Recidivism and the "First Offender"

RESEARCH SERIES ON THE RECIDIVISM OF

FEDERAL GUIDELINE OFFENDERS

ELEASE 2

RECIDIVISM

AND THE “FIRST OFFENDER”

A COMPONENT OF THE

FIFTEEN YEAR REPORT

ON THE

U.S. SENTENCING COMMISSION’S

LEGISLATIVE MANDATE

May 2004

RUBEN CASTILLO

Vice Chair

WILLIAM K. SESSIONS, III

Vice Chair

JOHN R. STEER

Vice Chair

RICARDO H. HINOJOSA

Commissioner

MICHAEL E. HOROWITZ

Commissioner

MICHAEL E. O’NEILL

Commissioner

EDWARD F. REILLY, JR.

(Ex officio)

EBORAH J. RHODES

(Ex officio)

Commission Staff

Linda Drazga Maxfield

Miles Harer*

Timothy Drisko

Christine Kitchens

Sara Meacham

*while serving at the U.S. Sentencing Commission

28 U.S.C. § 994(j).

U.S. Sentencing Commission, 1989; 1990; 1991a and 1991b; 1997; Ricca, 1997; 2001a; O’Neill, 2001.

28 U.S.C. § 994(j)) describes a “first offender” as one who “has not been convicted of a crime of violence or

an otherwise serious offense.”

he “first offender” philosophy in sentencing policy generally encourages lower

sentences for offenders who have little or no prior criminal conduct. This philosophy,

which can be derived directly from the guidelines’ Chapter Four introductory

commentary, postulates that first offenders are less culpable and less likely to re-offend. As such,

they are deserving of reduced punishment. Congress itself in the Sentencing Reform Act of 1984

promotes the first offender philosophy, citing the relevance of reduced sentencing levels for first

offenders under the federal sentencing guidelines.

Under federal sentencing guidelines, offenders with the least number of prior criminal

convictions are classified into Criminal History Category I (“CHC I”). As developed by the initial

Commissioners, CHC I is more broadly defined than “first offenders.” While offenders with zero

criminal history points are included, CHC I also applies to offenders with a prior conviction

receiving one criminal history point. Further, any offender in CHC I may have one or multiple

additional prior convictions if such convictions are excluded from receiving points under the

Chapter Four definitions.

Following the adoption of federal sentencing guidelines in 1987, the U.S. Sentencing

Commission initiated several investigations to refine a workable “first offender” concept within the

guideline criminal history structure.

Staff working groups started by analyzing the first offender

in terms of CHC I. Using district court data supplied through the Commission’s sentence monitoring

function, subsequent working groups concluded that a narrower first offender definition better

identified those deserving the least severe sentence than did the broad definition of first offender

appearing in 28 U.S.C. § 994(j).

Each successive working group built upon the previous efforts and

designed its set of first offender eligibility criteria. Over time, the proposed first offender definitions

grew generally more restrictive, with one definition even requiring that first offenders have no

criminal events, not even prior arrests.

The efforts of these past criminal history staff working groups have not resulted in a

guideline amendment to establish a first offender provision. In part, this stems from unresolved

policy and legal issues, such as the accuracy and appropriateness of prior arrest reporting or the

availability of prior court disposition data. The inaction also is tied to incomplete information on

the impact of a first offender provision.

Prior research by staff working groups documented how the various first offender definitions

reflected low culpability characteristics central to the guidelines’ Chapter Four. Their proposed first

U.S. Sentencing Commission, 1989.

Page 2 of 31

offender definitions captured offenders with few or no prior criminal history events, where prior

events that were present were not serious. What the prior staff working group research did not – and

could not (because of data unavailability) – address was an evaluation of recidivism behaviors under

various alternative first offender definitions.

This report responds to the previous lack of empirical information on offender recidivism

by drawing upon the Commission’s 2003 recidivism project data. Extrapolating from the results of

previous working groups’ first offender analyses, this report presents a set of alternative first

offender definitions. It documents the characteristics of offenders meeting the proposed alternatives,

illustrating that those meeting the definitions are unique among federal offenders in terms of low

culpability. The analysis then underscores the distinctive characteristics of the proposed first

offender groups by presenting their consistent patterns of lower recidivism rates. The research goal

is to identify criteria that define a low culpability and low recidivism first offender grouping as a

first step in the development of a possible guideline revision.

A. Commission Focus on First Offender Issues

he first offender adjustment was first studied at the Commission in late 1989 as part

of a larger proposed revision of the criminal history chapter. This effort suggested

that an additional CHC for “true” first offenders be developed in order to differentiate

between offenders whose instant offense was their first contact with the criminal justice system and

offenders whose instant offense was not their first contact with the criminal justice system.

A staff

criminal history working group generated two alternative proposals for handling first offender

issues. The first proposal classified all defendants with zero criminal history points as “true” first

offenders and redistributed all remaining offenders among the six pre-existing CHC categories. The

second proposal classified only those defendants with no prior “criminal trouble” as “true” first

offenders. The staff defined “prior criminal trouble” to mean any prior arrest, whether or not it led

to a conviction. However, two factors caused the criminal history working group to recommend that

the Commission further study the issue prior to any action. The first factor was the cited

unreliability of arrest records at that time. The second factor was the existing §4A1.3 (adequacy of

the CHC) guideline that expressly stated that the presence of a prior arrest by itself was not

sufficient justification for a sentence enhancement.

Approximately one year later, in late 1990, the Commission staff again addressed the subject

of the first offender as part of a larger review of the criminal history chapter. A criminal history

working group noted that CHC I offenders represented a wide variety of characteristics and offenses.

There even were great differences among offenders with zero criminal history points: some had

convictions that did not receive points, while others had no prior arrests at all. Taking these

differences into account, the working group defined the “true” first offender as any defendant who

U.S. Sentencing Commission, 1990. The “true” first offender also included offenders whose only convictions

were for minor offenses listed in guideline §4A1.2(c)(2) as never receiving criminal history points under the guidelines:

hitchhiking, juvenile status offenses and truancy, loitering, minor traffic offenses such as speeding, public intoxication,

vagrancy, and offenses “similar” to these, “by whatever name they are known.”

U.S. Sentencing Commission, 1991a.

Offenders in this third group, even though having zero criminal history points, could have prior convictions

that did not receive points because of the guideline exclusion rules (e.g., decay factor, juvenile status offense, etc.).

U.S. Sentencing Commission, 1991b.

28 U.S.C. §994(j) states in part that, “the Commission shall insure that the guidelines reflect the general

appropriateness of imposing a sentence other than imprisonment in cases in which the defendant is a first offender who

has not been convicted of a crime of violence or an otherwise serious offense . . . ”

U.S. Sentencing Commission, 2001a.

O’Neill, 2001.

Page 3 of 31

had no prior convictions nor prior criminal conduct.

Due to constitutional and policy

considerations, the criminal history working group did not advocate excluding defendants from a

first offender adjustment based solely on prior arrests or charges not leading to convictions.

However, the criminal history working group maintained that provisions could be drafted in such

a way that any “proven” criminal conduct would exclude a defendant from first offender status.

In October of 1991, another staff criminal history working group circulated its report.

After

an extensive case review, the criminal history working group divided the defendants with zero points

into three groups: (1) “true” first offenders,” (2) first offenders, and (3) other offenders with zero

criminal history points. Offenders in the first group, “true” first offenders, were defendants with no

prior contact with the criminal justice system of any kind. Those in the second group, first offenders,

were offenders with prior arrests or dismissed charges. All remaining zero point offenders were in

the third group.

Simultaneously with the release of this criminal history report, a separate

preliminary report by a sentencing alternatives working group discussed a possible sentencing

reduction for the first offender:

a two-level reduction for offenders who had zero criminal history

points and used neither violence nor weapons during the instant offense. The significance of this

proposal was that it both responded to the intent of 28 U.S.C. §994(j) and finessed the need to create

a new “first offender” CHC.

The sentencing alternatives working group also described another first

offender approach permitting first offenders access to probation or prison alternatives.

A criminal history working group established in fiscal year 2000 again examined first

offenders and other issues regarding the criminal history chapter.

In a more empirical context, this

working group pursued similar definitions of first offenders. Most recently, in a 2001 law review

article, Commissioner Michael O’Neill identified and described first offenders sentenced under the

federal sentencing guidelines.

Among other things, the article applied the first offender label to

those without any prior convictions and suggested that first offender provisions could be

incorporated into the guidelines either as a new first offender CHC or as a guided downward

departure. The article ended by recognizing the limitations of a first offender analysis without the

empirical advantage of recidivism data, recognizing that recidivism data permit complete

U.S. Sentencing Commission, 2001b.

Page 4 of 31

descriptions of the past criminal behavior of potential first offender groupings. In addition, the

access to recidivism data allows researchers to fully analyze recidivism risk under alternative first

offender definitions and compare this risk to repeat offenders.

