Truth-in-Sentencing in Virginia, Ostrom et all, 2001

Truth-in-Sentencing in Virginia, Ostrom et all, 2001

• Feb. 22, 2016 • Locations: United States of America -> Virginia • Topics: Sentencing

Download original document:
Brief thumbnail

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

The author(s) shown below used Federal funds provided by the U.S.
Department of Justice and prepared the following final report:
Document Title:

Truth-in-Sentencing in Virginia

Author(s):

Brian J. Ostrom, Fred Cheesman, Ann M. Jones,
Meredith Peterson, Neal B. Kauder

Document No.:

187677

Date Received:

April 5, 2001

Award Number:

96-CE-VX-0005

This report has not been published by the U.S. Department of Justice.
To provide better customer service, NCJRS has made this Federallyfunded grant final report available electronically in addition to
traditional paper copies.

Opinions or points of view expressed are those
of the author(s) and do not necessarily reflect
the official position or policies of the U.S.
Department of Justice.

This document is a research report submitted to the U.S. Department of Justice. This report
has not been published by the Department. Opinions or points of view expressed are those
of the author(s) and do not necessarily reflect the official position or policies of the U.S.
Department of Justice.

Prepared for the
National Institute of Justice

Authored by
Brian J. Ostrom, Project Director
Fred Cheesman
Ann M. Jones
Meredith Peterson

Nationd Center for State Courts

Funded by the National Institute of Justice

N]
-ma

NCSC

ISBN 0-89656-2034

,t+
This report was developed under a grant from the National Institute of Justice
(Grant 96-CE-VX-0005).The opinions and points ofview in chis report are those
of the authors and do not necessarily represent the official position or policies of
the National Institute of Justice or the Virginia Criminal Sentencing Commission.

Table of Contents
.-

Executive Summary ...............................................................................................................
Chapter 1

Introduction

.....................................................................................................................

Chapter 2

T h e Path to Reform .........................................................................................................

Chapter 3

The Design 0fTI.S Guidelines in Virginia ........................................................................

Chapter 4

The Impact ofTIS on Prison Population in Virginia ........................................................

Chapter S

The Impact ofTIS on Judicial Compliance ......................................................................

Chapter 6

Estimating Preventable Crime Under TIS ........................................................................

Chapter 7

Assessing the Impact ofTIS on Recidivism

......................................................................

Chapter 8
Conclusion ......................................................................................................................

Bibliography ........................................................................................................................

Acknowledgements
This project benefited greatly from the advice and guidance of many individuals.
First and foremost, we wish to thank everyone associated with the Virginia Criminal
Sentencing Commission (VCSC). Their cooperation and participation enabled us
to gather the data and information required to complete this study. In particular, we
gratefully acknowledge the support of the Honorable Ernest I? Gates, Chairman of
the VCSC, and Richard I? Kern, VCSC Executive Director. This project drew heavily

on the insight and knowledge of the VCSC staK Meredith Farrar-Owens helped
immensely with the design and interpretation of the recidivism study. National Center for State Courts project staff also thank Carolyn Williamson, James C. Creech,
Jody T Fridley, and Ann A. Jones for their willingness to meet and assist the evaluators throughout the life of the study.
George Allen (Governor 1994-1998), Frank Atkinson (General Counsel to the Governor), and Richard Cuflen (Attorney General) all graciously made time to participate in project interviews. Their candor and willingness to answer detailed questions
were important for documenting how and why sentencing reform occurred in Virginia. In addition, several officials from the Department of Corrections helped staff
during the study of the Offender Notification Release Program (ONRP). Patrick
Gurney, Classification and Records Manager, was instrumental in helping project
staff reach representatives from throughout Virginia's correctional field units.
We also wish to acknowledge the considerable contribution of our National Center
for State Courts colleagues Neil LaFountain and Margaret Fonner for their assistance
with data analysis and project administration. This publication also benefited from
the careful editing of David Morrill. The graphic design and report publishing was
provided by Lorraine HOE
Finally, the support of the National Institute of Justice must be recognized. In particular, the encouragement and monitoring provided by Jordan k i t e r and Janice
Munsterman brought the project to a happy conclusion. The National Center for
State Courts and the Virginia Criminal Sentencing Commission thank NIJ for recognizing the need to document and evaluate the major shifts in sentencing practice
and policy that have occurred in Virginia.

EXECUTIVE S U M M A R Y

Reform and Truth-inSentencing in Virginia
Truth-in-sentencing (TIS) is the most prominent sentencing reform movement of
the 1990s. The primary objective ofTIS is to more closely align the sentence imposed by the judge with the actual amount of time served in prison by restricting or

eliminating parole eligibility and good time. In many instances, these reforms are
accompanied by significant increases in the penalties for violent offenders. TIS is
based on a “just deserts” philosophy in which sentence length varies directly in proportion to the severity of the offense and allocates penalties as a deserved punishment rather than as a means for rehabilitation and treatment.
This report is the result of an 18-month partnership project funded by the National Institute of Justice to evaluate the development and impact ofTIS in Virginia.
The successful completion of this project required both intimate knowledge of the
process underlying the changes to Virginia sentencing law and the capacity to conduct the evaluation in an appropriate and impartial manner. For this reason, a partnership was developed to bring together the historic and institutional knowledge of
rhe Virginia Criminal Sentencing Commission (VCSC)
and an objective, third-party
evaluation team from the National Center for State Courts (NCSC).
The evaluators view the purposes and features of sentencing reform in Virginia
as given, and no value judgments are made about the goals of TIS. No effort is
made to advocate specific sentencing structures and strategies. As such, the purpose of this evaluation is to (1) analyze the approach used in Virginia to develop
and implement one of the nation’s pioneering efforts in TIS, inchding the abolition of parole and the iniriative to increase prison sentences for violent offenders;

(2) critically evaluate the analyses conducted to forecast the impact of TIS on sentencing outcomes and prison population; and (3) begin the process of conducting

an evaluation of the impact of Virginia’s sentencing reforms on recidivism among
violent offenders. With the exception of the recidivism analysis, all analyses referred to in this report were conducted by the VCSC. The role of the NCSC was to
evaluate the work of the VCSC.

Central findings Include:
a TIS became effective in Virginia on January 1, 1995. Virginia’s sentencing reforms

abolished parole, reduced good time allowances to ensure that inmates serve a
minimum of 85% of their imposed sentence, and increased prison sentences for
violent and repeat offenders.
m Virginia, a long-time proponent of structured sentencing, implementedTiS through
a revision of the state’s existing voluntary system of sentencing guidelines. The

benefit of the sentencing guideline approach is that it allows for a more accurate
assessment of the likely impact of changes in sentencing and/or parole policy.

Guidelines systems are arguably the most cost-effective means of providing racional structure, relevant data, and the ability m accurately monitor and forecast
D

sentencing outcomes.
Along with the federal government, Virginia is one of eight stares that have abolished parole and implemented TIS legislation that requires almost all violent and
nonviolent offenders to serve 85% of the imposed sentence. Under TIS, violent
and repeat offenders receive sentences two to six times longer than previously. The

amount of time served by nonviolent offenders was not changed by the move to

TIS. Judge-imposed sentences for nonviolent offenders are lower under TIS, but
the time served in prison remains the same because sentences are no longer re-

duced dramatically by parole and good time allowances.
D

Under TIS, offenders are expected to serve an average of 89.7% of the judicially
imposed sentence. Although parole was abolished for all offenders convicted after
January 1, 1995, parole remains in effect for individuals incarcerated prior to TIS
reform. The parole grant rate (for eligible offenders) has dropped from 46% in

1991 to 5% in 1998.
The judicial sentencing recommendations under Virginia‘s TIS guidelines remain
voluntary, but are usually followed by judges. Currendy, judicial compliance rates
are equal to or exceed overall pre-TIS guideline compliance rates of 78%.
Jury trial rates, predicted by some to rise as a result ofTIS, have fallen steadily over
the past 12 years. The most significant drop came at the time when bifurcated
trials and TIS were implemented. Jury trials currently make up less than 3% of
felony dispositions.
m Analysts in Virginia forecast that more than 26,000 violent and 94,000 nonviolent felonies are expected to be averted between 1995 and 2005 by the passage of
TIS-a proposition that was important for building institutional support for sentencing reform, Evaluators found that while analytically complex, the methods
employed for determining preventable crime were conceptuallysound and conservative in their estimates.
Prison population under TIS has been lower than originally forecasted. Evaluators
cite several possible reasons for these overestimates,including lower-than-expected
crime rates and inaccurate estimates of new admissions to prison.
n A deterrence effect is one way for TIS to reduce recidivism in Virginia. The Offender Notification Release Program (ONRP) is designed to educate inmates leaving Virginia prisons specifically about TIS reforms. AI1 inmates leaving the prison
system are given a type of “exit interview” during which they are informed about
the abolition of parole and the old good conduct credit system. Each departing
inmate receives a wallet-sized “notification card” that contains the possible sentencing consequences of being arrested and convicted of a new felony offense.

Tntb-in-Smtencingin Wrginia

As part of a long-term recidivism analysis, project d h a v e establiihed baseline recidivism races for the offender population rdeased from prison prior to the introduction

ofTIS. Half (49.3%)ofall offenders released from prison in 1993were re-arrested
for any new crime within three years. The number of persons who recidivate drops
quickly as the measure of recidivism becomes more conservative (e.g., of rhose
released from prison, 22% were reconvicted of a new felony).
Recidivism, if it does occur, is likely to happen sooner rather than later. For those
who recidivate, the average time until first re-arrest for any crime was about 12
months, and 75% recidivate within 19 monrhs.

rn Property offenders have the highed rates of recidivism, followed by drug offenders, then violent offenders. There is some evidence of offense specialization for
property and drug offenders: 75% of those re-arrested for a property offense were
originally incarcerated for a property crime and 59% of those re-arresred for a

drug offense were originally in prison for a drug crime.

Executive Summary

CHAPTER ONE

Introduction
R e f o r m and Truth-in-Sentencing in Virginia
Strategies for reducing violent crime dominated Virginia politics during the 1993
gubernatorial race. George Men, the republican candidate, made the elimination of
parole and the institution of harsher punishment for violent offenders the centerpiece of his campaign. After winning the election, Allen established the Sentencing
and Parole Abolition Commission, which moved quickly to recommend that Vir.-

ginia establish Truth-in-Sentencing (TIS) through a major restructuring of the state’s
existing system of sentencing and parole. Determining the exact dimensions of sentencing reform occupied the political process throughout the first nine months of
the Allen administration, and at a special session of the General Assembly in September, 1994, Virginia’s legislature passed the most significant and comprehensive sentencing reforms in the state’s history.
These reforms, which became effictive on January I, 1995, were designed to achieve
three objectives:
m

Increase prison terms for violent and repeat offenders;
Abolish parole;

rn Reduce aHowances of “good time” to ensure that inmates serve

85% of their im-

posed sentence.
The abolition of parole and the restructuring of good time were accomplished by
statute. In addition, the legislature created the Virginia Criminal Sentencing Commission (VCSC) to oversee the development, implementation, and maintenance of
TIS guidelines. It became the responsibility of the VCSC to “retool” Virginia’s exist-

ing guidelines so that violent. and repeat offenders would now receive significantly
harsher penalties. But the purse strings were not completely loosed. The legislative
mandate to the VCSC also required that the demand for prison space under the new
“hard time for hard crime” sentencing strategy be fiscally responsible. The VCSC used
a reasoned and innovative approach to both increase incarceration periods for violent
offenders and keep control over prison expenses under the new TIS guidelines.
What is Truth-in-Sentencing?
Truth-in-sentencing policies are designed to ensure that the amount of time an
offender actually serves in prison is closely aligned with the original judicially imposed sentence.’ Many states seek to achieve this goal by significantly iestricting or
Although the term truth-in-sentmcingcame to prominence in the 1990s,jurisdictions began
moving in that direction in the early 1980s.The first TIS law was passed in Washington State
in 1984. Congress mandatedTIS at the federal level with the SentencingReform Act of 1984,
which established a sentencing commission as an independent agency to recommend prescriptive sentencing guidelines, to eliminate parole, and to require that inmates serve at leasr
85% of their sentence (good time would be limited to 54 days per year). Discretionaryparole’

Titb-in-Sentencing in Virginia

eliminating parole eligibility and good-time credits. The precise definition of “significant restrictions” has been strongly influenced by rhe federal government. Under
the 1994 crime bill,’ Congress authorized incentive grants to states for construction
or improvement of correctional facilities to “free conventional prison space for the
confinement of violent offenders, LO ensure that prison cell space is available for the
confinement of violent offenders, and to implement truth-in-sentencing laws for
sentencing violent offenders.” To qualify for the TIS grants, states must require that
violent offenders will serve at least 85% of the imposed prison sentence.
The 85% rule has become so commonplace that in 1998 27 stares (including
I

Virginia) qualified for the federal grant program.j-Though eligible for TIS Incentive
Grants, many of rhese states have indeterminate sentencing systems; serving 85% of
the minimum term in a sentence of 5 to 20 years would satisfy the TIS requirements
of the federal legislation. A more conservative definition of TIS calls for sentences
imposed in a guidelines or determinate sentencing structure where the 85% calculation can be made on a definite or “fixed sentence. States also differ in the scope of

TIS legislation. In many states TIS applies only to violent offenders. The federal
government and eight states, including Virginia, apply an 85% TIS requirement to
all felony offenders. This definition reflects the philosophy of TIS that all offenders
serve a prison term that is closely aligned with the original sentence.
Proponents argue that TIS policies restore public confidence in the criminal justice system and further such concepts as predictability, proportionality, deterrence,
victims’ rights, and consistency in the sentencing process. TIS is deeply rooted in the
determinate sentencing philosophy that dominated the 1980s. Generally, the determinate model holds that the authority to set sentence length resides with the court
and that sentences should be served in full. Only modest reductions in sentence
length based on satisfactory behavior while incarcerated (good time) are acceptable.
The determinate model is based on a “just deserts” philosophy in which sentence
length varies directly in proportion to the severity of the offense and, to a lesser
extent, prior criminal history. The “just deserts” model emphasizes allocating scaled
pendries as a deserved punishment rather than as a means for rehabilitation and
treatment.4 This philosophy contrasts with indeterminate models that split authority over final sentence length between the court and the department of corrections.
Under an indeterminate system, the court typicallysets a minimum sentence in conjunction with a statutorily determined maximum sentence, with the actual release
date determined by the parole board.’
Opponents claim that TIS reforms are simply the latest in a long line of ill-conrelease was first abolished in Maine in 1975 (inmates in Maine currently serve 50 to 67% of
their sentences based on good-time accrual). For more on stare and federal reform efforts, see
Ostrom, Kauder, Rottman, and Peterson (1998) and Greenfeld, Beck, and Gilliard (199G).
*Violent Crime Control and Law Enforcement Act of 1994.
)Ditron and Wilson (1999).
4 V ~ Hirsch
n
(1976).
Wilkins, Newton, and Steer (1 993).

Introduction

ceived “get tough on crime“ policies. This camp argues that some discretionary release mechanism should be retained by a paroling authority and that, in the long
term, incarcerating offenders for longer periods of time simply wastes resources and

will have little positive effect on public safety, The National Council on Crime and
Delinquency (NCCD), for example, holds that the main factor for deciding release
time should be an updated assessment of the inmate’s risk to the community-determined once a specified fraction of the custodial term has been served. NCCD also
supports retaining the possibiliry of parole for serious offenders given maximum
terms or life sentences! Critics also contend that TIS leads to creative, if not deceptive, charging and sentencing practices. Disparity may result from selective
prosecutorial charging practices, or if pleas in certain jurisdictions are obtained by
threatening to charge a particularly harsh statutory provision?
The optimal design of a just and equitable sentencing system that also makes
efficient use of public resources will long be argued. People will continue to disagree
as to whether particular sentencing policies are good or bad. What can’t be argued is
that the implementation ofTIS in Virginia has had a substantial impact on judicial
sentencing practices and corrections policies.
Evaluating the Wrginia Experience
with Wuth-in-Sentencing
Early in 1997, the Virginia Criminal Sentencing Commission (VCSC) agreed to
participate in a systematic evaluation ofVirginia‘s new TIS reforms to be conducted
by the National Center for State Courts (NCSC). To answer the fundamental question, “What impact did the implementation ofTIS have on sentencing in Virginia?”
the evaluators examine sentencing in Virginia from 1980 through the first three
years ofTIS reform (January, 1995, to January, 1997).The evaluation findings cover
three distinct aspects of sentencing reform in Virginia and incorporate both a process
and outcome orientation.
First, the evaluation focuses on the process by which the new TIS system was
developed. In so doing, we define TIS and clarif) precisely what TIS was meant to
accomplish in Virginia. For the judiciary, the cornerstone of the 1993 sentencing
reforms was a major redesign of the existing sentencing guidelines. Prior to reform,
Virginia employed a set of voluntary, descriptive guidelines that, in combination
with existing parole policies, ensured that the sentence imposed would be very different from the sentence actually served. Under TIS, parole was abolished and new
guidelines were configured to more closely align imposed sentences with actual time
served. This section examines the deliberations of the Governor’s Commission and
the legislative committees responsible for implementing the ultimate design of sentencing reform in Virginia. Specific questions addressed in Chapters 2 and 3 include:

‘National Council on Crime and Delinquency (1992).
‘Tonry (1996).

Truth-in-Sentmcingin Krgznia

rn How has sentencingreform evolved in Virginia since 1980?Where doesTIS fit within

the historical antext of sentencing reform in Virginia? What features characterized
Virginia’s sentencing guideline system prior to TIS?What operational and political factors contributed to the adoption or rejection of specific reform components

and policies?
m What is the precise design of

Tis in Virginia?What is the current status of parole

and good time in Virginia?How are violent and nonviolent offenders created under
Virginia‘s TIS? How were the new guidelines and sentencing ranges developed?
Second, the evaluation analyzes the effect of the TIS reforms against a set of ex-

plicit and implicit standards. The Governor and the Virginia legislature believed
that judicial compliance with the new TIS guidelines would have two specific results:

I) relatively little change in the actual time served by nonviolent offenders; and
2) a need for more prison space due to significant increases in prison sentences for
violent offenders. Also, they hoped that longer prison sentences for violent and repeat offenders under TIS would reduce violent crime and lead to fewer victims and
lower costs of crime. VCSC staff conducted numerous analyses to estimate the costs
associated with the implementation ofTIS as well as the benefits of crime prevented
under the new system. This stage of the evaluation assesses the outcomes of TIS
against the expectations of the system designers. In addition, the methods used by
the VCSC to forecast the potential impact ofTIS on sentencing practices and corrections resources are reviewed and critiqued. Specific questions addressed in chapters
5 4 , and 5 include:

What is the impact of TIS on prison population? What techniques were used to
forecast prison popularion under TIS? What was the estimated impact of TIS?
How accurate was the forecast?
m What is the impact ofTIS on judicial compliance?How is judicial compliance mea-

sured? Has judicial compiiance changed with the introduction of TIS? How does
compliance in jury sentencing compare with compliance in nonjury sentencing?
m How much new crime is prevented by the harsher penalties under TIS? How did

Virginia estimate the level of preventable crime under TIS?What is rhe estimated
“cost of crime” avoided through extended incarceration of violent offenders? IS
there a beneficial “incarceration effect?”
Third, this evaluation includes the first half (or baseline) of a recidivism analysis
for use in assessing the impact of TIS. The full recidivism study will be designed to
compare recidivism of inmates released one year prior to the inception of the new
sentencing laws with that of inmates released under TIS. However, because it is still
too early to conduct an effective evaluation of the impact

of TIS on the rate of

recidivism of violent offenders, only the first haffwill be completed during this evaluation. At this stage, the NCSC, in close collaboration with the VCSC, has examined
the background characteristics and prior conviction hisrories of offenders released
from Virginia prisons in 1993. Records were then examined to determine whether
offenders had been re-arrested or re-convicted wirhin three years of their release from
Inwoduction

prison. Multiple measures of recidivism are calculated and discussed.

In addition, this stage of the evaluation also examines the creation and implementation of a recent VCSC innovation, the Offender Notification Release Program
(ONRP), which was implemented early in 1997. The ONRP is intended to enhance
the specific deterrent effect of the tougher sentencing laws by informing inmates
what their likely sentence will be if they commit other crimes after their release.
Specific questions addressed in chapters G and 7 include:
m What were the patterns of recidivism prior to the implementation ofTIS?How is
recidivism calculated and measured? How was the necessary data gathered?What
is the rate of recidivism for offenders released prior to the 1994 reforms?
a What is the design and purpose of the ONRP? How does Virginia attempt to
educate inmates about to exit state correctional facilities about TIS reforms? How
has the ONRP been implemented by the Department of Corrections?
In summary, the purpose of this evaluation is (1) to analyze the approach used in
Virginia to develop and implement one of the nation’s pioneering efforts in TIS,
including rhe abolition of parole and the initiative to increase prison sentences for
violent offenders; (2) to critically evaluate the analyses conducted by rhe VCSC to
forecast the impact ofTIS on sentencing outcomes and prison population; and (3) to
begin evaluating the impact of Virginia’s sentencing reforms and Offender Notification Release Program (ONRP) on recidivism among violent offenders.
Who Benefits from this Evaluation?
The evaluators view the purposes and features of sentencing reform in Virginia as
given, and no value judgments are made about the goals ofTIS. No effort is made to
advocate specific sentencing structures and strategies.As such, the general objectives

of this evaluation are (1) to increase our knowledge about the various sentencing
policy alternatives considered in Virginia and (2) to ciarifjr the outcome of particular
choices. The knowledge gained from this approach is primarily designed to benefit
Virginia policymakers and practitioners interested in an objective analysis of the
development and implementation of the new sentencing reforms in their state. However, given the ongoing interest in sentencing reform elsewhere, especially in TIS and
abolition of parole, there is considerable nariond interest in Virginia’s experience.
Additionally, an understanding of how sentencing reform operates in practice may
help others advocate policies in sync with their objectives. Hence, this evaluation has
been designed and written to clarify how sentencing reform efforts could be improved if initiated in other states.
Because many policymakers agree with the objectives ofTIS, it is easy to overlook
how outcomes might differ from intent. Desired objectives are not the same as workable solutions. For example, other states contemplating TIS reforms may benefit from
a description and analysis of how Virginia (I) determined its new sentencing ranges

under TIS, preserving historical time-served amounts for nonviolent offenders while
increasing time served for violent offenders; (2) estimated the probable impact of its

Titb-in-Sentenring in V;:rgipzia

sentencing reforms on avertable crime and the need for additional prison space; and

(3) is measuring the impact ofTIS on recidivism. Sound analysis will help policymakers
evaluate more accurately whether a sentencing policy alternative will, in fact, accomplish the desired outcome.
VCSC/NCSC Evaluation Partnership
The evaluation focuses on the process of sentencing reform in Virginia and criti-

cally examines the primary analyses and impact assessments conducted on behalf of
the Governor’s Commission as well as the legislative committees involved in sentenc--

ing reform. The majority of these analyses were conducted by the Criminal Justice
Research Center (CJRC) within the Department of Criminal Jusrice Services. Several key staff of the CJRC, including the director Richard Kern, accepted permanent
positions at the VCSC when it was established officially on January 1 , 1935.’ The
studies referenced and reviewed in this evaluation were collected from the files held
at the current VCSC and were found in their original formats as printouts, graphical
presentations, and various types of information and report packets (sometimes termed
“fugitive”research and analysis).
The successful completion of this project required both intimate knowledge of the
process underlying the changes to Virginia sentencing law and the capacity to conduct the evaluation in an appropriate and impartial manner. For this reason, a partnership was developed to bring together the historic and institutional knowledge of
the VCSC with an objective, third-party evaluation team ftom the NCSC. The part-

ners believe that the best (and arguably only) way to ensure that this evaluation had
access to the necessary data and program documentation underlying Virginia‘s implementation process was to involve the VCSC and its staff throughout the evaluation
process. VCSC involvement included identifying the fundamental issues that drove
sentencing reform; assisting in gaining access to and preparing databases; claritjling
any data problems, details, and nuances; and providing evaluators with other relevant information that affected Virginia’s sentencing reform egorts. Ongoing communication between the NCSC and the VCSC helped close important gaps in the
evaluation. At the same time, while cooperation between the VCSC and the evaluators was critical during certain stages, the evaluation team also acted independently.
The NCSC evaluation team was given a free hand to design and conduct the evaluation and, as a consequence, bears responsibility for the evaluarion results.

