Federal Drug Offenses: Departures from Sentencing Guidelines and 
Mandatory Minimum Sentences, Fiscal Years 1999-2001 (24-OCT-03,  
GAO-04-105).							 
                                                                 
Created in 1984, the United States Sentencing Commission (USSC)  
was charged with developing the federal sentencing guidelines to 
limit disparities in sentencing among offenders with similar	 
criminal backgrounds found guilty of similar crimes. Judges	 
determine a specific sentence based on an applicable sentencing  
guideline range, such as 57 to 71 months, provided in the	 
guidelines. Judges may impose sentences that fall anywhere within
the range, above it (upward departures), or below it (downward	 
departures). For some offenses, Congress established mandatory	 
minimum sentences. Judges may also sentence below the minimum in 
certain circumstances. We examined the differences in drug	 
offense departures from sentencing guidelines and mandatory	 
minimum sentences among federal courts and the documents the USSC
used to record and analyze sentences.				 
-------------------------Indexing Terms------------------------- 
REPORTNUM:   GAO-04-105 					        
    ACCNO:   A08765						        
  TITLE:     Federal Drug Offenses: Departures from Sentencing	      
Guidelines and Mandatory Minimum Sentences, Fiscal Years	 
1999-2001							 
     DATE:   10/24/2003 
  SUBJECT:   Courtroom proceedings				 
	     Crimes or offenses 				 
	     Criminal procedure 				 
	     Criminals						 
	     Drugs						 
	     Federal courts					 
	     Judges						 
	     Comparative analysis				 
	     Criminal sentencing				 

******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO Product.                                                 **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
******************************************************************
GAO-04-105

United States General Accounting Office

GAO

                       Report to Congressional Requesters

October 2003

FEDERAL DRUG OFFENSES

 Departures from Sentencing Guidelines and Mandatory Minimum Sentences, Fiscal
                                Years 1999-2001

GAO-04-105

October 2003

FEDERAL DRUG OFFENSES

Departures from Sentencing Guidelines and Mandatory Minimum Sentences, Fiscal
Years 1999-2001

Generally, downward departures are defined as (1) substantial assistance
departures, made at the prosecutor's request because the offender provided
substantial assistance to the government; and (2) other downward
departures made for other reasons, such as a plea agreement, a judge's
consideration of mitigating factors, or early disposition, i.e., "fast
track" programs initiated by prosecutors for low-level drug trafficking
offenses.

Of federal sentences for drug-related offenses in fiscal years 1999-2001,
the majority (56 percent) was within applicable guideline ranges. Downward
sentencing departures were more frequently due to prosecutors' substantial
assistance motions (28 percent) than for any other reasons (16 percent).

For federal drug sentences that carried a mandatory minimum term of
imprisonment, more than half of the drug sentences imposed fell below a
mandatory minimum. Of these, half fell below a minimum due to prosecutors'
substantial assistance motions and half due to other reasons.

After adjusting for differences in offense and offender characteristics
among judicial circuits and districts, our analysis showed variations
among certain circuits and districts in the likelihood an offender
received a substantial assistance departure, other downward departure, or
a sentence falling below a mandatory minimum. However, these variations
did not necessarily indicate unwarranted sentencing departures or
misapplication of the guidelines because data were not available to fully
compare the offenders and offenses for which they were convicted.

For drug sentences nationally, USSC receives 96 percent or more of the
three key documents, including the statement of reasons (SOR), used to
record sentence length and departures. For a small percentage of drug
cases in USSC's database, information is missing, incomplete, or too
difficult for USSC to interpret, principally affecting sentencing analyses
in districts where the missing or incomplete data are most prevalent.

Downward Departures for Federal Drug Sentences Imposed in FY 1999-2001

Highlights of GAO-04-105, a report to congressional requesters

Created in 1984, the United States Sentencing Commission (USSC) was
charged with developing the federal sentencing guidelines to limit
disparities in sentencing among offenders with similar criminal
backgrounds found guilty of similar crimes. Judges determine a specific
sentence based on an applicable sentencing guideline range, such as 57 to
71 months, provided in the guidelines. Judges may impose sentences that
fall anywhere within the range, above it (upward departures), or below it
(downward departures). For some offenses, Congress established mandatory
minimum sentences. Judges may also sentence below the minimum in certain
circumstances.

We examined the differences in drug offense departures from sentencing
guidelines and mandatory minimum sentences among federal courts and the
documents the USSC used to record and analyze sentences.

We recommend that USSC and the Administrative Office of the U.S. Courts
(AOUSC) continue to collaborate on education programs to encourage judges
and other court officials to use AOUSC's standard statement of reasons
form (SOR), to complete the SOR more effectively; and to revise the
standard SOR to better meet the

                       data collection needs of the USSC.

                     www.gao.gov/cgi-bin/getrpt?GAO-04-105.

To view the full product, including the scope
and methodology, click on the link above.
For more information, contact William O. Source: GAO analysis of USSC
data.
Jenkins at (202) 512-8757 or
[email protected].

Contents

  Letter

Results in Brief
Background
Federal Drug Sentences Departed Downward More Often Due to

Substantial Assistance than for Other Reasons
More Than Half of Drug Sentences Fell below an Otherwise
Applicable Mandatory Minimum Sentence
Likelihood of Sentences Departing Downward or Falling below a
Mandatory Minimum Varied by Judicial Circuits and Districts
Opportunities Exist to Improve Sentencing Data Collection and

Reporting
Conclusions
Recommendations
Agency Comments

                                       1

                                      3 5

11

14

19

21 24 24 25

  Appendix I Objectives, Scope, and Methodology 28

Objectives 28
Scope 28
Methodology 29

Appendix II 	Percentage of Sentences Falling below an Applicable Guideline
Range or An Otherwise Applicable Mandatory Minimum

Appendix III 	Likelihood of Sentences Falling below a Guideline Range or
an Otherwise Applicable Mandatory Minimum

  Appendix IV 	Data for Analyzing Departures and Mandatory
  Minimum Sentences at Circuit and District Court
  Levels Limited 61

Documents USSC Requests from District Courts 61
How USSC Uses Documents to Create Its Database 61
USSC Receives Most of Requested Sentencing Documents 62
USSC Takes Steps to Reduce Missing Document Rate 63

Missing or Difficult to Interpret Information Can Affect Analysis of

Departures 64 USSC Actions to Improve Coding 65 Multiple Reasons Cited for
Missing Documentation and

Information 65 Other Downward Departures Do Not Solely Reflect Judicial

Discretion 67

Appendix V Comments from U.S. Sentencing Commission

Appendix VI 	Comments from the Judicial Conference Committee on Criminal
Law

  Appendix VI GAO Contacts and Staff Acknowledgments 81

GAO Contacts 81 Staff Acknowledgments 81

  Tables

Table 1: Comparing Downward Departures Due to Substantial Assistance and
Other Reasons for All Federal Sentences with Federal Drug Sentences,
Fiscal Years 1999-2001 12

Table 2: Numbers and Percents of Drug Cases Involving Different Types of
Offenses or Offenders, and Percentages of Each Type Receiving Sentences
Departing Downward from the Guidelines 29

Table 3: Numbers and Percents of Drug Cases Involving Different Types of
Offenses or Offenders, and Percentages of Each Type Receiving Sentences
Departing Downward from the Guidelines 32

Table 4: Hypothetical Example Showing Effect of Two Different Methods of
Computing Percentages for "Other Downward Departures" 34

Table 5: Percentages of Drug Cases between 1999 and 2001 Resulting in
Upward Departures, Within Range Sentences, and Downward Departures, by
Circuit 39

Table 6: Percentages of Drug Cases between 1999 and 2001 Resulting in
Upward Departures, Within Range Sentences, and Downward Departures, by
District 40

Table 7: Percentages of Mandatory Minimum Drug Cases between 1999 and 2001
Resulting in Sentences at or above a Minimum, and below An Otherwise
Applicable Minimum Due to Substantial Assistance or for Other Reasons, by
Circuit 44

Table 8: Percentages of Mandatory Minimum Drug Cases between 1999 and 2001
Resulting in Sentences at or above the Minimum, and below An Otherwise
Applicable Minimum Due to Substantial Assistance or For Other Reasons, by
District 45

Table 9: Odds on Substantial Assistance Departures and Other Downward
Departures, and Odds Ratios Indicating the Differences between Circuits,
before and after Adjusting for Offense and Offender Characteristics 50

Table 10: Odds on Substantial Assistance Departures and Other Downward
Departures, and Odds Ratios Indicating the Differences between Districts,
before and after Adjusting for Offense and Offender Characteristics 51

Table 11: Odds on Mandatory Minimum Sentences Falling below an Otherwise
Applicable Mandatory Minimum for Substantial Assistance and for Other
Reasons, and Odds Ratios Indicating the Differences between Circuits,
before and after Adjusting for Offense and Offender Characteristics 56

Table 12: Odds on Mandatory Minimum Sentences Falling below an Otherwise
Applicable Mandatory Minimum for Substantial Assistance and for Other
Reasons, and Odds Ratios Indicating the Differences between Districts,
before and after Adjusting for Offense and Offender Characteristics 57

Table 13: Percent of Missing Drug Sentencing Documents, by Circuit, as
Shown in USSC's Database for Fiscal Years 1999-2001 63

Table 14: Percent of Missing Drug Sentencing Information on Documents USSC
Received, by Circuit, Fiscal Years 1999-2001 64

Table 15: First Reason Provided by Judges for Other Downward Departure in
Drug Sentences Nationwide, Fiscal Years 1999-2001 68

Table 16: Second Reason Provided by Judges for Other Downward Departure in
Drug Sentences Nationwide, Fiscal Years 1999-2001 72

Table 17: Third Reason Provided by Judges for Other Downward Departure in
Drug Sentences Nationwide, Fiscal Years 1999-2001 74

  Figures

Figure 1: Sentences Departing from Guideline Ranges and Sentences Falling
below a Mandatory Minimum 2 Figure 2: Geographical Boundaries of the 12
Regional Circuit Courts of Appeals and 94 District Courts 6 Figure 3:
Illustration of a Sentencing Guideline for an Offense of Conviction
Carrying a Mandatory Minimum 8

Figure 4: Percent of Substantial Assistance and Other Downward Departures
for Drug Sentences, by Circuit, Fiscal Years 1999-2001 13

Figure 5: Number of Districts by Percent of Drug Sentences Departing
Downward Due to Substantial Assistance or for Other Reasons, Fiscal Years
1999-2001 14

Figure 6: Percent of Drug Sentences Meeting or below an Otherwise
Applicable Mandatory Minimum, Fiscal Years 1999-2001 15

Figure 7: Summary of Mandatory Minimum Drug Sentences below an Otherwise
Applicable Mandatory Minimum and Departing Downward Compared with
Sentences that Did Not Carry a Mandatory Minimum, Fiscal Years 1999-2001
17

Figure 8: Percent of Drug Sentences Falling below an Otherwise Applicable
Mandatory Minimum due to Substantial Assistance or for Other Reasons,
Fiscal Years 1999-2001 18

Figure 9: Number of Districts by Percent of Sentences Falling below an
Otherwise Applicable Mandatory Minimum Due to Substantial Assistance or
Other Reasons, Fiscal Years 1999-2001 19

Abbreviations

AOUSC Administrative Office of the U.S. Courts
J&C Judgment and Commitment Order

PSR Presentence Report
SOR Statement of Reasons
USSC U.S. Sentencing Commission

This is a work of the U.S. government and is not subject to copyright
protection in the United States. It may be reproduced and distributed in
its entirety without further permission from GAO. However, because this
work may contain copyrighted images or other material, permission from the
copyright holder may be necessary if you wish to reproduce this material
separately.

United States General Accounting Office Washington, DC 20548

October 24, 2003

The Honorable F. James Sensenbrenner
Chairman
Committee on the Judiciary
House of Representatives

The Honorable Howard Coble
Chairman
Subcommittee on Crime, Terrorism,

and Homeland Security Committee on the Judiciary House of Representatives

The Sentencing Reform Act of 19841 established the independent U.S.
Sentencing Commission (USSC) and charged it with developing the Federal
Sentencing Guidelines. The guidelines were intended to avoid unwarranted
disparities in sentences among defendants with similar criminal records
who were found guilty of similar criminal conduct without eliminating the
thoughtful imposition of individualized sentences.2 To foster this goal,
the guidelines specify sentencing guideline ranges-a range of time (in
months)-that offenders should serve given the nature of their offense and
other factors but also permit sentences to depart upward or downward from
guideline ranges due to aggravating or mitigating circumstances. For
certain types of offenses, including certain drug and weapons offenses,
Congress has statutorily specified mandatory minimum sentences that
supplant the otherwise applicable guidelines range (see fig. 1). Judges
may impose sentences that depart downward from established guidelines or
fall below a mandatory minimum at the request of the prosecution because
the defendant has provided substantial assistance to the government. USSC
designates sentences departing downward for this reason as substantial
assistance departures. 3 A judge may also depart downward as a result of
accepting a plea agreement or after considering characteristics of the
offender or the offense not fully covered in the

118 U.S.C. 3551 et seq., 28 U.S.C. 991-998. 228 U.S.C. 991(b)(1)(B); see
also S. Rep. No. 98-225, p. 52 (1983). 3See 18 U.S.C. 3553(e).

guidelines, such as extraordinary rehabilitation efforts after the offense
but prior to sentencing. To distinguish such sentences from substantial
assistance sentences, USSC designates them as "other downward departures"
as part of its data gathering efforts.

Figure 1: Sentences Departing from Guideline Ranges and Sentences Falling below
                              a Mandatory Minimum

                Sources: USSC 2002 guidelines and GAO analysis.

As you requested, this report discusses downward departures for federal
drug offenses. We identified the extent to which federal drug sentences
departed downward from a guideline range determined by the court or fell
below an applicable mandatory minimum in fiscal years 1999-2001. Our
objectives were to (1) identify the percentage of federal sentences, and,
specifically, those for drug-related offenses, departing downward due to
substantial assistance motions or other reasons; (2) identify the
percentage of federal drug sentences that fell below an applicable
mandatory minimum due to substantial assistance motions or other reasons;
(3) compare the likelihood across judicial circuits and districts that
offenders received downward departing sentences or sentences below a
mandatory minimum; and (4) identify limitations, if any, of USSC's
sentencing data for drug offenses.

To meet these objectives, we obtained USSC's sentencing database for
fiscal years 1999 through 2001 and used its coding scheme to identify (1)
sentences departing downward due to a prosecutor's substantial assistance
motion (substantial assistance departures); (2) sentences that

departed downward for other reasons (other downward departures); and (3)
sentences that fell below an applicable mandatory minimum. We used the
offender and offense characteristics in USSC's database to statistically
control for major differences among circuits and districts in the types of
offenders sentenced and the offenses for which they were sentenced.
However, it is important to note that a variety of other variables, such
as an offender's extraordinary rehabilitation efforts after the offense
but prior to sentencing, could affect departure decisions, but data for
these variables are not readily available.

We also reviewed the types of documents USSC staff used to identify
departures, the reasons for those departures, and the number of missing
documents by circuit and district. We interviewed officials at USSC and
the Administrative Office of the U.S. Courts (AOUSC), the administrative
arm of the federal judiciary, and the Chair, Committee on Criminal Law of
the Judicial Conference of the United States, the judiciary's principal
policymaking body. We conducted our work between March and August 2003 in
Washington, D.C., in accordance with generally accepted government
auditing standards. Appendix I includes a more detailed description of our
scope and methodology. Appendixes II and III provide more detailed
descriptions of how we applied our methodology.

Federal judges imposed 175,245 sentences subject to the federal sentencing
guidelines in fiscal years 1999-2001. Of the 162,090 sentences for which
complete departure information was available, the majority of all federal
criminal sentences (64 percent) as well as drug sentences (56 percent)
were within applicable guideline ranges. Of the federal sentences that
departed downward, 18 percent departed downward due to prosecutors'
substantial assistance motions and 17 percent departed downward due to
other reasons. Federal drug sentences imposed during the same period
departed downward more frequently due to prosecutors' substantial
assistance motions (28 percent) than for other reasons (16 percent).4

4Other reasons that sentences departed downward included early
disposition, that is, "fast track," programs initiated by prosecutors;
plea agreements; and judges' consideration of mitigating circumstances.
See appendix IV for more information on the frequency of reasons cited for
other downward departures.

  Results in Brief

Of 68,670 federal drug sentences imposed during fiscal years 1999-2001
with complete departure and sentence length information, 41,861 carried a
mandatory minimum term of imprisonment. Of these, more than half (52
percent) fell below an otherwise applicable mandatory minimum sentence. An
equal proportion of sentences fell below an otherwise applicable mandatory
minimum due to substantial assistance (26 percent) as for other reasons,
such as the safety valve (26 percent).

After taking into account differences in offender and offense
characteristics across the 12 judicial circuits and 94 districts, our
statistical analysis showed major variation among certain judicial
circuits and districts in the likelihood of an offender receiving a
substantial assistance departure, other downward departure, or a sentence
falling below a mandatory minimum. For example, offenders sentenced in the
Third Circuit during fiscal years 1999-2001 were over 3 times more likely
to receive a substantial assistance departure at a prosecutor's initiative
than offenders sentenced in the First Circuit. Further, the odds that an
offender would be granted an "other downward departure" also varied
substantially across circuits. For example, offenders sentenced in the
Ninth Circuit were over 18 times more likely to have received an other
downward departure than offenders sentenced in the Fourth Circuit. We
analyzed the extent to which sentences departed below applicable
guidelines on average nationally; however, we did not compare the degree
to which the length of the departures varied across circuits and
districts. Finally, it is important to note the departure differences
among circuits and districts that we found may not, in and of themselves,
indicate unwarranted sentencing departures or misapplication of the
guidelines. Empirical data on all factors that could influence sentencing
were not available, and so an analysis that could fully explain why
sentences varied was not possible.

USSC data were generally sufficient for our analyses of downward
departures and mandatory minimum sentences across circuits and for the
great majority of districts. USSC's sentencing data are based on
information from five documents usually produced for each case during the
sentencing process, but it principally relies on three of them-the
Judgment and Commitment Order (J&C), Statement of Reasons (SOR), and
Presentence Report (PSR)-to identify the length of sentences imposed,
departures, and the reasons for departures. For drug sentences nationally,
USSC receives 96 percent or more of these three key documents. For a small
percentage of drug cases nationally, information regarding departures or
the reasons for departures is lacking in USSC's database because documents
are missing, incomplete, or too difficult for USSC to interpret.
Opportunities for improving USSC data exist.