B. Looking at First Offenders with Recidivism Data

t is, in fact, this lack of access to a recidivism datafile that kept the Commission from

acting on a first offender proposal. The previous criminal history working groups

described potential first offender groups using the following types of available

information: number of offenders in CHC I, number of criminal history points, number of prior

convictions, or types of prior convictions. Although theoretically and conceptually rigorous, these

analyses could not access re-offending outcomes for first offenders. In order to address all relevant

sentencing goals with criminal history measures, estimates are needed of recidivism rates and re-

offending risks associated with any potential first offender grouping. Whereas these data did not

previously exist, in the intervening years the Commission launched a major study of recidivism

under the guidelines.

The Commission’s recidivism project datafile contains details on the offenders’ prior

criminal history, including both arrests and convictions, characteristics of the instant offense,

demographic information, and subsequent re-offending. This made it possible for the first time to

examine and compare in-depth background and recidivism behavior among plausible first offender

groupings. The recidivism project collected data for more than 6,000 U.S. citizen offenders

sentenced in fiscal year 1992.

The recidivism data analysis reveals, across different groupings of offenders, that some types

of offenders recidivate at higher rates than others. Assessing this information helps in differentiating

among the first offender definitions not only for descriptive classifications but also for additional

predictive criteria relevant to first offender analyses.

C. Proposed First Offender Definitions

rior criminal history working groups identified the best theoretical and practical source

of first offenders as defendants in CHC I with zero criminal history points. The annual

proportion of federal offenders with zero criminal history points is substantial. In

fiscal year 2001, more than four out of every ten federal offenders (42.2%) had zero criminal history

points for their Chapter Four computation.

This proportion was even higher in fiscal year 1992,

when nearly half (49.9%) of citizen offenders had zero criminal history points.

Even within the group of offenders with zero points, additional distinctions are possible.

A prior conviction may be excluded from the criminal history computation if, for example: it occurred outside

of a five- or ten-year specified time frame; if it was a Native American or foreign conviction; if it was related to the

instant offense; if it were a minor offense and met specified conditions; etc.

The §4A1.2(c)(2) listed offenses that never receive points under the guidelines are: hitchhiking, juvenile status

offenses and truancy, loitering, minor traffic offenses such as speeding, public intoxication, and vagrancy. Offenses

“similar” to these, “by whatever name they are known,” also never receive criminal history points. A previous criminal

history working group suggested that offenders with only §4A1.2(c)(2) “never count” convictions be included in the first

offender definition. These “never count” minor offenses are consistent with a first offender criterion because their

presence does not improve recidivism prediction, as documented in the companion recidivism project report, “The

Exclusionary Rules.”

Page 5 of 31

Past research at the Commission differentiates among zero point offenders based on the offenders’

numbers of prior arrests and prior convictions.

Synthesizing the definitions of first offenders that

arose from prior Commission working group studies, the following three groups emerge.

Three Proposed Federal First Offender Groups

Proposed

First Offender

Groups

Zero Criminal

History

Points?

Prior

Arrests?

Prior

Convictions?

% of All

Citizen

Offenders

(no arrests)

Yes No No 29.8

(no convictions)

Yes Yes No 8.4

(minor conv. only)

Yes Yes

§4A1.2(c)(2)

offenses only

1.5

All three groups require that the offenders have zero criminal history points. Group A

offenders have no prior arrests. Group B offenders have prior arrests, but no prior convictions.

Group C offenders can have prior convictions, but they must fall under the minor offenses listed in

guideline §4A1.2(c)(2) that never count toward criminal history points.

Of the 1992 annual total of 29,749 citizen offenders, 8,854 offenders (29.8%) are in group

A. A total of 2,499 fiscal year 1992 offenders (8.4%) are in group B, with 460 offenders (1.5%) in

group C. The substantial number of offenders potentially meeting a first offender definition is

impressive:

• for all citizen offenders sentenced in fiscal year 1992, roughly 40 percent

(39.7%) meet the definition of group A, group B, or group C, and

• for all citizen offenders sentenced in fiscal year 1992 and having zero

Note that for completeness, Exhibit 1 through Exhibit 6 include a “remaining zeroes” group of citizen

defendants. These offenders are those with zero criminal history points but not covered by the definitions of groups A,

B, or C (i.e., they have non-minor prior convictions that do not receive any points due to guideline exclusionary rules).

In total, this “left-over” zero group constitutes 10.2 percent of all citizen offenders. The “remaining zeroes” group

appears on the exhibits in this report to provide a comparison with the three potential first offender groups.

CHC II through CHC VI are combined in the exhibits of this report. However, the characteristics by the

individual CHCs appear in a companion recidivism project report, “Background and Methodology of the U.S. Sentencing

Commission’s 2003 Recidivism Study.”

Page 6 of 31

criminal history points, roughly 80 percent (79.5%) meet the definition of

group A, group B, or group C.

Note that the impact of the definitions is cumulative. If group A is accepted as a first offender

definition, the tabulation above shows that nearly 30 percent (29.8%) of all citizen offenders would

be eligible for first offender status. However, if group B is accepted as a first offender definition,

then group A (having no prior criminal history arrests) is also accepted by default. Thus the sum

of offenders in either group A or group B accounts for 38.2 percent of all citizen offenders.

D. Characteristics of the Proposed First Offender Groups

his section of the report highlights the uniqueness of offenders in the proposed first

offender groups by describing their demographic, personal, social, instant offense,

prior arrest history, conviction history, and sentencing characteristics. These

comparisons serve as a prelude to the later description of recidivism rates, underscoring the

distinctive nature of offenders who have only minimal experience with the criminal justice system.

Comparisons are made among the three groups themselves, between the three groups and

CHC I offenders with only one criminal history point, and between the three groups and all

remaining offenders with more than one criminal history point (and therefore in CHC II through

CHC VI).

The characteristics discussed include demographic and personal characteristics, instant

offense and sentencing characteristics, and recidivism risk.

1. Demographic, Personal, and Social Characteristics

Exhibits 1, 2, and 3 provide descriptions of a wide range of offender characteristics at the

time of the guideline sentencing.

Gender

Exhibit 1 displays gender differences between group A, B, and C offenders and offenders

with criminal history points. In every column the greater percentage of offenders are male.

Page 7 of 31

However, the proportion of women reaches nearly 40 percent (37.4%) in first offender group A,

decreasing to 26.1 percent in group B and 22.3 percent in group C These are all substantially higher

female proportions than for offenders with one criminal history point (14.4%) or all offenders in

CHC II through CHC VI (9.1%).

Race and Ethnicity

White offenders comprise a higher proportion (between 66.1%, 58.9%, and 62.0%,

respectively) of offenders in groups A, B, and C than offenders in CHC II through VI (50.0%).

While the proportion of White offenders decreases with increasing prior criminal history, the reverse

is true for Black offenders. Offenders in CHC II through VI are more likely (41.8%) to be Black

than are offenders in groups A, B, or C (22.8% , 29.5%, and 27.2%, respectively). Group A has the

highest percent of White offenders (66.1%) and the lowest percent of Black offenders (22.8%).

Hispanic proportions are generally comparable across the columns of Exhibit 1.

Age

There is a positive association between age and the amount of prior criminal history.

Offenders in group A are unique in their high concentration of older offenders: over 35 percent

(35.4%) are age 41 years or older. The proportion of offenders over 50 years of age is higher in

groups A, B, and C (between 8.6% and 12.5%) than for offenders with one criminal history point

(4.2%) or with more than one criminal history point (5.4%).

Illicit Drug Use

Exhibit 2 presents information on social behaviors by first offender group status. In all cases,

offenders in proposed first offender groups A, B, and C generally exhibit what are traditionally

considered the most socially positive characteristics. For instance, 45.6 percent of offenders in the

higher CHCs used illicit drugs in the year prior to their instant offense. For group A, B, and C

offenders, however, illicit drug use in the year prior to their instant offense never exceeds 27.8

percent and is lowest for group A offenders (20.9%).

Employment Status

Exhibit 2 also shows that employment rates for groups A, B, and C are comparatively high

(greater than 80%) as are employment rates for all offenders with zero or only one criminal history

point at 79.5%. In contrast, the employment rate for offenders in CHC II through CHC VI is lower

at 67.6 percent.

Bauman and Graf, 2000, report that 25 percent of the adult population age 25 and over in 1990 did not have

a high school degree.

Page 8 of 31

Marital Status

Offenders in all groups A, B, and C of Exhibit 2 are most likely to be legally married. This

is true especially for group A, where almost half (48.0%) are married. In contrast, offenders in CHC

II through CHC VI are most likely to have never married (40.2%).

Mental Illness

Exhibit 3 shows that the percentages of mental illness do not substantially change between

the first offender groups and offenders with more criminal history. Data from the PSRs indicate that

reported mental illness rates are nearly constant across the columns of Exhibit 3, varying between

8.3 percent and 11.6 percent.