8Given the considerable overlap of key staff at the CJRC (prior to 1995) and at the VCSC
(after 1995), this evaluation uses the shorthand of VCSC to refer to research and analysis
conducted by both the CJRC and the VCSC.

Introduction

CHAPTER T W O

The Path to Reform
Virginia has been actively involved in sentencing research and reform since the
early 1980s. Initially driven by concern over sentencing disparity, Virginia has been a
consistent innovator and strong proponent of the sentencing guideline concept. The
new TIS guideline structure is better understood when presented in the context of
earlier reform efforts. In reviewing sentencing reform in Virginia over the past two
decades,’ this chapter also underscores the critical importance of relevant data and
effective staff to explain how decisionmaking during the 1994 reform process could
be both informed and fast. One fundamental, though often under-appreciated, component of rational sentencing reform is rhe creation and maintenance ofa sentencing
database. Virginia policymakers recognized that detailed and accurate information
on past sentencing practices greatly enhances a state’s ability co design and implement a specific set of sentencing reforms-and

accurately estimate the possible im-

pacts and associated cost. In addition, staff must have the capacity to knowledgeably
assess and explain the expected differences between alternative reform packages. The
rapidity with which the new TIS system was developed and approved by the legislature (as compared to many other states adoptingTIS policies) was directly related to
the extensive VCSC staff experience with sentencing-relared research prior to 1994.

Early 1980s-Beginnings of Reform in Wrginia
In 1982, Governor Charles S. Robb appointed the Task Force on Sentencing to
study current sentencing policies and to recommend changes if appropriate. This
study followed a series of newspaper articles and reports claiming the inconsistency
and disparity of sentencing decisions in Virginia. The Task Force issued a final report
in 1983, concluding that variation in the use of incarceration and length of prison
terms for similarly situated offenders did exist across Virginia.’ These differences
were found to be partially attributable to such factors as offender race, socioeconomic status, and location of the court. Based on these conclusions, the task force
recommended that the Supreme Court of Virginia take steps to improve statewide
consistency in sentencing through the development of historically based (or “descriptive”)sentencing guidelines.” The guideline concept did not have unequivocal
support among the Virginia judiciary. In the absence of judicial oversight of the
study methods and procedures, many judges were reluctant to accept earlier findings

of unwarranted sentencing disparity. Concern centered on the belief rhat disparity
studies conducted by the Governor’s Task Force and the Richmond Times-Dispatch

’Governor’sTask Force on Senrencing (1 983).

l o Similar findings/condusions had been reached in several other stares (Minnesota, Pennsylvania, Washington, Michigan), all of which established sentencing guidelines as a possible
remedy for disparate sentencing decisions.

9 0

Tmtb-in-Sentencingin Wgiaiu

were based on a nonrepresentative sample of cases and that not enough factors were
used to develop a rigorous statistical analysis of sentencing practices.” At this point
the Executive Committee of the Judicial Conference of Virginia decided that a more
comprehensive profile of sentencing in Virginia was necessary before appropriate
sentencing guidelines could be developed.
1984-1985 Building a Database
In 1984, the Secretary of Public Safety authorized the development of a fdly auto-

mated Pre-sentence Investigation (PSI) system for collecting detailed information on
--

almost all felony convictions. At that time, no database existed in Virginia to capture
the offense and prior record information needed to conduct a comprehensive analysis
of sentencing. Initially, this database would provide information for a statewide disparity study and, if required and appropriate, would serve as the basis for descriptive sentencing guidelines. Descriptive guidelines are based on actual past sentencing practices
of judges. Suggested sentences under this style of guideline reflect a careful analysis of
the sentences actually imposed by judges for particular combinations of offense and
offender characteristics. The goal is to eliminate the atypical or unusual sentence (e.g.,
the high and low extremes at both ends of the sentencing spectrum).

A key to understandingstatewide sentencing practice is having valid and reliable data on
past sentencingoutcomes. VCSC staffindicate that the lack ofsuch data made many judges
wary of previous sentencing disparity studies as well as the process of guideline development
in other states. Since Virginia’s guidelines were to be purely descriptive, their quality and
appropriateness would be tied directly to the data that underlie their development.
Pre-and post-sentence investigation (PSI) reporting formats were redesigned to
measure 212 objectively coded offense and offender variables.” Critical to the success ofVirginia’s PSI database was the adoption of standard codes for probation and
parole officers to record offense-specific information. These Virginia Crime Codes
(VCCs) are a nine-digit alpha-numeric offense identification system based on the

Code ofVirginia and include approximately 1,300 misdemeanor and felony crimes.
This new system replaced the use of “free format” descriptions (i.e., unstructured,
longhand attempts to describe the nature of past and current convictions).The VCC
database is maintained by the sentencing commission and is updated annually to
reflect changes in statute or the addition of new laws. The VCC database includes
the following information on each crime in the Code o f Virginia:
a unique Virginia Crime Code (VCC);
m a concise offense description, guided by the elements of the offense;

rn the Code of Krginia section correspondingto the VCC;

the statutory penalty range;
the State Police and Department of CorrectionsNCIC code correspondingto the VCC.
” “Justice For All,” (1983). This study examined sentences handed down for robbery cases
and found the existence of unwarranted sentsncing disparity.
”Roughly 20,000 new cases have been added to the PSI database each year since 1985; the
current system contains about 200,000 cases.

The Path to RefDnn

The PSI system and the establishment of the VCC coding system is particularly
noteworthy for this evaluation because this rich source of information underlies al-

most all sentencing guidelines data analysis, research, and monitoring from 1983 to
the present. Given the importance of the VCC system and PSI database, the state
should be cognizant of at least two related issues. First, consideration should be given
to developing a unique offender identifier to be used consistently across Virginia's
numerous criminal-justice-related data systems. Such an identifier would ease significantly the effort and cost associated with merging the PSI database with additional sources of data. For example, without a unique identifier, ie is currently dificult to supplement PSI data with criminal history information for analysis related to
--

recidivism, juvenile justice, or risk assessment. Second, the VCSC must be diligent
and clear in communicating their rationale for maintaining the PSI database. One
method of preserving the PSI database is by initiating and supporting efforts to improve efficiency through automation and quality control. Otherwise, efforts to scale
down or even eliminate the PSI data collection citing the ongoing expense required
to collect, edit, and sustain the system may surface. As compared to nonguideline

states, an advantage for Virginia (and other states that have developed and maintained guidelines) is the substantial collection of reliable data sources. The lesson
learned is that any meaningful attempt at structured guidelines development must
be accompanied by improvements in existing data systems.
19864987 Statewide Disparity Study
In April of 1986, the Chief Justice of the Virginia Supreme Court appointed the

Judicial Sentencing Oversight Committee to oversee a statewide study of judicial
sentencing practices within the Commonwealth. The study uncovered evidence of
unwarranted sentencing disparity: statistical analysis showed that a variety of extralegal factors influenced sentencing outcomes, including offender race and gender,
type of criminal defense attorney, jury vs. bench trial, and level of oflender education." The influence of these factors was also found to vary according to offense
type, sanction (i.e., probation, jail, prison), and geographical area of che state. According to VCSC staff, these findingswould later be the primary impetus for moving
forward in sentencing guidelines development.

During 1987, the Executive Committee of the Judicial Conference of Virginia
voted to present the results of the disparity study to all circuit court judges during a
series of regional meetings. The outcome in each region was a vote by the circuit
judges recommending the implementation of voluntary sentencing guidelines. Unlike other states considering guidelines as a way to curtail rising prison populations
or as means for implementing non-incarcerative sentences, Virginia's sole purpose
for guidelines developmenr.was to reduce unwarranted sentencing disparity.
O n the basis of the findings from the disparity study, the Chief Justice appointed
a new committee to oversee the development of sentencing guidelines. Although a
l3

Judicial Sentencing Guidelines Oversight Committee (1987).

Tnctb-in-Sentencingin Virginiz

departure from the practice in most other states where sentencing commissions
include representation from each branch of government, this committee was comprised solely of judges.14 The general belief of the Virginia committee was that sentencing is a judicial function; and since the guidelines were to be voluntary, only the
judiciary needed to be involved in their development.
1987-1988 Guidelines Developed
The Judicial Sentencing Guidelines Committee USGC) was responsible for

all

policy decisions regarding sentencing guidelines development and operation. The
first step was to operationally define “appropriate sentence length” so that the effectiveness of the guidelines could be measured. The JSGC determined that the sentencing guidelines ranges would encompass the middle 50% of historical sentence
lengths and that a judicially imposed sentence was defined as appropriate if it fell
within this range and “inconsistent” (and possibly disparate) if it fell below or above
this mid-range. Therefore, the highest 25% and the lowest 25% of all historical

sentences fell outside rhe guidelines ranges. The basic characteristics ofVirginia’s first
set of descriptive sentencing guidelines included:
m

Use of a judicial sentencing worksheet as opposed to a sentencing grid;

rn Presentation of eight specific offense groups (i.e., assault, burglary, drugs, fraud,
homicide, larceny, robbery, sexual assault) with individual sets of scoring factors
and worksheets;
m A bifurcated worksheet design beginning with an intout decision (prison v. no

prison), followed by length of incarceration, if appropriate;
m Presence on the worksheets ofonly legally reievant offense- and offender-related factors

found to be statistically significant in the analysis of historical sentencing practices;
m Recommendation of “effective time sentences” defined as

the length of the judi-

cially imposed sentence minus any suspended time;
a Strictly voluntary sentencing guidelines where judicial compliance would not be

mandated and there would be no opportunity for appellate review based on a
challenge to the guidelines.
Because Virginia’s guidelines were to be descriptive of historical patterns across the
commonwealth and based on legally defensible criteria, VCSC staff analyzed the PSI
database to determine normal sentencing practice as well as the specific offense and
offender-related factors significant in predicting judicial sentences. Thus, no “normative” adjustments were made to the observed sentencing patterns to enhance (or
reduce) the recommended punishment for specific crimes and only statistically significant offense- and offender-related factors were used to create the guidelines. In
this manner, the influence of extralegaI factors (e.g., race, gender, identity of the
judge or judicial circuit, method of adjudication) was reduced so that those factors
would no longer exert a systematic influence in sentencing decision^.'^

Kauder and Ostrorn (1398).
I s Judicial Sentencing Guidelines Oversight (1 989).
l4

The Path to Rejam

In developing the pilot guidelines, VCSC st& used three statistical procedures to
analyze PSI data on 33,573 felony cases sentenced between February, 1985, and
June, 1987. All felony convictions resulting in probation and/or a suspended sentence, a jail term, or a prison term were examined. The results of this comprehensive
andysis were used to design a sentencing guidelines framework consisting of three
worksheets:
worksheet A: used to determine whether a person would receive a prison or a
nonprison sentence;
Worksheet B: used to determine whether a person would receive probation or jail
(if nonprison sentence indicated on Worksheet A);
WOrLbeet C: used to determine the length of a prison sentence (ifprison sentence

indicated on Worksheer A).
Eight sets of offense-specific guideline worksheets were formulated and a manual
was created to explain their application.1G
Worksheets A and B were designed using multiple discriminate function analysis.
In keeping wich a bifurcated design, this analysis was used to determine the factors
influential in judicial decisions of whether or not an offender was to be incarcerated.
A second statistical technique called “probit” was used in the initial pilot guideline
development to refine the proportional weights of the factors for Worksheets A and

B. This technique aIlows one to compare each specific factor’s importance in the
sentencing decision. For example, assume that the coefficient (i.e., the numerical
representation of a factor‘s “imporrance” in a sentencing decision) for “use of a firearm” was the same as that for “serious injury of a victim.” This indicates that judges
have historically given about the same weight (sentence outcomes have been equally
influenced by these two offense factors) for firearm use as they have for serious victim
injury when considering whether or not someone should go to prison.
The offense- and offender-related characteristics linked to the length of prison
sentence (Worksheet C )were uncovered using ordinary least squares mdtiple regression (OLS).Coescients associated with each factor in &e analysis translate roughly
into months of incarceration. For example, a drug offender who scored “6l”on
Worksheet C under the fictor “Counts of Primary Offense” implies that the historic
sentence for a drug offender convicted of four counts of selling drugs was about 61
months (five years) longer than a person convicted of one count of selling drugs, all
other factors being equal.”The factors found to be statistically significant, and their
relative impact, were critical elements for future guidelines development, and, eventually, the establishment of the current TIS guidelines.
Interviews with VCSC st& and a review of published and unpublished source
materials document the analytical process for guideline worksheet development. The
-

I6Themost recent version ofVirginiaSentencim Guidelines still emdovs these three worksheets.
but now apply them to 12 categories of offenses. See, Virginia Criminal SentencingCommisI

sion 61998b).

I’JudiciaI Senrencing Guidelines Oversight Committee (1989).

Truth-in-Sentencing in Virgtnia

research staff responsible for conducting the disparity analysis and pilot guidelines
development operated in a team environment supervised by a project director with

Voluntary Sentencing GuidelinesCompliance Rates
January to September, 1989

previous experience in sentencing guidelines development and other criminological
research. Individual researchers were responsible for different segments of the guideline development, while results were checked independently through blind repeat
analyses using the same data. This process verified findings across analysts with the
results and any inconsistencies being reported during regularly scheduled staff meetings. Researchers were well qualified to conduct the analyses, possessing advanced

degrees in social science and criminal justice research and statistics, while also having
various levels of previous applied resear4 experience. Evaluators note that guideline
development in Virginia benefited greatly from comprehensive data sources, adequate
resources, and staff expertise.

By Circuit
1988-1990 Sentencing Guidelines Pilot Study

Virginia's judiciary voted to pilot test the voluntary guidelines before recommending statewide implementation. Six judicial circuits (out of a possible 3 1) representing
a mix of rural and urban courts were selected as pilot sites. A judge from each of these
six circuits sat on the Judicial Sentencing Guidelines Committee (JSGC), which
provided policy oversight during the process. After a series of regional training seminars, guidelines went into effect in July, 1988, with a plan to pilot the system for one

Circuit12
- 1

70%

Circuli9
- \

76%

Circuit19
- '

78%

Circuit- 21

78%

Circul4
! - '

80%

C i i t 16 ~-182%

Total -78%

year. Judges in the pilot sites were asked to consider the guidelines in almost all
felony cases, explain any reasons for departure, and return the completed forms for
monitoring and evaluation.
Because the purpose of the first set of guidelines was to reduce disparity, the JSGC
directed staff to evaluate the effects of guidelines on sentencing consistency and neutrality. Consistency was defined as the extent to which similarly situated offenders
who committed similar crimes received similar sentences. The JSGC chose to measure rhe effect of the guidelines on consistency by judicial compliance: the percentage of sentences that were within sentencing guidelines ranges before and after guidelines were implemented. Compliance races were examined in pilot and nonpilot sites
to provide a comparative control group. As shown in the bar chart, compliance rates

(percentage of judicial sentences that fell within recommended ranges) during the
pilot program ranged from 74% to 88% depending on the offense group, and ranged
from 70% to 82% depending on the pilot site. Overall, the average compliance rare
was 78%, with departures more likely to be mitigated (1 5%) than aggravated (7%).
Neutrality, or impartiality, was assessed by examining whether variation in sentence length was explained by differences in Iegally relevant factors (e.g., offense
severity, prior record) and not by extrdegal factors such as race or gender. Neutrality
was measured by applying the same statistical techniques used for guidelines developmenc to determine which extralegal factors, if any, exerted influence in sentencing
decisions in both pilot and nonpilot sites. Using consistency and neutrality as a framework for evaluating the existence of sentencing disparity has been documented in

The Pa& to Refirm

Voluntary Sentencing Guidelines
Percentageof Sentences within GuidelinesRanges,
January to September, 198g

past sentencing literature, and both terms still appear as meaningful performance
indicators in this area.'s

A year-long pilot study revealed that judges were using the guidelines, complying
with guideline recommendations, and that the guidelines effectively reduced unwar-

ilI

ranted disparity. As an example, the top bar chart shows the percentage of sentences

&Y
!E!!

for assault and burglary within the guidelines ranges in the pilot and nonpilot sites

both before and after guidelines were introduced. For both offenses shown, the com-

61%

pliance percentage is notably higher in pilot sites than nonpilot sites. An illustration
of the extent to which neutrality was achieved is depicted in the bottom bar chart.

Following the introduction of guidelines, the influence of extralegal factors in ex-

plaining variation in sentence length for prison-bound burglary offenders was re-

duced substantially in the pilot sites (10% of explained variance was accounted for
by extralegal factors in pilot sites as compared to 54% in nonpilot sites after guide-

lines implementation).

After
Wlii

Before
Guiddnes

Aft%
Guidehnes

Before

Guidefines

The evaluation also attempted to measure judicial perception and attitudes toward

the pilot guidelines. A survey conducted during the pilot program found strong ac-

ceptance of the voluntary guidelines among participatingjudges.'!'The survey showed

90% of judges believed the guidelines had increased consistency in sentencing, while
affecring judicial discretion minimally or not at all. Almost all judges (31 of32 judges

surveyed) felt that having the guidelines available as a reference tool was preferable to

Proportion of Sentence DecisionAccountedfor by Legal and
-Legal
Facto& Burglary CasesSentenced to Prison
not having guidelines. The same number said that the guidelines should be expanded

ILegal tactocs

statewide. The one judge who did not want to see guidelines expanded also indicated

a Extra-legalMoa

he did not believe in the existence oEunwarranted sentencing disparity.20

1991-1993 Statewide Voluntary Guidelines
After viewing the results of the pilot study, the JSGC, with the approval of the

Non-PilotSites

Chief Justice, recommended that the sentencing guidelines be implemented statewide: Virginia's circuit judges voted to adopt the sentencing guidelines statewide
54%

effective January, 1991 The sentencing guidelines were monitored and adjusted anI

nually over the next three years to reflect current judicial practice. Interviews with
commission members and staff reveal that a key to program acceptance by the judiciary was the descriptive and voluntary nature of the guidelines. In addition, the
comprehensive and yearly re-analysis of felony conviction and sentencing data to
ensure that the guidelines were based on current trends in judicial sentencing was

10%

unique to Virginia. Although many states make adjustments (largely normative ones)

to their sentencingguideline grids and/or worksheets to reflect the changing purposes

Gudelines
After

Eelare

ARar

&idallnes

G,,,delmes

'ROstromand Kauder (1 998), pp. 22-23; Westing (1982);BureauofJusticeAssistance (2996).
Interviews with VCSC staff indicate that some judges in nonpiiot sites requested and received guideline manuals and worksheet copies during the pilot study period. Judges were
provided manuals at the direction of the Oversight Cornmitree and the chief justice, since the
system was viewed as a valuable decision aid that was only voluntary in nature. The existence
and use of these manuals may have had contaminating effects on the study results, although
staff conversations with several judges indicate that there was no reason to believe guidelines
were being used sysremacically in nonpilot sites.
2o Judicial Sentencing Guidelines Oversight Committee (1989).

Truth-in-Sentencingin Virginia

or goals of sentencing, no state has kept and maintained such an exhaustive andytical approach to the guideline revision process as Virginia.”
The earliest years of guidelines development in Virginia (1985-1988) were supported almost entirely by Bureau of Justice Assistance grants that were later replaced
by state general fund monies. Late in 1990, Virginia‘s legislature passed House Joint
Resolution 46 encouraging the use of sentencing guidelines statewide and appropriated money for a full-time sentencing committee staff. Over the last ten years, staff
size has ranged from five to ten full-time employees (in addition to occasional grantfunded or temporary staff designated for special projects). Although other states have
-

seen staff size grow in more recent years, this level cf staff commitment was unusual
during the mid- and late-1980s. States currently have, on average, five to six employees assigned to staff a sentencing commission and to maintain a guidelines system,
although several states also use those positions for nonguideline-related
Virginia provides one instance where federal seed or start-up money was used to
initiate a long-term project, later supported by state revenue based on a proven need
and commitment to the program.
1994: Introducing Tk-uth-ln-Sentencing
and Parole Abolition
At the time of Governor Allen’s election in November, 1993, judges in Virginia were

using judicially controlled voluntary sentencing guidelines with an average compliance
rate of 76%? Though the judiciary was satisfied that the guidelines were accomplishing their intended purpose (to reduce unwarranted disparity) and with the design of
the guidelines (voluntary and descriptive), there was rising concern about large differences between judicially imposed sentences and the amount of time an offender actually served in prison. Public opinion in Virginia was strongly negative toward the perceived leniency of the parole board’s release decisions during the early 1 9 9 0 Fear
~ ~ ~of
crime was heightened by media coverage showing violent crime rates at record highs.

As the gubernatorial race was heating up in late 1993, both candidates increasingly
stressed specific crime and public safety issues in their respective platforms. Mary Sue
Terry, the Democratic candidate, focused on gun control, specifically, a five-day waiting period for handgun purchases. The Republican candidate, George Allen, made
parole abolition and TIS his primary public safety, if not his overall, campaign theme.
When the campaign season began, Allen was well behind in pre-election polls, but he
won the race by a wide margin. One of his first major actions after taking ofice was the
signing in January, 1994, of an anticrime package and the creation of the Commission
on Parole Abolition and Sentencing Reform.25
Part of this commitment can be attributed to sufkienr funding levels during different phases
of guidelines development. This also allowed guidelines staff to conduct numerous training
seminars and to provide ongoing presentations and technical assistance for judges, probation
officers, and attorneys.
22 Kauder, Ostrom, Peterson, and Rottman (1997).
Z3VirginiaCriminal Sentencing Commission (1995).
24 Survey Research Laboratory (1993).
25 “Governor Allen Signs Sweeping Anticrime Package,” (1994).

The Path to Reform

Allen charged this commission with “developinga plan to abolish parole, establish
truth-in-sentencing, and ensure that violent and repeat criminals stay in prison for
much longer periods of time.”2GThecommission had 32 political appointments, and
was cochaired by former U.S. Attorney General William

P. Barr and Former U.S.

Atrorney of the Eastern District of Virginia, Richard Cullen. The commission was
staffed administratively by an additional 18 persons representing the executive branch,
the attorney general’s otfice, and several private consultants.This administrative body
provided general policy direction for the Criminal Justice Research Center (CJRC)
within the Department of Crimina! Justice Services, the group responsible for the
research and impact analyses associated with commission recommendations. The
Research Center was headed by Richard E Kern, who was aIso serving simultaneously
as executive director for the existing JSGC.

At about this same time, the democratically controlled general assembly created
their own study group called the Sentencing and Parole Reform Commission. The
legislative commission, which was also receiving analytical and staff support from

the CJRC, was exploring a broad menu of potential reforms. What distinguished the
two commissions earPy on was the predetermined decision by the Governor‘s com-

mission to implement TIS and abolish parole.
In the summer of 1994, the Governor announced a special session of the General
Assembly to be convened in the fill for the sole purpose of considering sentencing
reform legislation. As the legislative session neared, the Governor’s commission and the
legislative commission solidified their respective reform packages. The Governor‘s package became known as Proposal X, while the legislative package was referred to as Proposal A. Policy stances formed and split along party lines, between the executive and
legislative branches, and by other competing special interest groups (includingprisoner
advocacygroups, the NRA, victims groups, the NAACP, etc.). The political wrangling
was intense as all seats of both General Assembly houses were up for election within a
year of the special sentencing and parole reform legislative session. Despite the rhetoric,
the final recommendations from each commission were ofien quite similar. Both agreed
to retain certain elements of Virginia’s pre-reform sentencing system, including:

A Sentencing Commission and the use of voluntary sentencing guidelines;
No appellate review of sentencing guidelines departures;
m Jury sentencing.
While there was also substantial agreement about the basic structure of sentencing
reform (e.g., abolishing discretionary parole release, curtailing good time, the proportion of imposed sentence to be served, and increasing time served for violent
offenders), there were important differences in the detaikZ7Thefollowing table depicts the main features Qf Proposals X and A and compares those features to the
system that was operating in 1994. The differences and similarities of the proposals
are analyzed and discussed in the next chapter.
D

Governor’s Commission on Parole Abolition and Sentencing Reform (1934).

