U.S. Department of Justice
Office of Justice Programs
Bureau of Justice Statistics

****************************************************************
This file is text only without graphics and many of the
tables. A Zip archive of the tables in this report in
spreadsheet format (.csv) and the full report including
tables and graphics in .pdf format are available on BJS
website at: http://www.bjs.gov/index.cfm?ty=pbdetail&iid=5642
****************************************************************


*******************
Special Report
*******************

****************************************
Recidivism of Offenders Placed on
Federal Community Supervision in 2005:
Patterns from 2005 to 2010
**************************************** 

Joshua A. Markman, Matthew R. Durose, 
and Ramona R. Rantala, BJS Statisticians
Andrew D. Tiedt, Ph.D., Former BJS Statistician
*******************************************

During fiscal year 2005, approximately 43,000 offenders
were placed on federal community supervision,
including nearly a quarter (23%) who were directly
sentenced to probation and more than three-quarters (77%)
who began a term of supervised release following a prison
sentence. Overall, 35% of these offenders were arrested within
3 years and 43% were arrested within 5 years of placement on
community supervision.

This report examines the extent to which offenders placed on
federal community supervision were arrested by federal and
nonfederal (i.e., state and local) law enforcement agencies prior
to and following their placement on community supervision. It
also compares the recidivism rates of former federal prisoners
to those of former state prisoners released in 2005.

********************************************
Criminal history and prison records 
were used to document recidivism patterns
********************************************

To measure recidivism, this study used a combination of
arrest charge, court disposition, incarceration sentence, and
custody information found in national criminal history records
and other data sources to measure recidivism. Depending
on the measure used, the percentage of offenders classified
as recidivists will decline as the recidivism measurement
progresses from arrest to imprisonment. In other words, of
those persons arrested, a smaller percentage are charged, and an
even smaller percentage are imprisoned. In the criminal history
records, the proportion of arrests that included information on
the offender’s adjudicated guilt or innocence and, if convicted,
on the sentence imposed (e.g., prison, jail, or probation) varied
across jurisdictions. This could be due to natural state-
variations in criminal justice practice or in criminal history 
reporting practices. The variations also may have been caused by 
a lack of reporting court dispositions to criminal history 
repositories or because the repository could not connect a 
reported court disposition to a specific arrest.

The analytical approach used in this research sought to minimize 
the effect on recidivism statistics posed by variations in 
criminal history reporting policies, coding practices, and
coverage. The analysis excluded traffic offenses because those
events were not commonly recorded on the criminal history
records across states. However, some variations in the content
of rap sheets remained and could not be remediated, such as the
nature of the charging decision. For example, when a person on 
parole is arrested for committing a burglary, some local
law enforcement agencies code the arrest offense as a parole
violation, some code it as a burglary, and others code both the
burglary and the parole violation. It is difficult to discern 
from the rap sheets, given that this is often a local coding 
decision, which charging approach was employed at each arrest.
Offenders placed on federal community supervision averaged five 
prior arrests.

*********************************************************
*************
HIGHLIGHTS 
*************

* Nonfederal (i.e., state and local) law enforcement agencies
were responsible for approximately three-quarters (76%)
of prior arrests of offenders placed on federal community
supervision in 2005.

* Nonfederal charges accounted for more than two-thirds
(68%) of all arrests that occurred during the 5 years following
placement on federal community supervision.

* Within 1 year following placement in 2005 on community
supervision, 18% of the federal offenders had been arrested
at least once.

* Among those conditionally released from federal prison,
nearly half (47%) were arrested within 5 years, compared to
more than three-quarters (77%) of state prisoners released
on community supervision.

* About 3 in 10 federal prisoners released to a term of
community supervision returned to prison within 5 years,
while nearly 6 in 10 state prisoners conditionally released
returned in 5 years.

* On average, offenders released from federal prison had
fewer prior arrests (5.6) than those released from state prison
(10.8).