In order to improve USSC's data on federal drug sentences, we are making
recommendations that USSC and AOUSC collaborate to improve the collection
and recording of sentencing information through education programs,
further revision of the standard SOR, and notification of Chief Judges of
unclear, incomplete or difficult to interpret information received from
their districts. On October 10, 2003, we received official written
comments on a draft of our report from USSC and the Chair of the Judicial
Conference Committee on Criminal Law that generally agreed with our
recommendations (see app. V). However, the Chair of the Criminal Law
Committee noted that our report did not sufficiently distinguish downward
departures that are due to judicial discretion from those that are due to
prosecutorial discretion. As a result, according to the Committee, the
category "other downward departures" invites confusion, and some may
mistakenly attribute all such departures to judges. We state in our report
that all other downward departures are not attributable to judicial
discretion and that data were not available to fully distinguish sentences
that are attributable to judicial discretion from those due to
prosecutorial discretion. USSC, AOUSC, and Department of Justice also
provided technical comments that we incorporated as appropriate.

The United States is divided into 94 federal judicial districts, each
containing the federal trial courts, where criminal and civil cases are
tried. Congress placed each of the 94 districts in 1 of 12 regional
circuits, each containing a court of appeals, to which district court
decisions may be appealed. Figure 2 is a map of the United States showing
the geographical boundaries of the 94 district courts and the 12 regional
circuit courts of appeals (including 11 numbered circuits and the District
of Columbia Circuit.). There is also a Court of Appeals for the Federal
Circuit with nationwide jurisdiction for specific types of cases, such as
patent appeals. This court does not hear cases involving the federal
sentencing guidelines.

  Background

 Figure 2: Geographical Boundaries of the 12 Regional Circuit Courts of Appeals
                             and 94 District Courts

Sources: Administrative Office of the U.S. Courts and GAO (data);
copyright (c) Corel Corp. All rights reserved (map).

In 1984, to help ensure that similar crimes committed by similar criminals
were punished with similar sentences, Congress, under the Sentencing
Reform Act, established the U.S. Sentencing Commission (USSC) and directed
that it develop a comprehensive sentencing scheme for federal crimes.5
USSC established guideline ranges for the length of federal prison

528 U.S.C. 994.

sentences taking into account offender and offense characteristics to
establish appropriate sentence terms. The sentencing guidelines cover more
than 90 percent of all felony and Class A misdemeanor cases in the federal
courts. The sentencing guidelines do not apply to Class B or C
misdemeanors or infractions, offenses with a maximum prison exposure of 6
months or less.6

Applying USSC's guidance, federal district judges in the 94 federal
district courts determine the appropriate sentencing guideline range for
an offender based on various factors related to (1) the offense and (2)
the offender. The offense is assigned an offense level, which for drug
offenses is based on several factors such as the quantity and type of drug
involved and whether the offense involved violence. The offender is also
assigned a criminal history category based on the number of criminal
history points. Criminal history points reflect the severity of an
offender's prior criminal record. Taken in combination, the offense level
and criminal history category correlate with a sentencing guideline range,
which is expressed in months. In addition, for some drug offenses where a
mandatory minimum sentence applies, the applicable mandatory minimum
sentence supplants the lower end of the applicable guidelines sentencing
range. For example, as shown in figure 3, a convicted offender whose
offense of conviction is assigned an offense level of 25 and who has a
"criminal history category I" should be sentenced between a maximum of 71
months and a minimum of 57 months under the sentencing guidelines unless a
mandatory minimum greater than 57 months (e.g., 60 months) is required. A
sentence less than 60 months falls below the applicable mandatory minimum.

6See USSC 2002 Guidelines, 1B1.9 and ch. 1, pt. A, intro. comment. 5.

Figure 3: Illustration of a Sentencing Guideline for an Offense of
Conviction Carrying a Mandatory Minimum

The guidelines also permit sentences that fall above or below an
applicable guidelines range, often called upward or downward departures,
respectively, in certain circumstances. As illustrated in figure 3, a
sentence of more than 71 months would depart upward from the applicable
guideline range while a sentence of less than 57 months would depart
downward, falling below the lower end of the guideline range.

At the request of the prosecution, the judge may depart downward because
the defendant has provided substantial assistance to the government-what
USSC designates as substantial assistance departures. But the guidelines
also provide that a judge may depart downward if the court finds certain
mitigating circumstances exist that were not adequately taken into
consideration by USSC in formulating the guidelines that should

result in a sentence below the guideline range.7 To assist sentencing
courts, the guidelines list both encouraged departure factors (such as
coercion or duress, diminished capacity, or aberrant behavior of
non-violent offenders) and discouraged though permissible departure
factors (such as age, physical condition, family responsibilities, or
prior good works).8 Judges may also consider other, unmentioned factors
that were not adequately considered by the guidelines (such as
extraordinary rehabilitation after the offense but prior to sentencing).
USSC designates consideration of encouraged, discouraged, and unmentioned
factors as "other departures." Judges are required to explain the reasons
for departing from the guidelines. The recently enacted PROTECT Act of
2003 makes clear that the reasons must be specific, written, and provided
to USSC.9

USSC maintains a database that records a variety of data on the offenders
and offenses for which sentences are imposed. Judges must comply with
USSC's data collection needs by furnishing a written report of the
sentence, and the PROTECT Act makes clear that specific sentencing
documents must accompany that report.10 Included in these data is
information on whether the sentences imposed fell within or outside the
applicable sentencing guidelines range as determined by the sentencing
judge. Information on the incidence of sentencing outside of the guideline
ranges is used by USSC to identify areas where the sentencing guidelines
may need adjustment. Congress, under the PROTECT Act, directed USSC to
review the grounds of downward departures that are authorized by its
sentencing guidelines, policy statements, and official commentary, and to
promulgate amendments to ensure that the incidence of downward departures
is substantially reduced. The required amendments to the sentencing
guidelines are due October 27, 2003.11

As previously noted, various drug offenses carry a mandatory minimum.12
For such offenses, the mandatory minimum precludes judges from

7See 18 U.S.C. 3553(b); USSC 2002 Guidelines, ch. 5, pt. K, sec. 2.
8See USSC 2002 Guidelines., 5K2.0 and ch. 5, pt. H, intro. comment.
9See P.L. 108-21 (April 30, 2003), sec. 401(c), amending 18 U.S.C.
3553(c).
10See id., sec. 401(h), amending 28 U.S.C. 994(w).
11See id., sec. 401(m)(1), (2)(A).
12See 21 U.S.C. 841, 844, 846, 960, 963.

sentencing at a lower guideline range minimum or from granting a downward
departure that might otherwise be available, unless one of two statutory
provisions applies. First, a judge may impose a sentence below the
applicable mandatory minimum if the government (the federal prosecutor)
files a motion with the court for such sentencing relief because of the
defendant's "substantial assistance" in the investigation or prosecution
of another person.13 The discretion to make such a motion rests solely
with the prosecutor. Second, in the absence of a substantial assistance
motion, the "safety valve" provision affords relief from any otherwise
applicable mandatory minimum sentence for drug offenders who have minimal
criminal history (i.e., no more than 1 criminal history point); were not
violent, armed, or high-level participants; and provided the government
with truthful information regarding the offense. In these cases, the court
is directed by statute to impose a sentence pursuant to the sentencing
guidelines without regard to a mandatory minimum.14

As incorporated in USSC's sentencing guidelines, both the substantial
assistance and the safety valve provision may affect sentencing for
offenders whose offense of conviction does not carry a mandatory minimum
sentence-that is, whose sentences are solely governed by the application
of the sentencing guidelines. For such offenders, a substantial assistance
motion permits the judge to depart downward from the applicable guidelines
range.15 With respect to the safety valve, the sentencing guidelines
provide offenders who are convicted of certain drug offenses and who meet
the legislative safety valve requirements a 2-level decrease to their base
offense level, for example, from level 25 to level 23.16

13See 18 U.S.C. 3553(e).
14See id. 3553(f).
15See USSC 2002 Guidelines, sec. 5K1.1.
16See id., secs. 2D1.1(b)(6); 2D2.1(b)(1).

  Federal Drug Sentences Departed Downward More Often Due to Substantial
  Assistance than for Other Reasons

The majority of federal sentences fell within an applicable guideline
range, but when sentences departed downward, or fell below a guideline
range, they did so about as often due to substantial assistance as to
other reasons. Of the 162,090 federal sentences from fiscal years
1999-2001 for which complete sentencing information was available, most
were within the guideline ranges determined by the court (64 percent), and
about an equal proportion of sentences departed downward due to
substantial assistance (18 percent) as for other reasons (17 percent).17

Similar to federal sentences overall, of the 69,279 drug sentences for
which complete departure information was available, we found that most
sentences were within guideline ranges (56 percent). Unlike federal
sentences overall, from fiscal years 1999 to 2001, federal drug sentences
departed downward more frequently due to substantial assistance (28
percent) than other reasons (16 percent), as shown in table 1. Other
reasons that drug sentences departed downward included early disposition,
that is, fast track, programs initiated by prosecutors; plea agreements;
and judges' consideration of mitigating circumstances. See appendix IV for
more information on the frequency of reasons cited for other downward
departures.

17Since circuits and districts may vary widely in the number of offenders
and type of offenses for which they are convicted, analyses that identify
the percent of cases that fell below an applicable guideline range or a
mandatory minimum do not provide a fair basis for comparing sentencing
practices across circuits and districts. For example, a greater proportion
of offenders appearing before a district court who possessed and shared
information of a crime that assisted the government in the investigation
or prosecution of others or whose offenses were less serious may influence
the frequency of sentences that fell below an applicable guideline range.
Taking various offender and offense characteristics into account to
identify the likelihood of a sentence falling below a guideline range or a
mandatory minimum provides a more accurate picture of variation in
sentencing across circuits and districts. We provide this statistical
analysis later in the report. See appendix I for more information on the
limitations of percentages as a basis for comparison.

Table 1: Comparing Downward Departures Due to Substantial Assistance and
Other Reasons for All Federal Sentences with Federal Drug Sentences,
Fiscal Years 1999-2001

All criminal sentencesa Number Percent Number Percent

All drug sentences

                       Sentences imposed   175,245            72,283   
                 Sentences with complete                               
                  sentencing information  162,090b    100    69,279b      100 
             Sentences within guidelines   104,389     64     39,138   
                       Upward departures     980       c       143     

                              Downward departures

Substantial assistance 29,247 18 19,107

Other reasons 27,474 17 10,891

Source: GAO analysis of USSC Data.

aIncludes only offenders who were sentenced under the Federal Sentencing
Guidelines.

bThe USSC database lacked departure information on a total of 12,630 (7
percent) of all federal sentences; of these, the USSC database lacked
departure information on 3,004 (4 percent) federal drug sentences. An
additional 525 sentences were missing offense type information.

cLess than 1 percent.

Prosecutors' substantial assistance motions resulted in downward departing
sentences that were on average 49 percent of the average lowest sentence
drug offenders otherwise would have received under the guidelines. Other
downward departures resulted in sentences that were on average 57 percent
of the average lowest sentence drug offenders otherwise would have
received under the guidelines. See appendix I for more detailed
information on sentence reductions.

The percentage of drug sentences that departed downward due to
prosecutors' substantial assistance motions or for other reasons varied
across judicial circuits in fiscal years 1999-2001, as shown in figure 4.

Figure 4: Percent of Substantial Assistance and Other Downward Departures
for Drug Sentences, by Circuit, Fiscal Years 1999-2001

The percentages of drug sentences departing downward differed notably
across the 94 districts, even in some cases among districts within the
same circuit. Figure 5 shows the 94 judicial districts grouped according
to the percent of sentences imposed in districts that departed downward
due to substantial assistance and for other reasons. In 55 districts, more
than 30 percent of sentences departed downward due to substantial
assistance, while in only 5 districts more than 30 percent of the
sentences departed downward due to other reasons. However, these
percentage differences do not take into account differences in offender
and offense characteristics that may contribute to differences among
circuits and districts.

Figure 5: Number of Districts by Percent of Drug Sentences Departing
Downward Due to Substantial Assistance or for Other Reasons, Fiscal Years
1999-2001

Number of districts

71

0 to 10% 11 to 20% 21 to 30% 31 to 40% 41 to 50% 51 to 60% Over 60%

Substantial assistance departures Other downward departures

Source: GAO analysis of USSC data.

More Than Half of

  Drug Sentences Fell below an Otherwise Applicable Mandatory Minimum Sentence

Of 41,861 federal drug sentences included in our analysis that carried a
mandatory minimum term of imprisonment, more than half (52 percent) fell
below an otherwise applicable mandatory minimum sentence. These sentences
were split equally among those that fell below an otherwise applicable
mandatory minimum sentence due to substantial assistance (26 percent) and
those that fell below for other reasons, such as the safety valve (26
percent).18 (See fig. 6.)

18If a sentence fell below a mandatory minimum sentence and involved a
substantial assistance motion, we described the sentence as a "substantial
assistance sentence falling below." If a sentence otherwise fell below, we
described that sentence as falling below because of "other reasons." The
data indicated that most, but not all, of sentences falling below for
other reasons involved the safety valve provision, which is the only
mechanism by which a judge may disregard a mandatory minimum in the
absence of a substantial assistance motion. Reasons for 681 drug sentences
falling below a mandatory minimum were not clearly indicated on the
sentencing documents, according to USSC officials. Our analysis included
these sentences as falling below a mandatory minimum for other reasons.

Figure 6: Percent of Drug Sentences Meeting or below an Otherwise
Applicable Mandatory Minimum, Fiscal Years 1999-2001

Below mandatory minimum because of substantial assistance

Meeting mandatory minimum

Below mandatory minimum for other reasons such as the safety valve Source:
GAO analysis of USSC data.

Nearly all of the mandatory minimum drug sentences carried either a 5-year
(48 percent) mandatory minimum or a 10-year minimum (49 percent). On
average, prosecutors' substantial assistance motions reduced drug
offenders' 5-year mandatory minimum sentence by 33 months. Sentences
lowered for other reasons, such as the safety valve, that would otherwise
be subject to a 5-year mandatory minimum were reduced by an average of 26
months. On average, prosecutors' substantial assistance motions reduced
drug offenders' 10-year mandatory minimum sentences by 63 months, and
sentences lowered for other reasons that would otherwise be subject to a
10-year mandatory minimum were reduced by an average of 52 months. See
appendix I for more detailed information on sentence reductions.

Figure 7 provides a summary of the number and percent of federal drug
sentences that fell below a mandatory minimum or guideline range compared
with sentences that did not carry a mandatory minimum. Almost all of the
sentences (99 percent) that fell below a mandatory minimum due to
substantial assistance also departed downward from an applicable guideline
range, whereas only a quarter of sentences that fell below an otherwise
applicable mandatory minimum for other reasons also

departed downward from an applicable guideline range. These percentage
differences do not take into account offender and offense characteristics
that may contribute to differences among circuits and districts.

Figure 7: Summary of Mandatory Minimum Drug Sentences below an Otherwise
Applicable Mandatory Minimum and Departing Downward Compared with
Sentences that Did Not Carry a Mandatory Minimum, Fiscal Years 1999-2001

Note: Of the 72,283 federal drug sentences imposed during fiscal years
1999-2001, 3,004 lacked complete information on departure status and 609
lacked sentence length information needed to identify sentences that fell
below a mandatory minimum. These sentences were omitted from this
analysis.

In 7 of the 12 Circuits, more sentences fell below a mandatory minimum due
to substantial assistance motions than for other reasons, as figure 8
shows.

Figure 8: Percent of Drug Sentences Falling below an Otherwise Applicable
Mandatory Minimum due to Substantial Assistance or for Other Reasons,
Fiscal Years 1999-2001

Percentage

50

40

40

30

20

10

0

National DC First Second Third Fourth Fifth Sixth Seventh Eighth Ninth
Tenth Eleventh Circuit Average

                     Substantial assistance sentences below

                       Sentences below for other reasons

Source: GAO analysis of USSC data.

In addition to these differences among the circuits, across the 94
districts the percentage of sentences meeting or below an otherwise
applicable mandatory minimum substantially varied. Figure 9 shows the 94
judicial districts grouped according to the percent of sentences imposed
that fell below an otherwise applicable mandatory minimum due to
substantial assistance and for other reasons. In 40 districts, more than
30 percent of sentences fell below an otherwise applicable mandatory
minimum due to substantial assistance whereas in 16 districts more than 30
percent fell below for other reasons. Appendix II has more details on our
analysis of sentences for which the offense of conviction carried a
mandatory minimum sentence.

Figure 9: Number of Districts by Percent of Sentences Falling below an
Otherwise Applicable Mandatory Minimum Due to Substantial Assistance or
Other Reasons, Fiscal Years 1999-2001

Number of districts

                                       45

0 to 10% 11 to 20% 21 to 30% 31 to 40% 41 to 50% 51 to 60% Over 60%

Sentences falling below a mandatory minimum due to substantial assistance
Sentences falling below a mandatory minimum for other reasons

                       Source: GAO analysis of USSC data.

  Likelihood of Sentences Departing Downward or Falling below a Mandatory
  Minimum Varied by Judicial Circuits and Districts

The percentage differences among circuits and districts suggest that
variation existed in the way courts sentenced offenders; however, as
discussed earlier, these percentages do not take into account factors such
as the offender and offense characteristics that may affect sentencing
within a circuit or district. For example, in addition to the number of
deportable aliens potentially affecting the percent of other downward
departures, some circuits and districts, when compared with others, could
sentence a greater proportion of offenders who possessed and shared
information of the crime that assisted the government in the investigation
or prosecution of others. Therefore, a larger percent of offenders in
those circuits and districts could have received a decrease in their
sentence due to substantial assistance. Recognizing that judicial circuits
and districts differed in the types of offenders sentenced and the
offenses for which the offenders were sentenced, our analysis adjusted for
differences such as race, gender, offense, criminal history, and offense
severity (see appendix I for a complete list of variables included in our
analysis). Although these

are the major factors that could affect the likelihood of an offender
receiving a departure, they are not all-inclusive.

We used adjusted odds ratios to estimate how circuits and districts vary
in sentencing practices. An adjusted odds ratio in this case indicates
whether a departure is statistically less likely, as likely, or more
likely to occur in one circuit as in another. We can describe how much
more likely or less likely a substantial assistance departure was to occur
in one jurisdiction versus another; for instance, we can estimate that a
substantial assistance departure was 20 percent less likely to occur or
3.6 times more likely to occur in one circuit compared with another. Our
analysis focused on adjusted odds ratios since they provide us with our
best estimates of differences across circuits and districts after taking
into account the differences in drug cases handled across jurisdictions.

Using adjusted odds ratios to estimate the likelihood of an offender
receiving a substantial assistance departure, we can see that percentage
differences can be misleading. In comparing percentages of substantial
assistance sentences in the Eighth Circuit (37 percent) with other
circuits, it appears that in 5 circuits -D.C. (30 percent), the Second (32
percent), Fourth (33 percent), Seventh (31 percent), and Eleventh Circuits
(30 percent)- fewer offenders received substantial assistance departures.
But taking into account differences of offenders and offense
characteristics, adjusted odds ratios show that in D.C., the Fourth and
Eleventh circuits, offenders were actually as likely to receive a
substantial assistance motion as in the Eighth Circuit. Although the
Second Circuit's percentage of substantial assistance departures was lower
than the Eighth Circuit's, after adjusting for offender and offense
characteristics, offenders in the Second Circuit were 1.4 times more
likely to be granted a substantial assistance departure than offenders in
the Eighth Circuit. In another example, the First Circuit appears to grant
more other downward departures (10 percent) than the Eighth (8 percent),
but the adjusted odds ratios imply that offenders in the First Circuit are
actually 22 percent less likely to be granted an other downward departure
than similar offenders in the Eighth Circuit.