”“Parole Abolition Sentencing Reform Pmposali’ (1994).

Truth-in-Sentencingin Erginia

Proposals for Sentencing Reform, 1994

Commission Structwe

Existing System

Proposal X (Executive)

ProposalA (Legislative)

Seven-member-,

ExecutiveBranch,
membersfrwnallbranches

Legislative Branch,members
from all branches

judsesonly
SentencingGuidelines

Voluntary descriptive,
based on historicaljudge
"effectivetime" sentencing

Voluntary, based on
historicaltime-served,
normative imeases
part of original reform

Voluntary, normative
adjustments (increases)
to be recomended
by commission
package and legislation

Worksheets

.ReqUeStd

Required

Average Time Served

21% - 47% of sentence

85%-100Y0of sentence

lCf.)% of sentence, plus
extendedtime for dangerous
offenders

Departures/appeals

No written reasonsMo appeal

Written reasonsMoappeal

Written reasons/No appeal

Jury sentencing

Bifurcated sentencing,
jury receives no
sentencing guidelines

Bifurcatedsentencing,
jury receives no
sentencing guidelrnes

Sentencing guidelines
also providedto jury

GoodTime

Muk@elevets/300days
per year average

Flat rate upto 54
davs per year

NMle; epplicatiwl to
extendadterm W b l e

Parole

Discretionary& mandatory

Abolished

Parole Supervision

Au on pardesupervision
uponexiting prison

Mandatay supervision
for 6months to 3 years

Long-termoomnunity
supervisionto blow

Prison bed space

Forecast variable

Forecast more predictable

Sentencing Guidelines Frameworkfor Truth-in-Sentencing
The NCSC evaluation team believes that one of the best design decisions made by
policymakers in Virginia was the retention of sentencing guidelines. The benefit of
the sentencing guideline approach is that it allows for a more accurate assessment of
the likely impact of a change in sentencing and/or parole policy. Guidelines systems
are arguably the most cost-effective means of providing rational structure, relevant
data, and the ability to accurately monitor and forecast sentencing outcomes.
Eight states (Ohio, Virginia, Arizona, North Carolina, Delaware, Kansas, Minnesota, and Mississippi) and the federal government have abolished parole and impiemented TIS legislation that requires almost all violent and nonviolent offenders to
serve 85% (75% in Delaware) of the imposed sentence. All but two states (Arizona
and Mississippi) introduced TIS into a sentencing guidelines system or developed
guidelines in conjunction with TIS reform. For example:
rn

North Carolina's sentencing reforms received considerable attention in 1994,when
parole was abolished, good time restricted, and a comprehensive community corrections plan developed. The North Carolina Sentencing Commission implemented
grid-based presumptive sentencing guidelines, increased sentences for violent offenders, and developed a structured system to divert nonviolent and most drug
offenders into alternative or intermediate sanction programs.

The Path to Refom

Oklahoma established aTruth-in-Sentencing Policy Advisory Commission in 1995
and proposed sentencing matrices (guidelines)and an 85% time-served minimum.
In conjunction with TIS, Oklahoma proposed two other major reform componenrs, which could free the prison space needed to accommodate the prisoners
who would now serve virtually the entire imposed sentence. The Pre-Adjudication
Atx provides services to substance-abusing offenders at the “front end” of the sys-

tem, and the Communiry Correction Act increases and enhances a continuum of
sentencing options at the community level. The Oklahoma commission also recommended abolishing jury sentencing as part of its overall TIS reform package.
B

Kansas established mandatory guidelines in 1993 and abolished parole releases
replacing post-supervision periods with a set 24- or 36-month supervision period.
Good time can be earned by participating in programs, but cannot reduce a sentence by more than 15%. Good time earned is further added to any period of postrelease. The Kansas grids contain border boxes that allow presumptive prison sentences to be replaced by explicit correctional/treatment programs only if readily
available to the offender.

However, the creation of a sentencing commission and the enactment of structured
sentencing guidelines is not a requirement for TIS. For example:
B Mississippi enacted legislation in 1995 that abolished hiscrerionary parole and
requires inmates to serve 85% of rheir imposed sentences without the introduction of sentencing guidelines. No adjustments were made to existing sentencing
ranges--judges still set a fixed term within the existing statutory ranges for particular felony classes.

m In Arizona, TIS requires offenders to serve 85.7% of their imposed “presumptive”
sentence. For most offenses, sentence lengths were “rolled back” to reflect the historical time served. However, offenders deemed to be “dangerous and repetitive”
did not have their sentence ranges adjusted. These offenders will serve longer periods of incarceration as a result of delayed release eligibility.28
The major problem for states without guidelines is the reduced ability to estimate
future prison bed space needs. The ability to forecast is particularly important in the
context of a major reform like TIS. Many commentators argue that the 85% rule
(with or without sentencing guidelines) will have greater impact on punishment and
the use of prison resources than other sentencing reform measures, including mandatory minimums and three-strikes legislation, because 85% policies are usually applied to all eligible offenders, regardless of prior criminal hi~tory.~’

The following timeline begins in 1385 and provides an overuiew ofthe major policy
initiatives & d i n g up to the initiation of TIS in 1995.

”National Institute of Corrections (1995a).
National Council on Crime and Delinquency (1 995).

Truth-in-Sentencing in Etginia

CHAPTER THREE

The Design of TIS
Guideline3 in Virginia
This chapter reviews the major policy issues and sentencing guideline design considerationsraised during deliberations over sentencing reform. Numerous data analyses
were conducted on behalf of the Governor’s Commission as well as the legislative
committees responsible for modifying statutes to codify the’intended reforms?’ The
primary targets of reform were (I) abolishing parole and (2) establishing TIS (including lengthier incarceration for violent felons). In addition, many other topics
were examined including the expanded role of alternative sanctions, the relationship
between age and recidivism, and mandatory minimum sentencing. The most pertinent studies are reviewed in the following three sections.
1)Abolition of Parole

The Structure ofparole Prior to TIS: Virginia’s system of parole came under fire
in the early 1990s.A commission was appointed by the General Assembly in 1990
to “determine specific reasons for Virginia’s low parole rate,” and make suggestions

for ref~rrn.~’
This move was motivated at least in part by severe overcrowding in
Virginia’s prisons. Just three years later, the parole board was being closely scrutinized on charges of undue leniency.

Optionsfor Parole and Good Time Reform: This debare focused on three basic
issues: (1) Should Virginia modify or completely eliminate discretionary release?;
(2) Should parole and good time reform apply to both violent and nonviolent
offenders?; and (3) Should post-release supervision be maintained?
2)Truth-in-Sentencing (Incorporating Longer Sentences for Violent Offenders)
m Sentence Time Served us. Sentence Time Imposed A necessary first step was to
determine the average difference between the judicially imposed sentence and the
actual time served in prison for violent and nonviolent offenders.

Sbifingfom ‘EffectiveTime”to “TimeServeXSentencingfor Nonviolent Offenders: To accommodate TIS and ensure that nonviolent offenders would serve the
same amount of time post-reform as pre-reform, the guideline recommendations
for nonviolent offenders were modified to reflect historical time served.

m Normative Sentence Enhancemenfifir Kolent Ofenhn: Violent offenders were targeted to receive and serve substantially longer sentences under TIS. The definition of

The Criminal Justice Research Center performed the majority of these analyses, most of
which have not been published orher than for the inrended audience. These studies were collected from the files held at the current VCSC and were found in their original formats as
printouts, graphical presentations, and various types of information and report packets (sometimes termed “fugitive”research and analysis).
31 “Report of the Joint Legislative Audit and Review Commission on Review of Virginia’s Parole Process to the Governor and General Assembly of Virginia” (1 992).
3”

The Design of TIS Guidelines in Virginia

(1) “violent offense” was expanded to include some burglary crimes and (2) “vio-

lent offender” to include the entire criminal history including juvenile delinquency
adjudications.

3) Related Analyses
m The Relationship Between Offender Age and Crime: Considerable debate took
place over whether the incapacitation ofyoung violent offenders during their “crime
prone years” would end a Iikefy cycle of recidivism and as such should be an explicit purpose of TIS reform.
m Mandatory Minimum Sentences under TIS: Issues of incorporating mandatory
minimum sentences within the rational framework of TIS guidelines were explored.

m Exp~ndAltPrnativePunisbmentlTreatment Options: Included in the comprehensive reform package was the legislative goal of using a risk assessment instrument
to identify and divert at least 25% of incarceration-bound drug and property offenders into alternative sanction programs.
Because of the analytical complexity and evaluation techniques applied, review of
the projected and actual impact of TIS on (I) correctional population and prison
bed space needs (Chapter 4), judicial compliance (Chapter 5 ) , and preventable
crime (Chapter 6 ) are discussed separately,
Yiginia Parole Grant Rate, 19911998

The Structure of Parole Prior to TIS
Changing public perception about Virginia’s discretionary release policies is linked

50%

to the 1994 gubernatorial campaign where the parole system was blamed for in-

45%

creased crime and waning public confidence in the criminal justice system.32One of
George Allen’s first actions as governor was to appoint a new parole board. As can be

40%.

seen in the adjacent trend chart, the impact was immediate: the parole grant rate fell
.............

sharply from about 20% in 1993 to 10% in 1994. Although parole was abofished by
Virginia’s General Assembly in 1994, the parole system continues to govern release

eligible for mandatory parole Si months prior to the expiration of their sentence. The

Survey Research Laboratory (1993).
In 1998, chree years after parole was abolished for new offenders, the parole grant rate for
offenders convicted prior to 1995 again dropped OK With only a few months of data, it is
difEcult to assign a reason for h e steep decline in the grant rare, except to point out that many
less serious offenders from the pre-1995 period have already been paroled, leaving a higher
proponion of serious offenders in the parole-eligiblepool.
32

Truth-in-Smtencing in Krgniu

prior to the 1994 reforms continue to earn good conduct allowance according to the

good time system in effect before parole reform. Each prisoner is assigned a good
conduct allowance class, which determines the accrual rate for good conduct credit
(i.e., Class I earns 30 days for 30 served; CIass II earns 20 for 30; Class I11 earns 10
for 30; Class IV earns 0 for 30). Full good conduct allowance is counted toward the
mandatory release dare and half of the good conduct allowance is credited toward
discretionary parole eligibility.
Once the Department of Corrections determines that an offender is eligible for

parole, the case is reviewed by the Parole Board. The parole review process consists of
an interview and recommendation by a parole examiner, after which each of the
board’s five members reviews the case individually. A consensus of three members
(four in first degree murder cases) is required to grant parole. If the offender is released, the board sets the conditions of parole, which may include residence in a
halfway house, day reporting, intensive supervision, electronic monitoring, and/or
drug testing. For offenders released on mandatory parole, a minimum of six months’
post-release supervision is required.
Criticism of the inconsistencies in parole decisionmaking led the Parole Board to
introduce a system ofparole guidelines in 1992.34These guidelines were an attempt
to structure parole decisions and base them on objective factors (e.g., present offense,
prior criminal record, personal and social history, community resources) as well as
subjective factors (e.g,, changes in motivation and behavior, impressions gained during interview^).^^ The guidelines were intended to increase consistency and accountability, give guidance to staff, make systematic use of experience, increase openness,
handle the increasing number of decisions, and make better prediction^.^^
Virginia’s parole guidelines considered four factors: felony risk, time served, institutional behavior, and “auxiliary” information. To determine felony risk, the guidelines
incorporate a risk assessment tool based on prior record, prison conduct, and offender characteristics (e.g., age, substance abuse, education). Each of these felony
risk factors is scored and the sum of all factors provides an indicator of felony risk,
which places the offender in one of four risk categories: low, medium low, medium
high, or high. To ensure appropriate punishment, consistency, and Cairness, the guidelines compare time served by the offender to the average for the governing offense.
The wide range of “average” time served for offenses is divided into four time-served
categories: low, medium low, medium high, and high. Also, the guidelines take into
account any disciplinary infractions that have occurred in the last year. Finally, auxiliary information such as special needs of the offender and input from the victim
and the inmate’s family is considered. These voluntary parole guidelines continue to
be used to assess parole-eligible offenders.

Joint Legislative Audit and Review Commission (1992)’p. 88.

Ibid.

’‘ Center for Effective Public Policy, quoted in Virginia Parole Board presentation materials.
The Design of TIS Guidelines in Virginia

Options for Parole and Good Ylme Reform
Options for sentencing and parole reform were considered and introduced by both
the legislative commission and the Governor’s Commission. Each group confronted
three basic issues: (1) Should Virginia modify or completely eliminate discretionary
parole release?; (2) Should parole and good-time reform apply to both violent and
nonviolent offenders?;and (3) Should post-release supervision be maintained?

Decision I: Shorrld Virginia mod$ or completely eIiminate dircrerionuy
parole release?
Many states have enacted legislation requiring offenders to serve a fiat percentage
(usually 85%) as part of a TIS reform package. Other states require a variable per-

9 centage based on characteristics of the offense or offender, such as prior record. For
‘

example, in Arkansas, the percentage of time that offenders must serve ranges from

33% to 70%according to the seriousness of the offense and whether the offender is
a habitual offender,” Another option used by some states was to modify release

policy by requiring offenders to serve a clearly articulated minimum Sentence before
becoming eligible for parole. New Hampshire, for example, has retained an indeterminate sentencing structure, but requires offenders to serve 100% of the minimum
sentence imposed before becoming eligible for parole.38The legislative commission
considering parole and sentencing reform met several times to consider these issues.
The final recommendation was to abolish parole entirely.
The Governor‘s Commission reached the same conclusion at its first meeting in
February 1994. The Governor asked the commission to remember that “parole
must be replaced by a system rhat deters crime by making punishment certain and
predictable.” Thus, it was a foregone conclusion that the Governor’s Commission
on Parole Abolition and Sentence Reform would recommend the elimination of
discretionary release.
Reform of good-time policies presented a similar set of options. All good conduct
dlowance could be eliminated, or the current system could be modified. Virginia‘s
good time credit allowance system was a complicated four-level structure, making it
difficult to reliably calculate release eligibility. Moreover, the system was considered
overly generous, allowing the average inmate to receive, on average, 300 days for 365
~erved.~’Modificationof the good rime system could mean simply reducing the number of good time levels or restricting offenders to a flat number of days per year.
Another option was to retain good time but not apply it to parole eligibility. Finally,
good time allowance could be incorporated up front by the judge, thereby reducing

the upper range of a sentence.
The two commissions reached different conclusions on good time reform. After
testimony and input from prison officials,“awareness of the dificuk task corrections
officials face on a daily basis, coupled with the responsibility to maintain discipline

Ostrom, Peterson, and Rottman, (1997).
National Institute of Corrections (1995b), p. 4.
Governor’s Commission on Parole Abolition and Sentencing Reform (1994), p. 41.

37 Kauder,
38

Truth-in-Sentencingin Wginia

and order'' led the Governor's commission to recommend replacement of the good
conduct allowance system with a flat rate of54 days a year that must be earned by the
offender.'"' The legislative commission chose to eliminate good time altogether, stating that "the beneficial effect of good time credits on correctional management appears to be arg~able."~'
Decision 2: Should Paroh and Good Time R@mt Apply to both Violent and
Nonvioht oflmders?
According to a 1995 survey by the National Institute of Corrections, 16 states
have eliminated discretionary parole release for all ~ E e n d e r sHowever,
.~~
several states
--

have opted to eliminate discretionary P l e only for targeted offender^.^^ In Virginia, both commissions recommended that parole and good time reform policies
should apply to all offenders. The governor's commission considered retaining the
current parole system for nonviolent offenders, but ultimately rejected it for three
reasons. First, the commission considered TIS an important reform in and of itself,
and as such, equally useful to judges and juries whether incarcerating violent or nonviolent offenders. Second, because most nonviolent offenders sentenced to prison in
Virginia face incarceration after several previous convictions, the commission felt
that they should be required to serve the full sentence imposed. Third, the commission questioned the efficacy of a system that combined real-timc sentences and pa-

role-eligible sentences for different offenders. Such a new two-tiered system-much
like the existing system it was replacing-would

be both difficult to administer and

confusing to che public.
Decision 3: Should post-release supervision be maintained?

In recommending the elimination of parole, Virginia joined a number of states
that have eliminated or limited parole release. Of the states that have abolished parole, only Maine has eliminated post-release supervision entirely.44Most states recognize the role a period of supervised release serves in helping the offender reintegrate
into the community successfully. For example, Minnesota incorporates a supervised
release period into the guidelines sentence where two-thirds of the sentence must be
served in prison and one-third is served on supervised relea~e."~
North Carolina requires all violent offenders to serve a nine-month period of post-release supervision,
with a five-year period required for sex offendersd6
___Ibid.
Commission on Sentencing and Parole Reform f 1999, p. 9.
42 Arkansas, Arizona, Delaware, Florida, Illinois, Indiana, Kansas, Maine, Minnesota, Mississippi, New Mexico, North Carolina, Ohio, Oregon, Virginia, and Washington (National Institute of Corrections, Status of Parole, 1995a, p. 6).
I' South Carolina has abolished parole eligibility for violent offenders. In Georgia, a consritutional amendment eliminates parole eligibility for offenders convicted ofcertain violent crimes.
New York has eliminated parole for second-time felons convicted of a violent felony (as defined by the legislature).
" Bureau ofJustice Assistance (199G).
'I Kauder, Ostrom, Peterson, and Rottman (1997), p. 19.
46 Ibid., p. 23.
4"

The Design of TIS GuiaWms in Virginia

In Virginia, both commissions recommended that some type of post-release supervision be retained. The Governor’s Commission called for a mandatory period of
supervision for six months to three years following release, the exact length of which
would be determined by the sentencing judge. The legislative commission’s plan
suggested that offenders receive an extended maximum term beyond the minimum
term imposed by the sentencing judge. Once the offender has served the mandatory
term, “clearly prescribed release criteria or other risk assessment tools” wouid be used
to evaluate the offender’s fitness to return to society. The legislative commission sug-

gested that a judicial entity (i.e., a public safety commission) determine whether the
offender must serve the extended term.47 In addition to this extended term, the
legislative commission recommended a period of post-release supervision for all re-

leased inmates.
In addition, the Governor’s Commission recommended that transitional policies
be developed for inmates as they approach their release dates. These programs would
provide for a “gradual step-down‘’ within the correctional facilities, including work
centers or drug treatment facilities.This recommendation was similar to legislation
in effect in other stares that provides transitional periods for inmates prior to release.

For example, in Ohio, offenders can be transferred from incarceration to community
sanctions. Ohio felons serving ten years or less are eligible for judicial release and, if
release is granted, the court can place the offender in any community-control sanction for up to five years. The percentage of time served before becoming eligible for
judicial release is determined on a sliding scale according to the original sentence
lengd~.~’In
Delaware, judges may sentence offenders to more than one level of punishment, allowing offenders to “flow down” from more to less severe sanctions.4’
lSS4 Special Sesslon legisiafion

The finalTIS legislation incorporated thc recommendations of the Governor‘s Commission regarding three major issues: parole, good time, and release supervision. Parole
was abolished and replaced with a period of post-release supervision similar to super-

vised probation. Good time accrual was bounded by a maximum of 4.5 sentence cred-

its (54 days per year) to be earned through program participation and adherence to
applicable rules and requirements. TIS legislation allows judges to impose a suspended
term of six months to three years for each felony count in addition to the term of
incarceration. This additional suspended term is imposed in conjunction with a sixmonth ro three-year period of post-release supervision (the length of the additional
term and the post-release supervision need not be the same). The additional term is
imposed if the offender does not adhere to the conditions of post-release supervision
(essentially the same as traditional pr~bation).~”
Judges can continue to suspend a portion of the imposed sentence and place the offender on probation after incarceration.
Commission on Sentencing and Parole Reform (1995), p. 10.
Kauder, Ostrom, Peterson, and Rortman (1997),p. 25.
49 Ibid., p. 6.
50 Virginia Criminal Sentencing Commission (1995a).
47
48

Tmth-in-Sentencingin Et-ginia

As a result of the TIS reforms, the caseload of the parole board has declined steadily
since 1996. The parole board decisions are now limited to those offenders whose
crimes were committed prior to January 1, 1995, and certain conditional release
decisions. Geriatric prisoners sentenced after January 1, 1995, can be considered for
conditional release by the parole board after serving a minimum of five years (offenders over 65) or ten years (offenders over 60) of the sentence i m p ~ s e d .As
~ ’ in other
states that have abolished discretionary parole, the parole board no longer retains any
role in supervising offenders after release, either on post-release supervision or parole. In mid-1996, the parole board support staff was reorganized under the Com--

munity Corrections division of the Department of Corrections, which supervises all
offenders released from Virginia prisons.

Truth-in-Sentencing (Incorporating
Longer Sentences for Violent Offenders)
All structured sentencing systems provide judges with guidance concerning the
appropriate sanctioning ranges for a particular set of case circumstances. Some systems provide only minimal guidance whereas others set rigid criteria for determining
a sentence. Since 1986, Virginia has used a very detailed set of factors (which are
different for each major offense group) to score a sentencing guidelines case. When
determining sentence length, the score serves as the midpoint for a sentencing range
that sets parameters for judicial compliance. Outlining how the ranges evolved from
the previous guidelines system is important for understanding the current TIS guidelines system.52
Prison Tsme Served vs. Sentence Imposed

The basic tenet ofTIS legislation is to more closely align imposed sentences with
time served. Felony offenders in Virginia are now required to serve at least 85% of
their prison sentence behind bars. Prior to the 1994 sentencing reforms, many argued that the combination of parole eligibility and good time credits meant that
time served was typically much less than the judicially imposed sentence. However,
the exact amount of time served by offense and offender type was not generally

known. One reason for this lack of information was that an acceptable time served
percentage had not been established in Virginia. The main reason, though, was the
inherent complexity of the calculation to determine eligibility for discretionary release. As discussed earlier in this chapter, multiple good-time accrual rates, parole
guidelines and risk assessment, and subjective impressions of rehabilitation made it
dificult to determine consistently, and with confidence, the amount of time offends’ “Sentencing and Parole Reform”, p. 33.
Individuals interested in the precise structure and content of the guideline scoring system

’’

and the ranges of senrence recommendations should see the Virginia Sentencing Guidelines
Manual ( 1 998). The Virginia sentencing guidelines incorporate individual case circumstances
that vary widely in terms of the nature of the offense, victim injury, extent and seriousness of
prior record, and prior terms of incarceration or legal restraint. All of these factors are used on
the guideline worksheets when determining a sentence recommendarion.

The Design of TIS Guidelines in Krginia

ers would serve in prison. It was simply very difficult, if not impossible, to articulate
the entire release eligibility process. Both democrats and republicans acknowledged
the inadequacy of a system in which actual time served could not be more easily
determined or predicted.
Despite the inherent difficulties, a necessary first step was to determine rhe actual
relationship between judge-imposed sentences and time served. A database was thus
created from the Offender Based Stare Correctional Information System (OBSCIS)
maintained by the Department of Corrections. Controlling for offense and prior
prison commitments, the average sentence and time served for offenders released

fiomprison between 1988 and 1992 was
The table below estimates the
average percent of sentence served under TIS (1995-1997) compared to the actual
time served for selected offense groups prior to TIS (1988- 1992).% During the years
prior to TIS, offenders in Virginia prisons typically served between 20% and 48% of
their imposed sentencedl
Estimatedlime Served (in years) in Prison by Virginia Felons, Before and After TIS
1988-1992
Average

mpased Average

1995-1997

Percentof Average

Sentence Imposed
Sentence TmeServed SBrVed
Sentence

Estinated
Average

Percentof

19.0

90.3

Estimated Sentence
Timeserved Served

Violent Offenses
~s,ciegree
murder
2Mdegree murder

16.7

5.7

34.1

[ 21

Robbery

13.8

. 4.4

31.9

9.8

8.9

90.8

6.6

2.2

33.3

6.1

5.5

89.4

4.1

3.7

89.4

-2.6

2.3

$93

1.6

1.4

88.9

1.6

1.4

9.4

1.8

1.6

88.3

Voluntary manslaughter
Aggravate

PropertyIDrug Offenses

6.2

1.9

30.6

Possession schedule VI1 drugs 5.4

1.4

25.9

0.9

20.5

Involuntarymanslaughter
edule 1/11tini

Larceny

Sale marijuana
Fraud

4.4

s3 The number of prior prison commitments is the only recidivism measure that statutorily
affects parole eligibility. OBSCIS contains a variable called “felon term indicator” or “FTL”
The FTI number equates to the number of times a person has received a prison commitment.
’’ Although offenders in Virginia prisons were serving significantly less than their imposed
sentence, the proportions were not f i r from the national average: felons sentenced in 1994
served between 32% and 55% of their sentences. See U.S. Department of Justice (1994).
5s Joint Subcommittees on Public Safety of the House Appropriations and Senate Finance
Committees (1 994).