*  Across demographic characteristics or extent of prior
criminal offending, state prisoners consistently had higher
rates of recidivism than federal prisoners within 5 years
after release.
*********************************************************

*****************************************
Offenders placed on federal community 
Supervision averaged five prior arrests
*****************************************

In 2005, the total correctional population was 7,055,600 of
which 304,500 were federal offenders. The total community
supervision population was 4,946,600 of which 117,900 were
offenders under federal community supervision.*** Footnote 1 See 
Correctional Populations in the United States, 2014, NCJ 249513, 
BJS web, December 2015***. Nationally, 8 in 10 offenders who 
were placed on federal community supervision in 2005 were male 
(80%) (table 1). More than a third (41%) were white and nearly a 
third (31%) were black. Approximately 30% were age 29 or younger 
and about 40% were age 40 or older. The average criminal history 
of these offenders included approximately five prior arrests. 
More than a third (38%) of the federal offenders were sentenced 
to community supervision for a drug offense (table 2).

During their criminal careers prior to being placed on federal
community supervision in 2005, these offenders were arrested
approximately 210,000 times (table 3). This total included
arrests made by federal, state, and local law enforcement
agencies in all 50 states, the District of Columbia, and U.S.
territories. Federal law enforcement agencies accounted for
approximately 24% of all prior arrests. State and local law
enforcement agencies were responsible for the other 76%
of prior arrests. Almost 45% of federal offenders placed on
community supervision in 2005 had 4 or more prior arrests.
The large majority (78%) of federal offenders had only one
prior federal arrest, which was the arrest that led to the 2005
community supervision sentence. Approximately 70% of
federal offenders had at least one prior nonfederal arrest,
with more than a third (35%) having four or more prior
nonfederal arrests.

**********************************************
2 in 5 offenders placed on federal community 
supervision were arrested within 5 years 
**********************************************

At the end of the 5-year follow-up period, 43% had been arrested. 
Nearly half of those who were arrested within the 5-year follow-
up period were arrested within the first year after being placed 
on community supervision in 2005. Within 1 year following 
placement on community supervision in 2005, 18% of federal 
offenders had been arrested at least once (table 4). By the end 
of the third year, 35% had been arrested.

Aggregate trends of the first arrest following placement on
federal community supervision appeared to diverge when
decomposing these arrests as federal and nonfederal charges.
Within 2 years after placement on community supervision,
18% of the federal offenders had been arrested for at least
one federal charge, and 19% had been arrested for at least
one nonfederal charge (figure 1). After 2 years, a higher
proportion of new first-time arrests were by nonfederal law
enforcement agencies. As a result, within 5 years of being
placed on community supervision in 2005, a quarter (25%)
of federal offenders had at least one federal arrest, while
more than a third (34%) of federal offenders had at least one
nonfederal arrest. During the 5-year follow-up period, 8%
had two or more federal arrests, while 18% had two or more
nonfederal arrests (table 5). When examining the first arrest
after placement on federal supervision, nearly two-thirds of
arrests were for nonfederal charges (65%). Among all arrests
that occurred within the 5-year follow-up period, 68% were
arrests for nonfederal charges (figure 2).

After being placed on federal community supervision in 2005,
the offense involved in the first arrest was related to the type 
of arresting agency. Charges for public order offenses were more
likely to occur among the federal arrests. Public order offenses
accounted for 33% of nonfederal arrests and 90% of federal
arrests, including 68% for a probation or parole violation
(table 6). After public order offenses, small proportions
of federal offenders on community supervision were first
arrested by federal law enforcement agencies for drug (6%),
property (2%), and violent offenses (1%). For nonfederal
arrests, the distribution was more uniform. Public order
offenses were the most common nonfederal offense type at
first arrest after placement on federal community supervision
(33%), followed by property (23%), violent (21%), and drug
(21%) offenses.

**********************************************
State prisoners have higher recidivism rates 
Than federal prisoners
**********************************************

BJS examined the recidivism patterns of the approximately
405,000 offenders released from state prison from 30 states
in 2005 and tracked for 5 years following release.***Footnote 2 
See Recidivism of Prisoners Released in 30 States in 2005: 
Patterns from 2005 to 2010, NCJ 244205, BJS web, April 2014***. 
That study found that recidivism rates varied with the attributes
of the inmate. For example, state prisoners released after
serving time for a property offense were most likely to be
arrested for a new crime compared to other offense types.
In addition, recidivism was highest among males and young
adults, and declined with age. The recidivism rates of state
prisoners after release increased as the extent of their criminal
history increased.