After adjusting for differences in drug offenses and offender
characteristics, the likelihood of an offender receiving a lower sentence
due to either a prosecutor's substantial assistance motion or for other
reasons varied substantially across the 12 U.S. circuits and the 94 U.S.
district courts. For example, drug offenders sentenced in the Third
Circuit from fiscal years 1999-2001 were over 3 times more likely to
receive a substantial assistance departure at a prosecutor's initiative
than drug

offenders sentenced in the First Circuit during that same time period. The
likelihood a drug offender would be granted an other downward departure
also varied substantially. Adjusting for differences in offender and
offense characteristics, drug offenders sentenced in the Ninth Circuit
from fiscal years 1999-2001 were over 18 times more likely to have
received an other downward departure than similar drug offenders sentenced
in the Fourth Circuit during that same time period.

The likelihood of courts to impose a sentence that fell below a mandatory
minimum due to substantial assistance or other reasons also varied
substantially across the 12 U.S. circuits and the 94 U.S. district courts,
even after taking into account offense and offender characteristics for
drug offenses. See appendix III for more details about the statistical
likelihood of drug sentences departing downward or falling below a
mandatory minimum and the variation in those likelihoods across all
circuits and districts.

Our analysis shows variation-in some cases substantial differences- among
circuits and districts in the likelihood that offenders convicted of drug
offenses would receive substantial assistance or other downward departures
in fiscal years 1999-2001. However, these differences, in and of
themselves, may not indicate unwarranted sentencing departures or
misapplication of the sentencing guidelines. Empirical data on all factors
that could influence sentencing were not available, and so an analysis
that could fully explain why sentences varied was not possible.

  Opportunities Exist to Improve Sentencing Data Collection and Reporting

USSC data were generally sufficient for our analyses of downward
departures and mandatory minimum sentences across circuits and for most
districts. USSC's sentencing data are based on information from five
documents usually produced for each case during the sentencing process.
USSC requires that district courts send them these documents for each
sentence imposed but principally relies on three of them-the Judgment and
Commitment Order (J&C), Statement of Reasons (SOR), and Presentence Report
(PSR)-to identify the length of sentences imposed, departures, and the
reasons for departures. Nationally for drug cases, USSC received 96
percent or more of each of these documents from the district courts. The
percentage of documents missing varied by circuit and districts within
circuits. For example, among the circuits the percentage of missing SORs
ranged from less than 1 percent to more than 7 percent. Among districts
within the Ninth Circuit, the percentage of missing SORs ranged from less
than 1 percent to more than 58 percent. Although USSC received most of the
requested documents, some were missing key

information or contained unclear information that was difficult to
interpret. For instance, among the 12 circuits, departure data were
missing for 1 percent to 7 percent of all drug sentences imposed in fiscal
years 1999-2001. See appendix IV for more detailed information.

In districts where the missing documents or information are concentrated
analysis of departures could be affected. Missing or unclear data also
limited our ability to determine when the safety valve was used as the
basis for sentencing below a mandatory minimum. For example, in our
preliminary analysis, we found that of the 11,256 federal drug sentences
for which the offense of conviction carried a mandatory minimum and fell
below that minimum, about 1,600 (14 percent) were coded by USSC as falling
below the applicable mandatory minimum but not involving either the safety
valve or substantial assistance. We discussed this issue with USSC. After
reviewing the underlying documents used for coding these 1,600 sentences,
USSC determined that over 900 sentences were miscoded. These miscoded
sentences were recoded in a variety of ways, including some coded as
involving the safety valve, some coded as involving substantial
assistance, some coded as having a changed drug quantity that affected the
applicable mandatory minimum, and some coded as missing safety valve
information. USSC did not recode 681 sentences; these sentences remained
coded as falling below a mandatory minimum but involving neither the
safety valve nor substantial assistance. In addition, safety valve
information was determined to be missing from 770 sentences for which the
offense of conviction carried and fell below a mandatory minimum.

AOUSC and USSC officials offered several explanations for missing
documents and information or unclear information on documents that was
difficult to interpret and code consistently. AOUSC officials noted that
judges may not submit documents due to security concerns in cases where
the record has been sealed or the offender placed in the witness
protection program. Processing a high volume of drug cases could
potentially affect document submission, they also noted. Of the four
circuits with the highest volume of drug cases, two-the Fourth and the
Ninth-also had the highest percentage of missing SORs, 7.4 percent and 6.6
percent, respectively. Although AOUSC developed a standard SOR form for
judges to use, USSC officials said that judges submit information to USSC
using a variety of forms and formats. In USSC's view, this may contribute
to missing information on documents (e.g., forms that do not prompt for an
applicable mandatory minimum) or unclear information on the forms that is
difficult to interpret.

In addition to these explanations, USSC and AOUSC officials said no
information had been provided to judges and other court officials on how
sentencing documents are used by USSC to create its database or how to
clearly and completely prepare forms such as the SOR to meet USSC's data
collection needs. USSC relies almost exclusively on the SOR to determine
whether the sentence imposed departed from the guidelines range, met a
mandatory minimum, or involved substantial assistance. Thus, missing,
incomplete, or difficult to interpret information on that form can affect
the completeness and accuracy of the data in USSC's database.

USSC has taken steps to reduce the number of missing documents and
information, but opportunities for improvements exist. For instance, USSC
sends an annual letter to the courts identifying those cases in which
there appear to be missing documents. However, USSC officials said they do
not inform courts of documents that, while received by USSC, contained
missing or unclear information. In addition, USSC collaborates with the
AOUSC and the Federal Judicial Center, the judiciary's research and
education body, to educate judges and court officials on how to apply the
sentencing guidelines, but they do not offer programs or workshops on how
to complete forms such as the SOR and other documents used by USSC.

Although the AOUSC has also taken steps to improve the quality of
sentencing data captured on the SOR, opportunities for improvements
remain. At its September 2003 meeting, the Judicial Conference of the
United States, the federal judiciary's principal policymaking body,
approved a new standard SOR form. The Conference designated the new form
as the mechanism by which courts comply with the requirements of the
PROTECT Act to report reasons for sentences to USSC. The form was revised
in part because AOUSC officials and the Chair of the Judicial Conference's
Committee on Criminal Law stated that the previous SOR provided an
imprecise measure of judicial discretion. It did not collect information
on other downward departures that are initiated by the prosecution. The
form was revised in consultation with USSC to better meet its data
collection needs. However, a USSC official said that the new SOR does not
specifically prompt for information on the application of the safety valve
or whether the offense of conviction carried a mandatory minimum. In
addition, judges will not be required to use this form although USSC
believes that the most effective step to improving the completeness of the
data the district courts report is for all courts to use a single,
standard SOR. AOUSC officials said that while the Judicial Conference has
endorsed the new form, they do not believe that the

Conference has the authority to require judges to use the new SOR.
However, AOUSC officials stated that with additional education they
believe judges will see the benefits of the new form and routinely use it.

  Conclusions

Recommendations

The judiciary provided and USSC collected and interpreted sentencing
information for the vast majority of the 72,283 drug sentences imposed
during fiscal years 1999-2001. The small percentage of documents and
sentencing information for drug cases that were lacking in USSC's database
did not affect the validity of our analyses at the national or circuit
level or for the vast majority of districts. However, the missing data
could limit analyses of sentencing practices in the few districts where
missing data are most prevalent. Reducing missing, incomplete, or
difficult to interpret information would improve USSC's data on departures
and reasons sentences fell below an applicable mandatory minimum. Unless
the judiciary's standard SOR is revised, judges are made aware of how to
effectively complete the SOR, and data submitted more consistently to USSC
by the courts, these data problems are likely to persist. More could be
done to help reduce the number of documents that are missing, incomplete,
or too difficult for USSC to interpret. Evaluating how USSC interprets
sentencing data was beyond the scope of this work. However, we believe
that without changes to the way USSC reports other downward departures,
that is, distinguishing, rather than combining other downward departures
initiated by the government with those initiated by judges, the benefits
of improved data collection of sentencing practices may not be fully
realized.

As USSC and the AOUSC work together to collect and record information on
federal sentences and provide additional education and information to
judges, we recommend that both USSC and AOUSC continue to collaborate to

o  	develop educational programs and information for judges and other
officers of the court to encourage the use of AOUSC's standard SOR and
more effective ways to complete the SOR and

o  revise the standard SOR to better meet the data collection needs of
USSC.

We also recommend, resources permitting, that USSC, in addition to
notifying courts of missing sentencing documents, notify the Chief Judge
of each district of documents for drug cases that were received but

  Agency Comments

contained information that was unclear, incomplete, or difficult to
interpret.

We requested comments on a draft of this report from USSC, the AOUSC, the
Judicial Conference Committee on Criminal Law, and the Department of
Justice (DOJ). We received written comments on October 10, 2003, from USSC
and the Judicial Conference Committee on Criminal Law. Both generally
agreed with our report and our recommendations, although the Criminal Law
Committee was concerned that the variation among districts we found in the
likelihood of other downward departures could mistakenly be attributed to
judicial discretion. Their official comments are reproduced in Appendix V.
We received oral technical comments from USSC and written technical
comments from DOJ and the AOUSC that we incorporated where appropriate.
DOJ, the Committee on Criminal Law, and USSC all noted that "fast-track"
sentences-prosecutor-initiated programs to encourage early case
disposition and reduce the burden on the courts-could account for some of
the variation we found among districts in other downward departures. We
have added new tables in appendix IV that show the reasons reported to
USSC for other downward departures and the frequency with which each
reason was cited.

USSC 	USSC generally agreed with our report and our recommendations. USSC
stated that it is already working to develop more detailed sentencing
documentation, submission procedures, and educational outreach to courts
and court personnel. USSC noted that while our recommendations are helpful
and consistent with their own thinking, implementation of such measures
may exceed their current resources given the increasing volume of
sentences to be processed and more detailed information for each sentence
required by the PROTECT Act.

    The Judicial Conference Committee on Criminal Law

The Judicial Conference Committee on Criminal Law generally agreed with
our report and our recommendations. The Committee noted that it has taken
significant steps to help USSC improve its data collection by revising the
standard Statement of Reasons form and endorsing the standard form as the
way to comply with PROTECT Act requirements.

The Committee commented that our report did not address the extent to
which judges themselves, absent a prosecutor's request, have imposed
sentences that fall below the sentencing guideline range. Further, the
Committee noted that our report did not sufficiently distinguish downward

departures that are due to judicial discretion from those that are due to
prosecutorial discretion. As a result, according to the Committee, the
category "other downward departures" invites confusion, and some may
mistakenly attribute all such departures to judges. We state in our report
on page 11 that other downward departures are attributable to prosecutors
as well as judges. Additionally, data are not recorded, coded, or reported
in ways that clearly delineate other downward departures due to judicial
discretion from those due to prosecutorial discretion.

    Impact of Fast Track Sentences on Analysis of Other Downward Departures

In addition, DOJ, USSC, and the Committee on Criminal Law stated that our
report did not sufficiently discuss the impact of early disposition or
"fast track" programs on rates of other downward departures in those
circuits and districts where such programs were in place. Fast-track
programs in the southwest border districts provide lower sentences
initiated by prosecutors for low-level drug trafficking offenses. DOJ
noted that these programs were developed in response to a dramatic rise in
immigration cases handled by federal prosecutors in districts along the
southwestern border and were designed to enhance public safety and
minimize the burden on the court system by processing these cases as
quickly as possible. All of the agencies took the position that some
circuits and districts departed downward more than others due to the
greater prevalence in some circuits and districts of cases involving fast
track programs.

It may ultimately be useful to distinguish fast track departures from
other downward departures, in the same way that we have distinguished
substantial assistance departures from other downward departures. However,
as currently coded in USSC database, fast track cases can be identified
only when a judge explicitly lists fast track as a reason for a downward
departure. Sentences citing fast track as a reason for departing downward
occurred almost entirely in one district--the Southern District of
California in which 2,171 sentences (58 percent of other downward
departures imposed in this district) were recorded by USSC as departing
due to the government's fast track program. In all of the remaining 93
districts combined, only 9 sentences were recorded as departing downward
due to fast track programs. Moreover, when we eliminate from our analyses
those other downward departures that list "fast track" as a reason for
departing, we obtain very similar results to those published in our
report; that is, a greater likelihood of other downward departures
occurring in the Southern District of California than in most other
districts. We do not include these results in detail in our report because
of our concern that fast track cases are not always reported by judges as

such or coded by USSC in their database. If fast track departures are to
be distinguished from other departures, then changes will need to be made
in how such cases are currently reported to USSC. USSC is completing a
report on departures, pursuant to a Congressional directive in the PROTECT
Act, and will address the impact of fast track programs on departures in
greater detail.

As agreed with your offices, unless you publicly announce its contents
earlier, we plan no further distribution of this report until 2 days from
the
date of this report. At that time, we will send copies of this report to
the
AOUSC and Judicial Conference; DOJ; USSC; and the Federal Judicial
Center. We will also make copies available to others upon request. In
addition, the report will be available at no charge on GAO's Web site at
http://www.gao.gov.

If you or your staffs have any questions about this report, please contact
David Alexander at (202) 512-8777 or at [email protected] or me at
(202) 512-8777 or at [email protected]. Major contributors to this report
are listed in appendix V.

William O. Jenkins, Jr.
Director, Homeland Security and Justice Issues

Appendix I: Objectives, Scope, and Methodology

Objectives 	Our objectives were to review and categorize all sentences
imposed under the federal sentencing guidelines by federal district judges
in fiscal years 1999 through 2001. Specifically, our objectives were to

o  	identify the percentage of federal sentences, and specifically, those
for drug-related offenses, departing downward from the applicable
guidelines range as determined by the court due to substantial assistance
motions or other reasons;

o  	identify the percentage of federal drug sentences that fell below an
applicable mandatory minimum due to substantial assistance motions or
other reasons;

o  	compare the likelihood across judicial circuits and districts that
offenders received downward departing sentences or sentences below a
mandatory minimum;

o  	identify limitations, if any, of the U.S. Sentencing Commission's
(USSC) sentencing data for drug offenses.

Scope 	To meet these objectives, we obtained USSC sentencing data for
fiscal years 1999 through 2001. During fiscal years 1999 through 2001,
federal judges imposed sentences on 175,245 criminal offenders. Of this
total, 11,584 sentences (6.6 percent) lacked information on whether there
was a departure from the guidelines range, and for 1,046 sentences (0.6
percent) the guidelines were not applicable. An additional 167 sentences
(0.1 percent) lacked information on the type of offender sentenced (drug
versus non-drug), and 358 sentences (0.2 percent) lacked information on
both departure status and type of offender. For the remaining 162,090
sentences (92.5 percent), table 2 shows the numbers and percents of
departure and non-departure sentences for drug and non-drug sentences.

                 Appendix I: Objectives, Scope, and Methodology

Table 2: Numbers and Percents of Drug Cases Involving Different Types of
Offenses or Offenders, and Percentages of Each Type Receiving Sentences
Departing Downward from the Guidelines

                                Departure status

                                             Substantial    Other    
                        Upward            No  assistance   downward  
    Type of sentence   departure  departure   departure   departure     Total 
        Non-Drug              837   65,251      10,140      16,583     92,811 
                             0.9%   70.3%       10.9%       17.9%    
          Drug                143   39,138      19,107      10,891     69,279 
                             0.2%   56.5%       27.6%       15.7%    
         Total                980  104,389      29,247      27,474    162,090 
                             0.6%   64.4%       18.0%       16.9%    

Source: GAO analysis of USSC data.

Of the total of 175,245 offenders sentenced, 72,283 offenders (41 percent)
were convicted of drug offenses, and of those, 69,279 offenders had
complete information on departure status. Roughly 11 percent of the
non-drug offenders and 28 percent of the drug offenders received sentences
below the guidelines range due to substantial assistance, while 18 percent
of the non-drug offenders and 16 percent of the drug offenders received
sentences that departed downward for other reasons. Of the 69,279 drug
sentences that had valid information for USSC's departure variable, 42,145
sentences (61 percent) carried a mandatory minimum. Our analyses of
mandatory minimum sentences excluded 284 of these sentences (0.6 percent
of all mandatory minimum sentences) that lacked the valid sentence length
information necessary to determine whether the sentence fell below the
minimum.

Methodology 	We had extensive discussions with knowledgeable USSC staff
about the definitions and use of the data elements in our analysis. USSC
takes many steps to ensure the reliability and completeness of the data it
receives from districts. We did not independently validate the data in
USSC's database; however, we did assess the quality of USSC data in our
analysis by testing and crosschecking selected data elements for internal
consistency. We discussed any anomalies we found from these tests with
knowledgeable USSC staff. On the basis of our tests and discussions with
USSC officials, we determined that the data were sufficiently accurate for
our reporting objectives.

Appendix I: Objectives, Scope, and Methodology

We defined sentences that fell above or below an applicable guideline
range in accordance with USSC's definition of "departures"-sentences
imposed that fall outside the sentencing guidelines range established by
the court. In USSC's database sentences are coded into five categories: no
departure, upward departure, downward departure, substantial assistance,
and inapplicable. We distinguished sentences coded as "substantial
assistance" as those that fell below sentencing guidelines due to
"substantial assistance" or prosecutorial discretion. For sentences that
fell below the guideline ranges for reasons other than substantial
assistance, generally attributed to judicial discretion, we used those
sentences coded as "downward departures." See appendix IV for a
description of other reasons, which include the government's early
disposition or "fast track" programs, cited by judges for downward
departures. Sentences for which departure information was not available or
coded "inapplicable" were deleted from our analysis.

We defined sentences that fell below an applicable mandatory minimum using
USSC's recorded information for sentence length. For convictions where a
mandatory minimum was recorded, those sentences with recorded lengths that
fell below the length of time stipulated by the mandatory minimum were
defined as "falling below the mandatory minimum." We identified 41,861
drug sentences that carried a mandatory minimum and had valid sentence
length data. Of those sentences we designated as falling below an
applicable mandatory minimum, we identified those sentences that involved
a substantial assistance motion and those that fell below the mandatory
minimum for other reasons, such as the safety valve. If a sentence fell
below and involved a substantial assistance motion, we interpreted that
sentence as a "substantial assistance sentence" that fell below a
mandatory minimum due to prosecutorial discretion. If a sentence otherwise
fell below, we interpreted it as falling below for "other reasons." Most
of these sentences (9,384 or 87 percent of them) involved offenders that
qualified for the safety valve provision allowing judges to grant
sentences below the mandatory minimum. The data do not indicate that all
of these sentences involve the safety valve, which may be the result of
coding errors or insufficient available data.