Truth-in-Sentencingin Erginiiz

The results of this analysis were fundamental to the Governor’s Commission’s efforts
to garner support for TIS.

Among all f e l o s rebared @om state prisons, those who had been convicted offirst
degree murder served an average of only 29% of their terms.. .those convicted of Pd

degree murder actually served slightly m o ~ averaging
,
34% of their terms.. . Of all
ofense categories, no group served, on average, as much as half of the sentence the
circuit court judge thought he or she was imposing.. ..56
The commission interpreted these findings as evidence of “across-the-board leniency:”
What is apparent is the absence of truth-in-sentencing in Virginia at any h e l . Early
release is not confined toparticular types of crimefor which one may suppose o f i d e r s
to be more amenable to treatment or lessprone to recidivate. Ifanything, the across-theboard bniency indicates a pervasive philusophy favoring rehabilitation of criminalr
rather than incapacitation...57
The right half of the table shows the expected average time served and the expected proportion of prison sentence served for felons sentenced between 1995 and
1997, the first three years ofTIS. It is estimated that compliance with the “85% rule”
now in place in Virginia will translate into offenders serving between 88% and 92%
of their imposed sentence^.^' The actual length of imposed prison sentence reflects
two crucial TIS guideline design considerations. First, the guideline ranges for non-

violent crimes were reduced from “effective time” to historical “time served.” Second, the guideline ranges for violent offenses were targeted for significant normative
increases from past “effective time.”
Shiiting From llEffecfive TimeNto
“7lrne Served” Sentencing tor Nonviolent Offenders

As discussed in Chapter 2, specific sentence recommendations on the pre-TIS
guideline worksheets were chosen based on careful analysis of past sentencing practices. The sentence ranges captured the middle 50% of past-time-served amounts for
groups ofsimilarly situated offenders. The highest 25% and lowest 25% ofsentences
being deemed “inconsistent” (and possibly disparate) were excluded. Hence, the sentencing worksheet recommendations reflected historical “effective sentences” (Le.,
the typical judicially imposed sentence for different groups of similarly situated offenders). Most importantly, these effective sentences under the pre-TIS guidelines
would be reduced by parole and good time policies.
In conjunction with parole abolition, the Governor’s Commission decided to transform the sentencing recommendations of the guidelines from historical “effective

sentencing” to historical “time-served” sentencing (Worksheet C).59No change was

Governor‘sCommission on Parole Abolition and Sentencing Reform (1 994), pp. 21-22.
Ibid., p. 22.
Actual time served figures reflect variation in average good time accrual r a m by offense.
s9 The design and purpose of each worksheet is discussed in more detail in Chapter 2. See the
Virginia Sentencing Guidelines Manual (1998) for the most current version of the worksheets
used to determine the sentencing recommendation for all crimes covered by the guidelines.
IG

The Des& of TIS Guidelines in Virginia 29

made to the guidelines with<respectto the determination of the historical prisonho
prison decision (Worksheets B and C): the rate at which offenders received a prison
versus a nonprison sentence would remain consistent with past practice.@ In addition,

the number of guideline worksheet offense groups was expanded from 8 to 12.'j1
The move from effective sentences (pre-TIS) to time-served sentences (post-TIS)
did not greatly change the amount of time nonviolent offenders would actually serve
in prison. The difference rested on the proportion of the imposed sentence that
would actually be served in prison (is., 85% under TIS). Public acceptance of the
Governor's plan hinged on clarifying why recommended sentences for nonviolent
offenders under TIS would sometimes appear substantially lower than in the past:
under the previous system, prison sentences were often reduced dramatically by pa-

role and good conduct allowance credits, while under the new system, the judge's
imposed sentence will be served in full (with the offender eligible for only 54 days or

1.5% good time credit). Therefore, judicial-imposed sentences for nonviolent offenders tend to be lower under TIS, bur the amount of time actually served in prison
remains about the same.
The following table compares the guideline sentence ranges recommended under
the previous parole system to those recommended under rhe TIS time-served guidelines for two typical nonviolent sentencing scenarios. Guideline sentence recommendations are calculated with great specificity depending on a variety of offense and offender factors. The TIS sentencing guidelines recommend a midpoint sentence (in
months) with an accompanying range that encompasses 50% of past-time-served
amounts for a group of inmates that were situated similarly in terms of offense and
offender characteristics. This normative decision mirrors practice under the previous
guideline system where the guideline ranges covered 509/0of past effective sencences.
Sentencing Recomrnendations-Comparing TIS to Previous Guidelines
"Efkctive"Sentence

Offense Scenario

Sell ScheduleI or II Drug:
1count, no addiwlal offenses,
no prior record
Grand Larceny from Person:.
2 counts,prior recordfw grand
hrceny, on probation at time of

"he Served"Sentence
RscwnmendedUnder
TIS Offenseor
RecmmendedUndernS
PrimGuidelines @ewe1rVa5)~ Offender Enhancement Guidelines(after i/ss)

4 yr, 11m.
(3yr. - 7 yr. 2 mo.)

No enhancement

1y.
(7 mo. - 1 yr. 4 mo.)

5 yr:
(2yr. 9 mo. 7 yr,3 mo.)

No enhancement

1yr. 8 m.
(11mo.-2 yr. 6 mo.)

offense

During a September 1998, interview, former Governor George Allen, Sentencing and Parole Abolition Chairman Richard Cullen, and Former Allen Chief Legal Counsel Frank B.
Atkinson stressed that having a system of sentencing guidelines in place meant ithat policymakers
and researchers would not have to start from scratch when devising the sentencing ranges
under TIS. In addition, circuit court (felony) judges were accepting of the use and purpose of
sentencing guidelines.
The 12 guideline offenses include murder/homicide, sexual assault, rape, robbery, assault,
larceny, burglary dwelling, burglary structure, kidnapping, drugs, fraud, and miscellaneous.
62 Judicial Sentencing Guidelines Committee (1 994).
@

30 Truth-in-Sentencingin Virpnia

Similarly, in the case of violent offenses, some recommended sentences may appear lower under the new system, but, if followed, the resulting length of stay in
prison will be significantly longer under the new system (as the previous table shows,
for example, in the cases of robbery and malicious wounding).
Violent Offenders: Normative Sentence Enhancements

During the September, 1994, Special Session, the General Assembly acted to enhance
sentence recommendations for certain categories of crimes beyond the level of historical
time served. These “normative” adjustments were made for violent crimes or in cases
involving a prior violent adjudication or conviction. The process began with VCSC staff
I

determining sentences imposed and actual time-served ainounts for violent offenders
who entered or left the system between 1988 and 1992. Historical time-served amounts
formed the basis for normative sentencing adjustments. However, prior to enhancement,
these hiscorical sentences were increased by 13.4% to incorporate the projected award of
sentence credits that might be earned under the new system.
For the crimes of first degree murder, second degree murder, rape in violation of
code 18.2-61, forcible sodomy, object sexual penetration and aggravated sexual battery, the recommended prison sentence was enhanced by:
m 125% for offenders without prior convictions for violent crimes;

300% for those with a criminal record that has at least one violent prior felony
conviction or juvenile adjudication with a statutory maximum penalty of less than

40 years, hereafter referred to as a Category I1 criminal record; and
rn 500% for those with a criminal record that has at least one violent prior felony

conviction or juvenile adjudication with a statutory maximum penalty of40 years
or more, hereafter referred to as a Category I criminal record.

For the crimes of voluntary manslaughter, robbery, aggravated malicious wound-

ing, malicious wounding, any burglary of a dwelling house or statutory burglary of a
dwelling house or any burglary committed while armed wirh a deadlyweapon or any
statutory burglary committed while armed with a deadly weapon, the recommended
prison sentence was enhanced by:
rn 100% for offenders with no prior violent convictions;

300% for Category I1 records; and
500% for Category I records.
For the crimes of manufacturing, selling, giving or distributing, or possessing with the

intent to do any ofthe former, of a ScheduleI or I1 controlled substance, the recommended
prison sentencewas not enhanced for those without a prior violent crime, but was increased
by 200% for Caregory I1 and 400% for Category I records. For any guidelines o&nse not
listed above, the recommended prison sentencewas not enhanced for those without a prior
violent crime, but enhanced 100% for Category I1 and 300% for Category I rec~rds.“~
Although the percentage enhancemenrs mentioned here are based on normative policy decisions, there was also empirical support for increasing time served for certain groups of violent and repeat violent offenders. This research is discussed in Chapter 3.

The Design of TIS Guidelines in Virginia 3 1

T i Sewed m Virginia Prisons
and the Effect of TIS Enhancements

7he Impact of Enhancements on Quidellne
Reoommendationsand Actual 79me Senred

Legislation that codified the enhancements to historical time-served amounts speciRrst Degree Murder

fied which offenses (both current and prior record) were to trigger increases. However, individual case circumstances vary widely in terms of the nature of the offense,

Basic Case

victim injury, extent and seriousness of prior record, and prior terms of incarceration

Category II
Prior Record

49.6
.4

Category I
Prior Record

or legal restraint. All of these factors are used on the guideline worksheets when
determining asentence recommendation. The table below illustrates the recommended

Me guideline ranges for two violent sentencing scenarios.
bb

Sentencing Recommendations Compahng TIS to Previous Guidelines
.-

Second Degree Murder

‘Eflectiw?”

sentence

RecommendedUnder
RiorGuidelinesLbecaawesp

TIS Offense or

Offender Enhancement

71me S e w ’ Sentence
RemmmendedUlderTlS
Eluldellnes (afterm 5 )

Robbery of residence:
Firearmuse, no bljury,
no prior record

11y. 8 mo.
(5yr 3 mo. 15 yr. 10rno.)

Vident offense
erhancement

5yr.5mo.
(3yr. 5 mo. - 6 yr. 7 mo.)

Rape:
Remuse, prior record,
Mcerlt r i l t i e s

30 yr. 6mo.
(14 yr. - 41 yr.)

violent offense
enhancwnent

( 1 5 ~ma.~~1
3 2 ~5m0.)
.

Offense Scenario

27 yr.

and prior record

enhancement

While it is dilticult to summarize how the normative enhancements afFect each

Rape

individual case, it is possible to examine past historical time-served amounts before

Basic Case

TIS with projectedand expectedactualtime-served amounts followingTIS. The pro-

category II
Prior Record

jected time-served amounts reflect the estimates used by policymakers during the
1994 reform process of what average judicially imposed sentences would be under

Category I
Prior Record

44 a

Forcible Sodomy

TIS. Expected actual time-served amounts are based on sentences actually imposed
by judges between 1995 and 1997. These figures are illustrated in the adjacent bars
for both a basic case and for cases involving Category I or I1 prior records.G5
As the bars show, both the projected and expected actual time-served amounts
under TIS are greater than past practice (1988-1992). However, the original projections of time served under TIS that informed the 1994 Special Session do not fully
track with expected actual time served based on sentencing practice during the first

three years ofTIS (1995-19971.‘‘ Offenders convicted of first degree murder, second
degree murder, and robbery with a firearm are all expected to serve more time than
was originally projected by the governor‘s commission. O n the other hand, offenders

Robbery wilh Firearm

Es
4
.
I1 138
-~

Basic Case

:i::472

Category
Prior Record

convicted of rape are expected to serve slightly less time while an offender with a
forcible sodomy conviction is expected to serve the projected time. The results also
vary by Category I or I1 prior record enhancements, with some offense groups expected to serve less time than anticipated (Category I1 prior record for first degree

-..Si&108

Sentencing Guidelines Commirtee (1 994).
A basic case is a case with no aggravating circumstances no multiple counts, no additional
offenses, no weapon use, and no prior record. Category I and I1 case definitions are explained
earlier in this section.
a The divergence b e m a projections and expected a d time served amounts are due primarily
to differences in the rate at which judges were expected to comply with guidelinerecommendations
and the rate at which they actually comply. See Chapter 5 for an analysis of judicial compliance.
64 Judicial

1988-1992lime Served
Pmpcteci under
Expected Actual UndwTlS

Tnrtb-in-Sentencingin f i r - n i a

murder and Category I for Robbery with a Firearm) and some serving more than
projected (Category I1 for Second Degree Murder). Rape offenders are expected to
serve slightly less time for a basic and Category I case and about GO% less for a
Category I1 case as compared to the projected time served. Despite these differences,
rapists are still expected to serve more than double the time under TIS as compared
to the old system.G7

R e l a t e d Analyses
The Relatlonahp Between Offender Age and Reoidivlsm

A major concern of the Governor‘s Commission on Parole Abolition and Sentence
--

Reform was the increase of young violent offenders. According to studies provided
by the Department of Criminal Justice Services, most <‘criminalcareers” begin around
age 14 and peak by age 21, with “retirement” by the late 20s or early 30s.The most
prevalent age of arrest for violent crime (e+, murder/manslaughter, robbery) was

Percent of offendersRecommitted to Prison by Age
and LengthofOriima1 Prison Stay

18.This was particularly trohbling to the commission given that recent increases in

Age admitted to prison

violent crime were occurring at a time when the most crime-prone age group (14 to

21) was at a ten-year low. Staff also conducted analyses to show the connection between age and time served and the likelihood of being recommitted to prison. As
shown on the right, young offenders convicted of violent crime who spend less than
three years in prison are more likely to be recommitted to prison as compared to
older offenders or young offenders who spend more than three years incarcerated.
The VCSC discussed specific straregies to target and “incapacitate” young violent
offenders through their most crime-prone years. Although the commission elected
not to use offender age as an explicit scoring factor within the guidelines structure,
they felt that much the same effect could be achieved by adding the number of prior
juvenile adjudications into the calcularion of prior record. Hence, sentence enhancements tied to prior record would apply more quickly to younger offenders with any
history of serious criminal activity. Prior to TIS reform, an offender’s juvenile record

18-19

32.1%

20-21 -%24.*%

-g%
;
25-29 -2;’,
22-24

30-34
Length of original prison stay

35-39
40t

LW than 3 p r s

u:\

mM~ethan3years

10%

20%

3096

40%

% ot oflenders recommittedto prison

was not scored on a guidelines worksheet.
Mandatory Minimum Sentences Under TIS

Mandatory minimum sentencing laws have existed in Virginia for almost 30 years
and are currently in effect for 45 discrete felony offenses. TheTIS sentencing guidelines make recommendations for almost 95% of Virginia’s felony offenders, includ-

G7 The reasons for these time-served variations may well be a function of data limitations and
noncomparable sample sizes for the different subgroups of offenders. For example, projected
time-served amountswere estimated on larger, more general, groups of offenders.The current
time-served figures are calculated on individualized offender groups that have actually been
sentenced under the new TIS system. Combining the more serious offender groups with specific offense and offender factors reduces the size of the samples that can be analyzed in a
comparable way. Judicialcompliancewith the guidelines may also impact time-served figures.
This can be seen with the rape category, where compliance is lower than all other offense
groups (most departures in rape cases ate mitigated sentences).This compliance issue has been
addressed in an ongoing fashion by the VCSC, with revised worksheets attempting to better
model the specific circumstances (e.g., victim age, relationship, etc) in rape cases.

The Design of TIS Guidelines in Krginiu 33

ing offenses for which mandatory minimum terms appyl@
.'

In these cases, the man-

datory minimum pendty supersedes the guideline recommendation.69An offender
convicted of a crime that carries a mandatory minimum penalty. must receive at least

the specified minimum sentence, which cannot be suspended in whole or in part.
However, prosecutors often avoid charging offenders with offenses that carry mandatory minimums. An offender who has cooperated in the prosecution of other cases
may not be charged with the mandatory minimum offense or a lengthy mandatory
minimum may be used as a bargaining chip in plea negotiations. For example, sex
offenses are among the hardest of cases to successfully prosecute, and certain concessions are sometimes made to ensure a felony conviction with accompanying prison

time. The VCSC estimates that applicable mandatory minimums are charged in
only about 50% of sex offense cases.70
In 1996, the General Assembly requested that the VCSC study the effects of mandatory minimum felony sentences on the use of prison beds and to identify deviations from the guidelines necessitated by the existence of mandatory minimum laws.
The commission developed a computer program to estimate the sentence expected
under the new TIS guidelines for all offenders affected by provisions of mandatory
minimums. Six categories of mandatory minimum offenses were analyzed by the
commission: injury to law enforcement officer, sale of drugs to minors, firearm use
in felonies, sexual assault (subsequent conviction), violent sexual assault (subsequent
conviction), and habitual traffic offender. The six categories of offenses cover 99% of
the total number of convictions which carry a mandatory minimum. The VCSC
determined that in most cases, the guidelines sentence must be adjusted upward to
satisfy mandatory minimum requirements.
Mandatory Minimum Penalties impact Analysis Results, 1995
Average QuWlines

Offense

Injury to Law Enforcement Officer
Saie of Drugs to Minor Three Years Junior
Use of Firearm in the Commission of Certain Felonies
Sexual Assault, Subsequent Conviction
SubsequentViolent FelonySexual Assault
Habitual Traffic Offender

Sentence Increase
Under Mandatory
Minimum (months)

1.9

Estimatedpercentageof

new p r b n sdmisstons
.2%

27.8

3.1

5.1

19.7
0
3.9

.3
.2
3.9

With respect to required prison space, the VCSC determined that the impact of
mandatory minimums needed to be evaluated in terms of the application of the law
as well as the severity of the penalty. For example, while the presumptive sentence

increase (relative to the guideline recommendation) is much greater for sexual assault
than habitual traffic, the VCSC study determined that the widely used habitual

Virginia Criminal Sentencing Commission (1996), p. 46.
Oklahoma and Utah have repealed mandatory minimum penalties as part of sentencing
reform. See Ostrom, Kauder, Rottman, and Peterson (1998).
Virginia Criminal Sentencing Commission (1996), p. 54.

34 Tmth-in-Sentencing in Virginia

traffic offender mandatory minimum had far greater impact on prison use than sexual
assault mandatory penalties. The total number of mandatory minimum convictions
for 1995 is shown below.
Of 21,756 Felony Convictions-l,605 Carried a Mandatory Minimum Penalty

Mandatofy

Nunber of 1995

Offense

MiPEnrdty

Convictions

Third Convktionfor a Violent Felony
Violent Sexual Assault, subsequent coMljction
D n g Kingpins
Sexual Assault, subsequent conviction
Fkearm Use During Felony
Drug Crimes wfiirearm
Assault Law EnforcementOfficer
Homicide (vehicular)
Traffic (habitualoffender)
All other Offense

Life
10years to life
20 years
5 to 20 years
3 to 5 years
2 to 5 years
6 mos. to 2 years
1Year
1 year

0
0

5
0
613
22
56
9
886
14

Most of Virginia’s mandatory minimums were enacted when parole was in effect.
When the General Assembly abolished parole and the earlier system of good time,
felons who formerly served between 20 and 50% of their sentences will now serve at
least 85% of their imposed prison term. The General Assembly has chosen not to
amend the general criminal statutes that delineate mandatory minimum penalties.

As a result, the actual penalty, as measured by time served, for felonies with mandatory minimum provisions occurring after January 1,1995,has increased significantly.
Use of Alternative PunishmenW7Peatment Optton8

One of the legislative requirements included in the comprehensive reform package
of 1994 was the goal of diverting 25% of prison-bound offenders to alternative sanctions. At the time sentencing reforms were being debated, policymakers were concerned about the rising prison population and that a significant share of the state
budget was being spent on corrections. In Virginia, as elsewhere, there was a great
deal of interest in identifying effective ways to punish nonviolent felons in a more
cost-efficient fashion. Alternative sanctions or so-called intermediate punishments
have been developed to address this need. However, many have raised the concern
that alternative punishments may be applied to the unintended offender population-those

who otherwise would receive probation (i.e., net widening). Also, there

is the issue of whether the use of intermediate sanctions, in lieu of traditional incarceration, is effective in protecting public safety, Given these issues and concerns lawmakers drafted language (Cod ofVirginia 417-235) that charges the VCSC to accomplish the following:
Prepare guidelines for sentencing courts to use in determining appropriate candidates for alternative sanctions;
Develop an offender risk assessment instrument for use in all felony cases, based
on a study ofVirginia felons, that will be predictive of the relative risk that a felon
will become a threat to public safety;

The Design of TIS Guidelines in ErgiEia 35

Apply the risk assessment instrument to nonviolent felony offenders and, with
due regard for public safety needs, examine the feasibility of achieving the god of
placing 25% of such offenders into alternative sanction programs.
The VCSC has interpreted its directive from the legislature as the diversion of

25% of nonviolent offenders into other means of punishment than incarceration.
Decisions about diversion are to be guided by the score obtained from a risk assessment instrument, prepared at the time of the pre-sentence investigation report for
use by the sentencing judge.
The VCSC has developed and is currently pilot testing a risk assessment tool to be

‘i

used by judges at the time of sentencing to identify the best candidates for diversion
based on past recidivism. The use of the risk assessment instrument is expected to
remain voluntary. Over the next 18 months, the NCSC and VCSC will expand their
partnership to indude a comprehensive evaluation of risk assessment and diversionary
policies that are now being implemented. The evaluation will have three goals: 1) to
evaluate the methods used to develop the risk assessment instrument; 2 ) ro evaluate the
use, workload implications, and effectiveness of the instrument: and 3) to establish a
methodology and baseline database to conduct a complete impact evaluation.
The intended goals of risk assessment can only be accomplished if adequate resources
and programs exist for offender diversion. Virginia currently uses boot camp, detention center, intensive supervision, day reporting, and electronic monitoring as alternative sanction options. On July 1, 1998, roughly 500 persons were in the detention,
diversion, and boot camp programs, up from 300 persons the same month in 1997. In

1998, however, there were more than 700 offenders on facility waiting lists.