To observe differences in recidivism outcomes, the following
section compares recidivism rates for prisoners placed on
federal community supervision to those conditionally released
from state prisons. (See Comparing federal and state prisoner
post-release recidivism patterns text box.)

Demographic characteristics and criminal histories differed
among federal and state prison populations released in 2005.
Four in 10 federal prisoners were age 40 or older, compared to
3 in 10 released state prisoners (table 7). Females were more
prevalent among federal prisoners (17%) than state prisoners
(10%). The racial distribution within both groups of prisoners
was similar.

*********************************************************
*************************************
Comparing federal and state prisoner
post-release recidivism patterns
*************************************

This report compares state prisoners released conditionally in
2005 and persons placed on federal community supervision
in 2005 after a period of incarceration. State prisoners
released conditionally include those released on discretionary
parole, mandatory parole, post-custody probation, and
other unspecified conditional releases and excludes those
released unconditionally (e.g., expirations of sentences,
commutations, and other unspecified unconditional releases)
on community supervision.

In addition to measuring recidivism as an arrest for a new
crime, this study examined recidivism as an admission to
prison for a technical parole or other community supervision
violation (e.g., failing a drug test or missing an appointment
with a probation officer) or a sentence for a new crime.
The availability of this information varied in the criminal
history records by state and federal jurisdictions. Given the
inconsistent reporting of such custody information in the
criminal history records, the return-to-prison measure relied
on a combination of data sources.

For released federal prisoners, BJS used the federal prison
admission records from the Bureau of Prisons (BOP) (part
of BJS’s Federal Justice Statistics Program) to supplement
the criminal history records by capturing returns to prison
both with or without a new sentence. For released state
prisoners, state prison admission records from the BJS
National Corrections Reporting Program (NCRP) were used
to supplement the criminal history data by capturing returns
to prison with or without a new sentence. The measure for
the state prisoners was based on those released from the
23 states in the study that had the necessary data.
*********************************************************

The two groups diverged regarding their prior offending
histories. Seven percent of released federal prisoners were
in prison for a violent crime, compared to 22% of released
state prisoners. Among most serious commitment offenses,
nearly half (46%) of federal prisoners were committed for a
drug offense, compared to 33% for state prisoners. Two in 10
(20%) federal prisoners were in prison for a property crime,
compared to about 3 in 10 (29%) state prisoners. On average,
the released federal prisoners (6 arrests) had fewer prior
arrests than released state prisoners (11 arrests).

Overall, the recidivism rates of released federal prisoners were
lower than those of state prisoners. Among federal prisoners
released on community supervision during 2005, 47% were arrested 
within 5 years, while 77% of the state prisoners conditionally 
released in 30 states in 2005 were arrested within 5 years (table 
8). Compared to state prisoners, released federal prisoners also 
had lower recidivism rates within demographic subgroups. 
Differences among federal and state prisoners also were observed 
when examining the number of prior arrests; however, the 
difference in post-arrest rates decreased as the number of prior 
arrests increased. For example, the gap between post-arrest rates 
among state prisoners with two or fewer prior arrests (52%) and 
federal prisoners with two or fewer prior arrests (27%) declined 
when comparing post-arrest rates among state prisoners with 10 or 
more prior arrests (86%) to federal prisoners with 10 or more 
prior arrests (72%).

While federal prisoners placed on community supervision
had lower overall recidivism rates than state prisoners
conditionally released, the pattern changed when
examining the types of arrests prisoners had following
release. Thirty-eight percent of federal prisoners had a
nonfederal arrest (by a state or local law enforcement
agency) within 5 years following release, compared to
75% of state prisoners (figure 3). However, 30% of federal
prisoners had a federal arrest within 5 years after release,
while 4% of state prisoners had a federal arrest. Within 2
years following release, federal prisoners were nearly as likely
to have had a nonfederal arrest as a federal arrest. Starting
at 3 years after release and extending through the end of the
5-year follow-up period, the likelihood for nonfederal arrests
increased at a rate faster than federal arrests.