Military Service

Rates of military service only vary slightly across the Exhibit 3 columns. Between 13.3

percent and 22.4 percent of offenders served in the military. Groups A, B, and C do not have the

highest rates of military service.

Educational Attainment

Offenders in groups A, B, and C are more likely to have a high school diploma, and even

some college training, than are offenders with one or more criminal history points. Exhibit 3

indicates that over 70 percent of those in the first offender groups graduated from high school, with

an additional approximate one-third of potential first offenders having at least some college

education. These high school graduation rates are equivalent to those of the general U.S. population

in 1990.

For offenders in CHC II through CHC VI, however, only somewhat more than half

(55.8%) have at least a high school diploma, with 15.1 percent having at least some college-level

training.

Financial Dependents

Over half (52.4%) of offenders in CHC II through CHC VI do not have any financial

dependents. Defendants in first offender groups A, B, and C are more likely to have dependents;

but still, only between 38.4 percent to 43.6 percent of these low criminal history offenders have no

dependents. The finding is consistent with higher rates of married offenders in groups A, B, and C.

Page 9 of 31

2. Instant Offense and Sentence Characteristics

Exhibits 4 and 5 examine the instant offenses of the group A, B, and C offenders and

compare them to offenders with more extensive criminal histories.

Primary Sentencing Guideline

Drug trafficking offenses (§2D1.1) are the most common instant offense types for all

columns in Exhibit 4. Group A is the least likely (26.1%) to have an instant offense under §2D1.1,

followed by group C (34.6%). All other columns have at least 40 percent of instant offenses

sentenced under drug trafficking guidelines.

Group A is unique in that a large proportion (almost half) of its instant offenses are sentenced

under §2F1.1 fraud (25.3%) and §2B1.1 theft (23.2%). Group B has 23.5 percent of its instant

offenses sentenced under these two guidelines, with group C having one third of its cases sentenced

under these two guidelines. For offenders in CHC II through CHC VI, only 16.9 percent of instant

offenses are under §2F1.1 fraud and §2B1.1 theft.

In contrast, almost one quarter of the instant offenses in CHC II through CHC VI are

sentenced under §2K2.1 weapon offenses (14.3%) or §2B3.1 robbery (9.4%). In groups A, B, and

C, these two guidelines account for less than 8 percent of sentencings. The proportion of weapon

and robbery instant offenses is even smaller in group A, where they account for only 2.3 percent of

sentencings.

Dangerousness of Instant Offense Conduct

During project data collection from the PSR, three types of dangerous behaviors were coded

from the section on instant offense conduct. These behaviors include weapon presence during the

offense, actual or threatened violence, and actual or threatened injury to a victim. Exhibit 4 shows

the occurrence of any one of these behaviors. Groups A, B, and C are the least likely to have

dangerousness conduct during the instant offense. The percentages range from a low of 9.3 percent

for group A to a high of 18.4 percent for group B. Dangerous instant offense behavior is far more

likely for offenders with criminal history points: 22.4 percent for offenders with one criminal history

point, and 32.9 percent for offenders in CHC II through CHC VI.

Culpability Criteria

A central tenet behind the first offender philosophy is that first offenders are less culpable

than other offenders. The bottom panel of Exhibit 4 displays four criteria generally associated with

less culpable criminal conduct: no use of violence or weapons; no bodily injury of a victim; a minor

These general criteria reflect those currently existing in the guidelines under §5C1.2 with respect to certain

drug trafficking offenders convicted under statutory minimum sentences. Here, however, their presence is measured for

all instant offenses as a quantifiable proxy for lessened culpability.

Page 10 of 31

role or minimal participation in the instant offense, and acceptance of responsibility.

As an offender meets more of the four of these criteria, the offender is considered less

culpable and thus might be more deserving of a reduced sentence. The last line of data in Exhibit

4 shows the percentage of each column meeting all four of these criteria. Group A offenders, by far,

are the least culpable offenders with 73.8 percent meeting all four criteria. In contrast, the most

culpable offenders are in the higher CHCs: only 49.5 percent of offenders in CHC II through CHC

VI meet all four criteria.

For groups B and C, and for offenders with one criminal history point, between 61.8 percent

and 71.2 percent of offenders meet all four criteria.

Type of Sentence

Exhibit 5 illustrates that offenders in groups A, B, and C are less likely to receive a straight

prison sentence than the offenders in CHC II through CHC VI (87.1%). Group A particularly stands

out with less than half (41.2%) receiving a straight prison sentence. Consistently, group A offenders

also stand out in that 30.3 percent received a probation only sentence. This is almost twice the rate

at which group B and group C offenders receive probation only sentences, and six times higher than

the 4.9 percent of offenders in CHC II through CHC VI. Both group A and C first offenders were

much more likely than any other grouping of offenders to receive probation with an alternative

sentence.

Length of Sentence

The length of imposed instant offense sentences are substantially higher for offenders in the

higher CHCs. For all offenders in CHC II through CHC VI, a total of 24.2 percent have sentences

between two and five years, with an additional 79.9 percent having sentences longer than five years.

This is in sharp contrast to first offender groups A and C where sentences over five years are given

to roughly 12 percent of offenders (12.8% and 11.4%, respectively) and sentences between two and

five years are given to 11.5 percent of group A offenders and 18.2 percent of group C offenders.

Mitigating Role

Exhibit 5 illustrates that the guideline mitigating role adjustment to reduce the offense level

was infrequently given: only 5.6 percent among all offenders in CHCs II through VI received this

adjustment. First time offenders were overall, only slightly more likely to receive this adjustment.

Due to alien deportation following conviction for criminal behavior, it is difficult to measure recidivism of

noncitizens convicted of federal crimes. A companion report in this recidivism project series, “Recidivism of Fiscal Year

1995 Noncitizen Federal Offenders,” addresses citizenship issues in recidivism research.

Details on the structure, methodology, and statistical techniques of the analysis are documented in the

project’s companion report, “Background and Methodology of the U.S. Sentencing Commission’s 2003 Recidivism

Study.”

Page 11 of 31

3. Summary of Offender and Offense Characteristics

Offenders in the proposed first offender groupings A, B, and C are readily distinguishable

from offenders with one or more criminal history points. The proposed first offender groups are

more likely to be White and older. Substantially higher proportions of first offenders are female

than for other offender groupings. They are more likely to have committed a fraud or larceny instant

offense. They have less violent instant offenses, receive shorter sentences, and are less likely to go

to prison. They are less likely to use illicit drugs, more likely to be employed, more likely to have

a high school education (or beyond), and more likely to have financial dependents. Finally,

offenders in groups A, B, and C, compared to other guideline offenders, have instant offenses that

are less culpable and less dangerous.

These distinctions demonstrate that any implementation of a first offender provision will not

only impact a large percentage of the federal caseload, but it will proportionally benefit offenders

in certain demographic, social, personal, and offense categories. In addition, tabulations from the

Commission’s datafiles indicate that many federal guideline offenders eligible for a possible first

offender reduction are also eligible for the §2D1.1 safety value adjustment and its often-overlapping

eligibility criteria.

E. Methodology

he sections below describe the recidivism research project and the definitions of

recidivism used in the analysis.

1. The Recidivism Project

The recidivism study data are a stratified, random sample of 6,062 U.S. citizens

who were

sentenced under the federal sentencing guidelines in fiscal year 1992.

Data on criminal behavior

prior to the federal instant offense, as well as demographic and offender characteristics, were

collected from the federal pre-sentence reports (PSRs) and other court documents submitted to the

Sentencing Commission by U.S. district courts. Prison release date information was extracted from

the SENTRY datafile of the Federal Bureau of Prisons in the U.S. Department of Justice.

Recidivism information was obtained from the “RAP” sheet data of the Federal Bureau of

Investigation’s Criminal Justice Information Services Division office.

While the two year recidivism follow-up period limit was set at June 1, 2001, the actual collection of FBI

“RAP” sheet data did not occur until October 2001. The four month delay between June and October assured that “RAP”

sheet data would reflect all events prior to June 1. The intervening months accounted for administrative processing time

to update fingerprint card information on the FBI datafile, thus minimizing any potential bias due to the states’

differential schedules for reporting data to the FBI.

This figure represents the sum of the sample offenders who by June 1, 2001, were still in prison (8.9%), who

had died (0.4%), or had not been released for at least two years (5.4%).

Page 12 of 31

The sample of offenders used in the recidivism analysis represents the 28,519 U.S. citizen

federal offenders in the Commission’s 1992 fiscal year sentencing datafile where the following “two

year window inclusion” conditions were met:

• a pre-sentence report (PSR) from a fiscal year 1992 sentencing was submitted

to the Sentencing Commission;

• a “RAP” sheet was located on the FBI datafile; and

• for offenders receiving prison sentences, the release from prison occurred at

least two years prior to June 1, 2001.

Given the project schedule for “RAP” sheet collection in October 2001, an offender must

have been released from prison by June 1, 1999. Because the sampled offenders were sentenced in

fiscal year 1992 (i.e., between October 1, 1991 and September 30, 1992), prison sentences as long

as seven years are available for study.