36 Truth-in-Sentencingin Wrginiu

CHAPTER FOUR

The Impact of TIS on Prison
Population in Virginia

The impact of TIS legislation on Virginia‘s correctional resources was a source of early
concern to lawmakers. During the 1994 Special Session, the Virginia General Assembly
passed legislation requirii the VCSC to estimate the impact of all proposed sentencing
legislation on correctional resource needs?’ The comprehensive sentencing reform package
included the followingfeatureswith the v e s t potential to affect correctional populations:
All felony offenders must serve at least 85% of their prison sentence;
a Violent offenders will serve substantially longer prison sentences (two to six times
longer in many cases);
Juvenile adjudications of delinquency for felony-level crimes are now scored as
part of an offender’s prior criminal record;

Local jails will now house offenders receiving sentences of six months or less rather
than 24 months or less;
The VCSC was charged by statute” to develop for judges’ use a risk assessment instrument that would be predictive ofthe relative risk that an offender poses to public safety.
The goal was to use this instrument to identify and divert to community corrections
up to 25% of nonviolent felons who would otherwise be incarcerared.
This chapter describes the specific techniques used by the VCSC to estimate the impact
of TIS on hture correctional populations in Virginia and compares the forecasted impact
to actual impa~c.7~
It is worth noting that the Virginia General Assembly went on to adopt
a very sensible constraint when TIS was implemented in 1995: All proposed sentencing
legislation in Virginia must be accompanied by a “Commission Prison Impact Statement.”
A bill w
i
l
l die in the legislature unless the necessary hnds are ap~ropriated.7~
§30-19.1:5 ofthe Codeof Virginia.
417-235, paragraphs 4, 5 , and 6 ofthe Codeof Virginia.
73 Following the 1994 reforms, Virginia joined the ranks of other states (e.g., Kansas, Minnesota,
North Carolina,Oregon,and Washingron)where enabling legislation requiredexplicit consideration by the sentencing commission of the impact of sentencing guidelineson correctional resources, Tonry (1997). When reviewing state sentencingcommission performance through the
early nineties, Tonry (1991, 1993) maintained that a necessary condition for success was the
legislativerequirement that “sentencingpolicy be meaninghlly related to correctionalresources.”
Tonry asserted that the ability of Minnesota, Oregon, and Washington to hold their prison
populations within capacity for extended periods after guidelines implementationwas attribucable to their “resource constraint” policies, Tonry (1 997). Success, though, has not necessarily
been long-lived. Prison populations in Minnesota and Washington rose rapidly following 1993legislated increases in penalties provoked by sensational crimes in each state.
74 Similarly, the sentencing commission in North Carolina has made skillhl use of “impact
statements”on a number of occasions to dissuade legislatorsfrom enacting punitive legislation
that would have taxed correctional resources well beyond their current capacities. North Carolina has successfully managed to constrain the growth of its state prison system by expanding
the use of intermediate sanctionsand community correctionsfor less serious offenders and still
increasing sentences for the most serious offenders,Wright (1 998). Effective management was
possible in North Carolina because t h e sentencing structure is effectively predicting the correctional resources that the State will need and is directing serious felons and misdemeanants
to longer prison terms while sending less serious felons to non-prison punishments” (p. 13).
71

The Impact of TIS on Prison Population in Virginia 37

Estimating the Impact of Truth-in-Sentencing
on Correctional Populations
Farecasting basics

Estimating the effect of TIS on the need for prison space in Virginia required forecasts of the correctional population incorporating different sets ofassumptions. In general, forecasts may be qualitative, qwntitative, or a blend of both approaches.” Qualitative forecasting methods generally use the opinions ofexperts to predict future events
subjectively. Such methods are used when historical data are either not available or of
questionable validity. Quantitative forecasting techniques analyze historicaldata to predict future values for a variable of interest (e.g., prison population).
Quantitative forecasting models can be grouped into two varieties-univariatemodels
and cawdmodels. Univariate models predict future values based solely on past values
of the time

When a univariate model (e.g., exponential smoothing, decompo-

sition methods, Box-Jenkins models) is used, historical data are analyzed to identify

and extrapolate patterns in the data to produce forecasts. For example, past levels of
prison population are used to forecast future levels of prison population. Univariate
forecasting models are most useful and accurate when conditions are expected to remain relatively constant or the time frame of the forecast is short. However, these
models are less useful when it comes to forecasting the impact of changes in poliyn

Causal forecasting involves identifying variables that are related to the variable
being forecast. Once these causal variables are identified, a model is developed that
describes the relationship between all the variables. For example, information on the
number of new admissions to prison, expected sentence length, and parole grant
rates could be used to forecast future levels of prison population. Causal models are
better suited than univariate models for assessing the impact of policy alternatives on
the future values of the variable of interest. This approach to modeling is often employed to produce forecasts with longer time horizons because it can incorporate
theoretical or other assumptions about future events.
One type of causal model that has seen extensive application to court and correctional policy modeling and alternatives forecasting is the simulation m ~ d e l Stochas.~~
tic-processsimulation (also called discrete everzt or Monte-Carlo simulation) refers to the
use of mathematical models to study systems that are characterized by the occurrence
of discrete, random events. These individual events are represented by random vari-

’’ Bowerman and O’Connell (1 993).
A time series is a chronologically ordered sequence of observationson a particular variable.
OConnell(1993).
78 Simulation is an activity whereby one can draw conclusions about the behavior of a given
system by studying the behavior of a corresponding model whose cause-and-effect relationships
are the same as (or similar to) those of the original sysrem, Gottfried (1 984). Sofmre to dwelop
simulation models has become increasingly available and progressively easier to use. Simulation
models historically were often developed from scratch using a programming language such as
FORTRAN or C++,though these were generally eclipsed by programming languages designed
specifically for simulation such as SLAM and SYMSCRIM: Increasing15 PC-based software
such as @Risk and rhe PGversion of SLAM are becoming available to develop simulation models. See, e.g., Kramer, Lubitz, and Kernpinen (1989); Flango and Osrrom (1996).
76

” Bowerman and

38 Truth-in-Sentencing in Virginia

ables whose values are generated by a computer. This approach synthesizes the randomness that is present in a real system, allowing the behavior of the original system
to be reproduced artificially.
Given the VCSC's need to examine numerous alternatives to implementing TIS,
the commission opted to develop a stochastic-process computer simulation forecast
model (Criminal Justice Research Center @Risk). The model was developed to simulate judicial decisionmaking and the demand for prison beds specifically within the
context of the new TIS guidelines. The program has the flexibility to model a wide
variety of alternative sentence ranges and recommendations. There are numerous
-

interrelated components of the simulation program:. Criminal justice system admissions, guidelines emulation, judicial compliance, rates of earned sentence credirs,
recidivism rates, and the offender-mix distribution. In addition, the model can accommodate anticipated changes in the crime prone "at-risk" age groups within the
admissions module of the program. The simulation model is programmed using the
Excel spreadsheet

and the @Risk software package.BO

CJRC @RfskSimulation Model

There are two central elements to simulating state prison population: stock population (Le., the number of inmates imprisoned at the beginning of the simulation)
and new admissions. The stock population was defined as the number of inmates in
Virginia prisons just prior to sentencing reform and the abolition of parole in Janu-

ary, 1995. It was assumed that the stock population ofprisoners sentenced prior to
the 1994 reforms would gradually decline over time at a rate largely determined by
the Parole Grant Rate (PGR). The higher the PGR, the faster the rate at which the
stock population will decline.
The @Risk model begins to estimate the number of new admissions by generating
a Length-of-Stay (LOS) for different categories of hypothetical offenders during each

month of the forecast period.*' This step differentiates the pool of new admissions
into offender groups and assigns an average sentence to the offenders in each group.
The generated LOS is then used to determine how many months each specific group
of offenders will remain in prison. The LOS generated for each hypothetical offender group (sentenced before and after reform) was then used to model the LOS
for all offenders admitted during a particular month. Admissions during a particular

month are described as a monthly admissions cohort.

The model uses special counting cells called

queuing celh to keep track of the

contribution that each category of offender from each monthly admissions cohort
makes to the prison population for all subsequent months in the forecast time horizon. For all months after the hypothetical offender has exited the system (because of
parole release or sentence completion), the offender's monthly admissions cohort

''Microsoft Excel Version 4.0.
@Risk Version 3.0.

*' Creech (1777).

The Impact of TIS on Prison Population in Virginia 39

adds nothing to the queuing cells. This process is repeated for every monthly admissions cohort. After the last month in the forecast time horizon is reached, the contributions of each admissions cohort to the prison population of each month are summed
and a forecast of prison population for each month is produced. A variety of summary statistical measures (e.g., mean, standard deviation, percentiles, minimum,
maximum, etc.) are produced as part of the process.
Initial validdon. A prerequisite to using simulation to model policy alternatives is
that the simulation model be valia!zted82This is typically accomplished by inputting
historical data for the model parameters, using the model to generate forecasts for a
time-period that has already passed, and then comparing the accuracy of the posthoc forecasts to the actual numbers. If the forecasts of historid data are accurate
according to pre-established criteria, the model is considered valid.

The historical approach to validation was not used because the type ofhistorical data
needed for the simulation were not available (e.g., compliance with theTIS guidelines,
rate of attrition of the stock population, etc.). Instead, the-model was validated by
comparing the forecasts produced by the CJRC @Risk model with forecasts derived
from a second model. This alternative model, the NCCD Pmphet simulation model,
was being used by the DOC to forecast how the stock population (on hand when TIS

reform was expected to be implemented in January, 1995) could be expected to e ~ i t . 8 ~

In this “prospective” validation, the two models were found to produce similar results
when they incorporatedsimilar assumptions.While validation with historical data would
have provided a less assailable assessment, the prospective method employed represented an informed attempt to address the essential step in model building of model
validation, especially given the limitations of their data.

Estimating the Effect of TIS on
Correctional Populations Using Slmulatlon
To prospectively assess the possible impact of TIS on the state-responsible prison
population, it was necessary to produce two different types of forecasts. The first assumed that the sentencing status quo would continue throughout the forecast time
horizon (called a baseline forecast). This assumption implies that the “effective time”
sentencing guidelines in use prior to reform in 1994 would continue to be used during
the entire forecast time horizon. The second forecast was based on the assumption that
sentencing reform and abolition of parole would occur as articulated by the Governor’s
Commission. The differencebetween the baseline (no reform) and Governor’s Commission (Proposal X reform) forecasts of prison population represents the expected
impact of sentencing reform and parole abolition on prison population,
Baseline forecasts were produced by the DOC using the NCCD Prophet Model.
The @Risk model was used to produce the forecasts of prison population under

*’Gottfried (1984).
The NCCD model used a truncated exponential distribution to determine LOSS for the
stock population. This is a common assumption in queuing models and has empirical support
in a variety of situations, Greenberg (1979).

Truth-in-Sentencingin Virginia

Proposal X. Stock prison population for both models was assumed to decline at a rate

that was determined by the parole grant rate incorporated in the forecast. Further,

both models used the same new commitment admissions forecasts developed by the
DOC (using Box-Jenkins models). Parole violators in both models were included

Confined Population

with other new commitments using the 1994 levels of parole revocation.

60,OM)

The Proposal X forecasts incorporated the normative adjustments in LOS for specified violent offenses as well as other changes in LOS for nonviolent offenders brought
on by the move to time-served guidelines. It was assumed that inmates under Pro-

Prison Population Forecast

Comparisonof The Commission’sPlan with Forecast Based
on a 15% Parole Grant Rate Amended Bll(1993 - 2005)

50,000.
40,OM) .

posal X would serve, on average, 88.2% of their total sentence. In addition, the

30,000.

Proposal X forecasts reflected the change in the definition of state responsible in-

20,oOo.

mates from any prisoner sentenced to more than two years to any prisoner sentenced
to more’than six months. It was also assumed that Proposal X would cake effect in
January, 1995.
The need for prison beds was forecast using two different assumptions about the

1993 1995 1991 1999 2001 2003 2005

parole grant race: 41.6 % and 15%. The 41.G% parole grant rate is the five-year
average over the period 1988 to 1992. The 15% figure was a “best-guess” estimate of
the future PGR made by the Parole Commission. This estimate was requested when
officials observed the PGR declining sharply following Governor Allen’s election in

Forecast Based on
41.6% Parole Grant Rate (1993 2005)

1993. The trend lines here show the state-responsible prison population forecasts for
both PGR assumptions. The baseline and Proposal X forecasts under both scenarios

Confined Population

indicate that between June, 1995, and June, 2005, prison population in Virginia will
approximately double.84Not surprisingly, the relationship between the expected im-

60,0001

pact of the baseline and the Proposal X forecasts on correctional population is con-

50,000.

tingent upon the assumptions made about the PGR.

Baseline41.6% PGR

Assuming a 41.6% PGR, the Proposal X forecast for June, 2005, exceeds the baseline
forecast by 2,929. On the other hand, assuming a 15% PGR, the baseline forecast
exceeds the Proposal X forecast by 3,733. The reason that the baseline forecast is
higher under the 15% PGR is that the model assumes all inmates will serve 85% of
their historical “effective time” sentence. Under these PGR assumprions, both sce-

lor
01993

1595 1997 1999 2001 2003 2005

narios show the expected impact of Proposal X on prison population to be relatively
modest, resulting in either a 5% increase over the baseline forecast if one assumed a

PreW by: Chinal JusticeResaarchCenter,DGJS

PGR of 41.6%, or a 6.7% decrease assuming a PGR of 15%. Therefore, the ultimate
impact of Proposal X was shown to be largely dependent on the PGR.
One point of agreement between the forecasts is the rapid, almost explosive growth
in prison population expected between 1995 and 2005. Both forecasts clearly implied that prison capacity would need to expand greatly over the next decade. A nonobvious result, assuming that the sharp dedine in the PGR following Governor Allen’s

election would continue indefinitely, is that the adoption of all TIS reforms would
actually reduce expected prison population relative to the status quo.

Criminal Justice Research Center (1994).

The Impact of TIS on Prison Popuhtion in Virginia 41

Actual B ForecastedPrison Population,19951997

The Impact of TIS on Corrections, 1995-1997
Forecat zz A m , ! Both the baseline and Proposal X forecasts predicted that prison
population would expand rapidly and significantly between 1995 and 2005. The
bars to the left compare actual and forecasted prison population for 1995-1997: the
forecasts exceed the actual population each year. The difference was marginal for
1995 (2.2%) but for 1996 (10.3%) and 1997 (20.8%), it was quite large. Contrary
to expectations, the net increase in prison population between 1995 and 1997 was
only 5% (compared to a forecasted increase of 24%). Indeed, there was no growth at

Actual & ForecastedPlison Admissions, 1995-1997

all in prison population between 1996 and 1997.85Clearlythe forecasts were in error.
The sourre of’recat emop. Errors in simulation typically result from three sources:86
(1) &e data, ( 2 ) an invalid model (resulting from improper specification or changes
in the system being modeled), and (3) implementation of the model (especially programming errors). The second and third sources of error were minimized, if not
eliminated, by the pre-implementation validation of the model (at least initially). In
this case, it appears that an inaccurate estimate of the admissions stream was the
source of error. The bar charts here compare the actual and forecasted admissions
(new commitments plus parole violators) and show that the forecasts substantially
exceeded the actual admissions for every year (by 22% for 1995,24% for 1996, and
33% for 1997). Contrary to expectations, the net increase in prison admissions between 1995 and 1997 was only 14% (compared to a forecasted increase of 24%).
At least two reasons can be identified for the inaccurate admissions forecasts: (1)
declining arrests for violent crime and (2) slower than expected growth in total arrest~.~’
As seen in the trend lines on the next page, the violent crime rate for selected
offenses declined for each crime type over the last five years. From 1993 to 1997,
murder and robbery rates decreased by 13%, rape by 18%, and assaults by 3%.
These unforeseen drops followed increases for each offense group during the late
1980s and early 1990s and contributed significantly to an overestimate of prison
admissions.88
In addition to the inaccurate admissions forecast, avo other potential sources of error
could come from invalid specification of the model. First, if the stock prison population
(not afEectedby TIS)is actually declining at a rate different than the assumed PGR, then
the prison population forecast will be inaccurate. As seen in the following table, data on
parole rates Since the implementation ofTIS suggest that theVCSC estimate of a postimplementation PGR of 15% was reasonably accurate. However, with respect to che
baseline forecast, it is questionable whether a PGR of 15% would have been sustained
indefinitely for all offenderssentenced under “effective time” guidelines.

Department of Public Safety (1997).
Gottfried (1984).
87 Department of Public Safety (1997).
One might also speculate that the drop in violent crime rates is in part the result of the
extended incapacitation of violent offenders incarcerated since the implementation of TIS
(Marvel1and Moody, 1994;Spelman, 1994; Levitt, 1996), though it is certainly controversial
and difficult to prove this hypothesis (Austin and Irwin, 1993).
85

Truth-in-Sentencing in Virginia

Violent Crime Rates in Virginia, 1985-1997
(per 100,000 population)

Discretionary Parole Decisions, 1995-1998
Fiscal Year
1995
1996
1997
1998

Caseload
19,643
21,589
16,461
14,031

Granted
2.810
3,853
3,208
2,373

Grant rate
14.3%
17.8%
19.5%
16.9%

Second, while the simulation approach employed by the VCSC was appropriate
given the types of policy alternatives it was required to evaluate, the method used by
the original @Risk model to estimate the LOS of each admission cohort incorporates
an assumption that is unrealistic at face value. Specifically, the assumption in ques--

tion is that the LOS of all admissions of the same category admitted during the same

Murder

241

1985

1987

1989

1991

1993

1995

1997

totally eliminate such variability. The initial @Risk model has been revised several

curred independently for each admission. The larest version of @Risk avoids problems of sampling strategy by using actual sentences for all admissions in a queuing

Rape

2010.

model framework.
Prison expansion. The forecasts produced by the VCSC were not used by the DOC

for planning in general and for facilities expansion in particular. Since 1987, Virginia

, / p

1985

1987

1989

1991

1993

1995

1997

has projected the size of its future prison and jail populations through a process
which combines technical forecasting expertise
known as “consensus f~recasting,”~~
with the judgment and experience of professionals working in all areas of the crimi-

150 -

nal justice system. Based on forecasts produced in this manner, Virginia expanded its

125-

prison capacity throughout the latter half of the 1980s and early 1990s. The recent

100 -

downturn in admissions has resulted in these forecasts missing their mark by a wide

75.

margin. As a consequence, the amount by which inmate population exceeds the
design capacity of the prison system declined from 52% to 37% between 1997 and
1998. Although prison population still exceeds technical capacity, Virginia currently

Robbery
’

25.

plans to lease as many as 3,290 prison beds to other states.

1985

1
1987

1989

1991

1993

1995

1997

Both estimares (i.e., by CJRC and by DOC) clearly overestimated the expected

sponsible inmates being held at local jails. However, the CJRC model accomplished its
primary objective in that it effectively demonstrated that TIS Sentencing Guidelines

150.

could be implemented without causing unmanageable pressure on the state-responsible prison population. In sum, the methodology employed by the CJRC to accom-

100-

plish this fairly complex demonstration was comprehensive and conceptually sound.

Assault

50.
0-

1985

Virginia Criminal Sentencing Commission ( I 397).

1987

1989

1991

1993

1995

1997

The Impact of TIS on Prison Population in Virginia 43

CHAPTER FIVE

The Impact of TIS on
Judicial Compliance
The primary goal of Virginia’s sentencing guidelines is to establish rational and
consistent sentencing standards subject to the state’s TIS laws.” A common measure
of sentencing guideline system performance is the extent to which sentences adhere’
to or are in compliance with, the guideline recommendations. High levels of statewide judicial compliance indicate that sentences are being meted out consistently.
Likewise, concern with unwarranted sentencing disparity is reduced when compliance is high. In addition, compliance and departure analyses provide an empirical
look at judicial satisfaction with the effect of guidelines on judicial discretion. One
interpretation is that high compliance rates, especially in a voluntary setting like
Virginia, indicate judicial acceptance and approval of the sentencing recommendations. In contrast, low compliance rates may indicate that judges are dissatisfied with
the limits being placed on their discretion. Departures become the way judges in-

form policymakers that the guidelines place undue constraint on discretion and do
not allow for appropriate or flexible sentencing decisions.
This chapter examines the effect on judicial compliance following the implementation
ofTIS legislation in Virginia. Judicial compliance with theTIS guidelines is
judges may depart fiom the guidelines and impose a sentence that is either more or less
severe than recommended. When a judge elects to sentence outside the guideline range,
the judge must submit a reason why to the commission. The first step in our assessment
of judicial satisfiction with the sentencing guidelines is to define judicial compliance.

Next,guideline compliance in the years just prior to reform (1991-1994)is compared
with compliance folowhg the passage ofTIS legislation (1995- 1998).The chapter concludes with a review of the most frequently cited reasons for departure.92

Defining Compliance
The VCSC examines compliance with Virginia’s guidelines using three general
measures: dispositional, durational and overall compliance. These alternative measures allow the commission to gain perspective on which elements of the guidelines
are functioning well and which have gained less acceptance among the judiciary.
”This statement reflects the stated goals ofthe Commission throughout the Commonwealth’s
experience with guidelines. See, for example, Report of the Ad Hoc Committee on Sentencing
Guidelines (1985), p. 1 and Virginia Criminal Sentencing Commission (1998).
” Judges use the guidelines as a reference but may choose to sentence outside them in particular cases. While compliance with guideline recommendationsis voluntary, completion of guidelines worksheets is now mandatory as stipulated in Q 19.2-298.01 ofThe Code of Virginia.
Also, in cases when judges choose to sentence outside the guidelines recommendations, judges
must, pursuant to 5 19.2-298.01(B),provide written explanations for the departures (Virginia Criminal Sentencing Commission, I995 Annual Report, p. 6).
92 All compliance analysis reviewed in this chapter was originally conducted by VCSC. See
Virginia Criminal Sentencing Commission (1998).
44

Truth-in-Sentencing in Virginia

Dispositionalcompliance is defined as the rate at which judges sentence offenders
to the same type of disposition recommended by the guidelines as follows: 1) probation/no incarceration, 2) incarceration up to six months, or 3) incarceration over six
months. Because the recommendation as to the type of disposition is the foundation
of the sentencing guideline system, the commission believes dispositional compliance is an important measure. The rate of dispositional compliance in FYI998 was

83% and has remained largely stable since the introduction of TIS in 1995.
Durationalcompliance is defined as the rate at which judges sentence offenders to
terms of incarceration that fall exactly within the recommended guideline range. In
Virginia, the measure of durational compliance considers only those cases for which
the guidelines recommend an active term of incarceration and the offender receives
an incarceration sanction of at least one day in jail. Durarional compliance among
-1998 cases was 76% and has varied by specific type of offense since the implementation ofTIS. This result indicates that judges more often agree with the recommended type ofsanction (dispositional compliance) than they do with the recommended sentence length in incarceration cases.
Overall compliance measures the extent to which Virginia’s judges concur with
recommended type of disposition and length of incarceration. Overall compliance
is the combination of sentences found to be in strictand generalcompliance. For a

case to be in strict compliance, the sentence must meet both dispositional and
durational criteria. General compliance is less exacting and “results from the
commission’s attempt to understand judicial thinking in the sentencing process,
and is also meant to accommodate special sentencing ~ i r ~ ~ m ~ t a For
n ~a ecase
~.”~~
to be in general compliance with the sentencing guidelines, it must meet one of
the following chree criteria:
m Compliance

by rounding provides an allowance in instances when the active sen-

tence handed down by a judge or jury is “very close” to the sentencing guideline
recommended range. For example, a judge is considered in general compliance
with the guidelines if he sentenced an offender to a two-year sentence based on a
guideline recommended range that goes up to one year eleven months.
rn

Time served compliance is intended to accommodate judicial discretion when a
judge sentences an offender to pre-sentence time served in a local jail when the
guidelines call for a short jail sentence. Even though the judge does not sentence
an offender to post-sentence incarceration time, the commission typically considers this type of case to be in general compliance.
Compliance due to alternative sanctioning arises most often in habitual traffic offender cases as the result of amendments to the law effective July 1, 1997. The
change allows judges, at their discretion, to suspend the mandatory minimum 12month incarceration term in habitual traffk felonies and sentence these offenders
to a Boot Camp, Detention Center, or Diversion Center Incarceration program.

93 Virginia

Criminal Sentencing Commission (1998). p. 23.
The Impact of

TIS on judicial Compliance 48

Sentencing Guidelines Compliance Rates,
Before and After TIS

Compliance

For cases sentenced since the e&ctive date of legislation, the commission considers

either mode of sanctioning to be in general compliance with the sentencing guidelines.

75.4%

25%

Jun Feb
1992 1593

Mar
1994

Nov Cki Mar
19% 1996 1998

Mar
1994

Nov Oct Mar
1995 1996 1998

Mar

Nov Oct Mar
1995 1996 1998

Mitigation

2ow

10%

Jun Feb
1992 1993

Aggravation

20%

-0% 1

Jun Feb
1992 1993

1994

Truth-in-Sentencingin Krginia

Overall Compliance and Departure
Overall compliance has remained relatively high since the inception of sentencing
guidelines in 1991. The overall compliance rate has ranged from 72 to 76% and
currentlysits at 75% (between 1/95 and 3130198 for 42,269 cases). BecauseVirginia’s
sentencing guidelines are designed to accommodate judicial discretion (they remain

TIS Sentencing Guidelines Compliance Rates
January 1995 March 1998

Larceny
- 1
Fraud

Dws

voluntary and there is no mandate to adhere to the guideline recommendations), the

@urglary Structure

commission does not view the attainment of 100% compliance as an ultimate goal.