Compared to the state prisoners conditionally released,
federal prisoners on community supervision were less likely to
return to prison following release (figure 4). This finding was
observable within a year after release and continued to the end
of the follow-up period. Among released federal prisoners,
32% returned to prison within 5 years for either a revocation
or an arrest that led to a new sentence. Among state prisoners
released in 2005 in 23 states with available recidivism
imprisonment data, 59% had either a parole or probation
violation or an arrest for a new offense within 5 years that led
to a return to prison.

In addition to having a lower recidivism rate following release,
federal prisoners had fewer arrests for a new crime overall
than state prisoners (table 9). For released federal prisoners,
18% had two to three arrests and 12% had four or more
arrests for a new crime. In comparison, 27% of released state
prisoners had two to three arrests and 31% had four or more
arrests for a new crime.

Both federal and state prisoners were most likely to have a
first-arrest charge for a public order offense, followed by drug,
property, and violent offenses (table 10).

For both federal and state prisoners on community
supervision, an arrest for a new crime following release was
more likely an arrest for a nonfederal charge than a federal
charge. Sixty-seven percent of all arrests for a new crime
among released federal prisoners were for nonfederal charges,
compared to 98% of arrests among state prisoners (table 11).

**************
Methodology
**************

Background
***********

In 2008, the Bureau of Justice Statistics (BJS) entered into
a data sharing agreement with the FBI’s Criminal Justice
Information Services (CJIS) Division and the International
Justice and Public Safety Network (Nlets) to provide BJS
access to criminal history records (i.e., rap sheets) through
the FBI’s Interstate Identification Index (III). A data security
agreement was executed between BJS, the FBI, and Nlets to
define the operational and technical practices used to protect
the confidentiality and integrity of the criminal history data
during data exchange, processing, and storage.

The FBI’s III is an automated pointer system that allows
authorized agencies to determine whether any state
repository has criminal history records on an individual.
Nlets is a computer-based network that is responsible for the
interstate transmissions of federal and state criminal history
records. It allows users to query III and send requests to
states holding criminal history records on an individual. The
FBI also maintains criminal history records that it has sole
responsibility for disseminating. The identification bureaus
that operate the central repositories in each state respond
automatically to requests over the Nlets network. Responses
received via Nlets represent an individual’s national criminal
history record.

Under the Criminal History Records Information Sharing
(CHRIS) Project (award 2008-BJ-CX-K040), Nlets developed
an automated collection system for BJS to use to retrieve
national criminal history records from the FBI and state
repositories on samples of offenders. Nlets produced software
to parse the fields from individual criminal history records
into a relational database. The database consists of state- and
federal-specific numeric codes and text descriptions (e.g.,
criminal statutes and case outcome information) in a uniform
record layout.

The Conversion of Criminal History Records into Research
Databases (CCHRRD) Project (grant 2009-BJ-CX-K058)
funded NORC at the University of Chicago to develop
software that standardizes the content of the relational
database produced by Nlets into a uniform coding structure
that supports national-level recidivism research. The
electronic records accessed by BJS through III for this study
are the same records used by police officers to determine the
current criminal justice status (e.g., on probation, parole, or
bail) of a suspect; by judges to make pretrial and sentencing
decisions; and by corrections officials to determine inmate
classifications, parole releases, and work furloughs.

Data collection
*****************

This study is based on the 42,977 offenders placed on federal
community supervision during fiscal year 2005 (October
1, 2004, to September 30, 2005). Federal supervision
includes offenders directly sentenced in the federal courts
to probation supervision in the community and offenders
entering supervision following release from prison to serve
a term of supervised release in the community. BJS obtained
information on these offenders from the Office of Probation
and Pretrial Services (OPPS) of the Administrative Office of
the U.S. Courts’ Probation/Pretrial Services Automated Case
Tracking System (PACTS). This information included each
offender’s date of birth, sex, race and Hispanic origin, most
serious conviction offense, length of community supervision
sentence, and the start date of the sentence. OPPS also
provided BJS with the FBI identification numbers needed to
obtain criminal history records on the federal offenders. The
first start date for a federal community supervision sentence
during 2005 was selected for those with multiple start dates
during the year. As part of the Recidivism of Offenders on
Federal Community Supervision Project (award 2010-BJCX-
K069), Abt Associates assisted BJS with assembling and
analyzing the data.