We also reviewed the types of documents USSC staff used to identify
departures, the reasons for those departures, and the potential effect of
missing or unclear documentation on the interpretation of the departure
data in USSC's database by district. We also interviewed officials at USSC
and the Administrative Office of the U.S. Courts (AOUSC), and the Chair of
the Judicial Conference Committee on Criminal Law.

Appendix I: Objectives, Scope, and Methodology

  Analyzing Numbers and Percents

We analyzed sentencing data using both descriptive statistics and
multivariate analytic methods. For fiscal years 1999-2001, we used USSC's
data to identify for each circuit and district the total number and
percent of sentences that fell above or within an applicable guideline
range and below a guideline range for substantial assistance or for other
reasons. We also identified all sentences that fell below an applicable
mandatory minimum due to substantial assistance or for other reasons for
each circuit and district. We provide these numbers and percentages in
appendix II.

The simple differences in the percentage of drug sentences that fall below
the guidelines range or below the mandatory minimum may not, without some
adjustment, provide an appropriate basis for making comparisons across
circuits and districts. Characteristics of the offenses and offenders
sentenced can vary from one circuit or district to the next, and these
differences may affect the number or percent of sentences that fall below
applicable guideline ranges or a mandatory minimum. Judges in some
circuits or districts, for example, may sentence a greater proportion of
offenders who possessed and shared information of the crime that assisted
the government in the investigation or prosecution of others or whose
offenses were less serious. Differences in these characteristics might
produce differences in sentences that have little to do with the exercise
of discretion. Therefore, the unadjusted differences in the percent of
sentences below the guidelines range or mandatory minimum might result
from judges sentencing different offenders, rather than from judges
sentencing offenders differently. Table 3 shows the offense and offender
characteristics we considered in the multivariate analyses we conducted to
adjust for such differences and re-estimate differences across circuits
and districts after taking them into account. It also provides the numbers
and percentages of all drug offenders or drug offenses that possessed each
of these characteristics.

Appendix I: Objectives, Scope, and Methodology

Table 3: Numbers and Percents of Drug Cases Involving Different Types of
Offenses or Offenders, and Percentages of Each Type Receiving Sentences
Departing Downward from the Guidelines

Offense/offender Number of all Percent of all Percent substantial Percent other
 characteristics drug cases drug cases assistance departure downward departure
                              Type of drug offense

                      Drug trafficking 66,616 96% 28% 16%

                Drug: Communication facilities 1,212 2% 20% 12%

                     Drug: Simple possession 1,451 2% 4% 2%

Drug type

            Cocaine              15,110         22%        31%            10% 
             Crack               14,562         22%        33%             8% 
            Heroine              5,101          8%         23%            14% 
       Marijuana/Hashish         21,699         32%        18%            28% 
             Other               11,208         17%        36%            10% 

                   Severity of offense score Weapon involved

        Less than 17           14,700         21%         13%             29% 
          17 to 22             13,822         20%         22%             19% 
          23 to 28             18,453         27%         30%             10% 
       Greater than 28         22,206         32%         38%              9% 

            No enhancement             58,779       90%       27%         16% 
         Enhancement applied           6,630        10%       35%          8% 
          Mandatory minimum                                          
         No minimum involved           27,133       39%       17%         24% 
      Mandatory minimum involved       42,145       61%       34%         10% 
             Guilty plea                                             
        No guilty plea entered         3,352        5%        2%          10% 
         Guilty plea entered           65,824       95%       29%         16% 
        Safety valve provision                                       
       No safety valve applied         45,788       73%       28%         16% 
         Safety valve applied          16,832       27%       30%         16% 
              Education                                              
        Less than high school          34,197       51%       23%         18% 
         High school graduate          21,502       32%       31%         12% 
             Some college              9,819        15%       34%         13% 
           College graduate            1,708        3%        31%         15% 

Appendix I: Objectives, Scope, and Methodology

       Offense/offender Number of all Percent of all Percent substantial

Percent other downward departurecharacteristics

                   drug cases drug cases assistance departure

Race Sex

         White                  17,405     25%          38%               13% 
         Black                  20,355     29%          32%                8% 
        Hispanic                29,811     43%          18%               23% 
         Other                   1,708      2%          32%               13% 

                            Male 59,801 86% 27% 15%

                            Female 9,467 14% 33% 19%

Citizenship

                        U.S. citizen 46,928 68% 33% 12%

                         Non-Citizen 22,351 32% 17% 24%

                           Criminal history category

                 38,798        56%               24%                      18% 
                  8,343        12%               31%                      12% 
                 10,010        14%               32%                      12% 
                  4,550        7%                32%                      13% 
                  2,185        3%                33%                      13% 

5,277 8% 35% 15%

Source: GAO analysis of USSC data.

Note: Severity Score is determined by the court. It ranges from 1 to 60
for all crimes, but the highest score for a drug crime is 46.

Our primary focus in this report involved understanding how departure
sentences vary across circuits and districts. Some of this variability
across circuits and districts in the percentages of substantial assistance
departures and other downward departures in the sentencing of drug
offenders results from differences in the characteristics of offenses and
offenders sentenced across circuits and districts. Moreover, the
prosecution has sole authority to initiate a downward departure for
substantial assistance, and all offenders are potentially eligible for
such consideration. If an offender receives a substantial assistance
departure, USSC codes the case as a substantial assistance departure and
does not reflect any other downward departures that the judge may have
granted in that particular case. Because of this coding convention, the
percentage of other downward departures is partly a function of the
percentage of downward departures for substantial assistance, which result
from prosecutorial motions. To understand how these percentages are
derived, it is useful to consider the following two hypothetical districts
and the numbers of sentences of each type in each district shown in table
4.

                 Appendix I: Objectives, Scope, and Methodology

Table 4: Hypothetical Example Showing Effect of Two Different Methods of
Computing Percentages for "Other Downward Departures"

Sentences above Substantial Other and within a assistance downward
District guideline range departures departures Total

                          Number of sentences imposed

A 40 20 40

B 20 50 30

Other downward departures as percent of total sentences imposed

                  A 40 (base 100) 20 (base 100) 40 (base 100)

                  B 20 (base 100) 50 (base 100) 30 (base 100)

Other downward departures as percent of total sentences less substantial
assistance departures

A 40 (base 100) 20 (base 100) 50 (base 80)

B 20 (base 100) 50 (base 100) 60 (base 50)

Source: GAO example.

The prosecution makes its selection for substantial assistance motions
after screening the total universe of 100 offenders sentenced in each
district. In our hypothetical example, the prosecutor offered and the
court accepted substantial assistance motions for 20 percent of 100
offenders in district A and 50 percent of 100 offenders in district B-or
30 percent less in district A. Because USSC's coding convention
distinguishes substantial assistance cases from other downward departures,
the universe of offenders who could be coded as receiving other downward
departures is equal to the number of offenders who did not receive
substantial assistance departures. In district A this would be 80
offenders (100 minus 20 substantial assistance departures) and in district
B it would be 50 (100 minus 50 substantial assistance departures). Using
this universe of offenders for our calculation, we would conclude that
district A involves 10 percent fewer other departures than district B
(40/80=50 percent versus 30/50=60 percent). While we offer percentages in
some of the following tables that, following standard procedures, are
based on the total number of offenders, we also use odds and odds ratios
to describe the likelihoods of sentences departing. These odds and odds
ratios have the advantage of utilizing the appropriate universe of
offenders in making comparisons across circuits and districts.

Appendix I: Objectives, Scope, and Methodology

  Estimating Departure Likelihoods

In order to make a fair comparison of sentencing patterns across circuits
and districts, we used logistic regression analysis to estimate the
likelihood that sentences would fall below an applicable guideline range
or a mandatory minimum, before and after adjusting for differences in
offender and offense characteristics across circuits and districts. Our
adjusted estimates of the differences in likelihoods across circuits and
districts involved controlling for the following offender and offense
characteristics:

o  	Offender: gender, race, education, citizenship, and criminal history
category score.

o  	Offense: type of drug involved; type and severity of offense; whether
the offense was eligible for mandatory minimum sentence; whether a gun was
involved in commission of the offense; whether the defendant was convicted
after trial or entered a guilty plea; and whether the safety valve was
applied.

Because they are somewhat more amenable to adjustment for offense and
offender characteristics, we use odds and odds ratios, rather than
percentages and percentage differences, to estimate the likelihood of
sentences falling below a guideline range and the variability in those
likelihoods across circuits and districts. We first calculated the odds on
substantial assistance falling below a guideline range among all
sentences, and then calculated the odds on other downward departures for
those sentences that did not involve departures for substantial
assistance. In both cases odds were compared across circuits and districts
by taking their ratios. In our simple two district example above, the odds
on substantial assistance departures in districts A and B would be
20/80=0.25 and 50/50=1.0, respectively, and the odds ratio of 1.0/0.25=4.0
indicates that the likelihood of substantial assistance departures were 4
times as great in district B as in district A. The odds on other downward
departures, excluding the substantial assistance departures, would be
40/40 =1.0 in district A and 30/20=1.5 in district B, and the ratio of
1.5/1.0=1.5 indicates that downward departures are 1.5 times as likely in
district B as in district A.

We conducted four regression analyses. First, we conducted two regression
analyses that estimated the likelihoods that drug sentences fell below an
applicable guideline range due to either prosecutors' substantial
assistance motions or for other reasons before and after controlling for
offense and offender characteristics. Second, we conducted two regression
analyses that estimated the likelihoods that drug sentences that

Appendix I: Objectives, Scope, and Methodology

carried a mandatory minimum fell below an otherwise applicable mandatory
minimum due to either substantial assistance or other reasons before and
after controlling for offense or offender characteristics.

Limitations

Our work was limited to drug sentences imposed during fiscal years
1999-2001, and excluded drug cases in that year that lacked information on
whether the case departed from the guidelines (4 percent of all drug
cases). We also excluded cases for which there was insufficient
information to indicate whether the sentence involved was below a
mandatory minimum (0.7 percent of all mandatory minimum drug cases), and
were unable to identify whether 7 percent of the cases that fell below the
mandatory minimum, and did not involve substantial assistance, involved
the use of the safety valve provision. Further, our ability to control for
differences in the likelihood of sentences departing from the guidelines,
or falling below a mandatory minimum, was also restricted to a reasonably
small number of characteristics for which we had data and was affected by
the amount of missing data on those characteristics. Empirical data on all
factors that could influence sentencing were not available, and so an
analysis that could fully explain why sentences varied was not possible.

Our analyses were also limited to determining whether sentences fell below
a guideline range or a mandatory minimum, and we did not investigate
whether there were differences across circuits or districts in how far
below a guideline range minimum or a mandatory minimum the sentences fell.
Nationwide, the average (mean) minimum sentence length, under the
guidelines, for drug sentences that departed downward for substantial
assistance reasons was 108 months (or about 9 years). Those sentences were
reduced as a result of the substantial assistance motion, on average, by
53 months, and the resulting sentence was, on average, 49 percent of the
average lowest sentence drug offenders otherwise would have received under
the guidelines.1 The average minimum sentence length under the guidelines
for drug sentences that departed downward for reasons other than
substantial assistance was 60 months (or 5 years). Those sentences were
reduced, on average, by 22 months, and the resulting sentence was, on
average, 57 percent of the average lowest

1The average percentage sentence reduction was calculated by taking the
percentage reductions for all sentences and averaging them. This average
does not necessarily equal the ratio of the average reduced sentence (as a
result of a departure) to the average expected sentence (the guidelines
minimum or mandatory minimum sentence length).

Appendix I: Objectives, Scope, and Methodology

sentence drug offenders otherwise would have received under the
guidelines.

Nearly all of the mandatory minimum drug sentences were for 5 years (48
percent) or 10 years (49 percent). The 5-year mandatory minimum sentences
that were reduced for substantial assistance were reduced by an average of
33 months, resulting in an average sentence that was 45 percent of the
mandatory minimum. The sentences lowered for other reasons, (primarily the
safety valve), that would otherwise be subject to a 5-year mandatory
minimum were reduced by an average of 26 months, resulting in an average
sentence that was 57 percent of the mandatory minimum. The 10-year
mandatory minimum sentences that were reduced for substantial assistance
were reduced by an average of 63 months, resulting in an average sentence
that was 47 percent of the mandatory minimum. The sentences lowered for
other reasons, (primarily the safety valve), that would otherwise be
subject to a 10-year mandatory minimum were reduced for other reasons by
an average of 52 months, resulting in an average sentence that was 57
percent of the mandatory minimum.

We also did not attempt to determine, for those sentences that fell within
the guideline range, across circuits and districts whether sentences fell
more frequently at the lower or higher end of the guideline range.
However, overall, 72 percent of drug sentences that were within the
guidelines range and did not depart were at the bottom of the range.

Appendix II: Percentage of Sentences Falling below an Applicable Guideline Range
or An Otherwise Applicable Mandatory Minimum

This appendix provides information on the percent of federal drug
sentences that fall below an applicable guideline range or an otherwise
applicable mandatory minimum. We show in tables 5 and 6 the variability
across circuits and districts in the percentages of drug sentences that
were (1) above the guidelines range, (2) within the guidelines range, (3)
below the range due to substantial assistance, and (4) below the range for
other reasons. We then show in tables 7 and 8 the variability across
circuits and districts in the percentages of mandatory minimum drug cases
that resulted in sentences (1) at or above a mandatory minimum sentence,
(2) below a mandatory minimum due to prosecutorial motions for substantial
assistance, and (3) below the mandatory minimum for other reasons.

Table 5 shows that the percentage of upward departures from the sentencing
guidelines for drug cases in fiscal years 1999-2001 was similar and
exceedingly small across all 12 circuits. However, the percentages of
within range sentences and downward departures varied substantially across
circuits. The percentage of all drug sentences that were within the
guidelines range varied from 34 percent in the Ninth Circuit to 69 percent
in the First and Fifth Circuits. The percentage of drug sentences that
involved downward departures for substantial assistance varied from 18
percent in the Ninth Circuit to 45 percent in the Third Circuit, and the
percentage that resulted in downward departures for other reasons varied
from 4 percent in the Fourth Circuit to 47 percent in the Ninth Circuit.
The Ninth Circuit was the only circuit in which the percentage of
departures for other reasons exceeded 20 percent and the only circuit in
which departures for other reasons were more common than departures for
substantial assistance.

Appendix II: Percentage of Sentences Falling below an Applicable Guideline Range
                  or An Otherwise Applicable Mandatory Minimum

Table 5: Percentages of Drug Cases between 1999 and 2001 Resulting in
Upward Departures, Within Range Sentences, and Downward Departures, by
Circuit

                           Sentenced     Substantial      Other 
                 Upward           within  assistance  downward   Total number 
    Circuit  departure  guideline range   departure   departure of drug cases 
     D.C.          0.3%              53%          30%       16% 
     First         0.3%              69%          21%       10%         2,166 
    Second         0.3%              48%          32%       20%         5,166 
     Third         0.2%              48%          45%        7%         2,750 
    Fourth         0.2%              63%          33%        4%         6,302 
     Fifth         0.1%              69%          20%       11%        15,102 
     Sixth         0.1%              53%          42%        5%         5,067 
    Seventh        0.3%              64%          31%        5%         3,017 
    Eighth         0.2%              55%          37%        8%         5,255 
     Ninth         0.3%              34%          18%       47%        12,116 
     Tenth         0.2%              63%          22%       15%         3,500 
Eleventh        0.3%              64%          30%        5%         8,457 

Source: GAO analysis of USSC data.

Note: 3,004 cases lacking information on departures were excluded.

Table 6 reveals that the percentages of sentences within the guidelines
range and the percentages of sentences departing downward from them
notably differed across the 94 districts, even in some cases among
districts within the same circuit.

Appendix II: Percentage of Sentences Falling below an Applicable Guideline
Range or An Otherwise Applicable Mandatory Minimum

Table 6: Percentages of Drug Cases between 1999 and 2001 Resulting in
Upward Departures, Within Range Sentences, and Downward Departures, by
District

Sentenced Substantial Other Upward within assistance downward Total number
District departure guideline range departure departure of drug cases

D.C. Circuit

District of Columbia 0.3% 53% 30% 16%

First Circuit Second Circuit Third Circuit

          Maine            0.0%        44%       52%       3%     
      Massachusetts        0.6%        51%       30%       19%    
      New Hampshire        0.5%        49%       43%       8%     
       Puerto Rico         0.1%        87%       7%        5%           1,006 
       Rhode Island        0.0%        80%       9%        11%    

         Connecticut          0.8%       44%       24%      31%    
      New York Eastern        0.3%       45%       24%      32%         1,961 
      New York Northern       0.2%       24%       68%       8%    
      New York Southern       0.2%       64%       24%      12%         1,642 
      New York Western        0.6%       42%       52%       5%    
           Vermont            0.0%       43%       37%      21%    

            Delaware             0.0%      52%      41%       7%           71 
           New Jersey            0.4%      60%      32%       8%          847 
      Pennsylvania Eastern       0.2%      36%      57%       6%        1,028 
       Pennsylvania Middle       0.0%      38%      56%       6%          392 
      Pennsylvania Western       0.0%      55%      37%       9%          352 
         Virgin Islands          0.0%      75%      17%       8%           60 
         Fourth Circuit                                             
            Maryland             0.2%      43%      43%      14%          517 
     North Carolina Eastern      0.2%      56%      41%       3%          619 
      North Carolina Middle      0.8%      63%      34%       2%          508 
     North Carolina Western      0.0%      32%      62%       5%          970 
         South Carolina          0.1%      60%      38%       2%          988 
        Virginia Eastern         0.3%      89%       8%       3%        1,360 
        Virginia Western         0.2%      56%      42%       2%          567 
     West Virginia Northern      0.3%      89%       9%       2%          339 
     West Virginia Southern      0.0%      81%      16%       3%          433 

Appendix II: Percentage of Sentences Falling below an Applicable Guideline
Range or An Otherwise Applicable Mandatory Minimum

Sentenced Substantial Other Upward within assistance downward Total number
District departure guideline range departure departure of drug cases

Fifth Circuit Sixth Circuit

       Louisiana Eastern        0.0%      74%      19%       6%    
        Louisiana Middle        0.0%      39%      58%       3%    
       Louisiana Western        0.0%      64%      32%       3%    
      Mississippi Northern      0.0%      42%      49%       8%    
      Mississippi Southern      0.0%      58%      39%       3%    
         Texas Eastern          0.0%      80%      14%       6%    
         Texas Northern         0.0%      65%      31%       4%    
         Texas Southern         0.1%      66%      26%       8%         4,863 
         Texas Western          0.1%      72%      12%      16%         6,750 