Assault

The rate at which judges sentence offenders more severely than the sentencingguide-

Burglary Dwellmg

line recommendation, known as the ‘‘aggravation’’rate, has ranged from a low of 9%

Homicide

(just prior to the implementation ofTIS) to the current level of 13%. The rate at which

Robbery

judges sentence offenders to sanctions below the guideline recommendation, or the

Kidnapping

“mitigation” rate, has dropped slightly since the introduction of TIS, declining from a

Sexual Assault

high of 17% to a current level of 11%. Isolating the departure cases between 1995 and

Rape

1998, 53% of the departures are cases of aggravation of the sentencing guideline recommendation, while 47% are cases of mitigation. These patterns of compliance and

-~
67%

Pre-TIS Sentencing GuidelinesCompliance Rates
January 1991 - March 1994

departure have been stable since the TIS guidelines were instituted.
Examining sentencing guidelines compliance rates by the 12 primary offense

Fraud

80%

groups reveals that compliance is neither consistent, nor the departure pattern

Drugs
%
771- ’

uniform, across the offense groups. The bars to the right show post-TIS compli-

Larceny 0

ance rates range from a high of 82% for larceny cases to a low of 62% for sexual

Burglary

assault cases. In general, higher races of compliance were found for property crimes
than the person offense categories-larceny, fraud, drugs, burglary (other than
dwellings) all had compliance above 70%. The sentences for person offense groups
(assault, burglary of a dwelling, homicide, rape, robbery, kidnapping, and sexual
assault) all had compliance rates below 70%.
Overall compliance within offense groups has not changed much as a result ofTIS

82%

Robbery
Murder/Homickie
-

Assault

Sexual Assault
1 Rape

72%

68%

66%

58%

legislation, although the changes that have occurred are more pronounced in the
crimes against the person categories. Under TIS, the person offense groups (including burglary of a dwelling and burglaries with weapons) receive statutorily mandated
midpoint enhancements that increase the guideline recommendation by a minimum
of 100-125%.74Further midpoint enhancements are applied in cases where theoffender
has a violent prior record, resulting in a sentence recommendation up to six times longer

than historical time served by violent offenders convicted of similar crimes under the old

parole laws. Undoubtedly, midpoint enhancements affect compliance rates, and the impact is likely not uniform across guideline offense groups. However, it is currently impossible to disentangle the role played by differential midpoint enhancements in overall
compliance.
Departures under TIS guidelines (measured by mitigation and aggravation rates) differ
significantly across offense groups. The table below shows that property crimes, fraud,
and burglaries of other structures (nondwellings) exhibit a marked mitigation pattern
among the departures, while drug and larceny offenses reveal patterns of aggravation.
517.1-805 ofCodeofVireinia.

The Impact of TIS an Judicial Compliance 47

Departures from the burglary of dwelling guidelines resulted in a mitigation rate
much higher than the other property offenses and similar to the rates of mitigation

among several of the person crime categories. The violent offenses of rape and robbery, and to a lesser extent assault and kidnapping, demonstrated strong mitigation
patterns. In fact, in more than one-fourth of the rape cases and over one-fifth of the
robberies, judges sentenced below the guideline recommendation. Despite the midpoint enhancement for violent current offenses and violent prior records, the guidelines offense groups of homicide and sexual assault showed stronger aggravation patterns from the guidelines than any other crime categories. To a certain degree, the
aggravation patterns for homicide and sexual assault offenses may reflect judicial
sentencing for “true” offense behavior in cases where a plea agreement resulted in a
less serious charge at c o n v i c t i ~ n . ~ ~
TIS Guidelines Departure Rates by Offense, 1995-1998

Total Cases
Examined

Mitigation Rate

Aggravation Rate

17.7%
19.8
15.5
10.2

14.0%
13.5
12.2
15.1

2,313
1,585
17,415

Assault
BurglaryDwelling
Burglary/OtherStructure

Dm9

2,001

FmUd

15.4

5.9

5,903

Kidnapping
Larceny

19.5

17.7
10.5

215
10,864

Murder/Homicide
Rape
Robbery

Sexual Assauk

7.1
12.6
29.0

21.9
11.4

22.3

610

8.8
14.5
26.8

468
1,928

938

With some notable exceptions, the implementation of TIS has not had a pronounced effect on compliance or departure rates (mitigations or aggravations). Furthermore, a majority of sentences fall within the guideline recommendations (i.e.,
for the case types listed, between 62% and 82% of the sentences complied with the
sentencing guidelines). The fairly high compliance rates may be, in part, an artifact

of the evolving nature of the sentencing guidelines. The VCSC updates the sentencing guidelines annually and continually fine-tunes the sentencing worksheets.% This
occurs by continually analyzing PSI data, completed guideline worksheet data, and
other information that comes before the commission. Some decisions to modifl
guideline worksheets are strictly data driven (as is the case with setting the ranges),

95 Offense scoring under Virginia’s sentencing guidelines is based solely on the conviction
offense, and unlike the United States Sentencing Guidelines, does not score the real offense
behavior in instances where a charge reduction occurs. Virginia’s guidelines do, however, account for elements of the crime such as victim injury and use of a weapon. Aggravation rate
for violent offenses, then, may reflect the desire on the part of judges to impose sentences
more closely in line with the actual offense committed rather than the offense to which the
offender plead guilty.
’K, Virginia’ssentencingguidelinesare based on a continuing analysis of judicial sentencingdecisions in the Commonwealth. This i s done to ensurethat judges are provided with guidelinesthat
reflect both historical sentencing decisions and changes in more recent sentencing decisions
(Judicial SentencingGuidelines Committee, 1993, p. 7).

Tmth-in-Sentencingin Virginia

and some are more qualitative (as is the case with increasing time-served amounts for TIS Guidelines Reasons For Departure
targeted offenders). However, most changes are a combination of quantitative and

JanUary 1995

- March 1998

qualitative input. For example, a relatively high departure for sexual assault cases 10 Most Frequent Mitigating Reasons
caused the commission to conduct a more in-depth study of convicted sex offenders.” As a result of this study, age of the victim was added to the sentencing worksheet

Alternativesanction
1

212%

Good rehabiiiion potential -16.7%

for sexual assault cases as a sentence enhancement.”’

Plea agreement

Judicial Departure Reasons
Compliance with the TIS guidelines, as with its predecessor (the guidelines in place
under the parole system) is voluntary. Hoyever, following the 1994 reforms, judges

Caoperativewith authwiti

were reguired to articulate and submit redons for sentencing outside the guideline

Sentenced by another court

recommendations. “The opinions of the judiciary, as reflected in their departure rea-

Minimal priw record

sons, are highly relevant to the Sentencing Commission as it deliberates on revision

No reason cited

recommendations. Unlike their counterparts in many other states using sentencing

Facts of the case

Weak case/evidence
Age of offender

guidelines, Virginia’s judges are not limited by any prescribed or standardized reasons

for departure set forth by the commission; they are free to depart for any reason they

to the General Assembly each December in the commission’s annual report, draw on
the opinions of the judiciary reflected in departure reasons. As a consequence, the
commission is active in encouraging judges to provide specific reasons for departure.
One important result is that, over time, judges are now more likely to give a reason
for their mitigated or aggravated sentences. “No reason cited went from being the
most common departure reason for both mitigation and aggravation (ranked number 1) at the end of 1995 to one of the least (ranked 9 of 10 for mitigation and
ranked 10 of 10 for aggravation) during the period 1995-1998.

Criminal Iifestyldorientation

7.3%

m5.5%
5%

=
=
=

4.8%
4.7%

4.4%

13.8%

Previous same offense
1-1

Plea agreement
Facts of the case
Recommendation too low

12.4%

12.2%

115%

7.9%

Jurylcommunilys e n t i i t -7%
Trudreal offense behavior

DrugamounVpurily
Sentencing consistency

No reason cited

During the first three years of TIS, mitigation cases reveal that the most com-

10.3%

10 Most Frequent Aggravating Reasons

find compelling and must only communicate that reason to the commission.”99

VCSC staff state that recommendations for revisions to the guidelines, submitted

10.3%

5.4%

m4.6%
.
I
4.2%

3.5%

monly cited reasons for departure were that an alternative sanction or community
punishment’00was imposed (21.2%)and that the offender had good rehabilitation
potentialioi (16.7%). For aggravated sentences, the most commonly cited reasons
-

The “ConvictedSex Offender”study found that three-founhs of all convictions in the sample
involved a sexual assault on a child under 18 years old, and almost half ofthe victims were under 13.
‘)I Recommendation 4: The sentencing guidelines for sexual assault offenses should be amended
by adding a factor to sectionsA and B to increase the total worksheet score in cases involving
victims who are under the age of 13 at the time of the offense. This modification significantly
increases the likelihood that sexual assault offenses involving victims under 13 will be recommended for prison, and, in rhe cases that will not result in a prison recommendation, this
modification ensures these offenders will receive a jail term. These recommendations do not
apply to rape, forcible sodomy and object penetration (Virginia Criminal Sentencing Commission (1 996), p. 79).
‘’Virginia’s Sentencing Commission Annual Report (1996), p. 20.
l W Detention Center Incarceration, Diversion Center Incarceration, Boot Camp Incarceration, intensive supervised probation, day reporting, and the drug court programs are examples
of alternative sanctions available to judges in Virginia.
Io‘ For instance, judges may cite the offender‘s general rehabilitation potential or they may cite
more specific reasons such as the offender‘sprogress in drug rehabilitation, a strong work record,
the offender’s remorse, a strong family background, or restitution made by the offender.
97

The Impact of TIS on/udicid Compliance 49

were the overall criminal liktyldorientation of the offender (13.8%)and that the offender had previously been convicted for the same offitnse (12.4%).The charts on the
previous page list the ten most frequently cited judicial reasons for sentence departure.

8%-

sible explanations for the downward trend. First, starting in 1987, data and analysis

surprising that criminal defense attorneys are increasingly reluctant to steer their clients

The Virginia General Assembly enacted provisions for a system of bifurcated jury trials that
became effective beginningJuly 1, 1994. In bifurcated trials, the jury establishes the guilt or
innocenceof the defendant in the first phase of the trial, and then, in a second phase, the jury
is presented with information on the offender’s background and prior record to assist jurors in
making a sentencing decision.
Virginia Criminal Sentencing Commission (1 997), p. 37.

50 Truth-in-Sentencingin Virginia

Since the implementation of TIS, the overall compliance rate of jury sentences
with the sentencing guidelines has been 43% compared to 76% in nonjury trials.
The majority of the departures have been aggravations (i.e., 45% aggravations for

Guideline Compliance Rates Comparing
Judge and Jury Sentences
(January 1995 -March 1998)

jury trials and 12% aggravations for nonjury trials). There has been virtually no
difference in the rate of mitigated sentences for jury and nonjury trials since the
implementation ofTIS.
In Virginia, judges are permitted by law to reduce a jury sentence they feel is
inappropriate. More often than not, however, they do not amend the sanction. For
example, just after the implementation ofTIS, judges modified about 29% of jury
--

sentencing cases. In cases modified when the jury was outside the guideline range,

hgravation

1129(45%
I Jury
I Judge

nearly half (45%) were cases where the find sentence was still outside the guidelines
recommendation. Judges brought a high jury sentence into compliance with the guide-

0% 20% 40% 60% 80%

line recommendation in 0 n h f . w out of ten modijcations. Unlike overall compliance
and departure rates, judicial modification patterns appear to have changed since the
implementation of TIS. Specifically, 86% of judicial modifications after TIS were
made to jury sentences outside the guideline recommendation compared to 69% of
judicial modifications in the last year of the old parole system.
Compliance rates in states with sentencing guidelines range from 75-100%. Comparing compliance rates across states is only useful for portraying differences in how
guideline systems have been developed or modeled. A 100% compliance rate (in
North Carolina) simply means judges are bound statutorily to adhere to guideline
recommendations. States with lower compliance rates may have drawn narrower sentencing ranges, or may measure compliance differently depending on the purposes
of monitoring.

To the extent that the goal of sentencing guidelines is to structure judicial discretion, not to eliminate it, then some level of departure is to be expected-if
couraged-in

not en-

order to account for atypical cases. This perspective differentiates sen-

tencing guidelines from mandatory sentencing. In Virginia, the majoriry (between

72-76%)of prison sentences handed down by judges pre- and post-TIS have complied with sentencing guidelines. The consistently high level of overall compliance
indicates that guidelines were developed and statistically modeled in a fashion consistent with past sentencing practices. In addition, a compliance rate in the 70-80%
range shows that judges are reasonably satisfied with guidelines recommendations.
The most recent figures (updated June 1999) show overall compliance at a high of

78%. Commission staff speculate that the recent increase in compliance may be
related to media reporting of compliance rates by name of judge.
It is important to note that patterns of judicial compliance vary when examined
for individual case types. Thus, it would appear that targeting adjustments to the
sentencing guidelines for specific case types and circumstances (eg., rape sexual assault, robbery) would be a reasonable way for the VCSC to maintain or increase
compliance rates. In fact, it is an ongoing strategy of the VCSC to target individual
offenses or specific scoring factors for revision on the worksheets.

The Impact of TIS on Judicial Compliance

CHAPTER S I X

Estimating Preventable
Crime Under TIS
Virginia, legislators wanted to know how the extended incapacitation of violent
offenders under TIS would effect crime rates. Specifically, they asked for information
on how Virginians would benefit from locking up violent offenders for longer peri-

ods of time. Implementing the governor‘s proposed sentencing reforms would re-

quire spending a larger share of the public treasury on housing violent offenders.
Under normal circumstances, imprisoned offenders do not pose a threat to the general public. But is the cost associated with giving certain offenders lengthier sentence~justified through a reduction in the amount of crime they might otherwise
commit if they had been released earlier? Is there a beneficial “incapacitation effect”
associated with TIS?
This chapter reviews two VCSC studies that estimate the “benefits of incarceration” in terms of the amount and value of crime prevented by sentencing reform.
n Estimate of preventable crime and recidivism under TIS: How much new crime is

prevented when certain offenders serve longer sentences?

m Estimate of the cost of crime prevented under TIS: What is the benefit (or cost
savings) to society from having fewer victims of crime?

There is no generally accepted method for determining the amount of crime prevented through longer prison sentences.’” Much of the literature on this subjectto5
focuses on ways to measure and calculate a theoretical criminal career parameter
lambda (A), which is the frequency (average annual rate) of offending by active offenders (sometimes referred to as an individual offending frequency). Given knowledge of 3L for a particular category of inmate (based on offense seriousness, prior
record, and other offender characteristics) and the expected Length-of-Stay, Ti, for
that inmate, the number of offenses prevented by incarceration of that inmate would
be equal to h(Ti).
The total number of preventable crimes (C) for N offenders of a
parricular category could be estimated as
N

C=Cax
i

A fUndamental unresolved issue wirh this approach to estimating the number of
crimes prevented by incapacitation is how to measure rates of offending.lW For example, controversy remains as to whether offending patterns vary with the age of the
offender’07or remain relatively constant over the offender’s active criminal career.’O*
lMSee, e.g., Gottfredson, and Hirschi (1986); Blumstein, Cohen, and Farrington (1988);
Zimring and Hawkins (1988).
’‘’See, e.g., &hen (1978);Horneyand Marshall (1991).
Cohen (1986);Visher (198G);HorneyandMarshall(l991); Marvel1 and Moody (1994).
IO7 Gottfredson and Hirschi (1986).
’” Blumstein, Cohen, and Farringon (1988).

Truth-in-Sentencingin Krgz’nia

Moreover, the information required to determine rates of offending for various classes
of offenders would likely be significant. Indeed, given these basic concerns, some reject

h. entirely as a useful
The staff of the VCSC opted to develop a methodology for counting “preventable” offenses that avoided the uncertainties associated with the measurement of ?L
by actually counting the number of offenses that occurred between the inmates’
actual release date and the later release date proscribed by Proposal X.”’ The VCSC
study was designed to identify preventable convictions based on an analysis of offenders released from prison between 1986 and 1991 who recidivated with a new
-

felony (nondrug) conviction between 1986 and 1993. Crimes committed by offenders released prior to 1986 were excluded from the study.
The study of preventable crime produced by the VCSC included: (1) developing a
framework to estimate “preventable”recidivism, (2) compiling a comprehensive database to study preventable recidivism (1 986-1993), (3) developing a projection of
preventable recidivism for 1995-2005, and (4) forecasting preventable crime from

1995 through 2005. For purposes of this analysis, recidivism was measured by a new
felony (nondrug) conviction. Measuring recidivism in this way provides a conservative estimate of preventable crime because felony convictions are only a fraction of
the number of crimes actually committed.“l

Step One: Developing a Framework to
Estimate Preventable Recidivism
The study began by estimating recidivism that would have been prevented between 1986 and 1993 by the extended incapacitation of violent offenders. A sample
was drawn consisting of offenders who would have been subject to normative sentence adjustments (due to the nature of their current offense and/or prior criminal
record) under the Governor’s plan (Proposal X) and who were released from prison
during the period 1986 to 1991. Because the last release dates for this offender group
occurred at the end of 1991, and subsequent criminal activity was tracked through

1993, all offenders in the sample were monitored for a minimum of two years following release.

To identify any felony convictions that occurred after the offender’s release, each
case in the sample was tracked using the Pre/Post-Sentence Investigation (PSI) database. First, a revised release date was calculated to approximate the date the offender
would have been released had Proposal X been in effect at the time of the offender’s
original conviction. The new release date was calculated using the midpoint value of
the recommended sentence range under the Proposal X sentencing guidelines for
each offense type.
“Preventable” offenses were identified based on whether they occurred after the
offender’s actual release and prior to the Proposal X release date. These offenses were
IO9

‘lo
‘I‘

See, e.g., Gottfredson and Hirschi (1986).
Criminal Justice Research Center (1994).
For a more complete discussion of measuring recidivism, see Chapter 7.

Ertimating Preventable Crime Under TIS 53

considered preventable because they presumably would not have occurred if the offender was still incarcerated. New felony convictions occurring after the Proposal X
release date were not considered preventable. Felony drug offenses were not considered preventable because the nature of the drug trade is such that “replacement”
effects would have almost certainly occurred.

Step 2: Compiling a Comprehensive Database to Study
Preventable Recidivism (1986-1993)
In this step, the necessary data were identified and assembled. This included:
Producing a distribution of historical time-served amounts under the pre-TIS guide--

lines (by offense type) for offenders who would be affected by the normative sentence adjustments.
Calculating a recidivism rate for this affected group of offenders by determining
the percentage of offenders in this category released from prison or jail who subsequently were convicred of a new felony (nondrug) offense (r).
Calculating the average number of preventable felony convictions (using the PSI
database) per recidivist offender in the affected sample (0.
Deriving two additional distributions showing the time across all recidivist offenders from (1) release date to a new violent felony offense and (2) release date to
a new nonviolent (nondrug) felony offense resulting in conviction.

Step 3: Forecast of Preventable
Felony Convictions (1995-2005)
VCSC staff began by forecasting the number of offenders who would be convicted
of offenses subject to the normative sentence adjustments under Proposal X. The
forecasts were produced by an ARIMA (Auto-Regressive Integrated Moving Average) model using monthly data on convictions for the targeted offenses from 1985
through 1993. Monthly forecasts were produced for the period from January 1995
through December 2005.

A release date for each offender in the forecast was determined using the average
historical time served for the offender’s offense class (derived from the distribution of
historical time served assembled in Step Two). An estimate of the total number of
offenders released each month in the forecast horizon was produced (Rj, where j
represents the month of release) by summing (across offense type) the forecasted
number of offenders (xij, where i represents the offense type and j represents the
month) convicted of offenses subject to the normative sentence adjustments under
Proposal X who were expected to be released during month j
Rj=& xij,

The recidivism rate (r, derived in Step Two) was applied to the forecast of offenders expected to be released during each month of the forecast horizon. The product is

an estimate of the number of offenders released for normatively adjusted offenses

Truth-in-Sentenringin Krginia

who will recidivate with a new felony (nondrug) conviction (Dj), for every month in
the forecast horizon, where
Dj=rRj
The number of preventable felony convictions per month (produced by the recidivist offenders affected by Proposal X) was estimated (Cj).The estimate was produced
by taking the product of the number of preventable felony convictions per recidivist
offender (f, calculated in Step 2) and the forecast of the number of recidivist offenders released each month of the forecast horizon (Dj) as follows
.-

Cj=fDj

At this stage, the forecast of these preventable felony convictions must be distributed across the months after the offenders’ release to simulate the pattern in which
these offenders actually’recidivate following their release horn prison. To this end,
preventable convictions were first disaggregated into violent and nonviolent preventable offenses and then by new offense type. This was accomplished by applying the
proportion of preventable violent convictions (pv) to the forecast of preventable total
felony convictions per month (Cj). Thus, the number of preventable violent felony
offenses that resulted in conviction per month (Vi) was equal to
Vj= pv(Cj)
while the number of preventable nonviolent felony convictions per month (Pj ) was
equal to
Pj=(l-pv)( Cj)
Once the number ofviolent and nonviolent preventable convictions for each month
between 1795 and 2005 was estimated, the next step was to distribute these convictions across time using the two distributions calculated in Step 2 (the time from release
date to either (1) a new violent felony offense or (2) a new nonviolent (nondrug)
offense that resulted in conviction). This step produced estimates of both the number
ofviolent [Nv(j)] and nonvioleiit [Np(j)] preventable offenses (resultingin conviction)
expected to occur each month between 1995 and 2005.
In summary, the specific types of iolent and nonviolent offenses expected to be

committed by recidivist offenders were estimated using proportions derived from
historical data (Step Two). The result was a forecast of the number of preventable
felony offenses (by offense type) expected to result in conviction during each month
between 1975 to 2005.’12These offenses were then distributed across future months
‘ I 2 If the (hiscorically derived) proportion of prevenrable violent felony offenses accounced for
by murder was represented by p(l), for rape by ~(21,
for robbery by p(3), and for assault by
p(4), then the number of preventable murders occurring during m o n t h j would be equal to
p(I) NuQ),the number of preventable rapes would be equal to p(2) NJ), the number of
preventable robberies would be equal to p(3) Nnfi),
and the number of prevenrable assaults
would be equal to p(4) 4.0).Similarly,if the (historically derived) proportion of preventable

Estimating Preventable Crime Under TIS 55

Forecast of PreventableFelony Crimes
Under the Commission'sPlan (1995 - 20051
Fmm 1995-2005an estimated 119,969 fehy crimes
would be prevented under the CommissiOn's h.

based on historical recidivism patterns. The final result was the number of both
violent and nonviolent preventable offenses (resulting in conviction) expected to
occur each month between 1995 and 2005.