BJS received approval from the FBI’s Institutional Review
Board (IRB) to access criminal history records through
III for this study. This study employed a 5-year follow-up
period. In February 2012, BJS sent the FBI identification
numbers supplied by OPPS to III via Nlets to collect the
criminal history records on the offenders. These records
included arrests and court dispositions from all 50 states and
the District of Columbia prior to and following the start of
the offender’s sentence in 2005. Juvenile offenses were rarely
included in the criminal history records unless the offender
was charged or tried in court as an adult. Over a 3-week
period, Nlets electronically collated the responses received
from the FBI and state criminal history repositories into a
relational database.

To ensure that the records received using the federal offenders’
FBI identification numbers were correct, BJS compared other
individual identifiers in the PACTS data to those reported in
the criminal history records. A federal offender’s date of birth
in the PACTS data exactly matched his or her birthdate in the
criminal history records 98% of the time. Nearly 100% of the
PACTS data and criminal history records matched on sex and
race at the person level.

BJS reviewed the composition of the criminal history
records for distributional differences and inconsistencies
in reporting practices and observed some variations across
states. During the data processing and analysis phases, steps
were taken to standardize the information used to measure 
recidivism and to minimize the impact these variations had
on the overall recidivism estimates. Administrative (e.g.,
a criminal registration or the issuance of a warrant) and
procedural (e.g., transferring a suspect to another jurisdiction)
records embedded in the arrest data that did not refer to an
actual arrest were identified and removed from the study.
Traffic offenses (except for vehicular manslaughter, driving
while intoxicated, and hit-and-run) were also excluded
from the study because the coverage of these events in the
criminal history records varied widely by state. Among
offenders placed on federal community supervision in
2005, approximately 2% of their arrests prior to and for 5
years after being placed on community supervision were for
traffic offenses.

The criminal history records from some states recorded
sentence modifications that occurred after the original court
disposition and sentence while records from other states did
not. To ensure that consistent counting rules were employed
when recidivism was measured across states, the initial
court disposition was captured for any arrest charge that
had subsequent sentence modifications reported within the
same arrest cycle. For instance, if a court adjudication was
originally deferred and then later modified to a conviction,
the deferred adjudication was coded as the disposition for that
arrest charge.

Deaths during the follow-up period
************************************

BJS determined that 934 of the 43,911 offenders placed on
federal community supervision in 2005 died during the 5-year
follow-up period. These offenders were removed from the
recidivism analysis. Initial identification of those who died
within the 5-year period was done using death information
contained in the criminal history record and OPPS’s PACTS
data. Additional deaths were identified by probabilistically
linking the federal offenders to individuals identified as dead
in the Social Security Administration’s (SSA) public Death
Master File (DMF).

Link Plus, a probabilistic record linkage program developed
by the Centers for Disease Control and Prevention (CDC),
was used to create and score potential matches between
the federal offenders’ records and the public DMF, using
common identifiers found on each file.*** Footnote 3 Centers for 
Disease Control and Prevention. (2006). Link Plus Version
2.10 probabilistic record linkage software. Atlanta, GA: Centers 
for Disease Control and Prevention.*** For persons with
multiple Social Security numbers and names and dates of
birth, all possible combinations were tested for matches. The
software computed a probabilistic record linkage score for
each matched record, with the score representing the sum of
the agreement and disagreement weights for each matching
variable (the higher the score, the greater the likelihood that
the match made is a true match). To differentiate true matches
from false matches, the scores of the linked records were
manually evaluated to ascertain the appropriate upper and
lower bound cutoff scores. During this review, records with a
score of 20.0 or higher were determined to be exact matches of
name, Social Security number, and date of birth, and scores of
less than 10.9 indicated that none of the personally identifiable
information matched.

Accordingly, these cutoffs were used as the upper and lower
cutoff scores to automatically designate true matches and
nonmatches. All remaining pairs that fell between the upper
and lower cutoff scores were manually reviewed by two
independent reviewers and independently categorized. All
discrepancies where the reviewers did not agree (less than 1%)
were jointly classified.