                   Page 41 GAO-04-105 Federal Drug Sentences
Kentucky 0.2% 57% 40% 2%  Kentucky 0.0% 81% 16% 3%  Michigan 0.0% 51% 43% 6%  Michigan 0.3% 49% 44% 6%    Ohio   0.3% 47% 44% 9%    Ohio   0.0% 38% 54% 7%  Tennessee 0.1% 57% 41% 2% 742 Tennessee 0.0% 53% 40% 7% 205 Tennessee 0.2% 56% 40% 4% 517 Seventh      Illinois 0.5% 38% 56% 5% 424 Illinois 0.1% 53% 38% 9% 687 Illinois 0.3% 90% 6% 4% 722 Indiana  0.2% 71% 25% 4% 455 Indiana  0.0% 39% 58% 3% 296 Wisconsin 0.4% 65% 32% 3% 280 Wisconsin 1.3% 79% 17% 3% 153 Eighth       Arkansas 0.0% 81% 16% 3% 321 Arkansas 0.0% 70% 27% 4% 168   Iowa   1.3% 52% 41% 6% 477   Iowa   0.1% 45% 44% 11% 692 Minnesota 0.2% 54% 31% 15% 590 Missouri 0.0% 59% 36% 5% 908 Missouri                     
Eastern                   Western                   Eastern                   Western                   Northern                  Southern                   Eastern                       Middle                        Western                      Circuit      Central                      Northern                     Southern                    Northern                     Southern                      Eastern                       Western                      Circuit      Eastern                      Western                      Northern                     Southern                                                     Eastern                      Western  0.1% 37% 59% 3% 914

Appendix II: Percentage of Sentences Falling below an Applicable Guideline
Range or An Otherwise Applicable Mandatory Minimum

                             Sentenced    Substantial     Other 
                   Upward          within assistance  downward   Total number 
     District   departure guideline range  departure  departure of drug cases 
     Nebraska        0.0%             65%         25%       10% 
North Dakota      0.0%             53%         38%        8% 
South Dakota      0.5%             69%         19%       11% 

                                 Ninth Circuit

                   Page 42 GAO-04-105 Federal Drug Sentences
                                                        California                   California                   California                   California                                                                                                                                                Northern                                           Washington                      Washington                       Tenth                                                                New                          Oklahoma                    Oklahoma                     Oklahoma                                                                           Eleventh      Alabama                     Alabama                      Alabama                      Florida                       Florida                      Florida                                                    
Alaska 0.0% 69% 19% 11%  Arizona 0.6% 28% 13% 58% 2,560  Central   0.2% 72% 16% 12%   Eastern   0.0% 64% 26% 10%   Northern  1.5% 60% 15% 23%   Southern  0.1% 17% 13% 70% 5,312 Guam 1.2% 43% 54% 1%  Hawaii 0.4% 58% 32% 10%  Idaho 0.0% 33% 62% 5%  Montana 0.0% 54% 40% 5%  Nevada 0.3% 63% 18% 18%  Mariana  0.0% 29% 71% 0%  Oregon 0.3% 55% 33% 12%   Eastern   0.0% 62% 14% 24% 249  Western   0.2% 52% 33% 15% 544 Circuit      Colorado 0.0% 41% 52% 7% 420 Kansas 0.4% 63% 29% 7% 560 Mexico 0.1% 66% 13% 21% 1,672 Eastern  1.4% 54% 5% 39% 74 Northern 0.8% 73% 21% 5% 131 Western  0.9% 74% 22% 3% 229 Utah 0.0% 67% 12% 21% 199 Wyoming 0.0% 59% 34% 7% 217 Circuit       Middle  0.4% 38% 55% 6% 279 Northern 0.0% 49% 49% 2% 417 Southern 0.0% 44% 49% 6% 342 Middle  0.2% 57% 37% 6% 2,007 Northern 0.9% 60% 38% 1% 565 Southern 0.1% 78% 16% 6% 3,237 Georgia 1.1% 55% 39% 5% 558
                                                                                                                                                                                                                                                                                                         Islands                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                         Middle                  

Appendix II: Percentage of Sentences Falling below an Applicable Guideline Range
                  or An Otherwise Applicable Mandatory Minimum

                               Sentenced    Substantial     Other 
                                     within assistance  downward        Total 
                      Upward                                           number 
      District     departure   guideline     departure  departure     of drug 
                                 range                                  cases 
       Georgia          0.3%            60%         30%       10% 
      Northern                                                    
       Georgia          0.0%            63%         33%        4% 
      Southern                                                    

Source: GAO analysis of USSC data.

Note: 3,004 cases lacking information on departures were excluded.

There were 6 districts in which the percentage of sentences departing
upward from the guidelines exceeded 1 percent of all cases-Wisconsin
Western (1.3 percent), Iowa Northern (1.3 percent), California Northern
(1.5 percent), Guam (1.2 percent), Oklahoma Eastern (1.4 percent), and
Georgia Middle (1.1). The percentage of sentences within the guidelines
range varied substantially, from 17 percent in the California Southern
District to 90 percent in the Illinois Southern District. Fewer than 10
percent of the sentences departed downward for substantial assistance in
Puerto Rico (7 percent), Rhode Island (9 percent), Virginia Eastern (8
percent), West Virginia Northern (9 percent), Illinois Southern (6
percent), and Oklahoma Eastern (5 percent). At the same time, the
percentage of cases departing downward for substantial assistance exceeded
50 percent in 16 districts, and was highest in the North Mariana Islands
(71 percent), New York Northern (68 percent), North Carolina Western (62
percent), and Idaho (62 percent) Districts. Sentences departing downward
for other reasons represented only 3 percent or less of all sentences in
24 districts but over 20 percent of the sentences in 10 districts; these
other downward departures were especially common in New York Eastern (32
percent), Oklahoma Eastern (39 percent), Arizona (58 percent), and
California Southern (70 percent) Districts. While the percentages of other
downward departures were fairly similar and involved 10 percent or fewer
of all cases in the various districts in the Sixth and Seventh Circuits,
the range in the percentage of other downward departures were sizable
across the districts in the Second Circuit (5 percent to 32 percent),
Ninth Circuit (none to 70 percent), and Tenth Circuit (3 to 39 percent).

Table 7 shows the differences across circuits in the percentages of
mandatory minimum drug sentences between 1999 and 2001 that were at or
above an applicable mandatory minimum, below an otherwise applicable
mandatory minimum due to substantial assistance, and below an otherwise
applicable mandatory minimum for other reasons. The percentage of
mandatory minimum sentences that were at or above an applicable mandatory
minimum sentence ranged from 35 percent in the Ninth Circuit to 64 percent
in the Fourth Circuit. The percentage of sentences that fell below an
otherwise applicable mandatory minimum

Appendix II: Percentage of Sentences Falling below an Applicable Guideline
Range or An Otherwise Applicable Mandatory Minimum

due to substantial assistance motions ranged from 19 percent in the First
Circuit to 40 percent in the Third Circuit. The percentage of mandatory
minimum sentences that fell below an otherwise applicable mandatory
minimum sentence for other reasons ranged from 12 percent in the Fourth
Circuit to 38 percent in the Ninth Circuit.

Table 7: Percentages of Mandatory Minimum Drug Cases between 1999 and 2001
Resulting in Sentences at or above a Minimum, and below An Otherwise
Applicable Minimum Due to Substantial Assistance or for Other Reasons, by
Circuit

                          At or     Below a          Below a  Total number of 
                        above a   minimum due to     minimum        mandatory 
                      mandatory   substantial      for other          minimum 
            Circuit     minimum    assistance        reasons       drug cases 
               D.C.         54%              30%         17%              236 
              First         51%              19%         30%            1,444 
             Second         37%              34%         29%            2,835 
              Third         37%              40%         22%            1,927 
             Fourth         64%              25%         12%            4,633 
              Fifth         46%              20%         34%            7,161 
              Sixth         49%              39%         13%            3,201 
            Seventh         59%              24%         17%            2,279 
             Eighth         53%              31%         16%            4,062 
              Ninth         35%              27%         38%            5,381 
              Tenth         44%              22%         34%            2,065 
           Eleventh         47%              23%         29%            6,637 

Source: GAO analysis of USSC data.

Note: 41,861 sentences carried a mandatory minimum and had complete
sentence length information; 284 sentences with a mandatory minimum lacked
complete sentence length information.

Table 8 provides these same percentages, classified by districts rather
than circuits, and shows that variability in the sentencing of mandatory
minimum offenders is considerable across the 94 districts. The percentage
of sentences falling below an otherwise applicable mandatory minimum for
substantial assistance reasons was very different across districts,
ranging from less than 10 percent of all mandatory minimum sentences in 7
districts to over 50 percent in 9 districts. The percentage of sentences
falling below an otherwise applicable mandatory minimum for other reasons
also varied greatly across districts, from less than 10 percent of all
mandatory minimum sentences in 11 districts to 50 percent or more in 3
districts.

Appendix II: Percentage of Sentences Falling below an Applicable Guideline
Range or An Otherwise Applicable Mandatory Minimum

Table 8: Percentages of Mandatory Minimum Drug Cases between 1999 and 2001
Resulting in Sentences at or above the Minimum, and below An Otherwise
Applicable Minimum Due to Substantial Assistance or For Other Reasons, by
District

                          At or above Below a minimum Below a Total number of
                 a mandatory due to substantial minimum for mandatory minimum
                         District minimum assistance other reasons drug cases

D.C. Circuit

District of Columbia 54% 30% 17%

First Circuit Second Circuit Third Circuit

             Maine                  31%             52%             17%       
         Massachusetts              43%             33%             23%       
         New Hampshire              37%             42%             21%       
          Puerto Rico               58%             6%              36%       
          Rhode Island              58%             7%              35%       

          Connecticut            52%         22%         27%     
       New York Eastern          34%         30%         36%     
       New York Northern         17%         70%         13%     
       New York Southern         43%         25%         33%            1,230 
       New York Western          40%         46%         14%     
            Vermont              31%         54%         15%     

              Delaware               53%        36%        11%             36 
             New Jersey              31%        30%        39%            577 
        Pennsylvania Eastern         32%        52%        16%            828 
        Pennsylvania Middle          53%        37%        11%            200 
        Pennsylvania Western         52%        31%        17%            263 
           Virgin Islands            52%        30%        17%             23 
           Fourth Circuit                                          
              Maryland               61%        28%        10%            443 
       North Carolina Eastern        70%        23%         7%            505 
       North Carolina Middle         70%        21%         9%            458 
       North Carolina Western        50%        41%         9%            822 
           South Carolina            61%        28%        11%            685 
          Virginia Eastern           76%         7%        17%            996 
          Virginia Western           49%        39%        12%            456 
       West Virginia Northern        82%         7%        12%             92 
       West Virginia Southern        70%        12%        18%            174 

Appendix II: Percentage of Sentences Falling below an Applicable Guideline
Range or An Otherwise Applicable Mandatory Minimum

                          At or above Below a minimum Below a Total number of
                 a mandatory due to substantial minimum for mandatory minimum
                         District minimum assistance other reasons drug cases

Fifth Circuit Sixth Circuit

        Louisiana Eastern          55%        17%        29%     
         Louisiana Middle          50%        40%        10%     
        Louisiana Western          83%        12%         5%     
       Mississippi Northern        61%        27%        12%     
       Mississippi Southern        50%        32%        18%     
          Texas Eastern            71%        11%        18%     
          Texas Northern           57%        25%        18%     
          Texas Southern           40%        21%        39%            2,702 
          Texas Western            36%        18%        46%            2,124 

                   Page 46 GAO-04-105 Federal Drug Sentences
Kentucky 32% 52% 15%  Kentucky 71% 14% 15%  Michigan 46% 40% 14%  Michigan 57% 32% 11%    Ohio   45% 42% 13%    Ohio   44% 47% 9%  Tennessee 52% 35% 13% 472 Tennessee 49% 40% 12% 136 Tennessee 59% 31% 10% 367 Seventh     Illinois 60% 35% 5% 304 Illinois 44% 31% 26% 570 Illinois 79% 4% 17% 547 Indiana  60% 22% 18% 274 Indiana  43% 45% 12% 272 Wisconsin 60% 23% 17% 217 Wisconsin 79% 9% 12% 95 Eighth      Arkansas 68% 15% 17% 207 Arkansas 63% 18% 18% 119   Iowa   64% 26% 10% 388   Iowa   54% 30% 16% 629 Minnesota 49% 27% 24% 472 Missouri 53% 33% 15% 663 Missouri                
Eastern               Western               Eastern               Western               Northern              Southern              Eastern                   Middle                    Western                  Circuit     Central                 Northern                 Southern                Northern                 Southern                  Eastern                   Western                Circuit     Eastern                  Western                  Northern                 Southern                                           Eastern                  Western  38% 53% 9% 654

Appendix II: Percentage of Sentences Falling below an Applicable Guideline
Range or An Otherwise Applicable Mandatory Minimum

                At or above  Below a minimum         Below a  Total number of 
                            due to substantial   minimum for        mandatory 
                a mandatory                                           minimum 
     District       minimum     assistance     other reasons       drug cases 
     Nebraska           57%                22%           21% 
North Dakota         44%                45%           11% 
South Dakota         63%                23%           14% 

                                 Ninth Circuit

               Alaska                 59%        19%       22%    
               Arizona                18%        20%       62%          1,157 
         California Central           52%        14%       34%    
         California Eastern           57%        23%       20%    
         California Northern          54%        16%       31%    
         California Southern          19%        34%       47%          1,357 
                Guam                  28%        52%       19%    
               Hawaii                 44%        33%       23%    
                Idaho                 54%        37%       8%     
               Montana                50%        40%       10%    
               Nevada                 51%        18%       30%    
      Northern Mariana Islands        22%        67%       11%    
               Oregon                 50%        27%       22%    
         Washington Eastern           51%        16%       33%            128 
         Washington Western           41%        35%       24%            381 
            Tenth Circuit                                         
              Colorado                32%        52%       16%            251 
               Kansas                 63%        22%       15%            382 
             New Mexico               29%        16%       55%            874 
          Oklahoma Eastern            57%        7%        37%             46 
          Oklahoma Northern           65%        21%       14%             72 
          Oklahoma Western            73%        17%       10%            151 
                Utah                  41%        16%       43%            107 
               Wyoming                54%        27%       19%            185 

Appendix II: Percentage of Sentences Falling below an Applicable Guideline Range
                  or An Otherwise Applicable Mandatory Minimum

                          At or above Below a minimum Below a Total number of
                 a mandatory due to substantial minimum for mandatory minimum
                         District minimum assistance other reasons drug cases

                                Eleventh Circuit

        Alabama Middle          43%         48%         9%      
       Alabama Northern         52%         37%         11%     
       Alabama Southern         49%         41%         10%     
        Florida Middle          53%         28%         19%             1,642 
       Florida Northern         59%         33%         8%      
       Florida Southern         35%         14%         50%             2,686 
        Georgia Middle          65%         25%         10%     
       Georgia Northern         57%         22%         21%     
       Georgia Southern         75%         19%         7%      

Source: GAO analysis of USSC data.

Note: 41,861 sentences carried a mandatory minimum and had complete
sentence length information; 284 sentences with a mandatory minimum lacked
complete sentence length information.

Appendix III: Likelihood of Sentences Falling below a Guideline Range or an
Otherwise Applicable Mandatory Minimum

This appendix provides odds and odds ratios to describe the differences
across circuits and districts in sentences falling below a guideline range
or an otherwise applicable mandatory minimum for substantial assistance
and other reasons, both before and after controlling for differences in
offender and offense characteristics. In the left columns of tables 9 and
10, we show the odds on substantial assistance departures across circuits
and districts and ratios indicating differences across circuits and
districts, before and after we adjust for characteristics of offenses and
offenders. In the right columns of tables 9 and 10, we show the odds on
other downward departures across circuits and districts and ratios
indicating differences between them, before and after we adjust for
characteristics of offenses and offenders. In the comparisons across
circuits, we used the Eighth Circuit as the reference category, so the
odds ratios reflect how much more or less likely other circuits were than
that circuit to depart in sentencing offenders. In comparisons across
districts, the Minnesota District was used as the reference category. The
offense and offender characteristics we controlled for were described
earlier in appendix I. The ratios that estimate differences before and
after adjusting for these characteristics were derived from logistic
regression models. We focus on adjusted ratios in the following
discussion, since they provide us with our best estimates of differences
across circuits and districts after taking into account the differences in
the drug cases handled across jurisdictions.

Table 9 shows that both the odds on substantial assistance departures and
other downward departures varied substantially across circuits. After
adjusting for differences across circuits in offense and offender
characteristics, the odds on substantial assistance departures were
significantly greater in three circuits than the Eighth Circuit. In the
Third Circuit, for example, substantial assistance departures were 2.2
times as likely as in the Eighth Circuit. Four circuits were not
significantly different from the Eighth Circuit in terms of the likelihood
of sentences departing for substantial assistance, and in the remaining 4
circuits substantial assistance departures were significantly less likely.
In the First Circuit, for example, substantial assistance departures were
less likely by a factor of 0.64, or 36 percent less likely, than in the
Eighth Circuit. The fact that some circuits are less likely than the
Eighth Circuit while others are more likely than the Eighth Circuit to
depart for substantial assistance implies that some differences between
other circuits are larger than those explicitly indicated by these ratios.
For example, these ratios imply that substantial assistance departures in
the Third Circuit are 2.2/0.64=3.4 times as likely as in the First
Circuit.

  Appendix III: Likelihood of Sentences Falling below a Guideline Range or an
                     Otherwise Applicable Mandatory Minimum

Table 9: Odds on Substantial Assistance Departures and Other Downward
Departures, and Odds Ratios Indicating the Differences between Circuits,
before and after Adjusting for Offense and Offender Characteristics

                  Odds on                       Odds on              
             substantial                            other            
              assistance  Unadjusted  Adjusted  downward  Unadjusted Adjusted 
    Circuit   departures    ratios      ratios departures   ratios     ratios 
    Eighth           0.59                            0.14            
     D.C.            0.44       0.75*     1.15       0.31      2.19*    2.16* 
     First           0.26       0.44*    0.64*       0.14       1.02    0.78* 
    Second           0.47       0.81*     1.4*       0.41      2.92*    2.39* 
     Third           0.83       1.41*     2.2*       0.15       1.05 
    Fourth           0.49       0.84*     1.01       0.06      0.43*    0.37* 
     Fifth           0.25       0.43*    0.85*       0.16       1.11    0.55* 
     Sixth           0.71       1.22*    1.68*       0.10      0.69*    0.57* 
    Seventh          0.45       0.78*     0.91       0.08      0.55*    0.48* 
     Ninth           0.22       0.38*    0.75*       1.38       9.9*    6.87* 
     Tenth           0.29       0.49*    0.78*       0.23      1.67*     1.2* 
Eleventh          0.43       0.74*     0.98       0.09      0.61*     0.5* 

Source: GAO analysis of USSC data.