Step 4: Forecast of Preventable Crime (1995-2005)
The number of reported crimes always exceeds the number of criminal convictions. During this step, the estimated number of preventable felony convictions was
used to estimate the overall reduction in reported felony offenses attributable to TIS. '
The ratio of the number of index crimes'" reported to the police to the number of
convictions for index crimes between 1991 and 1993 was calculated. For example,
--

during this period, there were 6.7 rapes reported to the police for every rape conviction. These ratios were then applied to the forecast of the number of preventable
convictions for each index offense category to produce estimates of future preventable index crime reported to the police. For example, if the ratio of the number of
rapes reported to the police to the number of convictions for rape is designated as rp,
then number of preventable rapes in month j [Pr (j)] is estimated to be equal to

Pr (j)

= rp

p(2) Nv(j),

where
p(2) = proportion of preventable violent felony offenses accounted for by rape
Nv(j) = the number of preventable offenses (resulting in conviction) in month j,
Using this methodology, the CJRC estimated that there was an average of 12 felony
offenses reported for each felony conviction (across all index offense categories, reported over a multiyear period). This average reported offense-to-conviction ratio
implies that for every future preventable felony conviction there would be an additional 12 index crimes prevented (and thus not reported) due to the extended incarceration of offenders under Proposal X.
The trend lines here show the forecast (1995-2005)of preventable reported felony
crimes under Proposal X. More than 26,000 violent and 93,891 nonviolent felonies
were expected to be prevented by the implementation of Proposal X between 1995
and 2005.'14

Conclusions
The methodology for estimatingpreventable crime just described is analytically com-

plex and makes numerous interrelated behavioral assumptions. As a consequence, the
nonviolent felony offenses accounted for by burglary was represented by p(5), for arson by

p(@, and for motor vehicle theft byp(7), then the number of preventable burglaries occurring
during month j would be equal to p(5) Npfi),
the number of preventable arsons would be
equal top(@ N,fi)and
, the number of preventable robberies would be equal top(3 Npfi);).
Murder, rape, robbery, assault, burglary, arson, and motor vehicle theft.
Note that the Criminal Justice Research Center also developed a more comprehensive estimate of the cost of recidivism in Virginia (Criminal Justice Research Center, 1994a) which also
included law enforcement, correctional and judicial, as well as victim, costs of d (i.e., not just
preventable) felony recidivism in 1993. These were estimated to total $670 million in 1994.
'I4

Truth-in-Sentencingin V&h'a

accuracy of the estimates may be affected by potential sources of statistical error and
possible challenges to the assumptions. First, whiie not a criticism, the choice of study
period constrains the results. Forecasted recidivism for theyears 1995-2005is based on
the patterns and experience of an earlier period of time (1986-1993)that may no
longer be representative. Falling crime rates and record employment b e l s after 1993
may indicate a change in many criminal careers. Second, primary results, such as the
number of violent felony convictions prevented by Proposal X are calculated by combining many separate estimates. Each estimate contains potential measurement error’15
that is exacerbated when the individual estimates are combined. Third, the manner in
which repeat crimes were distributed across time (i.e., using average time to recidivism)
.-

could be challenged as unrealistic.
O n the other hand, the estimates were produced-quickly using a carefully conceived method designed to make the most out ofavailable data. The approach avoided
attempts to measure complex, theoretically challenging quantities such as h through
expensive and time-consuming longitudinal research. In addition, there are several
reasons to believe that these estimates met a basic goal of producing conservative
estimates of preventable crime. Several other studies on this subject use much higher
ratios to estimate the actual number of crimes committed by an offender compared
to each felony conviction. Zedlewski (1987),in his analysis of the costs and benefits
of confinement, cited a Rand Corporation survey of inmates in California, Michigan, and Texas that found the average number of crimes committed per year by an
offender was 187,with a median of 15 crimes per year. DiIulio (1990), in a survey of
425 Wisconsin inmates, found the average number of crimes committed per year to
be 141, with a median of 12 crimes per year.
Estimated Cost Savings Resulting
from Preventable Crime
The primary benefit of prevented crime under TIS is that there are fewer victims
of crime. The legislature asked the VCSC to estimate the “costs of crime” avoided by
individuals who did not become crime victims due to the extended incapacitation of
violent offenders under Proposal X. As in the case of preventable crime, there is no
widely accepted method to make such a determination.”’ Miller, Cohen, and
Wiersema contend that the costs of crime to victims are mainly (1) out-of-pocket
expenses such as medical bills and property losses; (2) reduced producrivity at work,
home, and school; and (3) nonmonetary losses-such as fear, pain, suffering, and
lost quality of life. While some of these losses are tangible and easily quantified, the

‘I5 For example, the ratio of reported felonies to convictions for each offense type ignores the
lag relationship between reported offenses and convictions (i.e., convictions must follow the
reported crime though not necessarily during the same year) which obviously causes some
measurement error.
‘I6 See, e.g., Haynes and Larson (1984); Zedlewski (1987);Zimring and Hawkins (1988);
Baird (1993);Levitt (1996);Miller, Cohen, and Wiersma (1996);Block (1997).

fitimating Preventable Crime Under TIS 57

intangible losses (such as quality oflife) may also be valued in dollar t e r r n ~ , ”though
~
there is less agreement on the best method for accomplishing this. The CJRC drew
on data provided by the Federal Bureau of Investigation (FBI), the Virginia State
Police, National Council on Compensation Information, Jury Verdict Research, Inc.,
and the National Fire Incident Reporting System, to develop “a very conservative”
estimate of the costs of crime (to victims) that would be prevented (or avoided)
under TIS.
The CJRC’s approach drew extensively on the methods used by Miller and Cohen,
and their associates, in a series of studies designed to measure the cost of crime.L18

CJRC identified and measured victim cost of crime by focusing on the following
victim “cost centers:’’”9
Medical Costs were derived from the Detailed Claims Information (DCI) database of the National Council on Compensation Insurance. This database longitudinally tracks medical costs for injured persons. The injury distribution from National Crime Viccims Survey {NCVS) was then applied to the cost figures.
Mental Health Costs associated with psychological injury were computed using a
study of 39 1 South Carolina victims (women) of violent crimes. The rate of injury
was then applied to “won” jury verdicts for emotional distress and severely disabling psychological injury. The rate that “psychological injury” occurs (as measured by the PSI database) is roughly the same for both men and women - 33.5%.
Emergency Response Costs were estimated at $144.00 per injury based on the
National Medical Care Utilization and Expenditure Survey, 1980.This figure likely
understates considerably the current costs of emergency response.
Productivity Losses were estimated in the short and long terms. Short-term esrimates were based on lost work days reported in the NCVS, combined with data
on average daily earnings for those who work. For students (victims under age 1’3,
the estimated value of lost school days (daily cost per pupil) was used. For longterm estimates, injury codes (ICDs) for victims of violent crimes were used in
conjunction with reported hospital status times.
Program Administration Costs were defined as the administrative costs of health
and disability insurance. These were estimated by multiplying the costs of health
and disability insurance by the percent reimbursed.
Lost Quality of Life was estimated using rwo approaches: willingness-to-pay and
jury awards for pain and suffering.
Willingness-to-pay, typically assessed by means of a survey insuument,lZ0measures the
amount that people are willing to pay for day-to-day safety and to maintain their
existing quality of life (defined across such dimensions as cognitive, mobility, sensory, and cosmetic that may be diminished by crime).

‘I7
’I8
‘I9
12’

Cohen (1988).
Miller (1990);Miller, Cohen,and Rossman, (1994);Miller, Cohen, andWiersema (1996).
Chabotar (1987).
Mitchell and Carson (1989).

Tnth-in-Sentencingin Krp’nia

a Jury awards for pain and suffering are also used to estimate the lost quality of life

resulting from crime. These are a function of medical care, productivity costs,
category of injury, and mental health care related to emotional distress. Jury awards
are based on a standard of compensation that has been defined by the courts as
“one which permits the jury to award a “fair” and “reasonable” amount that com-

EstimatedVictim Costs’ Associated With Preventable
Crime Under The Commission’s Plan, 1995-2005
Violent Crime
$ Millions
2500

pensates for pain and suffering.This was measured by examining actual amounts

awarded by juries. Data were taken from Jury Verdict Research, Inc., which col-

2000

lects award information on virtually all personal injury cases in civil proceedings.

1500

The company claims that it can predict court awards within + or - 7%.

1000

Results of Cost Analysis
The study of preventable crime under Proposal X forecasted compensatory damages by crime type based on the relationship between medical cost and productivity

500

0
1995

1997

1999

2001

2003

2005

losses and jury awards. For example, the estimated cost (or value of a statistical life)
for a murder was calculated as follows:
Medical ...................................................
Emergency Service
Productivity

........................................

Non-Violent Crime
$ Millions

$6,467
$520

500

- -

...........................................$656,192
cnme is esiimatedal$4400 mdkon

...................................
$663,179
Mental Health .....................................................
0
QualiryofLife .....................................
$1,715,918

Total Monetary

Total Cost

........................................

$2,379,097

Based on the preventable crime and victim cost analyses, the commission estimated
that the value of crime prevented by the implementation of Proposal X between

1995

1997

1999

2001

2003

2005

‘1989dollars adpsted fcr inilati,

1995 and 2005 would yield a cumulative savings to victims and society of $2.7
billion. The trend line here shows the estimated victim costs (1989 dollars adjusted
for inflation) associated with forecasted preventable violent and nonviolent crime,
respectively, under Proposal X, 1995-2005.

Conclusions
The cost analysis, based on a highly regarded methodology developed by Miller,
Cohen, and their associates, incorporated a number of elements designed to keep the
estimates conservative. First, the cost estimates did not include a number of preventable crimes because cost data were not available. In addition, certain cost centers suggested by Miller, Cohen, and Wiersema’2’were absent from the CJRCg estimates (e.g.,
sociallvictim services) due to a lack of data, while others (e.g., mental health cost estimates) are based on estimates that likely understate the true costs. Finally, lost quality
of life is the largest cost component in the estimates and also the most difficult to
measure. The use of both willingness to pay and jury awards are conservative and

Miller, Cohen, and Wiersema (1996).

Ertimating Preventable Crime Under TIS 59

reduce concern over this source of measurement error. As a consequence, it appears
that the CJRC approach produced an estimate that can be viewed as a lower-limit to
the costs of crime to victims avoidable by the implementation of TIS in Virginia.
Additional analyses using the cost-savings estimates may have been useful to
policymakers. While a comprehensivebenefit-cost analysis of preventable crime would

be extremely difficult, a more limited comparison of the costs associated with extended incarceration of offenders with the cost savings to victims and society could
have been attempted. Extended incarceration, while increasing correctional costs,
reduces court and law enforcement expenditures associated with arrests for preventable crimes. These additional savings to government, along with the savings to victims and society, could be compared to the costs of extended incarceration and used,
for example, to justifv new prison construction. While officials laid the groundwork
for such a comparison, it was never actually conducted.

60 Truth-in-Sentencingin Virginia

CHAPTER SEVEN

A fundamental issue in evaluating Virginia‘s new TIS policies is their impact on the
amount of new crime being committed in the Commonwealth. Although the major
objective of TIS reform was to ensure more certain punishment and longer prison
terms for violent offenders, policymakers also raised the difficult issue of “what impact
the new sentencing system may be having Qn Virginia’s crime rate.”’**Have the new

k
laws helped to deter some persons from committing
new crime because of the knowledge of tougher penalties under TIS?Does incarcerating violent offenders for longer

periods of time under TIS help reduce the chances that they will commit new crimes
when they are eventually released from prison?This chapter examines one critical aspect of the relationship between sentencing reform and the crime rate in Virginia: Has

TIS helped reduce the level of offender recidivism in Virginia? Criminological research
shows that a relatively large share of crime is committed by a small pool of known and
repeat offenders. IfTIS policies are successful in reducing offender recidivism, then it is
likely that these policies will help reduce the crime rate generally.

As a first step in assessing what, if any, impact TIS is having on the level of offender
recidivism, this chapter establishes the recent historical baseline of crime in Virginia.
The second section discusses a new initiative-the Offender Notification Release
Program-designed to inform offenders being released from prison about Virginia’s
new sentencing laws. The final part of this chapter reviews the design of a long-range
recidivism study and analyzes the pattern of recidivism for offenders released from
prison prior to the implementation ofTIS.
The Current Level of Crime in Virginla
Benveen 1993 and 1997, reported crime in Virginia declined. The overall rate of

“index crime”’23in Virginia (per 100,000 population) dropped by over 8% from

4,210 in 1993 to 3,870 in 1977. While there was a slight increase in four of the
index crimes berween 1976 and 1997, the rates of ail eight index crimes have declined over the past five years.
index Crlmes In Vlrghla, 1993-1997

1993
1994
1995

1996
1997

3,Qn

3,870

’” Virginia Criminal Sentencing Commission (1997),p. 73.
Index crimes are defined as murderhon-negligent manslaughter, forcible rape, robbery,
aggravared assault, burglary, larceny, motor vehicle theft, and arson.

Assessing the lmpact of TIS on Recidivism

Index Crime in Virglnia by Crime Type, 1993-1997
Percent
Change
1993

1994

1995

1996

1997

1993-1997

-14 7

Forcible Rape
Aggravatd Assault

193

192

-19.2

133

122

124

-14.1

197

183

185

-4.0

602

582

562

-17,O

-. ’

2,832

2,785

2,767

2,744

2,657

-6.2

-4.8

-4.4

Arson

The cause of this dedine is dificult to interpret. O n one hand, the dedine in the
rate of violent crime in Virginia is in line with a pattern observed nationally. The rate
of index crime in the United States has fallen from 5,483 in 1993 to 4,923 in 1997
sparking a debate over why and how long this trend will last. On the other hand, the
implementation ofTIS in Virginia and a drop in the state crime rate raises the possibility that the two events are related. The.issue of whether the drop in Virginia2
crime rate can be attributed to sentencing reform or some other combination of
initiatives is complex and requires considerable longitudinal data that are simply
unavailable at this time. The following sections of this chapter take important first
steps in addressing this issue by examining an innovative new approach to reducing
hture offender recidivism and establishing baseline recidivism measures for offenders released from prison prior to the TIS reforms.

Offender Notification Release Program
A deterrence effect is one way for TIS to reduce recidivism in Virginia. It may be
that knowledge of the tough new penalties deters some previous offenders who would
otherwise have broken the law again from committing new crimes, or at least certain
types of crime. The criminological literature refers to this concept as specific deterrence: the degree to which the threat or actual application of punishment will deter
an individual who has committed a crime from engaging in crime again.”* The
Offender Notification Release Program (ONRP) was developed in 1996 as a joint
effort of the VCSC and the Department of Corrections (DOC) to educate inmates
leaving Virginia prisons specifically about the TIS reforms. The program provides
exiting inmates an overview of the sentencing system since the abolition of parole
and the institution of tougher sentencing laws for violent and repeat offenders. On
average, a returning violent offender sentenced under the new guidelines should
expect to serve two to six times longer than under the state’s old law.
This concept is distinct from generaldeterrence,which is the degree to which knowledge of
criminal penalties deters members ofthe general population, not just those convicted of crimes,
from engaging in criminal behavior. General deterrence effects are very hard to measure because of the difficulty of assessing the depth of knowledge people have of criminal punishments and what, if any, impact this knowledge has in preventing them from committing
crime. At this time, the VCSC is not undertaking any study of general deterrence under TIS.

62 Twth-in-Sentencingin Krginia

The program has two purposes: 1) to inform inmates about to re-enter society of the
changes in Virginia’ssentencing and parole laws, and 2) to reduce the likelihood of recidivism. A number of criminological studies of the deterrent value of new punishment
initiatives have produced mixed results, with some researchers concluding that many
offenders were unaware of the change in sanctions designed to influence their behavior.
From a theoretical perspective, the VCSC and the DOC believe that the deterrent
value of specific punishments under TIS might be increased if the targeted population
(released inmates) is adequately informed of the new sanctions for hture misconduct.

As part of the offender notification program, all inmates leaving the prison system
are given a cype of “exit interview” where they are informed about the abolition of
parole and the old good conduct credit system. Each departing inmate receives a wdlet-sized “notification card” that contains the possible sentencing consequences of being arrested and convicted of a new felony offense. The program became operational
statewide in January, 1997. Virginia’s ONRP is the first of its kind in the nation.

The ONRP Process
Each correctional facility in Virginia has a supply of white and yellow cards that
indicate the amount of time an offender can expect to serve if convicted of a new
murder, rape, robbery, or aggravated assault after release from prison. White cards
are given to inmates with a nonviolent record and the yellow cards are given to
inmates with a violent record. The two cards show different expected time-served
amounts because sentences are increased for offenders with violent prior records.
The time is compared with the average time served under the previous sentencing
laws that allowed for early release on parole. The next page shows the front and back
of the ONRP cards (redrawn from the originals).
The Community Release Unit located within the DOC Division of Operations
determines which card the inmate will receive based on a review of the inmate’s
record. This review is triggered in part on a form obtained from the Court and Legal
Services Unit that predicts a release date based on good rime and parole eligibility
(for those offenders serving sentences under the old parole system). The record review helps to determine if the inmate has any outstanding charges to answer, other .
sentences to serve, or whether the inmate will be transferred out-of-state for similar
reasons. In addition, the review identifies whether the inmate has a history of violence and therefore should receive a yellow card. All correctional facilities have been
provided with a comprehensive list of all violent offenses. A “Notification of Release
Post/Probation Supervision” form is then fixed to the facility indicating which card
is to be assigned.
After the institution receives the DOC release information, correctional staff review the inmate’s on-site records to make sure the correct card is assigned. OEicials
are required to give the card to the inmate as close to the day of release as possible,
ONRP cards are handed out to inmates convicted of felonies who are classified as
state responsible (rhose given state prison sentences of six months or more). This

Assessing the Impact of TIS on Recidivism 63

ONRP White C
a
d

Front

WARNING: Viiinia has abolished parole and inposed mudl longer prison sentences on crbrjnalswithpastrecords.

m Virginia has made big changes inthe way we sentence convictedcriminals. Put Simply, IF YOU
COMMIT A VIOLENT CRIME IN VIRGINIA IN THE FUTURE, YOU WILL LIKELY BE SENT
BACK TO PRISON FOR A VERY LONG PERIOD OF TIME.
I There is no more parole. The entire sentence bnposed by the judge or jury wili be served, with
good time credits limited to five weeks per year at most.

Most importantly of all, should you commit a burglary or any other vident crime you will serve
FAR MORE HARD TIME than under the dd system. The back of this card shows some ex-amp& of the ACTUAL PRISON TIME you Wa face if you are convictedin Virginia.

I We expect you to obey the hws and build a productivelife after release. But we want you to
understand the very serious consequences if you commit future violent crimes in Virginia.

ONRP White Card

- Eack

Actual Prison Tme to Serve Under Virginia’s Guidelines
These recommendationscan be increasedbased on your prior recordand the facts of the case.

old System

New No Parole System

First Degree Murder

11Years

28 Years - Life

serious Assault

1,SYears

3 Years 9 years

Robbery

2 Years

5 Years - 14 Years

Rape

5 Years

13Years 33 Years

of Conviction

ONRP Yellow Card

- Back

Actual Prison Time to Serve Under Virginia’s Guidelines
These recommendationscan be increased based on your priir recordand the facts of the case.
Type of Conviction

Old System

New No Parole W e m

First Degree Murder

11Years

50 Years - Life

Serious Assauk

1.5 Years

6 Years- 9 years

Robbery

2 Years

9 Y e m - 14 Years

Rape

5 Years

22 Years - 33 Years

64 Truth-in-Sentencingin Virginia

includes persons at work release centers but excludes those in local jails and the state
boot camp (boot camp is considered a probation sentence).

DOC Response
The Director of DOC informed all regional directors, wardens, and superintendents of the ONRP implementation, stating:
“Issuingwarning cards is a serious matter for all Department of Corrections employees. The cards are designed to inform all inmates of the consequences of committing future violent crime in Virginia. Inmates must be aware that in Virginia,
they are likely to serve a much longer sentence fbr committing a violent crime.
S& responsible for issuing the cards need to explain the card to the inmate.
Counseling staffshould also spend time explaining the consequences of change in
the sentencing guidelines prior to the inmate being released. When issued properly the warning card can act as a deterrent to committing a future violent crime.”
(DOCmemorandumporn Ron Angebne, Director, December 3,1936)
The director assigned the manager of DOC Classification and Records to oversee
implementation and training for the ONRP program. Each institution, field unit,

and work release center was required to send at least one representative to a training
session conducted by officials from the VCSC. Training occurred at four regional
locations with an average attendance of 30 people. Training sessions were short, with
attendees being given general program information and working through some hypothetical release scenarios.
The NCSC evaluation team interviewed a number of individuals who work for
D O C about the implementation of the ONRl? Support for the offender notification
concept was strong, with several recommendations made to enhance the overall effectiveness of this program:
Provide a video tape explaining the ONRP to inmates. Several DOC managers at

local facilities suggested a video tape to ensure a consistent and accurate explanation
of the system. Inmates currently view videos on other matters, and those interviewed
feel that an ONRP video could be easily integrated into existing release procedures.
Provide ongoing training. Managers indicate the need for ongoing training on
program goals and how best to administer the card. In particular, a process should
be developed to inform new correctional officers of the program.
Make it easier to get ONRP cards. Oficials at one institution have found it difficult to keep an adequate supply of ONRP cards. They mentioned having to ask
for cards from a nearby institution when their own supply ran out.
Review the card more than once with exiting inmates. Oficials at several institutions stated that inmates were busy thinking of other things upon release, including living arrangements, transporcation from the facility, personal finances, etc.
The ONRP card was seldom a high priority as offenders prepared to leave prison.
Officials mentioned a strategy of reviewing the card several days prior to release
and again at release to increase awareness of Virginia’s new sentencing laws.

Assessing the Impact of TIS on Recidivism 65

The VCSC and other’policymakers are interested in determining whether the

ONRP increases the potential deterrent effect ofVirginia’s sentencing reforms among
offenders being released from prison. This issue will be examined as part of a broader
two-phase study ofoffender recidivism in Virginia. The first phase establishes a baseline
recidivism rate for a cohort of offenders who were released from prison before TIS
and the ONRP went into effect (the cohort is composed of oflenders released in FY

1994). The second phase of the study (funds permitting) will begin at a yet-to-bedetermined date and examine recidivism rates for offenders released after the implementation of sentencing reform.

Recidivism in Virginia
To determine whether TIS and ONW policies have affected offender recidivism,
project staff have established baseline recidivism rates for the offender population
released from prison prior to the introduction of reform in January, 1995. The longrange plan is to compare the recidivism rate of offenders released pre-TIS (phase
one) with the recidivism rate of offenders released post-TIS (phase two). The VCSC

is now deliberating on when the second phase, measuring recidivism for those released after exposure to TIS and the ONRP, should begin.

Sampling Methodology
The baseline recidivism rate was developed by examining recidivism among a sample
of offenders released from the Virginia Department of Corrections in FY1993. The

sampling frame was prepared as follows:
Offenders appearing on the release file who died in prison or were executed during

FY1993 were excluded (53 cases).
Offenders who had previously been released from prison for the current incarceration term (parole violators) were excluded (1,722 cases). The results of the analysis, therefore, reflect recidivism among offenders after their first release from prison
for the current term of incarceration.
Offenders imprisoned for offenses other than completed or attempted person, property or drug crimes (offenses such as habitual traffic, weapons, arson, gambling,
conspiracy to commit a felony) were excluded (1,742 cases). Over half (54%) of the
offenders excluded in this step were convicted of habitual t&c offenses and were
imprisoned under Virginia’s 1&month mandatory minimum penalty law.
Offenders admitted prior to January 1, 1985, were excluded, since these cases
predated the statewide standardization of the Pre-/Post-Sentence Investigation (PSI)
report (243 cases). The PSI system will serve as the source of extensive prior record
and socio-demographic data for the offenders included in the study sample. The
offenders excluded in this step comprise less than 3% of released offenders remaining at this stage. Their exclusion affects a larger portion of the violent offense
groups than the property and drug groups: 42% of the remaining murderers an2
26% of the remaining kidnappers were admitted to prison prior to 1985, compared to less than 1% of drug offenders.

66 Truth-in-Sentencingin Virgiinia

A disproportionate stratified random sample of 1,400 cases was drawn from the
sampling frame (N= 8,089).
Offenders convicted of crimes against the person (murder, manslaughter, kidnapping, forcible rape/sodomy, robbery, assault, sexual assault offenses) were oversampled
relative to their proportion in the sampling frame to comprise 50% of the sample
cases (as seen in the table below). Within the person crime category, sampling was
performed using proportionate stratification by offense group to ensure that the person offense groups are represented in the sample in the same proportions as they
appear in the sampling frame. Offenders convicted of property offenses (burglary,
larceny and fraud/forgery) and drug crimes have been undersampled relative to their
proportion in the sampling frame to comprise the remaining 50% of the sample. As
with the person offenses, property and drug cases were sampled using proportionate
stratification by offense group, such that the offense groups are represented in the
sample in the same relative proportions as in the sampling frame.
Sampling Design

Cases

PERSON
Murder/Homicide
Manslaughter

Kidnapping

Forcible RapeiSodomy

114

ACTUAL
whn
i
Category
Proportion

Poplation
Proportion

WtMn
Category
Proportion

0.19

0.063

SAMPLE
Total
Sample

Sample

Pmpoei

cases

0.500

700

0.063

0.047

0.042

0.073

Robbery

480

0.308

216

Assault

526

0.337

236

Sex Offenses

202

(3.129

0.129

Burglary

1,123

0.172

121

-nY
Fraud/Forgery

1,997

0.306

214'

564

0.306
0.086

0.086

Drugs

2,845

0.436

0,436

305

PROPERTY AND DRUGS

0.81

0.500

700

Once the sample was drawn, matching the sample cases to the automated PSI
report data base was attempted, first by social security number (SSN) and offense,
and, for cases unmatched by SSN, by CCRE (Central Criminal Records Exchange)
number and 0 f f e n ~ e . IOverall,
~~
69.5% of the sample cases were matched successfully, resulting in the ability to track 973 released inmates for evidence of recidivism.