The number of released prisoners who were identified as dead
in the DMF likely represents an undercount of the actual
number of deaths within the sample. This is partly due to
the limitations of the public DMF. Specifically, due to state
disclosure laws, the public DMF does not include information
on certain protected state death records (defined as records
received via SSA’s contracts with the states). This change,
which occurred in November 2011, resulted in SSA removing
more than 4 million state-reported death records from the
public DMF and adding more than 1 million fewer records
annually to the public DMF thereafter. As a result, the public
DMF contains an undercount of annual deaths.

How extensively the public DMF undercounts the annual
number of deaths is not known. Preliminary analyses of
deaths in the public DMF compared to those reported via the
CDC’s mortality counts suggest that, in 2005, the public DMF
undercounted the overall number of deaths in the United
States by around 10%. The undercount has increased each year
since 2005. As of 2010, the public DMF contained around half
(45%) of the deaths reported by the CDC.

Furthermore, the coverage of the public DMF differs by
decedent age, with younger decedents being less likely to
appear in the public file. Because of this, the death count of
offenders placed on federal community supervision in 2005 is
likely an undercount of the actual number of deaths.

Missing criminal history records
**********************************

Among the 42,977 eligible offenders known to have not died
during the follow-up period, BJS did not obtain criminal
history records on 2,961 offenders because either OPPS was
unable to provide their FBI or state identification number
or the offender had an FBI identification number that did
not link to a criminal history record either in the FBI or
state record repositories. To account for the missing data
and to ensure the recidivism rates were representative of
all 42,977 offenders in the analysis, the data regarding the
40,016 offenders with criminal history information required
statistical adjustments to account for those offenders without
criminal history information. A joint distribution of the
offenders was produced based on sex, race and Hispanic
origin, age, and most serious conviction offense. This 
distribution documented the number of offenders who fell
into each of 236 specific subpopulations defined by crossing
5 categories of age, 2 categories of sex, 4 categories of race
and Hispanic origin, and 4 categories of conviction offenses.
Within each of the subgroups, statistical weights were applied
to the offenders with criminal history information to account
for those without criminal history information. When this
weight was applied to data on the federal offenders, the
standard errors around the estimates were relatively small.
For instance, an estimated 43% of persons placed on federal
community supervision in 2005 were arrested for a new crime
within 5 years. The standard error around this estimate was
0.004%.

Comparison group of released state prisoners
**********************************************

To examine the differences in the recidivism rates of the
federal offenders in this study compared to those of state
offenders, this report used the findings from a previous BJS
study that collected national criminal history records on a
sample of offenders released from state prisons in 30 states in
2005. Using data reported by state departments of corrections
to BJS’s National Corrections Reporting Program (NCRP),
68,597 released prisoners were randomly selected to represent
the 404,638 inmates released at that time. In 2005, these 30
states were responsible for about three-quarters of all state
prisoners released nationwide. States were selected for the
study based on their ability to provide prisoner records and
the FBI or state identification numbers of offenders released
from correctional facilities in 2005. For more information on
this study, see Recidivism of Prisoners Released in 30 States
in 2005: Patterns from 2005 to 2010 (NCJ 244205, BJS web,
April 2014).

Conducting tests of statistical significance
***********************************************

This report examines how the recidivism rates of offenders
released from federal prisons in 2005 compared with the
recidivism rates of offenders released conditionally from
state prisons in 30 states during the same year. Because the
recidivism study of state prisoners was based on a sample
and not a complete enumeration, the estimates are subject to
sampling error (i.e., a discrepancy between an estimate and
a population parameter based on chance). One measure of
the sampling error associated with an estimate is the standard
error. The standard error can vary from one estimate to the
next. In general, for a given metric, an estimate with a smaller
standard error provides a more reliable approximation of
the true value than an estimate with a larger standard error.
Estimates with relatively large standard errors are associated
with less precision and reliability and should be interpreted
with caution. BJS conducted tests to determine whether
differences in the estimates based on the state prisoners
were statistically significant once sampling error was taken
into account.

All differences discussed in this report are statistically
significant at or above the 95% confidence interval. Standard
errors were generated using SUDAAN and SPSS statistical
software packages that estimate sampling error from complex
sample surveys.

Recidivism measures
**********************

This study used three types of events to measure recidivism:
arrest, incarceration, and return to prison.