The selection of these specific dates reflects the

Commission’s interest in the recidivism impact of mandatory minimum sentences. The effect of five

year mandatory minimum sentences can be analyzed using the project’s sampling strategy.

With the study’s prison release deadline established as June 1, 2001, there were offenders

in the sample who either were still in prison on this date, or had at this time been released prison for

less than two years. For the entire recidivism sample, 14.7 percent of sample offenders are not in

the analysis because they had not finished serving their prison time or did not have a two year “at

risk” window of opportunity.

With respect to the first offender analysis, however, the focus is

upon CHC I. A total of 91.3 percent of offenders in CHC I meet the two year window inclusion

conditions and are in the analysis. This results in an unweighted sample with 1,389 offenders in

CHC I, of whom 1,128 had zero criminal history points.

2. Definitions

Recidivism results are presented using two substantive definitions. The first, or “primary,”

definition includes the first of any one of the three following events during the offender’s first two

years back in the community:

• a re-conviction for a new offense;

Disposition information was obtained from the FBI’s criminal history record (“RAP” sheet).

Spohn and Holleran, 2002. The methodological arguments supporting various empirical definitions of

recidivism are contained in the companion project report, “Background and Methodology of the U.S. Sentencing

Commission’s 2003 Recidivism Study.”

The re-conviction recidivism rate for zero point offenders is 3.5 percent.

The re-conviction recidivism rate for one point offenders is 5.5 percent.

Page 13 of 31

• a re-arrest with no conviction disposition information available on the post-

release criminal history record;

• a supervision revocation (probation or post prison supervision).

The second “re-conviction only” recidivism definition limits recidivism to the first re-

conviction during the two year follow-up period. It uses only the first of the three events in the

primary recidivism measure listed above.

The use of two different recidivism definitions addresses the state of post-release criminal

behavior records. The recidivism literature recognizes that the FBI offender “RAP” sheets are the

most accurate and readily available data source for repeat criminal behavior. However, “RAP”

sheets can contain errors or partial information. For example, “RAP” sheets only contain

information on offenses for which offender fingerprints were obtained. Additionally, depending on

the reporting policies and practices of local jurisdictions, arrest dispositions may not always be

transferred to the FBI for inclusion on “RAP” sheets. “RAP” sheets will underreport actual criminal

behavior, and will underreport convictions resulting from arrests. For this reason, recidivism studies

(including all of the Commission’s previous recidivism study reports) commonly cite recidivism

rates separately for what is termed here the “primary” and “re-conviction” definitions. However,

these same studies also argue that, compared to the “re-conviction” definition, the “primary”

recidivism definition is a more reliable and valid measure for the probability of actual re-offending

because of its inclusiveness and its high association with actual re-offending.

Also, because these

two definitions are highly correlated, the research findings resulting from the two definitions are

highly related and lead to nearly identical conclusions, although the magnitude of their comparative

findings differs.

Data cited throughout this paper are weighted to represent the entire comparable population

of fiscal year 1992 U.S. citizen offenders sentenced under the federal guidelines.

F. Recidivism Rates for the Proposed First Offender Groups

xhibit 6 shows two-year recidivism rates for the various offender groupings. All

offenders with zero criminal history points have a primary recidivism rate of 11.7

percent.

This zero point offender recidivism rate is substantially lower than the

recidivism rates for offenders with only one criminal history point (22.6%),

or for offenders with

The re-conviction recidivism rate for two or more point offenders is 10.3 percent.

The re-conviction recidivism rate for group A offenders is 2.5 percent.

The re-conviction recidivism rate for group C offenders is 2.9 percent.

The re-conviction recidivism rate for group B offenders is 5.3 percent.

The re-conviction recidivism rate for Remaining Zero group offenders is 5.2 percent.

All offenders in group C have at least one conviction for a minor prior offense under §4A1.2(c)(2). The

arrests appearing in Exhibit 7 for group C involve only arrests not resulting in a conviction.

Page 14 of 31

two or more points (36.5%)

combined in the CHC II through CHC VI column.

Among the potential first offender groups, group A has the lowest primary recidivism rate

at 6.8 percent,

followed by offender group C with a rate of 8.8 percent.

Offenders in group B

have a recidivism rate of 17.2 percent,

which is more than twice the rate of group A and nearly

twice the rate for group C. Even with the comparatively higher recidivism rate, group B’s rate is

still several percentage points lower than offenders in the remaining zeroes group (21.7%).

Its low recidivism rate makes first offender group A stand out. Recall that group A offenders

have no prior criminal events, not even a prior arrest. Compared to group B, which has prior arrests

but no convictions, and a higher recidivism rate, it appears that prior arrests are predictive of

recidivism for offenders with minimal prior criminal experience.

G. Pre-Instant-Offense Criminal Histories

his report section highlights differences in the prior criminal histories of groups A, B,

and C and the link between arrests and recidivism. Following from the analysis

above, it appears that arrests themselves, independent from convictions, predict risk

of recidivism. This relationship is supported in a comparison of group A and group B. Neither

group has prior convictions, but offenders in group A with no prior arrests have dramatically lower

recidivism rates (6.8%) than offenders in group B with prior arrests but no convictions (17.2%).

Several exhibits focus on how recidivism rates vary by the number, types, and characteristics of

prior arrests.

Exhibit 7 shows the distribution of prior arrests for the proposed first offender groups and

the comparison groups. The data for first offender group A are consistent with the group’s

definition: offenders in group A have no prior arrests. For group B first offenders, 58.6 percent have

one prior arrest, 22.8 percent have two prior arrests, and 18.6 percent have three or more prior

arrests. The group B arrest history is more consequential than the history of offenders in group C.

For group C,

only a small proportion have more than three prior arrests. Those in Group B have

higher levels of contact with the criminal justice system, with nearly one out of ten (10.3%) of group

B offenders having four or more arrests.

This information was not collected for CHC II through CHC VI and thus cannot not appear in the exhibit.

The sum of 26.5 percent and 15.7 percent is 42.2 percent.

The small sample size of the group C category constrains confident citing of the proportions for most of the

reasons why points were not assigned to prior offenses. However, the 69.8 percent proportion for dismissed prior arrests

can be used.

Page 15 of 31

Information on the shape of the relationship between number of prior arrests and recidivism

appears in Exhibit 8. The first offender groups A and B represent all guideline sentenced federal

offenders who have never had a prior conviction. The exhibit illustrates that recidivism rates

increase with the number of prior arrests. Those with only one prior arrest (and no convictions) have

a recidivism rate nearly twice that of offenders with no prior arrests (13.2% versus 6.8%,

respectively). Those with two or more prior arrests (and no convictions) have a recidivism rate

nearly double that of those with only one prior arrest (13.2% versus 23.2%, respectively). Clearly,

arrests without convictions are predictive of re-offending behavior.

An additional inquiry focuses on the types of offenses charged for past arrests that did not

result in a conviction. Exhibit 9

again reflects the fact that first offender group A has no prior

arrests. For group B first offenders, arrests without convictions are concentrated in drug trafficking

(11.9%), drug possession (9.7%), assault (8.2%), fraud (8.6%), and theft (10.2%) offenses.

Unfortunately, at this level of detail, the sample size for group C first offenders does not support

comparisons. However, patterns across all columns of Exhibit 9 are comparable. For the different

groupings of offenders in CHC I, the distribution of nonconviction arrests is similar.

These data suggest that the types of non-conviction arrests themselves do not predict

recidivism. However, the reasons why these arrests did not result in convictions do show

meaningful differences. Exhibit 10

addresses these findings. The comparison involves groups B

and C because group A has no prior arrests.

For group B offenders, prior arrests are dismissed or otherwise not prosecuted less than half

of the time (45.4%). In contrast, more than a quarter (26.5%) of the time the disposition of the prior

arrest is pending and for another 15.7 percent of arrests, the offender is under warrant for the prior

arrest, with the sum of these two categories over 40 percent.

Arrests either pending or under

warrant will be either prosecuted or not, and when prosecuted will either result in a conviction or

not. Consequently, for some portion of the pending or warranted arrests, the offender will end up

with a subsequent conviction. This means that in four out of every ten prior arrests, the disposition

of the group B prior arrest may result in a pre-instant offense conviction.

The situation is quite different for group C offenders, where the most common reason that

a prior arrest did not receive points is that the arrest charges were dismissed. Almost 70 percent

(69.8%) of arrest charges of group C offenders are dismissed or otherwise not prosecuted.

There

is essentially no chance that any of these prior dismissals will result in a subsequent conviction.