Note: Odds on substantial assistance departures were calculated using all
cases. Odds on other downward departures were calculated using only cases
not involving downward departures for substantial assistance. Ratios were
calculated using the Eighth Circuit as the referent category. Adjusted
ratios are from logistic regression models that control for offense and
offender characteristics.

*Odds ratio coefficients that are significant at the 0.05 level.

Table 9 also shows that for those sentences that do not involve
substantial assistance departures, (and again after adjusting for offense
and offender characteristics), other downward departures are significantly
more likely in 4 circuits than in the Eighth Circuit, significantly less
likely in 6 circuits than in the Eighth Circuit, and no different in the
other one. The fact that other downward departures are 6.87 times more
likely in the Ninth Circuit than in the Eighth Circuit, but less likely by
a factor of 0.37 in the Fourth Circuit than in the Eighth Circuit, implies
that such departures are 6.87/0.37 = 18.6 or 19 times as likely in the
Ninth Circuit as in the Fourth Circuit.

Table 10 shows that both the adjusted odds ratios on substantial
assistance departures and other downward departures also varied
substantially and significantly across districts. Substantial assistance
departures were significantly more likely in 41 districts than in the
Minnesota District. In the small Northern Mariana Islands District, for

Appendix III: Likelihood of Sentences Falling below a Guideline Range or
an Otherwise Applicable Mandatory Minimum

example, substantial assistance departures were 10 times more likely than
in the Minnesota District, and in the large New York Northern District,
they were 5 times more likely. Twenty-three districts were not
significantly different from the Minnesota District in terms of the
likelihood of sentences departing for substantial assistance, and in the
remaining 29 districts substantial assistance departures were
significantly less likely. In the Illinois Southern District, for example,
substantial assistance departures were less likely by a factor of 0.11,
which implies that the likelihood of substantial assistance departures
were 9 times higher in the Minnesota District than they were there. Other
districts, these odds imply, were even more disparate from one another.
For example, these ratios imply that substantial assistance departures in
the New York Northern District were 5.5/0.11=49.5 or 50 times more likely
than in the Illinois Southern District.

Table 10: Odds on Substantial Assistance Departures and Other Downward
Departures, and Odds Ratios Indicating the Differences between Districts,
before and after Adjusting for Offense and Offender Characteristics

                   Odds on                      Odds on              
               substantial                          other            
                assistance Unadjusted Adjusted  downward  Unadjusted Adjusted 
    District    departures     ratios   ratios departures     ratios   ratios 
Minnesota          0.45                           0.27            

D.C. Circuit

District of Columbia 0.44 0.97 1.19 0.31 1.12

First Circuit Second Circuit

         Maine         1.10     2.44*     2.79*     0.07     0.27*      0.28* 
     Massachusetts     0.42     0.93      1.01      0.37     1.35    
     New Hampshire     0.75     1.66*     2.23*     0.15     0.56       0.42* 
      Puerto Rico      0.08     0.17*     0.17*     0.06     0.23*      0.19* 
     Rhode Island      0.10     0.23*     0.28*     0.13     0.48*      0.42* 

        Connecticut        0.32    0.71*    0.69*    0.69    2.56*      2.42* 
     New York Eastern      0.31    0.68*     0.97    0.71    2.61*      2.48* 
     New York Northern     2.13    4.75*    5.51*    0.34     1.23       0.95 
     New York Southern     0.32     0.7*    0.78*    0.19    0.69*      0.66* 
     New York Western      1.10    2.44*     4.5*    0.11     0.4*      0.49* 
          Vermont          0.58     1.28    1.47*    0.48    1.78*      1.81* 

Appendix III: Likelihood of Sentences Falling below a Guideline Range or
an Otherwise Applicable Mandatory Minimum

                  Odds on                       Odds on              
              substantial                           other            
               assistance Unadjusted  Adjusted  downward  Unadjusted Adjusted 
District    departures      ratios   ratios departures     ratios   ratios 

                          Third Circuit Fourth Circuit

          Delaware         0.69     1.54    1.78*    0.14     0.50      0.37* 
         New Jersey        0.47     1.04    1.29*    0.12    0.46*      0.45* 
        Pennsylvania                                                 
          Eastern          1.33    2.96*    3.41*    0.17    0.62*       0.6* 
    Pennsylvania Middle    1.25    2.79*    3.44*    0.17     0.62      0.56* 
        Pennsylvania                                                 
          Western          0.59     1.30     1.25    0.16    0.57*      0.49* 
       Virgin Islands      0.20    0.44*     1.00    0.11     0.41   

          Maryland          0.75    1.66*    1.76*    0.33     1.20   
       North Carolina                                                 
           Eastern          0.71    1.57*    1.37*    0.05    0.18*     0.15* 
       North Carolina                                                 
           Middle           0.52     1.15     1.05    0.03    0.11*     0.09* 
       North Carolina                                                 
           Western          1.66     3.7*    3.06*    0.16     0.6*     0.44* 
       South Carolina       0.62    1.38*    1.35*    0.03    0.11*     0.09* 
      Virginia Eastern      0.09     0.2*     0.2*    0.03    0.12*     0.12* 
      Virginia Western      0.72    1.61*    1.43*    0.04    0.14*     0.11* 
        West Virginia                                                 
          Northern          0.10    0.22*    0.24*    0.02    0.09*     0.06* 
        West Virginia                                                 
          Southern          0.19    0.41*    0.46*    0.04    0.15*     0.12* 
        Fifth Circuit                                                 
      Louisiana Eastern     0.24    0.52*    0.53*    0.09    0.32*     0.34* 
      Louisiana Middle      1.36    3.02*    3.94*    0.08     0.28      0.32 
      Louisiana Western     0.48     1.07     1.03    0.05    0.19*     0.16* 
    Mississippi Northern    0.98    2.17*    2.02*    0.19     0.71      0.62 
    Mississippi Southern    0.63    1.41*    1.42*    0.05    0.19*     0.18* 
        Texas Eastern       0.16    0.37*    0.38*    0.07    0.28*     0.23* 
       Texas Northern       0.44     0.98     0.97    0.07    0.25*     0.22* 
       Texas Southern       0.36     0.8*    1.28*    0.11    0.42*     0.35* 
        Texas Western       0.13     0.3*    0.55*    0.22     0.81      0.6* 

Appendix III: Likelihood of Sentences Falling below a Guideline Range or
an Otherwise Applicable Mandatory Minimum

                  Odds on                       Odds on              
              substantial                           other            
               assistance Unadjusted  Adjusted  downward  Unadjusted Adjusted 
District    departures      ratios   ratios departures     ratios   ratios 

                         Sixth Circuit Seventh Circuit

     Kentucky Eastern      0.68    1.51*    1.96*    0.04    0.16*      0.17* 
     Kentucky Western      0.19    0.43*    0.47*    0.03    0.12*      0.14* 
     Michigan Eastern      0.76    1.68*    2.02*    0.12    0.44*      0.43* 
     Michigan Western      0.80    1.78*    1.91*    0.13    0.47*      0.42* 
       Ohio Northern       0.77    1.72*     1.8*    0.20     0.74   
       Ohio Southern       1.19    2.65*    3.18*    0.19     0.70   
     Tennessee Eastern     0.69    1.54*    1.56*    0.04    0.14*      0.12* 
     Tennessee Middle      0.65    1.45*    1.58*    0.14    0.51*   
     Tennessee Western     0.68    1.51*     1.4*    0.07    0.26*      0.22* 

                   Page 53 GAO-04-105 Federal Drug Sentences
Illinois 1.27 2.82* 2.62* 0.14 0.52* 0.37* Illinois 0.60 1.34* 1.52* 0.17 0.62* 0.67* Illinois 0.06 0.14* 0.11* 0.04 0.15* 0.11* Indiana  0.34 0.76 0.97 0.05 0.18* 0.18* Indiana  1.39 3.08* 2.69* 0.09 0.32* 0.35* Wisconsin 0.47 1.05 0.96 0.04 0.14* 0.12* Wisconsin 0.20 0.46* 0.35* 0.03 0.12* 0.1* Eighth        Arkansas 0.19 0.43* 0.35* 0.04 0.14* 0.12* Arkansas 0.37 0.81 0.83 0.05 0.19* 0.18*   Iowa   0.69 1.54* 1.03 0.12 0.44* 0.41*   Iowa   0.79 1.76* 1.20 0.24 0.89 0.77 Missouri 0.57 1.26* 1.19 0.09 0.32* 0.31* Missouri 1.44 3.21* 2.63* 0.09 0.33* 0.3* Nebraska 0.33 0.73* 0.52* 0.16 0.58* 0.54* North  0.62 1.38 1.16 0.15 0.56 0.65 South  0.24 0.54* 0.57* 0.15 0.56* 0.58*  Ninth        Alaska 0.24 0.54* 0.61* 0.17 0.61 0.69 Arizona 0.15 0.34* 0.61* 1.98 7.3* 7.41* California 0.19 0.42* 0.55* 0.17 0.63* 0.69 California 0.35 0.78 0.86 0.15 0.55* 0.51* California 0.18 0.4* 0.48* 0.38 1.4* 1.68* California                                     
Central                                    Northern                                   Southern                                   Northern                                 Southern                                    Eastern                                   Western                                   Circuit       Eastern                                    Western                                  Northern                                  Southern                                Eastern                                   Western                                                                              Dakota                               Dakota                                   Circuit                                                                                        Central                                     Eastern                                    Northern                                   Southern  0.15 0.33* 0.65* 4.18 15.39* 14.69*

Appendix III: Likelihood of Sentences Falling below a Guideline Range or
an Otherwise Applicable Mandatory Minimum

                 Odds on                        Odds on              
               substantial                          other            
               assistance  Unadjusted Adjusted  downward  Unadjusted Adjusted 
    District   departures    ratios     ratios departures     ratios   ratios 
      Guam            1.18      2.63*    2.12*       0.03       0.1*    0.13* 
     Hawaii           0.47       1.05     1.06       0.16       0.6* 
      Idaho           1.62      3.61*    4.16*       0.16       0.57 
     Montana          0.68       1.5*    1.76*       0.10      0.36*    0.41* 
     Nevada           0.23       0.5*    0.55*       0.28       1.02 
    Northern                                                         
     Mariana                                                         
     Islands          2.40      5.34*   10.08*       0.00       0.00 
     Oregon           0.48       1.07     1.02       0.22       0.80    0.59* 
Washington         0.16      0.36*    0.44*       0.39       1.43 
     Eastern                                                         
Washington         0.50       1.11    1.34*       0.28       1.04 
     Western                                                         

Tenth Circuit

         Colorado          1.07    2.38*    3.82*    0.17     0.64   
          Kansas           0.41     0.92     0.96    0.11     0.4*       0.4* 
        New Mexico         0.15    0.34*    0.54*    0.31     1.14   
     Oklahoma Eastern      0.06    0.13*    0.11*    0.71     2.6*      3.28* 
     Oklahoma Northern     0.27     0.6*    0.55*    0.07    0.27*      0.26* 
     Oklahoma Western      0.28    0.62*    0.52*    0.04    0.15*      0.11* 
           Utah            0.14    0.31*     0.4*    0.31     1.13   

         Wyoming          0.52     1.15      0.73    0.13    0.46*      0.43* 
     Eleventh Circuit                                                
      Alabama Middle      1.23     2.74*    3.84*    0.16     0.58   
     Alabama Northern     0.96     2.13*    2.11*    0.04    0.14*      0.15* 
     Alabama Southern     0.98     2.17*     2.1*    0.15    0.54*       0.5* 
      Florida Middle      0.59     1.3*     1.28*    0.10    0.36*      0.31* 
     Florida Northern     0.62     1.39*     1.25    0.02    0.06*      0.05* 
     Florida Southern     0.19     0.42*    0.44*    0.07    0.26*      0.24* 
      Georgia Middle      0.64     1.42*    1.88*    0.09    0.33*      0.31* 
     Georgia Northern     0.43     0.96      0.89    0.16     0.6*      0.55* 
     Georgia Southern     0.49     1.08      1.09    0.07    0.26*      0.22* 

Source: GAO analysis of USSC data.

Note: Odds on substantial assistance departures were calculated using all
cases. Odds on other downward departures were calculated using only cases
not involving downward departures for substantial assistance. Ratios were
calculated using the Minnesota district as the referent category. Adjusted
ratios are from logistic regression models that control for offense and
offender characteristics.

*Odd ratio coefficients that are significant at the 0.05 level.

Appendix III: Likelihood of Sentences Falling below a Guideline Range or
an Otherwise Applicable Mandatory Minimum

Table 10 also shows that for those sentences that do not involve
substantial assistance departures, other downward departures are
significantly more likely in 7 districts than in the Minnesota district,
significantly less likely in 62 districts than in the Minnesota district,
and no different in the remaining 23 districts.1 The fact that other
downward departures are 15 times more likely in the California Southern
District than in the Minnesota district, but less likely by a factor of
0.09 in the South Carolina District than in the Minnesota District,
implies that such departures are 15/0.09 = 167 times as likely in the
California Southern District as in the South Carolina District.

Tables 11 and 12 pertain to mandatory minimum sentences and show that
substantial and often significant variation in the likelihood of sentences
falling below an otherwise applicable mandatory minimum exists even after
controls for differences in offense and offender characteristics across
circuits and districts. If we focus on the adjusted ratios in table 11
first, which estimate the differences among circuits after controls, we
find that there were some circuits in which the odds on sentences falling
below an otherwise applicable mandatory minimum due to substantial
assistance were significantly higher than in the Eighth Circuit, and
others in which those odds were significantly lower.

1 As noted in table 10, USSC data contained no "other downward departure"
sentences in drug cases in the Northern Mariana Islands District for
fiscal years 1999 through 2001.

  Appendix III: Likelihood of Sentences Falling below a Guideline Range or an
                     Otherwise Applicable Mandatory Minimum

Table 11: Odds on Mandatory Minimum Sentences Falling below an Otherwise
Applicable Mandatory Minimum for Substantial Assistance and for Other
Reasons, and Odds Ratios Indicating the Differences between Circuits,
before and after Adjusting for Offense and Offender Characteristics

                    Odds on                      Odds on             
                                                 falling             
             falling below                        below              
            for substantial Unadjusted Adjusted for other Unadjusted Adjusted 
Circuit       assistance   ratios     ratios   reasons   ratios     ratios 
    Eighth             0.44                          0.36            
     D.C.              0.42       0.95    1.54*      0.33       0.91    2.55* 
    First              0.23      0.53*    0.67*      0.64      1.78* 
    Second             0.51      1.16*     1.7*      0.85      2.38*    1.53* 
    Third              0.68      1.53*    2.29*      0.75      2.08*    1.54* 
    Fourth             0.33      0.74*     0.99      0.23      0.64* 
    Fifth              0.25      0.55*    0.71*      0.85      2.37* 
    Sixth              0.63      1.43*    1.74*      0.33       0.93 
Seventh             0.31      0.71*    0.88*      0.33       0.93 
    Ninth              0.37      0.83*     1.04      1.21      3.37*       2* 
    Tenth              0.29      0.65*    0.78*      0.89      2.47* 
Eleventh            0.30      0.68*     0.9*      0.75      2.09* 

Source: GAO analysis of USSC data.

Note: The odds on sentences falling below for substantial assistance were
calculated using all cases. The odds on sentences falling below for other
reasons were calculated using only cases that did not fall below for
substantial assistance. Ratios were calculated using the Eighth Circuit as
the referent category. Adjusted ratios are from logistic regression models
that control for offense and offender characteristics.

*Odds ratio coefficients that are significant at the 0.05 level.

The same is true of the likelihood of sentences falling below an otherwise
applicable mandatory minimum for reasons other than substantial
assistance. The adjusted ratios in table 11 suggest that the biggest
difference in the likelihood of mandatory minimum sentences falling below
an otherwise applicable mandatory minimum due to substantial assistance
involved the Third and First Circuits (such sentences were 2.29/0.67=3.4
times more likely in the former circuit than in the latter), while the
biggest difference in the likelihood of mandatory minimum sentences
falling below an otherwise applicable mandatory minimum for reasons other
reasons, such as the safety valve, involved the D.C. vs. the Fourth and
Sixth Circuits (such sentences were 2.55/0.94=2.7 times more likely in the
former circuit than in the latter two).

Table 12 shows, similarly, that in many districts judges were much more
likely than in the Minnesota district to issue sentences below a mandatory

  Appendix III: Likelihood of Sentences Falling below a Guideline Range or an
                     Otherwise Applicable Mandatory Minimum

minimum to offenders facing a mandatory minimum, both for reasons of
substantial assistance and for other reasons; and, at the same time,
judges in many other districts were less likely to do so, overall.