Arrest: An arrest within 5 years of being placed on federal
community supervision in 2005. Information presented on
the number of arrests is based on unique arrest dates, not
individual charges. An arrest can involve more than one type
of charge. For instance, one arrest could include a charge for
a violent crime and a charge for a drug crime. BJS uses the
most serious offense charge when characterizing the arrest
offense type.

Return to prison--offenders from federal prison: An arrest
within 5 years of being released from a federal prison in 2005
that resulted in a return to prison. When the type of facility
(e.g., prison or jail) where an incarceration sentence was to
be served was not reported in the criminal history records,
a sentence of a year or more was defined as imprisonment.
Information on the number of prison sentences is based on
each unique arrest date that led to a prison sentence, not the
date that the sentence was imposed.

A return to prison also may result from an arrest for a new
crime or a technical violation of a condition of release within
5 years of being released from a federal prison in 2005. This
recidivism measure incorporates the criminal history records
from the FBI and state repositories, and the federal prison
admission records obtained from the Bureau of Prisons
(BOP). The criminal history records provided information
on arrests that resulted in a prison sentence during the 5-year
follow-up period. Using a common set of identifiers, BJS
supplemented the criminal history records with information
on the federal prisoners who returned to prison for a technical
violation that did not involve a sentence for a new crime. 

Prisoners released from Maryland, Nebraska, Nevada,
Ohio, Pennsylvania, and Virginia were excluded from the
return-to-prison analysis because the individual identifiers
or complete prison admission data needed to locate returns
to prison during the entire 2005 through 2010 observation
window were not available. Louisiana prisoners were also
excluded from the return-to-prison analysis because the
sentencing information in the criminal history records from
this state was generally not linked to the associated arrest.

Measuring recidivism as an arrest for 
a new crime compared to a revocation
***************************************

This report focuses on the arrests and incarcerations found
in official criminal history records. Recidivism may also be
defined as a revocation of the community supervision when
the person violates terms or conditions which may or may
not result in a return to prison and which may or may not be
recorded in criminal history records. As part of this study,
BJS obtained revocation information from OPPS’s PACTS
data on the 43,000 offenders placed on federal community
supervision in 2005. To understand the extent to which
revocations affect the recidivism patterns based solely on
criminal history records, BJS separately tracked recidivism
rates of federal offenders based on a combination of arrests
in the criminal history records and revocations in the PACTS
data. BJS employed the same methods to standardize arrest
measures, including removal of administrative, procedural,
and traffic violation-related records. The differences in the
level of observed recidivism using the two measures were
consistent during the follow-up period. Within 1 year of being
placed on community supervision in 2005, approximately
18% of offenders were arrested while 20% had either an arrest,
revocation, or both (figure 5). At the end of the 5 years, 43%
had an arrest and 45% had an arrest, a revocation, or both.
In summary, while independent revocation information
may add some insight, criminal history records on their
own provide a nearly complete assessment of arrest-based
recidivism rates and trends.

Offense definitions
*********************

Violent offenses include homicide, rape or sexual assault,
robbery, assault, and other miscellaneous or unspecified
violent offenses.

Homicide includes murder, nonnegligent manslaughter,
negligent manslaughter, and unspecified homicide offenses.

Rape or sexual assault includes (1) forcible intercourse
(vaginal, anal, or oral) with a female or male, (2) forcible
sodomy or penetration with a foreign object (sometimes
called deviate sexual assault), (3) forcible or violent sexual
acts not involving intercourse with an adult or minor,
(4) nonforcible sexual acts with a minor (such as statutory
rape or incest with a minor), and (5) nonforcible sexual
acts with someone unable to give legal or factual consent
because of mental or physical defect or intoxication.

Robbery is the unlawful taking of property that is in the
immediate possession of another, by force or the threat
of force. It includes forcible purse snatching, but excludes
nonforcible purse snatching.

Assault includes aggravated and simple and unspecified
assault. Aggravated assault includes (1) intentionally and
without legal justification causing serious bodily injury,
with or without a deadly weapon, or (2) using a deadly
or dangerous weapon to threaten, attempt, or cause
bodily injury, regardless of the degree of injury, if any.
The category also includes attempted murder, aggravated
battery, felonious assault, and assault with a deadly
weapon. Simple assault includes intentionally and without
legal justification causing less than serious bodily injury
without a deadly or dangerous weapon, or attempting
or threatening bodily injury without a dangerous or
deadly weapon.