Page 16 of 31

The differences in group B and group C reasons for non-conviction prior arrests might be

a factor in the groups’ differences in recidivism rates. For group C, the prior arrests are usually

dropped and not prosecuted. For group B, however, the high level of pending or warranted prior

arrests may indicate latent conviction events, spared from receiving criminal history points only

because of the timing of the instant offense sentencing. If this were the case, it suggests that some

group B offenders may have recidivism behavioral tendencies in common with federal offenders

who actually experience one, or more, prior convictions and receive the associated criminal history

points. Not only is this explanation consistent with the higher recidivism rates of group B offenders,

but in addition the data from the last line of Exhibit 10 also suggest that the prior arrests of group

B offenders indicate greater offender culpability. A total of 16.8 percent of group B nonconviction

arrests involve weapons, injury, or violence, the highest percentage among groups in CHC I.

In summary, first offender group A has the lowest recidivism rate and the least culpable

criminal history. Focusing on groups B and C, however, the data reveal an interesting twist. The

recidivism rates in Exhibit 6, combined with prior arrest characteristics depicted in Exhibits 7

through 10, suggest that group B offenders may have greater prior involvement, and possibly more

culpable involvement, with the criminal justice system than is represented by their possession of

zero criminal history points. Although group C first offenders have prior convictions (albeit for

minor traffic and public order offenses which are excluded from receiving criminal history points),

their prior arrest history and recidivism rates indicate that they better reflect the reduced culpability

motivation underlying a first offender philosophy than do offenders in group B.

For prior arrests among first offenders, the following findings emerge. For group B first

offenders there is a large representation (40.0%) of drug trafficking, drug possession, violence, and

larceny offenses. For group C first offenders relatively few (18.7%) of these prior arrests were for

drug trafficking, drug possession, violence, and larceny offenses. Less than two percent (1.4%) of

these prior nonconviction arrests resulted in a not guilty disposition. The number of prior arrests

among those with no prior convictions appears to be directly related to recidivism risk.

These results may go some way in helping to explain why offenders with prior arrests but

no prior convictions, have a recidivism rate nearly three times as large as offenders with no prior

arrests. The prior arrests of group B offenders suggest a correlation with serious criminal conduct

prior to the sentencing of their federal instant offenses.

These arrests were often for more serious

offenses and typically had the charges dismissed, pending, or under warrant for almost nine out of

every ten (87.6%) arrests. These data suggest that at least for some of the prior arrests for these

offenders, the prior charges relate to actual prior offending, and thus generate their observed higher

recidivism rates.

H. Conclusion

his report examines recidivism risk for three plausible first offender groupings

consistent with historical Commission definitions of possible first offenders. The

three first offender groups all come from offenders with zero criminal history points,

and are defined as follows: group A contains offenders with no prior arrests; group B contains

offenders with prior arrests, but no prior convictions; and group C contains offenders with only prior

Page 17 of 31

convictions that are to never count towards criminal history. Among these groups the lowest

recidivism rate is for group A with a rate of 6.8 percent. Group B has a recidivism rate of 17.2

percent. Group C has a recidivism rate of 8.8 percent.

Comparing recidivism rates for the three groups leads to the following conclusions

regarding a first offender definition. Group B offenders are the most problematic as candidates for

first offender status. While group B offenders have no prior convictions, it appears that the absence

of prior convictions is often due to prosecutorial declinations, arrests that were pending prosecution

at the time of sentencing for the instant offense, or arrests with issued outstanding warrants.

Moreover, group B, with its recidivism rate of 17.2 percent, has a substantially greater risk for future

offending compared to groups A and C.

Group C offenders, having been previously convicted of minor traffic or public order crimes,

have only limited contact with the criminal justice system prior to their instant offense. Also, with

a recidivism rate of 8.8 percent, the overwhelming majority of group C offenders – almost nine out

of ten – are not likely to recidivate.

From both culpability and recidivism risk perspectives, group A offenders, with no prior

arrests, most strongly meet the conceptual definition of the first offender category. Offenders in

group A have had no recorded contact with the criminal justice system prior to their instant federal

offense. Moreover, as indicated by their extremely low recidivism rate of 6.8 percent, they are

easily the most empirically identifiable group of guideline federal offenders who are the least likely

to re-offend.

The analysis delineates recidivism risk for offenders with minimal prior criminal history and

shows that the risk is lowest for offenders with the least experience in the criminal justice system.

Offenders with zero criminal history points have lower recidivism rates than offenders with one or

more criminal history points. Even among offenders with zero criminal history points, offenders

who have never been arrested have the lowest recidivism risk of all. Regardless of what definition

is adopted, however, there are a myriad of legal and policy decisions to make. For example, a “no

prior arrest” first offender definition raises issues about the acceptability of using arrest data – often

with incomplete disposition information – and the accuracy of prior history information sources.

A “no prior conviction” definition raises issues about how to count prior convictions that are

otherwise excluded from the criminal history computation due to juvenile, tribal, foreign, or decay

provisions. Other general issues involve the first offender definitions and their empirical

associations with offender characteristics: certain first offender definitions will disproportionally

impact demographic groups (such as women, minorities, or perhaps, noncitizens) or certain instant

offense categories (such as fraud and larceny offenses). Further, an important policy decision is

required to address any overlap between first offender sentencing provisions and safety valve

provisions for drug traffickers under guideline §2D1.1.

Finally, the adoption of some first offender definitions may require the court to collect

information that is not currently included in the criminal history computation and thus may not be

consistently reported in PSRs. Specifically, this information may involve data on prior arrests

Page 18 of 31

without dispositions and prior convictions that do not receive points. If a first offender definition

requires that PSRs collect this information, a standardized protocol may be needed to ensure that

facts are collected completely and accurately. For this report, a small comparison determined that

currently arrest data from PSRs are more complete than arrest data from FBI “RAP” sheets, but it

is not known what level of new information would be additionally needed on the PSRs.

This report is the first step in an empirical discussion regarding a first offender provision.

As with prior Commission efforts, first offender definitions can be developed and their sentencing

outcomes can be estimated. It is only now, however, with the Commission’s recidivism data, that

the analysis will include the impact of alternative first offender provisions on the predictive power

of the guidelines’ criminal history measures.

Page 19 of 31

References

Bauman, Kurt J. and Nikki L. Graf

2000 Educational Attainment. U.S. Census Bureau.

Delong, Elizabeth R., David M. Delong, and Daniel L. Clarke-Pearson

1988 Comparing the areas under two or more correlated receiver operating curves: A

nonparametric approach. Biometrics 44:837-845.

Hosmer, David W. Jr. and Stanley Lemeshow

1999 Applied Survival Analysis: Regression Modeling of Time to Event Data. New York:

Wiley.

Klein, John P. and Malvin L. Moeschberger

1997 Survival Analysis: Techniques for Censored and Truncated Data. New York: Springer.

Maltz, Michael D.

1984 Recidivism. Orlando, FL.: Academic Press.

Mariano, Marianne

2001 Summary of Past Reports on Criminal History Categories “0” and “VII.” United States

Sentencing Commission, Memorandum: June 15, 2001.

O’Neill, Michael E.

2001 Abraham’s Legacy: An Empirical Assessment of (Nearly) First-Time Offenders in The

Federal System. Boston College Law Review 42:291-348.

Ricca, Caryl

1997 Simplification of Chapter Four: Comments from the Probation Officers Advisory Group

to the U.S. Sentencing Commission. Federal Sentencing Reporter 9(4):209-212.

Schmidt, Peter and Ann Dryden Witte

1988 Predicting Recidivism Using Survival Models. New York: Springer-Verlag.

Sphon, Cassia and David Holleran

2002 The Effect of Imprisonment on Recidivism Rates of Felony Offenders: A Focus on Drug

Offenders. Criminology 40:329-357.

Page 20 of 31

Swets, John A., Robyn M. Dawes, and John Monahan

2000 Better Decisions Through Science. Scientific American 283(4):82-87.

United States Sentencing Commission

1989 Memorandum from Working Group on Criminal History to Phyllis Newton concerning

the Proposed Amendments to Criminal History Chapter, December 20. Washington,

D.C.: United States Sentencing Commission.

United States Sentencing Commission

1990 Criminal History Working Group: Discussion Concerning Categories “0” and “VII,”

November 20. Washington, D.C.: United States Sentencing Commission.

United States Sentencing Commission

1991a Criminal History Working Group Report: Category 0, Category VII, Career Offender,

October 17. Washington, D.C.: United States Sentencing Commission.

United States Sentencing Commission

1991b Preliminary Report to the Commission: Staff Working Group on Alternatives, October

23. Washington, D.C.: United States Sentencing Commission.

United States Sentencing Commission

1997 Simplification Draft Paper on Chapter Four. Washington, D.C.: United States

Sentencing Commission.

United States Sentencing Commission

2001a Federal Sentencing Register Notice of Action, BAC2201-40/2211-01, listed under

tentative priority (13). http://ussc.gov/FEDREG/fedr07031.htm.

United States Sentencing Commission

2001b Sourcebook of Federal Sentencing Statistics. Washington, D.C.: United States

Sentencing Commission.

United States Sentencing Commission

2004 Measuring Recidivism: The Criminal History Computation of the Federal Sentencing

Guidelines. Release 1 of the Research Series on the Recidivism of Federal Guideline

Offenders. Washington, D.C.: United States Sentencing Commission.