Table 12: Odds on Mandatory Minimum Sentences Falling below an Otherwise
Applicable Mandatory Minimum for Substantial Assistance and for Other
Reasons, and Odds Ratios Indicating the Differences between Districts,
before and after Adjusting for Offense and Offender Characteristics D.C.
Circuit

                    Odds on                       Odds on            
                    falling                                          
                  below for                       falling            
                                                    below            
                substantial Unadjusted Adjusted for other Unadjusted Adjusted 
District      assistance   ratios    ratios    reasons   ratios     ratios 
Minnesota           0.37                          0.54            

Dist of Columbia 0.42 1.13 2.78 0.33 0.6*

First Circuit Second Circuit

          Maine              1.08    2.91*       1.02       0.72      1.33    
      Massachusetts          0.50    1.34*       2.55*      0.57      1.05    
      New Hampshire          0.72    1.94*       0.69       0.82      1.50    
       Puerto Rico           0.07    0.18*       0.27*      0.65      1.20    
      Rhode Island           0.07    0.2*        0.31       0.63      1.16    

                   Page 57 GAO-04-105 Federal Drug Sentences
                                              New                              New York                             New York                               New                                                                      Third                                                New                                 Pennsylvania                                 Pennsylvania                                  Pennsylvania                               Virgin                            Fourth                                                  North                                   North                                      North                                     
Connecticut 0.28 0.75 3.69* 0.57 1.05 3.89*  York   0.43 1.17 3.5* 1.17 2.15*  Northern 2.33 6.26* 2.75* 0.98 1.8*  Southern 0.33 0.88 6.16* 0.80 1.48*   York   0.85 2.29* 1.02 0.43 0.80  Vermont 1.19 3.19* 6.71 0.53 0.97 3.95 Circuit       Delaware 0.57 1.52 0.43 0.29 0.53 1.50 Jersey 0.43 1.16 0.8 1.50 2.77* 0.87   Eastern    1.07 2.88* 2.33* 0.62 1.13 1.37    Middle    0.59 1.58* 0.32* 0.33 0.60 4.49*   Western    0.45 1.2 1.18 0.36 0.66* 0.92 Islands 0.44 1.18  0.33 0.61 0.34 Circuit       Maryland 0.40 1.07 0.44* 0.23 0.42* 1.87 Carolina 0.30 0.8 0.28* 0.15 0.27* 0.44 Carolina 0.27 0.72* 0.46* 0.15 0.28* 0.38* Carolina                             
                                            Eastern                                                                                                      Western                                                                                                                                                                                                                                                                                                                                                                                            Eastern                                  Middle                                    Western  0.70 1.88* 0.53* 0.36 0.66* 3.04*

Appendix III: Likelihood of Sentences Falling below a Guideline Range or
an Otherwise Applicable Mandatory Minimum

                      Odds on                     Odds on            
                      falling                                        
                    below for                     falling            
                                                    below            
                  substantial Unadjusted Adjusted     for Unadjusted Adjusted 
                                                    other            
      District     assistance   ratios    ratios  reasons   ratios     ratios 
South Carolina        0.39       1.05  0.39*      0.23      0.43* 
      Virginia           0.08       0.2*   0.83      0.23      0.42*     0.5* 
      Eastern                                                        
      Virginia           0.63      1.69*   1.92      0.29      0.54* 
      Western                                                        
West Virginia         0.07      0.19*             0.15      0.27* 
      Northern                                                       
West Virginia         0.14      0.37*   0.39      0.29      0.53* 
      Southern                                                       

                          Fifth Circuit Sixth Circuit

      Louisiana Eastern       0.20  0.53*     0.57    0.58     1.07   
      Louisiana Middle        0.68   1.83    0.23*    0.33     0.61   
      Louisiana Western       0.14  0.37*    0.12*    0.08    0.14*     0.24* 
    Mississippi Northern      0.37    1      0.36*    0.27    0.49*     2.88* 
    Mississippi Southern      0.48   1.28    0.27*    0.45     0.83     0.34* 
        Texas Eastern         0.12  0.33*    0.39*    0.28    0.52*   
       Texas Northern         0.33   0.88    0.43*    0.40     0.74   
       Texas Southern         0.27  0.72*     0.63    1.07    1.96*   
        Texas Western         0.22  0.59*    0.45*    1.44    2.65*     1.76* 

     Kentucky Eastern        1.10   2.97*    2.46*    0.52     0.96   
     Kentucky Western        0.16   0.44*    0.16*    0.23    0.42*      0.4* 
     Michigan Eastern        0.67   1.79*     0.85    0.38    0.71*      0.89 
     Michigan Western        0.46   1.24     0.16*    0.32    0.58*      0.79 
       Ohio Northern         0.73   1.96*     0.85    0.38     0.7*      1.37 
       Ohio Southern         0.87   2.35*     0.6     0.29    0.54*      0.59 
     Tennessee Eastern       0.54   1.44*    0.45*    0.32    0.59*      0.57 
     Tennessee Middle        0.66   1.77*     1.4     0.30    0.54*      0.65 
     Tennessee Western       0.46   1.23      0.69    0.22     0.4*      0.65 
      Seventh Circuit                                                 
     Illinois Central        0.54   1.46*    0.38*    0.14    0.27*      0.37 
     Illinois Northern       0.45    1.2      0.71    0.71     1.30      1.07 
     Illinois Southern       0.04   0.11*     0.42    0.22     0.4*      0.78 
     Indiana Northern        0.29   0.77      0.59    0.33     0.6*      0.85 
     Indiana Southern        0.81   2.19*     0.55    0.39     0.71      1.49 
      Wisconsin East         0.29   0.78      0.66    0.34    0.63*      0.77 
      Wisconsin West         0.10   0.28*    0.09*    0.17    0.31*      0.1* 

Appendix III: Likelihood of Sentences Falling below a Guideline Range or
an Otherwise Applicable Mandatory Minimum

            Odds on falling                       Odds on            
                  below for                       falling            
                                                    below            
                substantial Unadjusted Adjusted for other Unadjusted Adjusted 
District      assistance   ratios    ratios    reasons   ratios     ratios 

                          Eighth Circuit Ninth Circuit

     Arkansas Eastern       0.18   0.47*     0.62    0.27     0.5*   
     Arkansas Western       0.23   0.61     0.15*    0.35     0.64   
      Iowa Northern         0.35   0.93     0.35*    0.21    0.39*   
      Iowa Southern         0.43   1.15     0.33*    0.42     0.77   
     Missouri Eastern       0.48   1.3*      0.6     0.33    0.61*   
     Missouri Western       1.12   3.02*     1.29    0.29    0.54*   
         Nebraska           0.27   0.74*     0.94    0.39    0.73*   
       North Dakota         0.83   2.24*     1.98    0.27     0.50   
       South Dakota         0.30   0.82      0.39    0.23    0.43*      0.16* 

                   Page 59 GAO-04-105 Federal Drug Sentences
                                                                         California                            California                          California                           California                                                                                                                                                                                                                              Northern                                                                      Washington                                Washington                                 Tenth                                                                                        New                               Oklahoma                              Oklahoma                                Oklahoma                                                                                                       
Alaska 0.24 0.64 0.67 0.38 0.70  Arizona 0.26 0.69* 1.87 3.73 6.87* 7.5*  Central   0.16 0.42* 0.53 0.73 1.35   Eastern   0.30 0.8 0.77 0.42 0.77   Northern  0.19 0.5* 1.07 0.60 1.11   Southern  0.51 1.37* 2.27* 2.68 4.94* 2.61* Guam 1.09 2.94* 1.55 0.72 1.33 0.2* Hawaii 0.49 1.31 1.36 0.56 1.03 0.3* Idaho 0.59 1.6 0.34* 0.25 0.46 1.07 Montana 0.66 1.77* 0.89 0.22 0.41* 0.81 Nevada 0.23 0.61* 0.48 0.63 1.15 0.65 Mariana  2.00 5.37* 0.28 1.00 1.84 0.93 Oregon 0.37 1.01 0.27* 0.56 1.02 0.66  Eastern   0.20 0.53* 0.66 0.70 1.29 0.73  Western   0.54 1.44* 1.23 0.69 1.27 1.48 Circuit       Colorado 1.07 2.89* 0.68 0.73 1.34 1.57 Kansas 0.28 0.76 0.3* 0.28 0.51* 0.41* Mexico 0.19 0.5* 1 1.99 3.66* 0.86 Eastern  0.07 0.19*  0.65 1.20 10.79* Northern 0.26 0.71 0.53 0.24 0.45* 1.50 Western  0.21 0.56* 0.88 0.16 0.29* 0.22* Utah 0.19 0.51*  1.05 1.92* 2.50                                
                                                                                                                                                                                                                                                                                                                                                                                                                                Islands                                                                                                                                                                                                                                                                                                                                                                                                                                                    Wyoming 0.37 1 0.18* 0.42 0.77 1.19

  Appendix III: Likelihood of Sentences Falling below a Guideline Range or an
                     Otherwise Applicable Mandatory Minimum

            Odds on falling                       Odds on            
                  below for                       falling            
                                                    below            
                substantial Unadjusted Adjusted for other Unadjusted Adjusted 
District      assistance   ratios    ratios    reasons   ratios     ratios 

                                Eleventh Circuit

      Alabama Middle        0.91   2.45*    0.46*    0.34     0.62   
     Alabama Northern       0.59   1.59*     0.63    0.27    0.49*   
     Alabama Southern       0.69   1.85*     0.75    0.27    0.49*       0.3* 
      Florida Middle        0.38   1.03     0.46*    0.44     0.82   
     Florida Northern       0.49   1.32      1.63    0.16     0.3*   
     Florida Southern       0.17   0.44*     0.56    1.56    2.87*   
      Georgia Middle        0.33   0.88      0.2*    0.21    0.39*      0.23* 
     Georgia Northern       0.28   0.75*    0.26*    0.47     0.86   
     Georgia Southern       0.23   0.62*    0.18*    0.12    0.21*      0.25* 

Source: GAO analysis of USSC data.

Note: The odds on sentences falling below for substantial assistance were
calculated using all cases. The odds on sentences falling below for other
reasons were calculated using only cases that did not fall below for
substantial assistance. Ratios were calculated using the Minnesota
District as the referent category. Adjusted ratios are from logistic
regression models that control for offense and offender characteristics.

*Odd ratio coefficients that are significant at the 0.05 level.

Appendix IV: Data for Analyzing Departures and Mandatory Minimum Sentences at
Circuit and District Court Levels Limited

  Documents USSC Requests from District Courts

o   o   o   o   o

  How USSC Uses Documents to Create Its Database

Overall, the data the U.S. Sentencing Commission (USSC) has received
from district courts and judges were generally sufficient for our analyses
of downward departures and mandatory minimum sentences across
circuits and for most districts. Missing data due to missing sentencing
documents or information posed few limitations for our analysis.
However, opportunities for improvement exist.

Under the authority of the Sentencing Reform Act of 1984, USSC required
courts to forward to it the following five sentencing documents in every
guidelines case.1 The PROTECT Act of 2003 codifies this data collection
requirement:

the Judgment and Commitment Order (J&C);
the Statement of Reasons (SOR);
the Pre-sentence Report (PSR);
any written plea agreements, if applicable; and
all indictments or other charging documents.

Under the PROTECT Act, courts are to send to USSC a "Report of
Sentence" enclosing the required sentencing documents within 30 days of
a judgment, and the Chief Judge in every district is to ensure that their
courts do so.2

Of the five sentencing documents submitted by district courts, USSC
officials told us they rely primarily on the J&C, SOR, and PSR to obtain
the
sentencing information that USSC staff code into USSC database. From
the J&C, USSC obtains data on the sentence, including the number of
months of any imprisonment, the statute of conviction, and whether any
mandatory minimum sentence applied. USSC officials also said that they
rely almost exclusively on the SOR to obtain data on the basis for the
sentence, such as whether the sentence imposed fell within or outside the
applicable sentencing guidelines range as determined by the court, the
reason(s) for any departure, and whether a substantial assistance motion
or safety valve adjustment was used. If the SOR is missing, USSC coding

128 U.S.C. 994(w), 995(a)(8). Courts are also to forward additional
documentation related to actions taken after sentencing, such as
revocations of probation or resentencing under Federal Rule of Criminal
Procedure 35(b).

2P.L. 108-21, sec. 401(h), amending 28 U.S.C. 994(w).

Appendix IV: Data for Analyzing Departures and Mandatory Minimum Sentences
at Circuit and District Court Levels Limited

procedures have required document analysts to record the departure status
as missing, although other documents, such as the plea agreement, may have
information that indicate whether and why the sentence departed. USSC is
initiating some changes in its coding procedures as discussed below. From
the PSR, which is drafted by a district probation officer, USSC obtains
demographic and other background information about offenders, an initial
sentencing recommendation according to the guidelines, and other
sentencing information such as whether the offense of conviction had a
mandatory minimum (should this information not be noted in the J&C), and
whether the safety valve could potentially be applied (in certain limited
circumstances where this information has not been recorded in a SOR).

Our analysis shows that district courts provided these five sentencing
documents to USSC for the great majority of drug sentences imposed in
fiscal years 1999-2001. Of 72,283 drug sentences imposed during this
period, district courts submitted between 96 and 99 percent of the three
key sentencing documents- the J&C (99 percent), SOR (96 percent), and PSR
(98 percent)-from which USSC obtains sentencing data. According to USSC
data, a lower percentage of plea agreements (89 percent) and indictments
(87 percent) were submitted during this time period. During the period of
our review, USSC did not primarily rely on these two documents for
departure information. Table 13 shows, by circuit, the percentage of each
type of sentencing document USSC did not receive in fiscal years
1999-2001.

  USSC Receives Most of Requested Sentencing Documents

 Appendix IV: Data for Analyzing Departures and Mandatory Minimum Sentences at
                   Circuit and District Court Levels Limited

Table 13: Percent of Missing Drug Sentencing Documents, by Circuit, as
Shown in USSC's Database for Fiscal Years 1999-2001

Sentencing  National                                              
                                                       Sixth                  
    documents  average  D.C. First  Third Fourth Fifth Seventh Tenth Eleventh
                             Second                    Eighth        
                                                       Ninth         
    Judgment                                                         
       and                                                           
commitment                                                        
                                                       0.5%                   
      order        0.7% 2.0% 0.2%    1.1%  1.4%  0.5%  0.3%    0.4%      0.6%
                             0.8%                      0.4%          
                                                       1.0%          
    Statement                                                        
       of                                                            
                                                       1.9%                   
     reasons       3.7% 0.5% 4.9%    1.4%  7.4%  1.6%  2.5%    4.5%      2.4%
                             5%                        1.3%          
                                                       6.6%          
Presentence                                                       
                                                       1.3%                   
     report        1.5% 1.3% 2.3%    1.8%  1.7%  1.2%  2.3%    1.8%      0.9%
                             1.8%                      1.1%          
                                                       1.6%          
      Plea                                                           
    agreement       11% 33%  12%       5%     7%    9%   7% 8%    9%       9% 
                             50%                         5% 8%       
                                                        3% 20%                
Indictment       13% 39%  3% 75%   32%  10%      5%   3% 5%    6%       7%

Source: GAO analysis of UUSC data.

Among the 12 circuits, the rate of missing SORs-the principal document
used to determine the reason for a sentencing departure-ranged from less
than 1 percent to about 7 percent. Two of the 4 circuits in which the
highest number of drug sentences were imposed were also missing the
highest percent of their SORs-the 9th Circuit at 6.6 percent and the
Fourth Circuit at 7.4 percent. A circuit's average can mask wide
differences among the districts within the circuit. For example, the
percentages of missing SORs among districts in the Ninth Circuit ranged
from less than 1 percent to 58 percent and in the Fourth Circuit from less
than 1 percent to 20 percent.

  USSC Takes Steps to Reduce Missing Document Rate

USSC reviews the documents it receives from the district courts and
annually sends a letter to each district court identifying the cases in
which documents appear to be missing. Additionally, in its annual report,
USSC discloses the overall document submission rate for all criminal cases
for the J&C, SOR, and PSR documents. USSC also attempts to identify
guidelines cases for which the courts may not have submitted any
sentencing documents. By linking data from a database maintained by AOUSC
with the data on cases in its database, USSC develops a list of cases for
which it has not yet received documentation. USSC sends this list of cases
to the relevant district courts and asks them to review the list and
forward any documents USSC should have received.

 Appendix IV: Data for Analyzing Departures and Mandatory Minimum Sentences at
                   Circuit and District Court Levels Limited

Missing or Difficult to In addition to missing sentencing documents, the
documents USSC

received in fiscal years 1999-2001 had missing information or information
Interpret Information that was difficult to interpret. As shown in table
14, among the circuits Can Affect Analysis of departure data were missing
for 1 percent to 7 percent of drug sentences

imposed in fiscal years 1999-2001. In addition, for 4 percent to 15
percentDepartures of sentences, information was missing on whether the
safety value was used as the basis for sentencing below a mandatory
minimum.

Table 14: Percent of Missing Drug Sentencing Information on Documents USSC
                  Received, by Circuit, Fiscal Years 1999-2001

Sentencing  National                                                     
information average  D.C. First  Third Fourth Fifth Sixth   Eighth Tenth Eleventh 
                          Second                    Seventh Ninth        
 Departure                                                               
information       4%  3%   5% 5%    2%     8%    2%   2% 3%  1% 7%    5%       3% 
  Safety                                                                 
valve                                                                 
                  8% 11%  10%      13%  15%      4%   5% 7% 6% 12%    7%       4% 
  applied                 10%                                            

Source: GAO analysis of UUSC data.

Nationally, of the 72,283 federal drug sentence imposed in fiscal years
1999-2001, 3,004 (4 percent) were coded as missing information necessary
to determine whether the sentence departed from an applicable guideline
range. Of these, 2,118 sentences were missing information because the SOR
had not been received, and for 570, the SOR was received but did not
include departure information.

Missing or unclear data also limited our ability to determine when the
safety valve was used as the basis for sentencing below an otherwise
applicable mandatory minimum. For example, in our preliminary analysis, we
found that of the 11,256 federal drug sentences for which the offense of
conviction carried a mandatory minimum and fell below that minimum, about
1,600 (14 percent) were coded by USSC as falling below the applicable
mandatory minimum but not involving either the safety valve or substantial
assistance. We discussed this issue with USSC. After reviewing the
underlying documents used for coding these 1,600 sentences, USSC
determined that over 900 sentences were miscoded. These miscoded sentences
were recoded in a variety of ways, including some coded as involving the
safety valve, some coded as involving substantial assistance, some coded
as having a changed drug quantity that affected the applicable mandatory
minimum, and some coded as missing safety valve information. USSC did not
recode 681 sentences; these sentences remained coded as falling below a
mandatory minimum but involving neither the safety valve nor substantial
assistance. In addition, safety valve information was determined to be
missing from 770 sentences for which the offense of

Appendix IV: Data for Analyzing Departures and Mandatory Minimum Sentences
at Circuit and District Court Levels Limited

conviction carried and fell below a mandatory minimum. A USSC official
said that there is no specific prompt on the SOR asking for information on
the application of the safety valve or whether the offense of conviction
carried a mandatory minimum.

On the basis of our analysis, missing or incomplete sentencing information
is unlikely to affect analyses nationally or by circuit but could affect
the analysis of departures in districts where the missing documents or
information are concentrated. Missing or incomplete sentencing information
may also affect USSC's records for individual judges and thus USSC's
ability to provide accurate judge-specific sentencing analysis were
Congress to request this information under the auspices of the PROTECT
Act.

  USSC Actions to Improve Coding

Multiple Reasons Cited for Missing Documentation and Information

USSC officials told us that they have not generally followed-up with
district courts to obtain information that is missing from submitted
documents or is unclear (e.g., whether the safety valve provision was the
basis for a sentence below an applicable mandatory minimum). USSC staff
does not use information from one document to substitute for missing or
unclear information in another document. As a result of coding issues we
identified during this review, USSC plans to implement new quality control
and review procedures for sentences where information on the SOR is
missing, incomplete, or unclear. These include identifying common errors
for coding staff, using technology to develop automatic edit checks for
apparently contradictory coding information for a sentence (e.g., those
below an applicable mandatory minimum whose reason for departure is not
substantial assistance or the safety valve), and having a staff attorney
review sentences in which the coding supervisor is unable to determine the
appropriate coding.

Officials from USSC, the AOUSC, and the Judicial Conference Committee on
Criminal Law cited several reasons that sentencing documents or
information on sentencing documents were missing. First, USSC and AOUSC
officials told us that some judges do not provide all the documents, in
part because judges may be unclear whether documents under court seal or
that pertain to individuals in the federal witness protection program are
to be forwarded to USSC.