Other violent offenses is a category that encompasses a range
of crimes, including intimidation, illegal abortion, extortion,
cruelty towards a child or wife, kidnapping, hit-and-run with
bodily injury, and miscellaneous or unspecified crimes against
the person.

Property offenses include burglary, fraud/forgery, larceny,
motor vehicle theft, and other miscellaneous or unspecified
property offenses.

Burglary is the unlawful entry of a fixed structure used for
regular residence, industry, or business, with or without the
use of force, to commit a felony or theft.

Larceny is the unlawful taking of property other than
a motor vehicle from the possession of another, by
stealth, without force or deceit. Includes pocket picking,
nonforcible purse snatching, shoplifting, and thefts from
motor vehicles. Excludes receiving or reselling stolen
property or both, and thefts through fraud or deceit.

Motor vehicle theft is the unlawful taking of a selfpropelled
road vehicle owned by another. Includes the theft
of automobiles, trucks, and motorcycles, but not the theft
of boats, aircraft, or farm equipment (classified as larceny).
Also includes receiving, possessing, stripping, transporting,
and reselling stolen vehicles, and unauthorized use of a
vehicle (joyriding).

Fraud/forgery includes using deceit or intentional
misrepresentation to unlawfully deprive persons of
their property or legal rights. It also includes offenses
such as embezzlement, check fraud, confidence game,
counterfeiting, and credit card fraud.

Other property offenses include arson, stolen property
offenses, possession of burglary tools, damage to
property, trespassing, and miscellaneous or unspecified
property crimes.

Drug offenses include possession, trafficking, and other
miscellaneous or unspecified drug offenses.

Drug possession includes possession of an illegal drug,
but excludes possession with intent to sell. It also includes
offenses involving drug paraphernalia and forged or
unauthorized prescriptions.

Drug trafficking includes manufacturing, distributing,
selling, smuggling, and possession with intent to sell.

Other drug offenses include offenses involving drug
paraphernalia, forged or unauthorized prescriptions, and
other miscellaneous or unspecified drug offenses.

Public-order offenses include weapons offenses,
driving under the influence, and other miscellaneous or
unspecified offenses.

Weapons include the unlawful sale, distribution,
manufacture, alteration, transportation, possession, or use
of a deadly or dangerous weapon or accessory.

Driving under the influence (DUI) is driving under the
influence of drugs or alcohol and driving while intoxicated.

Other public-order offenses are those that violate the
peace or order of the community or threaten the public
health or safety through unacceptable conduct, interference
with governmental authority, or the violation of civil rights
or liberties. The category also includes probation or parole
violations, escape, obstruction of justice, court offenses,
nonviolent sex offenses, commercialized vice, family
offenses, liquor law violations, bribery, invasion of privacy,
disorderly conduct, contributing to the delinquency of a
minor, and other miscellaneous or unspecified offenses.

**********************************************************
The Bureau of Justice Statistics of the U.S. Department of 
Justice is the principal federal agency responsible for measuring 
crime, criminal victimization, criminal offenders, victims of 
crime, correlates of crime, and the operation of criminal
and civil justice systems at the federal, state, tribal, and 
local levels. BJS collects, analyzes, and disseminates reliable 
and valid statistics on crime and justice systems in the United 
States, supports improvements to state and local criminal
justice information systems, and participates with national and 
international organizations to develop and recommend national 
standards for justice statistics. Jeri M. Mulrow is acting 
director.

This report was written by Joshua A. Markman, Matthew R. Durose, 
and Ramona R. Rantala, BJS Statisticians, and Andrew D. Tiedt, 
Ph.D., former BJS Statistician. Alexia D. Cooper, Ph.D., verified 
the report.

Lynne McConnell and Jill Thomas edited the report, and Tina 
Dorsey and Barbara Quinn produced the report.

June 2016, NCJ 249743
**********************************************************

*************************************************
Office of Justice Programs
Innovation * Partnerships * Safer Neighborhoods
www.ojp.usdoj.gov
*************************************************

**********************
6/2/2016 1:00pm JER 
**********************