Page 21 of 31

Exhibit 1

Demographic Characteristics of Offenders, by Criminal History Category and Criminal History Points

With Details for Zero Point Offender Categories

Recidivism Study 2003

CRIMINAL HISTORY CATEGORY I

CATEGORIES

II – VI

ZERO POINT OFFENDERS

ONE

POINT

OFFENDERS

TWO OR

MORE POINT

OFFENDERS

Demographic

Characteristics

Total

Zeroes

Group A

no arrests

Group B

no convictions

Group C

never-count conv

Remaining

Zeroes

N% N% N% N% N% N% N%

TOTAL 14,845 100.0 8,854 100.0 2,499 100.0 460 100.0 3,032 100.0 3,231 100.0 11,672

100.0

Gender

Female 4,459 30.0 3,308 37.4 653 26.1 103‡ 22.3 395 13.0 463 14.4 1,057

9.1

Male 10,386 70.0 5,546 62.6 1,846 73.9 357 77.7 2,637 87.0 2,758 85.6 10,615

90.9

Race/Ethnicity

White 9,365 63.5 5,849 66.1 1,471 58.9 270 62.0 1,775 59.8 1,930 61.0 5,773

50.0

Black 3,798 25.7 2,016 22.8 738 29.5 119 27.2 926 31.2 959 30.3 4,838

41.8

Hispanic 1,173 8.0 756 8.5 237 9.5 23‡ 5.2 157 5.3 195 6.2 703

6.1

Other

420 2.8 233 2.6 52‡ 2.1 24‡ 5.6 110‡ 3.7 78‡ 2.5 239

2.1

Age

Under 21 1,056 7.1 644 7.3 218 8.7 21‡ 4.6 173 5.7 295 9.1 720

6.2

21 – 25 2,183 14.7 1,223 13.8 516 20.6 80‡ 17.4 364 12.0 443 13.7 2,173

18.6

26 – 30 2,294 15.5 1,379 15.6 469 18.8 68‡ 14.8 379 12.5 793 24.6 2,472

21.2

31 – 35 2,252 15.2 1,234 13.9 425 17.0 86‡ 18.6 507 16.7 530 16.4 2,358

20.2

36 – 40 2,103 14.2 1,243 14.0 299 12.0 92‡ 20.0 469 15.4 372 11.5 1,641

14.0

41 – 50 2,944 19.8 1,777 20.1 358 14.3 56‡ 12.1 753 24.9 662 20.5 1,681

14.4

Over 50 2,012 13.6 1,354 15.3 214 8.6 57‡ 12.5 387 12.8 136 4.2 627

5.4

Prior Arrests with no dispositions and convictions for only the “never-count” offenses specifically listed in §4A1.2(c)(2).

“Other” race category includes Native Americans and Asians.

‡ Indicates fewer than 10 sample subjects. Findings may not be statistically significant.

SOURCE: U.S. Sentencing Commission, FY1992 Recidivism Sample (U.S. Citizens), 2003, weighted data. Missing data are excluded, unless specified.

Page 22 of 31

Exhibit 2

Social and Personal Characteristics of Offenders, by Criminal History Category and Criminal History Points

With Details for Zero Point Offender Categories

Recidivism Study 2003

CRIMINAL HISTORY CATEGORY I

CATEGORIES

II – VI

ZERO POINT OFFENDERS

ONE

POINT

OFFENDERS

TWO OR

MORE POINT

OFFENDERS

Social/ Personal

Characteristics

Total

Zeroes

Group A

no arrests

Group B

no convictions

Group C

never-count conv

Remaining

Zeroes

N% N% N% N% N% N% N%

TOTAL 14,845 100.0 8,854 100.0 2,499 100.0 460 100.0 3,032 100.0 3,231 100.0 11,672 100.0

Illicit Drug Use

No Illicit Drug Use 11,072 74.6 7,002 79.1 1,804 72.2 359 78.0 1,907 62.9 2,184 67.6 6,345 54.4

Illicit Drug Use 3,773 25.4 1,852 20.9 695 27.8 101‡ 22.0 1,125 37.1 1,046 32.4 5,325 45.6

Employment Status

Unemployed 2,301 15.5 1,175 13.3 461 18.4 44‡ 9.5 622 20.5 447 13.8 3,778 32.4

Employed 12,544 84.5 7,679 86.7 2,038 81.6 416 90.5 2,411 79.5 2,784 86.2 7,894 67.6

Marital Status

Never Married 4,004 27.0 2,131 24.1 842 33.7 131 28.5 900 29.7 896 27.7 4,688 40.2

Legal Marriage 6,509 43.8 4,253 48.0 848 33.9 204 44.3 1,204 39.7 1,277 39.5 2,694 23.1

Divorced 1,975 13.3 1,079 12.2 309 12.4 56‡ 12.1 532 17.6 380 11.8 1,897 16.2

Other

2,357 15.9 1,391 15.7 501 20.0 70‡ 15.1 396 13.1 677 21.0 2,393 20.5

Prior Arrests with no dispositions and convictions for only the “never-count” offenses specifically listed in §4A1.2(c)(2).

Illicit drug use during the year prior to the instant offense. Missing values counted as “No”illicit drug use.

Employment status during the year prior to the instant offense. “Employed” includes alternative forms of employment and “Unemployed” includes missing values.

“Other” marital status includes “Co-habitating,” “Widowed,” and “Separated.”

‡ Indicates fewer than 10 sample subjects. Findings may not be statistically significant.

SOURCE: U.S. Sentencing Commission, FY1992 Recidivism Sample (U.S. Citizens), 2003, weighted data. Missing data are excluded, unless specified.

Page 23 of 31

Exhibit 3

Social and Personal Characteristics of Offenders, by Criminal History Category and Criminal History Points

With Details for Zero Point Offender Categories

Recidivism Study 2003

CRIMINAL HISTORY CATEGORY I

CATEGORIES

II – VI

ZERO POINT OFFENDERS

ONE

POINT

OFFENDERS

TWO OR

MORE POINT

OFFENDERS

Social/ Personal

Characteristics

Total

Zeroes

Group A

no arrests

Group B

no convictions

Group C

never-count conv

Remaining

Zeroes

N% N% N% N% N% N% N%

TOTAL 14,845 100.0 8,854 100.0 2,499 100.0 460 100.0 3,032 100.0 3,231 100.0 11,672 100.0

Mental Illness

No 13,170 90.4 7,874 90.5 2,201 90.0 416 90.5 2,679 90.3 2,931 91.7 10,087 88.4

Yes 1,407 9.7 832 9.6 244 10.0 44‡ 9.5 288 9.7 266 8.3 1,327 11.6

Prior Military Service

No 12,326 83.0 7,437 84.0 2,166 86.7 371 80.7 2,352 77.6 2,666 82.5 10,053 86.1

Yes 2,520 17.0 1,417 16.0 333 13.3 89‡ 19.3 680 22.4 564 17.5 1,620 13.9

Educational Attainment

< Than High School 3,872 26.2 1,961 22.2 703 28.2 70‡ 15.1 1,139 37.8 1,066 33.0 5,122 44.2

High School 5,820 39.4 3,541 40.1 987 39.7 230 50.1 1,062 35.3 1,138 35.2 4,355 37.6

Some College 3,396 23.0 2,122 24.1 597 24.0 103 22.3 575 19.1 835 25.9 1,758 15.1

College Graduate 1,691 11.4 1,196 13.6 203 8.1 57‡ 12.5 235 7.8 192 5.9 362 3.1

Financial Dependants

0 5,729 40.6 3,261 38.4 996 43.0 200 43.6 1,272 44.9 1,336 43.2 5,599 52.4

01 – 04 7,827 55.5 4900 57.7 1,247 53.8 249 54.2 1,431 50.5 1,647 53.3 4,642 43.5

05 – 10 544 3.9 329 3.9 75‡ 3.2 10‡ 2.3 129 4.6 110‡ 3.5 442 4.1

Prior Arrests with no dispositions and convictions for only the “never-count” offenses specifically listed in §4A1.2(c)(2).

Mental Illness is defined as any mention of a mental illness regardless of whether or not the claim was verified by a doctor. Alcohol and illicit drug abuse/addiction were excluded.

Educational Attainment at the time of the instant offense.

Number of individuals to whom the offender provided financial support during the year prior to the instant offense.

‡ Indicates fewer than 10 sample subjects. Findings may not be statistically significant.

SOURCE: U.S. Sentencing Commission, FY1992 Recidivism Sample (U.S. Citizens), 2003, weighted data. Missing data are excluded, unless specified.

