, by assembling all federal and state records, the database
provides a complete criminal record for all individuals for which valid records
could be found, across any jurisdiction maintaining data on that individual.
After the receipt of data from NORC, the Commission again reviewed all records
for completeness and questionable entries. From the 36,007 offender records
initially processed through ATLAS, 1,812 records were eliminated for reasons
including apparent death during the study period, missing criminal history
records, and missing or suspect information as to United States citizenship.
Additionally, 4,013 offenders were released from BOP on detainer, which
ordinarily indicates transfer of custody to state court or transfer to a state
correctional facility following completion of their federal sentence. These
detained offenders were not released into the community during the study
period, and their whereabouts are not recorded in the data; therefore, they
were also withheld from the study. Over thirty thousand (30,182) usable
records remained for analysis.

Appendix B

1

Of the 262,284 arrest records processed for 2005 releases, these records were almost entirely U.S. state and federal records. However, also included are 345 (0.1%) territorial
(non-federal) arrests provided by U.S. territories, primarily by Puerto Rico and the U.S. Virgin Islands.
2

The guidelines sentencing table provides sentencing ranges based on an offender’s offense level and Criminal History Category. See USSG, Ch.5 Pt.A. The offense level is
calculated using offense specific aggravating and mitigating factors prescribed by the guidelines. See USSG §1B1.1(a)(1)-(5). The Criminal History Category is based on the
recency and severity of an offender’s prior sentences and supervision status. See USSG §4A1.1. The sentencing range is determined by “[t]he intersection of the Offense Level
and Criminal History Category” on the sentencing table. USSG, Ch.5 Pt.A, comment. (n.1). The sentencing table is subdivided into four zones (A, B, C, and D) that determine
confinement options for each sentencing range. Zone B authorizes probation with condition of community confinement (e.g., home detention) and Zone C authorizes a “split”
sentence of imprisonment and community confinement. USSG §§5B1.1(a)(2), 5C1.1(d)(2).
3

In 2003, Congress enacted the Prosecutorial Remedies and Other Tools to end the Exploitation of Children Today Act (“the PROTECT Act”), Pub. L. No. 108-21 (2003), which
restricted the use of departures by sentencing courts and changed the appellate standard of review for cases in which departures were imposed. The PROTECT Act restricted the
availability of departures. The PROTECT Act period began May 1, 2003.

4

See MATTHEW DUROSE, ALEXIA COOPER & HOWARD SNYDER, BUREAU OF JUSTICE STATISTICS, RECIDIVISM OF PRISONERS RELEASED IN 30 STATES IN 2005: PATTERNS FROM
2005 TO 2010 (2014) (hereinafter PRISONERS RELEASED IN 30 STATES IN 2005), http://www.bjs.gov/content/pub/pdf/rprts05p0510.pdf.
5

Instances of arrest or sentencing that appeared to be duplicates of existing events were removed by NORC. Certain administrative records were removed after review of each
state’s procedures demonstrated that some entries were likely to be either reports of non-offending contact with law enforcement (e.g., registration as a sex offender) or other
information that pertained to the offender but was not a criminal justice event. Some states included historical data within cycles which caused issues with correct parsing of
criminal history information; in these cases, the first court disposition for that cycle was retained and all others removed. Arrest entries that occurred outside of the eight-year
follow-up period were removed from the datafiles and were not used to ascertain recidivism.

6

See PRISONERS RELEASED IN 30 STATES IN 2005, supra note 4, at 22.

7

The major arrest charge codes, as ranked by the Commission beginning with the most serious, were homicide, rape or sexual assault, robbery, assault, other violent offense, drug
trafficking, burglary, larceny, motor vehicle theft, fraud, other property offense, drug possession, other drug offense, weapons offense (e.g., unlawful sale, etc.), driving under the
influence, and other public order offense.
8

However, unlike the BJS major and detailed charge code rankings, the Commission chose to rank drug trafficking offenses immediately behind violent offenses in order of
severity.

Appendix B