Second, USSC relies almost exclusively on the SOR to determine whether the
sentence departed, met a mandatory minimum, or involved substantial
assistance. If the SOR is missing, USSC's coding procedures require

Appendix IV: Data for Analyzing Departures and Mandatory Minimum Sentences
at Circuit and District Court Levels Limited

document analysts to record the departure status as missing, even if other
documents, such as the plea agreement, suggests that a departure may by
recommended by the government. As a result, incomplete information
prevents USSC from collecting some sentencing data, as illustrated below
by two examples drawn from drug sentences imposed during fiscal years
1999-2001:

o  	In one case, the SOR did not indicate the reason the court sentenced
the offender to 97 months-a sentence below the applicable 10-year (120
month) mandatory minimum. Without this information on the SOR, under the
coding conventions used, USSC document analysts could not record
substantial assistance as the reason that the sentence of 97 months fell
below a mandatory minimum even though the plea agreement (prepared by the
parties) and the PSR (prepared by the probation officer) indicated that a
substantial assistance motion was to have been made.

o  	In another case, the SOR stated that the court was crediting the
offender for time served but failed to state the specific amount of time
being credited. Unable to determine the amount of time being credited, and
thus the sentence length being imposed, USSC document analysts could not
determine whether the sentence departed or met an applicable mandatory
minimum.

Third, judges report the information using different versions of the SOR
forms that can make consistent interpretation more difficult. For example,
some jurisdictions provide one-page, single-spaced narratives that report
the sentence and, in rare cases, others provide a transcript of the
sentencing hearing instead of an SOR. According to USSC officials,
interpreting multiple forms that report sentencing information in
different ways and in different locations complicates the process of
coding sentencing data such as departure status and use of the safety
valve and may lead to missing sentencing information. USSC officials
stated the single most effective step towards improving the completeness
of data the courts report and USSC's ability to code it would be the
increased use of a standard SOR. The Judicial Conference at its September
2003 meeting accepted revisions to the standard SOR. The Conference
designated the revised form as the mechanism by which courts comply with
the requirements of the PROTECT Act to report reasons for sentences to
USSC. The Committee plans to encourage judges to use it through education
about the benefits of its use, but the Chair of the Committee stated that
the Committee does not believe it has the authority to require the use of
the new SOR. Officials from AOUSC and the Committee said they believe that
with additional education judges will routinely use the

Appendix IV: Data for Analyzing Departures and Mandatory Minimum Sentences
at Circuit and District Court Levels Limited

new standard SOR, resulting in more useful and higher quality data
reported to USSC.

Last, according to officials from AOUSC and the Criminal Law Committee,
judges and other court officials lack an awareness of how to complete the
SORs with a level of detail that would allow USSC to collect sentencing
information. The Committee official said that education for judges and
other court officials is needed on how to properly complete the SOR. In
addition, no feedback mechanism is in place to inform judges that
information on the SOR was incomplete or unclear to USSC and, therefore,
cases are coded as missing sentencing information. Although USSC contacts
the courts to request missing sentencing documents be submitted, it does
not provide a similar list of documents that contained information coded
as missing. Without knowing which cases are coded as missing sentencing
information, judges cannot clarify or complete information needed by USSC.

While USSC and the Federal Judicial Center offer programs and workshops on
application of the guidelines to judges and other court officials, no
education programs are provided on how to complete the SOR in ways that
provide clear, complete information. Officials from USSC, AOUSC, and
Criminal Law Committee said that education on how to apply increasingly
complex guidelines has been their focus, not educating judges and other
officials to correctly complete the SOR. Officials also said that in the
future it would be possible to provide programs at judicial workshops or
through the Federal Judicial Center that educates judges and other court
officials on how to provide clear, complete reports on sentencing.

The category "other downward departures" generally thought to represent
judicial discretion may also reflect downward departures resulting from
prosecutorial discretion and initiative. In this report we classified
departures as either "substantial assistance" or "other downward"
departures. Substantial assistance departures can be viewed as a measure
of prosecutorial discretion because only the prosecutor has the authority
to initiate and recommend to the court that an offender be given a reduced
sentence for substantial assistance to the prosecution. Neither the judge
nor defense counsel may do so. The remaining departures, "other downward
departures," are generally considered to be an indication of judicial
discretion. AOUSC officials suggested, however, that the category "other
downward departures" provides an imprecise measure of judicial discretion.
For example, AOUSC officials noted that some departures classified in USSC
database as other departures may actually arise from

  Other Downward Departures Do Not Solely Reflect Judicial Discretion

Appendix IV: Data for Analyzing Departures and Mandatory Minimum Sentences
at Circuit and District Court Levels Limited

agreements, particularly plea bargains, that either were initiated or
supported by the government. We did not confirm this statement with
federal prosecutors. USSC documents in its database up to three reasons
judges provide for an other downward departure. According to USSC database
for drug sentences in fiscal years 1999-2001, the first reason provided
for an other downward departure in 18 percent of the sentences was the
government's fast track programs;3 in 16 percent, plea agreement; and in 4
percent, deportation. Tables 15, 16, and 17 detail for drug sentences the
number and percent of other downward departures associated with the first,
second, and third reasons provided for those departures.

 Table 15: First Reason Provided by Judges for Other Downward Departure in Drug
                    Sentences Nationwide, Fiscal Years 1999-

                                          Number of other    Percent of other 
                                             downward     downward departures 
                First reason                departures     for drug sentences 
              No Reason given                          16                   * 
(5G1.3) Convictions on related counts               10                   * 
                (5H1.1) Age                            40                   * 
     (5H1.2) Educational and vocational                 1                   * 
                   skills                                 
        (5H1.3) Mental and emotional                   55                  1% 
                 conditions                               
         (5H1.4) Physical condition                   223                  2% 
    (5H1.4) Drug dependence and alcohol                   
                   abuse                               21                   * 
     (5H1.5) Previous employment record                 5                   * 
          (5H1.6) Family ties and                     360                  3% 
              responsibilities                            
           (5H1.6) Community ties                      14                   * 
     (5K1.1) Substantial assistance at                  1                   * 
                   motion                                 
    not 5K1.1 Cooperation without motion               30                   * 
        Cooperation (motion unknown)                   34                   * 
      (5K2.0) Several persons injured                   1                   * 
          (5K2.2) Physical injury                       1                   * 

(5K2.3) Extreme psychological injury 1 *

(5K2.6) Weapons and dangerous
instrumentalities 1 *

3 "Fast-track" or other early disposition programs in the southwest border
districts provide lower sentences initiated by prosecutors for low-level
drug trafficking offenses.

Appendix IV: Data for Analyzing Departures and Mandatory Minimum Sentences
at Circuit and District Court Levels Limited Appendix IV: Data for
Analyzing Departures and Mandatory Minimum Sentences at Circuit and
District Court Levels Limited Appendix IV: Data for Analyzing Departures
and Mandatory Minimum Sentences at Circuit and District Court Levels
Limited

                   Page 69 GAO-04-105 Federal Drug Sentences
                                                                                           (5K2.12)                                                                                                (2D1.1)        (2Q1.2,      (4A1.3)                 General     does not                 Significance                represents                                                    Other                                                                                                                   Guidelines                                     Put                                                                   Time or cost                                             
 Number Percent                      First              for drug (5K2.10)     (5K2.11)     Coercion        (5K2.13)          Fast Track            (2A1.1) Death       (2A6.1)                    Unusually       2Q1.3)       Pattern     (4A1.3)     adequacy     reflect                      or          Criminal       the               Pursuant               Due to           plea        Mule/Role                            Local         Adequate     purposes                        Prey to       do not      of the        No prior       defendant's   defendant's.                                                 involved in      Acceptance of                        
of        of  downward            reason departures sentences Victim's 2 *  Lesser  4 *   and    74 1% Diminished 207 2% -Immigration 1,969 18%  not caused   2 * Factors not    guideline 4 * high drug 16 *   Harm    1 *   of    2 * Pending 1 *    of      seriousness   history 7 *  similarity  6 * history    defendant's 1,169 11% to a plea 1,680 16% stipulations 10 * agreement 1 *  in the   89 1% Deportation 472 4% conditions 4 * punishment       of     82 1% Deterrence 36 *  other  3 * reflect the   offense 8 * record/first 6 * sentence in      Reduce    9 *          Lack of             defendant 5 *      the      3 * responsibility 91 1%               2 *
 other    other             downward                             conduct        harm        duress         capacity                                intentionally     incorporated                  amount        resulting     conduct      cases      criminal   of criminal                 of past        category   involvement           agreement                              reason        offense                                           to meet     sentencing                       inmates     seriousness                 offender        line with     disparity.      culpability/accountability                 investigation                          Limited/minor   
                          departures                                                                                                                                      in                                     from risk                             history;                               conduct         over-                                                                                                                                    the                                                                                                   co-                                    of                                                                    prior record    

                   Page 70 GAO-04-105 Federal Drug Sentences
                                                                                                                                                                                 Dollar                                                                                                  Not                                            Guidelines                                                                           Defendant                                                 Lower     chance to                                                                                                                                                             Source:   
 Number Percent                      First              for drug                                                          Sufficient      Nature/seriousness       First         amount                  Currently      state or                                       Military     representative        Guidelines                        too                         (5K2.0)      mitigating              Other           is a law      or ex-law      Offense                    sentence       be a                   (5K2.16)      Lack of                                   Time        Child abuse                                                               GAO     
of        of  downward            reason departures sentences Rehabilitation 191 2% Restitution 2 * Incapacitation 4 * punishment 23 *   of the offense   5 *   felony   2 * involved   (general) 2 * receiving      federal    1 *   Defendant's     character 2 *  record  1 *     of the     93 1%     too        represented 2 * low/offense   represented 1 *   General     circumstance 1,671 15% (SPECIFY) 804 7% enforcement   enforcement 1 * behavior   incident 995 9%   gives     productive   society 1 * Voluntary  7 * youthful 2 *   Delay in     concerns 2 * served 28 *     (child       syndrome) 4 * Remorse 2 * Missing/indeterminable 191 2% analysis  
 other    other             downward                                                                                                                             conviction     in crime                 punishment   jurisdiction        positive                                   "heartland"         high/offense                      level                      aggravating                                             officer       officer        was an                    defendant   member of                disclosure     guidance     prosecution;                            abuse/battered                                                           of USSC   
                          departures                                                                                                                                                                       under                       background/good                                                   level over-                      under-                          or                                                                              isolated                       a                                                            evidentiary                                 child                                                                 data.    

Note: These percentages are based on 10,891 drug sentences imposed during
fiscal years 1999-2001 for which departure information is available and
the sentence was coded an other downward departure. USSC codes up to three
reasons provided by judges as the bases for departing.

*Less than 1percent.

Appendix IV: Data for Analyzing Departures and Mandatory Minimum Sentences
at Circuit and District Court Levels Limited

Table 16: Second Reason Provided by Judges for Other Downward Departure in
Drug Sentences Nationwide, Fiscal Years 1999-2001

                   Page 72 GAO-04-105 Federal Drug Sentences
                           (Second                                     (5H1.2)        (5H1.3)                         (5H1.4)                (5H1.5)                                              not 5K1.1                                   (5K2.12)                                            (2A6.1)                     (2D1.1)       (2D1.1)      Significance        Criminal                                                                         Other                                                       Adequate                                                                                  
Second Number of   Percent Reason              No       (5H1.1)      Educational     Mental and       (5H1.4)           Drug                 Previous       (5H1.6) Family          (5H1.6)      Cooperation     Cooperation     (5K2.10)     Coercion       (5K2.13)        Fast Track         Factors not                  Unusually     Unusually          or              history           the           (4B1.1)      Pursuant           Due to          plea        Mule/role                         Local        punishment     purposes                     Charge/plea    seriousness                   
reason   drug      of drug  field  8,986 82% reason 1 *   Age   20 *     and     2 * emotional  34 * Physical  71 1% dependence   abuse 7 * employment 4 *     ties and     105 1% Community 3 *   without   7 *   (motion   4 * Victim's 1 *   and    15 * Diminished 34 * -Immigration 206 2% incorporated  * guideline 1  high drug 1 * high drug 1 *  similarity  1 *    category       defendant's 72 1%  Career  1 * to a plea 61 1% stipulations 2 * agreement 1 *  in the   44 * Deportation 39 * conditions 3 *  to meet    *     of     9  Deterrence 13 *  does not    *   of the    1    No prior   7 *
       sentences sentences  left             given                   vocational      conditions      condition          and                   record       responsibilities          ties          motion         unknown)       conduct       duress        capacity                                in                       purity        amount         of past        over-represents   involvement       offender     agreement                         reason        offense                                          the        sentencing                    reflect the      offense      record/first   
                           blank)                                      skills                                         alcohol                                                                                                                                                                                                                              conduct                                                                                                                                                                                                                                   offender     

 Appendix IV: Data for Analyzing Departures and Mandatory Minimum Sentences at
                   Circuit and District Court Levels Limited

               Second reason               Number of drug     Percent of drug 
                                              sentences             sentences 
Put defendant's sentence in line with                                    * 
      codefendant's. Reduce disparity.                    4 
Lack of culpability/accountability of                                    * 
                 defendant                                1 
             Lack of available                            1                 * 
          facilities/overcrowding                           
        Time or cost involved in the                      2                 * 
               investigation                                
        Acceptance of responsibility                     18                 * 
               Rehabilitation                            46                 * 
               Incapacitation                             1                 * 
           Sufficient punishment                         12                 * 
          First felony conviction                         1                 * 
    Defendant's positive background/good                                    * 
                 character                                2 
              Military record                             1                 * 
Not representative of the "heartland"                 18                 * 
       (5K2.0) General aggravating or                       
                 mitigating                                 
                circumstance                            282                3% 
              Other (SPECIFY)                           160                1% 
                  Unknown                                 2                 * 
      Offense behavior was an isolated                  567                5% 
                  incident                                  
       (5K2.16) Voluntary disclosure                      1                 * 

Lack of youthful guidance 6 *
Time served 2 *
Child abuse (child abuse/battered child *
syndrome) 3
Missing/indeterminable 4 *

Source: GAO analysis of USSC data.

Note: These percentages are based on 10,891 drug sentences imposed during
fiscal years 1999-2001 for which departure information is available and
the sentence was coded an other downward departure. USSC codes up to three
reasons provided by judges as the bases for departing.

*Less than 1percent.

Appendix IV: Data for Analyzing Departures and Mandatory Minimum Sentences
at Circuit and District Court Levels Limited

Table 17: Third Reason Provided by Judges for Other Downward Departure in
Drug Sentences Nationwide, Fiscal Years 1999-2001

                   Page 74 GAO-04-105 Federal Drug Sentences
                           (Third                          (5H1.2)        (5H1.3)                       (5H1.5)                                                          (5K2.12)                                      (2D1.1)         Criminal                                                                                                                      Put                                                                                      Defendant's            Not             (5K2.0)                                      Offense                    
Third  Number of   Percent Reason            (5H1.1)     Educational     Mental and      (5H1.4)        Previous       (5H1.6) Family       Cooperation     (5K2.11)     Coercion      (5K2.13)       Fast Track      Unusually         history        defendant's    Pursuant      Mule/role                       Local                         No prior       defendant's    codefendant's.    Acceptance of                                                positive         representative       General                         Other        behavior                 
reason   drug      of drug field  10,705 98%   Age   7 *     and     1 * emotional  4 * Physical  15 * employment 3 *     ties and     20 *   (motion   2 *  Lesser  1 *   and    3 * Diminished 6 * -Immigration 5 * high drug 1 *    category      * involvement 9  to a plea 6 *  in the   8 * Deportation 3 * conditions 1 * Deterrence 3 * record/first 3 * sentence in  *     Reduce     1  responsibility 3 * Rehabilitation 12 * Incapacitation 7 * background/good 1 *     of the     6 * aggravating  * circumstance 17  (SPECIFY) 13 *  was an  24 * Lack of  1 *
       sentences sentences  left                         vocational      conditions     condition        record       responsibilities       unknown)         harm        duress       capacity                        amount       over-represents                   agreement      offense                                                      offender        line with       disparity                                                                    character         "heartland"           or                                         isolated      youthful   
                           blank)                          skills                                                                                                                                                                         the                                                                                                                                                                                                                                                      mitigating                                     incident      guidance   

                       Source: GAO analysis of USSC data.

Note: These percentages are based on 10,891 drug sentences imposed during
fiscal years 1999-2001 for which departure information is available and
the sentence was coded an other downward departure. USSC codes up to three
reasons provided by judges as the bases for departing.

*Less than 1percent.

                   Page 75 GAO-04-105 Federal Drug Sentences

Appendix V: Comments from U.S. Sentencing Commission

Appendix VI: Comments from the Judicial Conference Committee on Criminal Law

Appendix VI: Comments from the Judicial Conference Committee on Criminal
Law

  Appendix VI: Comments from the Judicial Conference Committee on Criminal Law

                       Copy of testimony is not included.

Appendix VI: Comments from the Judicial Conference Committee on Criminal
Law

Appendix VI: GAO Contacts and Staff Acknowledgments

GAO Contacts

  Staff Acknowledgments

(440188)

William O. Jenkins, Jr. (202) 512-8757 David Alexander (202) 512-4223

In addition to the persons named above, the following persons made key
contributions to this report: William W. Crocker, III, Christine Davis,
Barbara Hills, David Makoto Hudson, E. Anne Laffoon, William Sabol, Doug
Sloane, Wendy Turenne.

  GAO's Mission

Obtaining Copies of GAO Reports and Testimony

The General Accounting Office, the audit, evaluation and investigative arm
of Congress, exists to support Congress in meeting its constitutional
responsibilities and to help improve the performance and accountability of
the federal government for the American people. GAO examines the use of
public funds; evaluates federal programs and policies; and provides
analyses, recommendations, and other assistance to help Congress make
informed oversight, policy, and funding decisions. GAO's commitment to
good government is reflected in its core values of accountability,
integrity, and reliability.

The fastest and easiest way to obtain copies of GAO documents at no cost
is through the Internet. GAO's Web site (www.gao.gov) contains abstracts
and full-text files of current reports and testimony and an expanding
archive of older products. The Web site features a search engine to help
you locate documents using key words and phrases. You can print these
documents in their entirety, including charts and other graphics.

Each day, GAO issues a list of newly released reports, testimony, and
correspondence. GAO posts this list, known as "Today's Reports," on its
Web site daily. The list contains links to the full-text document files.
To have GAO e-mail this list to you every afternoon, go to www.gao.gov and
select "Subscribe to e-mail alerts" under the "Order GAO Products"
heading.

Order by Mail or Phone 	The first copy of each printed report is free.
Additional copies are $2 each. A check or money order should be made out
to the Superintendent of Documents. GAO also accepts VISA and Mastercard.
Orders for 100 or more copies mailed to a single address are discounted 25
percent. Orders should be sent to:

U.S. General Accounting Office 441 G Street NW, Room LM Washington, D.C.
20548

To order by Phone: 	Voice: (202) 512-6000 TDD: (202) 512-2537 Fax: (202)
512-6061

Contact:

To Report Fraud, Web site: www.gao.gov/fraudnet/fraudnet.htm

  Waste, and Abuse in E-mail: [email protected]

Federal Programs Automated answering system: (800) 424-5454 or (202)
512-7470

Jeff Nelligan, Managing Director, [email protected] (202) 512-4800

Public Affairs 	U.S. General Accounting Office, 441 G Street NW, Room 7149
Washington, D.C. 20548
*** End of document. ***