Page 24 of 31

Exhibit 4

Instant Offense Characteristics for Offenders, by Criminal History Category and Criminal History Points

With Details for Zero Point Offender Categories

Recidivism Study 2003

CRIMINAL HISTORY CATEGORY I

CATEGORIES

II – VI

ZERO POINT OFFENDERS

ONE

POINT

OFFENDERS

TWO OR

MORE POINT

OFFENDERS

Instant Offense

Characteristics

Total

Zeroes

Group A

no arrests

Group B

no convictions

Group C

never-count conv

Remaining

Zeroes

N% N% N% N% N% N% N%

TOTAL 14,845 100.0 8,854 100.0 2,499 100.0 460 100.0 3,032 100.0 3,231 100.0 11,672 100.0

Primary Sentencing Guideline

§2D1.1 (drug traf.) 4,924 33.2 2,311 26.1 1,228 49.1 158 34.6 1,226 40.5 1,413 43.7 4,761 40.8

§2F1.1 (fraud) 3,144 21.2 2,241 25.3 370 14.8 68‡ 14.8 465 15.3 486 15.0 1,326 11.4

§2B1.1 (theft) 2,578 17.4 2,049 23.2 217 8.7 82‡ 18.2 230 7.6 426 13.2 640 5.5

§2K2.1 (weapon) 256 1.7 87‡ 1.0 57‡ 2.3 23‡ 4.9 89‡ 2.9 122 3.8 1,667 14.3

§2B3.1 (robbery) 316 2.1 115 1.3 84‡ 3.4 12‡ 2.6 105‡ 3.5 94‡ 2.9 1,094 9.4

Dangerous Instant Offense

Yes 1,941 13.1 824 9.3 461 18.4 45‡ 9.8 611 20.2 723 22.4 3,836 32.9

Culpability Criteria

1. No Viol. or Weap. 12,925 87.1 8,051 90.9 2,028 81.1 404 87.9 2,442 80.5 2,552 79.0 7,908 67.8

2. No Bodily Injury 14,749 99.3 8,833 99.8 2,457 98.3 460 100.0 2,999 98.9 3,210 99.4 11,498 98.4

3. Mitigating Role 13,655 92.0 8,119 91.7 2,298 92.0 458 99.6 2,779 91.6 2,918 90.3 10,683 91.5

4. Acceptance of Resp. 12,373 83.3 7,562 85.4 2,050 82.0 374 81.4 2,386 78.7 2,644 81.8 8,819 75.6

Meet All Four Above

Culpability Criteria

10,356 69.8 6,532 73.8 1,623 64.9 327 71.2 1,873 61.8 2,000 61.9 5,782 49.5

Prior Arrests with no dispositions and convictions for only the “never-count” offenses specifically listed in §4A1.2(c)(2).

Dangerousness indicates whether or not an offender during the Commission of their instant offense either threatened or caused actual violence or injury and or a weapon was involved. Missing values are counted

as “No” dangerousness present.

§5C1.2 - Limitation on Applicability of Statutory Minimum Sentences in Certain Cases. Examines Safety Valve eligibility criteria - §5C1.2(a)(2), §5C1.2(a)(3), §5C1.2(a)(4), §5C1.2(a)(5), excluding §5C1.2(a)

and §5C1.2(a)(1) exclusions.

‡ Indicates fewer than 10 sample subjects. Findings may not be statistically significant.

SOURCE: U.S. Sentencing Commission, FY1992 Recidivism Sample (U.S. Citizens), 2003, weighted data. Missing data are excluded, unless specified.

Page 25 of 31

Exhibit 5

Sentence Characteristics for Offenders, by Criminal History Category and Criminal History Points

With Details for Zero Point Offender Categories

Recidivism Study 2003

CRIMINAL HISTORY CATEGORY I

CATEGORIES

II – VI

ZERO POINT OFFENDERS

ONE

POINT

OFFENDERS

TWO OR MORE

POINT

OFFENDERS

Sentence

Characteristics

Total

Zeroes

Group A

no arrests

Group B

no convictions

Group C

never-count conv

Remaining

Zeroes

N% N% N% N% N% N% N%

TOTAL 14,845 100.0 8,854 100.0 2,499 100.0 460 100.0 3,032 100.0 3,231 100.0 11,672 100.0

Type of Sentence

Fine Only 134 0.9 121 1.4 0‡ 0.0 0‡ 0.0 12‡ 0.4 12‡ 0.4 3‡ 0.0

Probation Only 3,646 24.6 2,685 30.3 364 14.6 87‡ 18.9 510 16.8 562 17.4 572 4.9

Probation +Alternatives 2,768 18.7 1,894 21.4 340 13.7 121 26.4 413 13.6 486 15.0 542 4.7

Prison + Alternatives 818 5.5 503 5.7 126 5.1 10‡ 2.3 178 5.9 168 5.2 392 3.4

Prison 7,467 50.3 3,650 41.2 1,657 66.6 241 52.4 1,919 63.3 2,003 62.0 10,163 87.1

Length of Sentence

0 6,560 44.4 4,701 53.3 716 28.8 208 45.3 935 31.0 1,060 33.0 1,117 9.6

01 – 05 1,059 7.2 640 7.3 157 6.3 31‡ 6.8 231 7.7 273 8.5 421 3.6

06 – 11 797 5.4 419 4.8 136 5.5 21‡ 4.6 220 7.3 241 7.5 800 6.9

12 – 23 1,678 11.3 912 10.3 315 12.6 63‡ 13.7 388 12.9 294 9.2 1,770 15.2

24 – 59 2,014 13.6 1,017 11.5 503 20.2 84‡ 18.2 409 13.6 461 14.4 2,809 24.2

60 or More 2,674 18.1 1,133 12.8 661 26.6 52‡ 11.4 828 27.5 881 27.4 4,704 40.5

Points for Mitigating Role

0 points 13,537 91.6 8,060 91.4 2,213 89.4 427 95.0 2,837 93.6 2,902 90.1 10,936 94.5

-2 points 879 6.0 527 6.0 199 8.0 23‡ 5.0 131 4.3 239 7.4 482 4.2

-3 points 31‡ 0.2 10‡ 0.1 10‡ 0.4 0‡ 0.0 10‡ 0.4 10‡ 0.3 46 0.4

-4 points 330 2.2 221 2.5 54‡ 2.2 0‡ 0.0 54‡ 1.8 70‡ 2.2 110 1.0

Prior Arrests with no dispositions and convictions for only the “never-count” offenses specifically listed in §4A1.2(c)(2).

The type of sentenced given for the offender’s instant offense.

The sentence imposed for the offender’s instant offense, presented in months.

‡ Indicates fewer than 10 sample subjects. Findings may not be statistically significant.

SOURCE: U.S. Sentencing Commission, FY1992 Recidivism Sample (U.S. Citizens), 2003, weighted data. Missing data are excluded, unless specified.

Page 26 of 31

Exhibit 6

Recidivism Rates, by Criminal History Category and Criminal History Points

With Details for Zero Point Offender Categories

Recidivism Study 2003

CRIMINAL HISTORY CATEGORY I

CATEGORIES

II – VI

ZERO POINT OFFENDERS

ONE

POINT

OFFENDERS

TWO OR MORE

POINT

OFFENDERS

Total

Zeroes

Group A

no arrests

Group B

no convictions

Group C

never-count conv

Remaining

Zeroes

N% N% N% N% N% N% N%

TOTAL

12,546 100.0 7,448 100.0 2,089 100.0 416 100.0 2,592 100.0 2,882 100.0 8,895 100.0

Primary Recidivism

Definition

Did Recidivate 1,465 11.7 508 6.8 359 17.2 36‡ 8.8 561 21.7 650 22.6 3,250 36.5

Did NOT Recidivate 11,081 88.3 6,940 93.2 1,730 82.8 380 91.2 2,031 78.3 2,231 77.4 5,646 63.5

Re-Conviction

Recidivism Definition

Did Recidivate 442 3.5 184 2.5 111‡ 5.3 12‡ 2.9 134 5.2 158 5.5 913 10.3

Did NOT Recidivate 12,104 96.5 7,264 97.5 1,978 94.7 404 97.1 2,458 94.8 2,723 94.5 7,981 89.7

Number of offenders with a 24 month period at risk of recidivating following either initiation of probation (for offenders receiving probation-only sentences) or release from confinement (for those offenders

receiving confinement sentences).

Prior Arrests with no dispositions and convictions for only the “never-count” offenses specifically listed in §4A1.2(c)(2).

Primary recidivism definition based on offender’s re-arrest, including supervised release/ probation violations, re-arrest, or re-conviction.

Re-conviction recidivism definition based solely on the offender’s re-conviction, excluding any supervised release/ probation violations or re-arrests.

‡ Indicates fewer than 10 sample subjects. Findings may not be statistically significant.

SOURCE: U.S. Sentencing Commission, FY1992 Recidivism Sample (U.S. Citizens), 2003, weighted data. Missing data are excluded, unless specified.

Page 27 of 31

Exhibit 7

Number of Prior Arrests,

by Criminal History Category I

With Details for Zero Point Offender Categories

Recidivism Study 2003

CRIMINAL HISTORY CATEGORY I

ZERO POINT OFFENDERS

ONE

POINT

OFFENDERS

Offender’s Number of Prior

Arrests

Total

Zeroes

Group A

no arrests

Group B

no convictions

Group C

never-count conv

Remaining

Zeroes

N% N% N% N% N% N%

TOTAL 14,845 100.0 8,854 100.0 2,499 100.0 460 100.0 3,032 100.0 3,231 100.0

0 Arrests 10,710 72.1 8,854 100.0 0‡ 0.0 303 65.9 1,553 51.2 1,722 53.3

01 Arrest 2,265 15.3 0 0.0 1,465 58.6 124 26.9 677 22.3 685 21.2

02 Arrests 992 6.7 0 0.0 570 22.8 23‡ 4.9 400 13.2 363 11.2

03 Arrests 387 2.6 0 0.0 207 8.3 10‡ 2.3 169 5.6 183 5.7

04 – 09 Arrests 469 3.2 0 0.0 257 10.3 0‡ 0.0 212 7.0 225 7.0

10+ Arrests 21‡ 0.1 00.0 0‡ 0.0 0‡ 0.0 21‡ 0.7 52‡ 1.6

Prior arrests that did not result in a conviction (i.e., these arrests were found not guilty, dismissed, were pending, on warrant status, or disposition records could not be located by P.O.).

Group A offender’s do not have prior arrests, by definition.

Prior Arrests with no dispositions and only for “never-count” convictions for the offenses specifically listed in §4A1.2(c)(2).

‡ Indicates fewer than 10 sample subjects. Findings may not be statistically significant.

SOURCE: U.S. Sentencing Commission, FY1992 Recidivism Sample (U.S. Citizens), 2003, weighted data. Missing data are excluded, unless specified.

Page 28 of 31

Exhibit 8

Recidivism Rates, by Prior Arrests for Offenders

With No Prior Convictions

(First Offender Group A and Group B)

Offender’s Number

of Prior Arrests

Two-Year

Recidivism Rates

Group A

0 arrests 6.8

Group B

1 arrest 13.2

2 or more arrests 23.2

SOURCE: U.S. Sentencing Commission, FY1992 Recidivism Sample (U.S.

Citizens), 2003, weighted data.

Page 29 of 31

Exhibit 9

Type of Counts That Did Not End in a Conviction,

by Criminal History Category I

With Details for Zero Point Offender Categories

Recidivism Study 2003

CRIMINAL HISTORY CATEGORY I

ZERO POINT OFFENDERS

ONE

POINT

OFFENDERS

Type of Counts That Did Not End

in a Conviction

Total

Zeroes

Group A

no arrests

Group B

no convictions

Group C

never-count conv

Remaining

Zeroes

N% N% N% N% N% N%

TOTAL

10,270 100.0 0 100.0 5,935 100.0 233 100.0 4,102 100.0 4,626 100.0

Drug Trafficking

Cocaine 309 3.0 265 4.5 0‡ 0.0 44‡ 1.1 99‡ 2.1

Crack 87‡ 0.8 77‡ 1.3 0‡ 0.0 10‡ 0.3 23‡ 0.5

Marijuana 171 1.7 108‡ 1.8 0‡ 0.0 63‡ 1.5 111‡ 2.4

Other 360 3.5 253 4.3 10‡ 4.5 96‡ 2.3 211 4.6

Drug Possession

Cocaine 224 2.2 150 2.5 0‡ 0.0 73‡ 1.8 105‡ 2.3

Marijuana 436 4.2

Group A

Offenders

Have No Prior

Arrests/Counts

229 3.8 12‡ 5.2 195 4.8 153 3.3

Other 361 3.5 200 3.4 0‡ 0.0 161 3.9 105‡ 2.3

Assault

Battery 370 3.6 161 2.7 21‡ 9.0 188 4.6 108‡ 2.3

Aggravated 234 2.3 117 2.0 0‡ 0.0 117 2.8 117 2.5

Simple 63‡ 0.6 42‡ 0.7 0‡ 0.0 21‡0.5 42‡ 0.9

Other 263 2.6 164 2.8 0‡ 0.0 99‡ 2.4 150 3.2

Fraud

Insufficient Check 354 3.4 278 4.7 33

‡ 14.2 44‡ 1.1 403 8.7

Other 295 2.9 230 3.9 0‡ 0.0 65‡ 1.6 120 2.6

Exhibit 9 Continued . . .

Page 30 of 31

Exhibit 9 Continued

CRIMINAL HISTORY CATEGORY I

ZERO POINT OFFENDERS

ONE

POINT

OFFENDERS

Type of Counts That Did Not End

in a Conviction

Total

Zeroes

Group A

no arrests

Group B

no convictions

Group C

never-count conv

Remaining

Zeroes

N% N% N% N% N% N%

TOTAL

10,270 100.0 0 100.0 5,935 100.0 233 100.0 4,102 100.0 4,626 100.0

Larceny

Grand 162 1.6 110‡ 1.8 0‡ 0.0 52‡ 1.3 117 2.5

Petty 181 1.8 113 1.9 0‡ 0.0 68‡ 1.6 91‡ 1.9

Shoplifting 166 1.6 101‡ 1.7 0‡ 0.0 65‡ 1.6 31‡ 0.7

Burglary 208 2.0 131 2.2 0‡ 0.0 77‡ 1.9 89‡ 1.9

Other 345 3.4 153 2.6 0‡ 0.0 192 4.7 239 5.2

Group A

Traffic Offenders

DUI 441 4.3 Have No Prior 225 3.8 0‡ 0.0 216 5.3 204 4.4

Reckless Driving 200 1.9 Arrests/Counts 155 2.6 23‡9.7 23‡ 0.5 21‡ 0.5

Driving w/o a License or Suspended 368 3.6 131 2.2 23‡ 9.7 215 5.2 164 3.5

Other 623 6.1 370 6.2 0‡ 0.0 253 6.2 248 5.4

Other

Failure to Appear/ Pay Fine 426 4.1 99‡ 1.7 0‡ 0.0 326 7.9 245 5.3

All Other Offenses 3,623 35.3 2,073 34.9 111‡ 47.7 1,439 35.1 1,427 30.9

The disposition of these counts were either dismissed, not guilty, pending, on warrant status, the disposition could not be located by the P.O., or anything else other than guilty. The N does not represent people,

cases, or arrests rather the number of counts for each variable.

Prior Arrests with no dispositions and convictions for only the “never-count” offenses specifically listed in §4A1.2(c)(2).

‡ Indicates fewer than 10 sample subjects. Findings may not be statistically significant.

SOURCE: U.S. Sentencing Commission, FY1992 Recidivism Sample (U.S. Citizens), 2003, weighted data. Missing data are excluded, unless specified.

Page 31 of 31

Exhibit 10

Prior Arrest Characteristics,

by Criminal History Category I

With Details for Zero Point Offender Categories

Recidivism Study 2003

CRIMINAL HISTORY CATEGORY I

ZERO POINT OFFENDERS

ONE

POINT

OFFENDERS

Prior Arrest Characteristics

Total

Zeroes

Group A

no arrests

Group B

no convictions

Group C

never-count conv

Remaining

Zeroes

N% N% N% N% N% N%

TOTAL

8,059 100.0 0 100.0 4,545 100.0 200 100.0 3,313 100.0 3,723 100.0

Disposition of Arrest

Not Guilty 187 2.5 56‡ 1.4 0‡ 0.0 131 4.0 75‡ 2.0

Dismissed 3,312 44.8 1,780 45.4 133 69.8 1,400 42.7 1,625 44.2

Pending 1,927 26.1 Group A 1,037 26.5 24‡ 12.8 866 26.4 1,208 32.8

Warrant 780 10.6 Offenders 617 15.7 0‡ 0.0 164 5.0 42‡ 1.1

P.O. Not Find Disposition Record 1,067 14.4 Have No Prior 369 9.4 33‡ 17.4 665 20.2 721 19.6

Other 119 1.6 Arrests/Counts 63‡1.6 0‡ 0.0 56‡ 1.7 10‡ 0.3

Dangerous Prior Arrest Conduct

Yes 1,176 14.6 764 16.8 0‡ 0.0 412 12.4 204 5.5

Prior arrests that did not result in a conviction (i.e., these arrests were found not guilty, dismissed, were pending, on warrant status, or disposition records could not be located by P.O.). The N represents the total

number of prior arrests not leading to convictions.

Prior Arrests with no dispositions and convictions for only the “never-count” offenses specifically listed in §4A1.2(c)(2).

Dangerousness indicates whether or not an offender in the Commission of their prior arrest either threatened or caused actual violence or injury and or a weapon was involved. Missing values are counted as “No”

dangerousness present.

‡ Indicates fewer than 10 sample subjects. Findings may not be statistically significant.

SOURCE: U.S. Sentencing Commission, FY1992 Recidivism Sample (U.S. Citizens), 2003, weighted data. Missing data are excluded, unless specified.