''I Comparison of the matched and unmatched sample cases reveals no significant differences
by offender race, gender, age at release, and number of prior prison terms served. However,
five significant differences between matched and unmatched cases (p<.05) exist by offense
group, judicial circuit, year of admission and number of prior Virginia felonies served. Post:
sampling weighring was applied to ensure that the data set of matched cases accurately reflects
the same distributions for offense.

Assessing the Impact of TIS on Recidivism 67

Due to incompatibility of data systems, the Virginia State P’oliceagreed to provide

hard copies of criminal history rap sheets from the Virginia Central Criminal History information system for each of the 973 offenders in the sample. Next, a trained
coder examined each rap sheet and recorded the recidivism data elements over a
three-year period that began with their release from prison. Data collection forms

were optically scanned, errors were detected and corrected, and a data file was prepared by VCSC s&.

Definlng Recidivism
Numerous definitions of offender recidivism have been employed to measure the
fiequency and extent of repeated contact with the criminal justice system. There is no
single or “correct” definition of recidivism; the choice depends on the issue of interest.
Potential definitions include rearrest for any new crime, re-arrest for a specific type of
new crime (e.g., identical offenses, felony offenses) re-convictions for any or for specific
types of new crime, re-incarcerations, time to new arrest, etc. In addition, recidivism

measures are used to analyze deterrence and incapacitation effects generally as well as to
assess the risk posed by individual offenders. For example, sentencing guideline systems always include prior criminal record in the sentencing calculation and will typically impose a harsher sanction on offenders who have recidivated. Project staff gathered the following information on 30 kctors relevant to measuring different aspects
of recidivism.
Re-Arrest Measures

Any new arrest - yeslno
Date of 1st non-felony arrest
Date of 2nd non-felony arrest
Dace of 1st felony arrest
Date of 2nd felony arrest
Number of misdemeanor arrests
Number of felony arrest events
Number of felony arrests - person
Number of felony arrests - property
Number of felony arrests - drug
Number of felony arrests - other
Arrests outside Vi4 - yeslno
All arrests outside VA- yeslno

Re-Conviction Measures

Any conviction - yeslno
Date of 1st non-felony conviction
Sentence for 1st non-felony conviction
Date of 2nd non-felony conviction
Sentence for 2nd non-felony conviction
Date of 1st felony conviction
Sentence for 1st felony conviction
Date of 2nd felony conviction
Sentence for 2nd felony conviction
Number of misdemeanor convictions
Number of felony conviction wents
Number of felony convictions - person
Number of felony convictions - property
Number of felony convictions - drug
Number of felony convictions - other
VCC code of conviction offense
Returned as technical violator

From this extensive set of information, four different measures of the frequency of
offender recidivism and the extent of penetration of a new criminal act into the
justice system were calculated:
rn

Any new arrest

rn Any new felony arrest
rn Any new conviccion
Any new felony conviction

68 Truth-in-Sentencingin Virginia

Basic ''quality of d
ad issues are assodated with both re-mest and reanviaion measures.
D

The most inclusive measure-"any

new arrest"-is

commonly used by researchers to

gauge recidivism and includes apprehensions for most crimes including misdemeanors
and felonies. In Virginia, however, not every arrest will show up in the Virginia Central
Criminal History (CCH) information system. For example, arrests for drunk in public, vagrancy, and other local ordinance violations that are not usually subject to jail
time will not be included on the rap sheet. Consequently, recidivism as measured by
any new arrest will result in some undercounting. O n the other hand, the use of arrest
may also overcount recidivism because some people who are arrested are released without being chacged or ultimately found to be innocent by the court.
a Recidivism measurement that relies on conviction is also subject to questions of

interpretation. One issue emerges due to plea bargaining: How does one count a
criminal event that is originally charged as a felony but is subsequently reduced to
a misdemeanor or even dropped entirely? Moreover, conviction measures often
result in some undercounting because case dispositions are not always reliably and
fully documented in case records.
Project staff believe these potential concerns only minimally affect the results of
the analysis.

Measuring Recldivism in Virginia:
A Multivariate Statistical Approach

A two-stage approach is used to conduct a preliminary analysis of recidivism for
offenders released from prison prior to the implementation of TIS. In this section,
the statistical technique of logistic regression is used to analyze the extensive set of
defendant-related variables in the recidivism database discussed above. The goal is to
determine which of the many potentially important factors do the best job of "explaining" the likelihood of recidivism. Once the most influential factors are identified, the next section employs a graphical analysis to illustrate the association among
many of the most significant factors and the various measures of recidivism.
The Statistical Model.

Whether an individual released from prison will recidivate with a new arrest or a new
conviction is very difficult to predict with any degree of certainty. However, it is
possible to reasonably estimate the probability of recidivism by examining the statistical relationship between the characteristics of the person being released and their
observed pattern of recidivism. The likelihood of recidivism is known to be influenced by factors such as age, race, gender, type of offense, and offense history
(Gendreau, Little and Goggin, 1996).'" The following table shows the eleven variables that were included in the current study because of their strong potential to
predict recidivism.

Iz6TheGendreau, Little, and Goggin study provides a convenient distillation of much of our
cumulative knowledge of the factors associated with adult recidivism and provides juscification for many of our choices of predictor variables. They used meta-analytic techniques to
Assessing tbe Impact of TIS on Recidivism 69

Predictors of Recidivism
Measurement
Variable Name
Levels

Explanation

Meansor
Percentages

0 = Non-Whie
1= Whiie

Race of released
inmate

0=34.7%

IncarFrationOffense

1= P e r m
2 = Property
3 = Druas

Type of offerne
for which inmate
was institutionalized

1=19.3%
2=45.2%
3=35.2%

LegalSatus

0 = No legal status Whether inmate
0=56.5%
I = Legal Status
had an official
1=43.5%
legal status with
the court (e.g.,
probation or parole)
at the time of the
incarceration offense

Felony Events

Number of felony Number of prior
1.22
sentencing events (to incarceration offense)
sentenchas for felonies

Race

misdemeanors

Standard
Deviation

1=65.3%

1.65

misdemeanor
convictions

determine which variables were the best predictors of adult offender recidivism. One hundred
and thirty-one studies produced 1,141correlationswith recidivism. They found significant ‘’mean
effect sizes” for age, race, gender, and adult criminal history (in this study measured by criminal
specialization, number of prior felony sentencing events, number of prior felony convictions
resulting in incarceration, and number of misdemeanor convictions).Their meta-analysis also
provides hrther confirmationof prior narrative reviews (e+ Gottfredson, 1987) which identified these variables as significantand potent predictorsof recidivism. Gabor (1 986), Gottfredson
and Gottfredson (19851, Gottfredson (1987), and Wilbanks (1987) review studies that found
juvenile record, type of incarceration offense, length of the prison term, and legal status at the
time of the incarceration offense to be imporrant predictors of adult recidivism.
12’ Means are reported for variables measured with continuous scales while percentages are
reported for nominal scale variables.
”* Standard deviations are reported for variables measured with continuous scales.

Tmth-in-Sentencingin Wrginia

A statistical model is developed to compare and conttast how these multiple defendant-related characteristics interact to explain recidivism. Such an analysis is necessary to control simultaneously for the influence of this set of factors (called independent or predictor variables) on the likelihood of recidivism. This statistical technique enables one to discern the unique contribution of each of the individual independent variables in explaining variation in recidivism rates (called the dependent
variable). The multivariate analysis technique used in the present study is logistic
regression, appropriate for use with dichotomous dependent variables. All four measures of recidivism, the four dependent variables, are dichotomous because they each
.-

have only two values: “one” if the inmate recidivated in the manner described or

‘‘zero”if they did not.
The results of the regression analysis and the variables that are statistically significant in explaining each type of recidivism are displayed in the table on the next: page.
The entries in the table are the regression coeficients (called logits in logistic regression) for each independent variable. The coefficients indicate the relative influence
of each independent variable on rhe probability that an inmate will recidivate in the
manner prescribed. A positive coeficient indicates that larger values of the independent variable are associated with an increased probability of recidivism, while a negative coefficient indicates a diminished probability of recidivism.
Overall Significance.

The last row of the table shows the overall success of each model in correctly
distinguishing whether an offender will recidivate (ie., the percentage of cases correctly predicted by the

This percentage is compared to the “null hypoth-

esis“, defined as the most frequent outcome within each measure of recidivism.I3’
Notice chat the regression model predicting new arrests (within three years of release)
considerably improves on our ability to identify the offenders most likely to recidivate over chance alone (66% vs. 51%). In addition, the ability of the regression
models to improve on chance when classifying inmates as recidivists or nonrecidivists
diminishes as criteria for recidivism becomes more stringent. While the models improved on the probability of correctly classifying inmates (relative to chance) by 15
percentage points when the criteria was simply a new arrest, improvement declined
to seven percentage points when

the criteria was stiffened to a new felony arrest. Our

ability to improve on chance when classifying inmates as having a new conviction
129The
primary measures of ‘goodnessof fit’ are displayed at the bottom of the table.The most
frequently used indicator in logit is called the ‘-2 log likelihood.’ Based on this measure, the
models are significant. In this case, the -2 log likelihoods are Chi-square variates with 16
degrees of freedom (because there are 16 explanatory variables in this model). It is the analog
of the F-statistic in linear regression and tests the hypothesis that all of the coefficients are
equal to zero. The table shows that each model meets the standard of significance, thereby
leading to a rejection of the null hypothesis, indicating that each model fits their particular
measure of recidivism well.
I3’That is, in about 39% ofthe cases in our sample,the offender wasarrested for a new felony
offense, meaning that in about 61% of the cases there was no new felony arrest. Therefore, the
null hypothesisor best guess would be to predict no new felony arrest and be right about 61%
of the time.

Asmsing the Impact of TIS on Recidivism

between 3-4 Yo,while the model provided no improvement on the ability to predict
new Felony convictions. These results are not unexpected since the ability to predict can

be expected to decrease as the probability of the phenomenon being predicted decreases.
The kdivldual Factors.

Essentially the same independent variables are significant in explaining recidivism
for a "new arrest," "new felony arrest," and "new conviction." These defendant characteristics indude Age,13' Race, Gender, Incarceration Offen~e,'~'
Number of Prior
Felonies Served,'33 Number of Prior Misdemeanors, and whether the inmate had a
Juvenile Record.
The model describing "new felony convictions" varied from the other three in that
Gender, Juvenile Record, and Prior Misdemeanors are not significant, but the Number of Prior Felony Sentencing Events is significant. In predicting the likelihood of a
'new felony conviction," measures related to offense seriousness and prior felony
sentencing history emerge as most significant. The independent variables LOS, Le-

gal Status, and Specialization were not related to probability ofrecidivism for any of
the measures, nor was there ever a difference in the probability of recidivism for drug
offenders relative to person offenders.
Logistic Regression Results
Variable Name

New Arrest

Measures of Recidivism134
New Felony New
New Felony
Arrest
Conviction
Conviction

-0.675**
-0.782*'*
-0.903***
-1.361'**
-1.450***
-0.876"'
0.866***

-0.619"
-0.878'''
-0,763"'
-1.342'"
-1.490'"
-0.958'"
0.598''

Age :13s

22-24
25 29
30 - 34
35 - 39
40 +

Race
Gender
IncarcerationOffense :I3
Property
0.382'
0.311
Drugs
-0.001
LOS
LegalStatus
-0.019
Specialization
-0.009
Felony Events
0.011
FeloniesServed
0.526'**
Misdemeanors
0.050"'
Juvenile Record
0.324"
Constant
-0.405
-2xLLR
1112.468*'*
%C
O ~ T(% Nul
~ l) C 66%~ (51%)
* Significant at the .10 level (p.10)

0.462**
0.221
-0.001
-0.004
0.006
0.024
0.328"'
0.037'"
0.308''
-0.547
1092.489"'
68% (61%)

** Sign%car?t
at the .05level (pc.05)

-0.423
-0.424
-0.289
-0.689"
-1.199-0.711***
0.821tW

-0.431
-0.648"
-0.482
1.1 51'"
1.658"'
-0.65?**'
0.390

0.375'
0.529''
-0.054
0.228
-0.005
-0.001
0.151
-0.004
-0.014
-0.020
0.022
0.129*'
0.313"
0,234'
0.055
0.020
0.335**
0.231
-1.297*** -1.478*'*
1054.122*"
885.607"'
69% (65%)
77% (77%)
"*Significant at the .Oilevel (pc.Ol)

Inmates aged (35-39) and (40 +) were dwap less likely to reddivare than inmates aged (14-21);
inmates aged (25-29) wem less likelyto recidivatethan inmatesaged (1621) f$r all the measures except
"newcanvictions;"and inmatesaged (22-24) and aged (30-34) were less likely to be a r d or d
b r a felonyoffense than inmates aged (14-21), but not to be wnvictedor wnvicpedofa felonyo h .

Pmperty Offenders were ahnays more likely to recidivatethan Person O&nden.
The more times an inmate had been incarcerated fbr Wony ofknses, the more lWy they were to
recidivate.
Within three years after release.

'''

Compared to Age 14-21.
Compared to Current Offense: Person.
72

Truth-in-Sentencingin Viq+nia

As another means of interpretation, this section graphically summarizes the results
of an analysis that examined the bivariate relationships between many of the offender characteristics found to be significant in the preceding multivariate analysis.

Half (49.3%) of all inmates released from Virginia prisons in 1993were re-arrested
for any new crime within three years. The number of persons who recidivate drops
quickly as the measure of recidivism becomes more conservative (e.g., of those released from prison, 22.4% were reconvicted of a new felony).

Overall Recidivism Rates Across Four Measures
--

The analysis covers 962 offendersreleased from prisonin 1993,
recidivismwas tracked for a period of three years.
$
Re-arrested
Rearrested - Felony

49.3%

39.6%

Reconvicted
- 1
Reconvicted- Felony

35.4%

22.4%

In general, males are more likely to be re-arrested and more likely to be recon-

victed than females, and nonwhites have higher rates of recidivism than whites.

Recidivism Rates by Gender
Re-arrested
Re-arrested - Felony
Reconvicted
Reconvicted -Felony

Recidivism Rates by Race
Re-airested
Re-arrested - Felony
Reconvicted
Reconvicted - Felony

26%

56%

1whner
Imks

Assessing tbe lmpact of TIS on Recidivism 73

Recidivism, if it does occur, is likely to happen sooner rather than later. For those
who recidivate, the average time until first re-arrest for any crime was 12.6 months,
with 750/0 recidivating within 19 months. As shown in the bottom area chart, for
those inmates re-arrested for a felony, 56% come back within one year.
For Those Who Recidivated,Average Time From Release to Re-Amst

(

75%
Recidivated within:

Time to first re-arrest

Mean
12.6mo

9.9mo

Time to first felony rearrest

13.6 mo

10.2 mo

19.1 mo
21.5 mo

For Those Who Recidivated,T i e to Re-Arrest

Re-arrested for any Crime
60% within 12 months
-...__.

Months

Re-arrested for Felony
56% wthin 12 months
___-....__..._..__._.~~..~.~~~~......

15%

10%

0%
3

Months

Truth-in-Sentencingin Vi'rginia

a Inmates released for property and drug offenses are more likely to recidivate as

compared to violent (person crimes) offenders. However, regardless of the original
offense, the percent of inmates recidivating stair-steps downward based on the recidivism measure used (re-arrest, felony re-arrest, reconviction, or felony reconviction). For example, the bar chart shows that 42% of inmates originally incarcerated
for a violent crime were re-arrested for any crime, 32% were re-arrested for a felony,

28% were reconvicted, and 16% were reconvicted of a felony.

~--

Four Measures of Recidivism
Originally
Incarceratedfor

Person

New arrest
New felony arrest
New conviction -28%
New felony conviction =16%

42%

32%

New arrest -152%
Property

mug

New felony arrest
New conviction
New felony conviction
-

44%
41%

27%

New anest
New felony arest
New conviction
New felony conviction -20%

50%

38%
32%

20%

40%

60%

Assessing the impact of TIS on Recidivism 75

n One issue raised during the debate over TIS reform was whether offenders who
recidivate tend to follow a consistent criminal path (Le., do violent offenders who
recidivate tend to commit additional violent crime?). The analysis shows that while
violent offenders had lower recidivism rates overall, those who were re-arrested were
most likely arrested for a violent crime (32% were re-arrested for a violent crime compared to 13%for property crime, 14% for drug crime). Likewise, persons released for
property crime were most likely to be re-arrested for another property crime (74%)and

drug offenders were most likely re-arrested for another drug crime (59%).
Are prsons released from prison likely to be re-arrested for the same type of felony offense
for which they were wiginally incarcerated?

Originally

Incarceratedfor

person

Property

New Person
New Property
New Drug
New Other

New Person
New P r o m
New Drug
New Other

Dnrg

New Person
New Property
New Drug
New Other

32%

13%
14%

16%

44%

57%

25%

13%

59%

27%

74%

28%

Tntb-in-Sentencingin Virginia

20%

40%

60%

80%

Examining the release cohort by offense, those originally incarcerated for larceny
had the highest recidivism rates, -both in terms of re-arrest (47%) and reconviction

(30%). Larceny was followed by burglary, fraud, assault, and drugs. Released inmates least likely to be re-arrested or reconvicted were those incarcerated for kidnapping, sex offenses (not including serious sexual assaults), manslaughter, and murder.
Two Measures of Recidivism
Originally
Incarceratedfor
.-

4 ~ e fdmy
w arrest

I .s%

Murder

New felony convicton

22%

Manslaughter

22%

Kidnapping

14%

Rape
Robbery

Assault

39%

Sex menses

17%

Burglary
47%

Larceny

Fraud
Drugs
I

10%

20%

30%

40%

50%

Assessing the impact of TIS on Recidivism 77

Re-Amled for a New Crime
Age upan pnson release

fenders recidivate at higher rates than old offenders. Released offenders less than 21
years of age had a 64% recidivism rate for any new crime, roughly 14 percentage

14-21 -64%

points higher than those ages 22-34. Offknders who were less than 21 years of age at

22-24%05-

time of release were three times more likely to be reconvicted of a new felony crime

25-29

51%

when compared to older offenders (those over the age of 40).

30-34%05-

rn The frequency of past criminal behavior is a good indicator of hture criminal

35-39 -43%

behavior. Data were examined that measured the number of prior felony sentencing

events an offender had before their incarceration for the released offense. Offenders

40t -:42%

with a history of felony convictions were much more likely to recidivate across all

Reconvictedfor a New Crime

four measures. The relationship between criminal history and recidivism was even

14-21 -42%
22-24

more pronounced when examining the seriousness of past criminal behavior. Seri-

32%

ousness was defined as the number of prior felony convictions that resulted in a

25-29 -37%

period of incarceration. For those who had served no prior periods of incarceration,
44% were re-arrested for any new crime. O n the other hand, eight out of ten (79%)
of those with three prior terms of incarceration were re-arrested for a new crime. The

30-34 %0435-39 -36%
4C-t

-26%

bars show a clear stair-step when examining the number of prior incarcerations for
each recidivism measure.

Re-Arrested for a New Felony Crime

14-21 -55%

Recidivism Rates Across Four Measures:
Examining Prior Felony Sentences
Resulting in incarceration

22-24%
14-1

25-29 %9330-34
35-39
40t

42%

33%

032%

Number of prior felony
sentenc6s resultig in
incarceration:
INone

lone

Two

@Three

Reconvicted for a New Felony Crime
14-21 -31%
22-24
25-29

30-34
35-39

23%

22%

28%

19%

40t =12%

Re-arrested for a
New Felony Crime
I?

25 68%

Rmnvictedfor a
New Felony Crme
42%

The pre-TIS data collected as part of the recidivism analysis will now serve as tht
baseline for measuring recidivism for those released during TIS.

Truth-in-Sentencingin Vi?p'nia

CHAPTER EIGHT

Conclusion
Virginia embarked on a major program of sentencing reform early in 1994. These
reforms, collectively called “Truth-in-Sentencing” (TIS), became effective on January 1, 1995, and substantially increased prison terms for violent and repeat offenders, abolished parole (except for chose already under sentence), and reduced good
-

timeallowances so that newly sentenced offenders would be required to serve at least
85% of their imposed sentence. Virginia was not alone in this regard. The 1994
federal Violent Crime Control and Law Enforcement Act authorized incentive grants
to the states in part for implementation of TIS laws, and by 1999, 27 states had
adopted the 85% rule. However, Virginia is one of just eight states and the federal
government that apply TIS to both violent and nonviolent offenders.
This evaluation traces the evolution of sentencing reform in Virginia since 1980,
describes how TIS has been designed, discusses its impact on prison population and
prevention of crime in the state, and begins an analysis of recidivism before and after
TIS. The knowledge gained through this study is primarily designed to benefit Virginia policymakers and practitioners interested in an objective analysis of the new
sentencing reforms in their state. However, given the ongoing interest nationally in
TIS and the abolition of parole, this evaluation has been designed and written to
clarify how sentencing reform efforts could be improved if iniriated in other states.
Lessons Learned
Five primary policy implications emerge from this evaluation. The first concerns

the comprehensive, inclusive, and ultimately effective process of reform used in Virginia. The remaining four implications focus on the rigorous, empirically-basedstrat-

egy used in Virginia to assess the impact of sentencing reform.
1. Manuging the process of reform. What is most striking to the NCSC evaluation
team is that since the early 1980s, even in a highly charged political environment,
sentencing reform has occurred in Virginia through a‘reasonablywell-planned process influenced heavily by data analysis. The initial impetus for reform was based on
the belief that judicial sentences were widely disparate, resulting in perceived injustices. A I982 task force recommended the development of historically based descriptive sentencing guidelines. As a result, a statewidedatabase of felony sentencing
was created, a study of sentencing disparity was commissioned by the Chief Justice
of the Supreme Court, and, in 1987-88, sentencing guidelines (entirely voluntary,
with no appellate review allowed) were developed. These guidelineswere pilot tested
in six judicial circuits between 1988 and 1990. An evaluation of the pilot project
showed that implementation of the voluntary guidelines had reduced sentencing
disparity, and that the judges involved believed that consistency and neutrality had
been improved while sentencing discretion had not been adversely affected. AccordConclusion 79

in& the guidelines weie adopted statewide beginning in 1391, and were revised on
an annual basis to continue to reflect the most current sentencing decisions handed
down by judges and juries. By the 1993 election, guidelines were being used in
Virginia under judicial control with an average compliance rate of 76%.
Newly elected Governor Allen moved quickly to convene a Commission on
Parole Abolition and Sentencing Reform, and the recommendations of that commission, as well as those of a competing legislative commission, were considered at
a special session of the Virginia legislature in 1994. The result was the enactment
of TIS legislation that abolished parole, reduced good time credits to a maximum
of 54 days per year, provided for a period of supervision following release, and
required felony offenders to serve at least 85% of their imposed prison terms. In
fact, it is currently estimated that offenders will serve between 88% and 92% of
the imposed term. The structuring of the guidelines and the 85% requirement
have achieved the dual legislative and executive goals of dramatically increasing
prison time for violent offenders while virtually guaranteeing predictability of
imposed sentences compared to actual time served.
The Virginia experience highlights the importance of ongoing planning and analysis
when confronting reform of an emotional and politically charged issue such as sentencing. Sentencing reform did not just happen in Virginia. Reform occurred through
a ten to twelve year process that included all three branches of government and was
supported by periodic analyses and evaluations. Regardless of one's philosophical
belief about the purposes and goals of sentencing, the process worked in that it
achieved a significant measure of predictability in sentencing, reduced disparity in
large measure, and, responding to public demand, increased prison t i e for violent
and repeat offenders.
2. Intpact of TIS on *on

population. The impact of TIS legislation on Virginia's

correctional resources was a source of early concern to state lawmakers. The VCSC
took seriously its charge by the Virginia General Assembly to examine the impact of
numerous alternatives to implementingTIS by developing a sophisticated simulation
forecast model (Criminal Justice Research Center @Risk model). The model was de-

signed to simulatejudicial decisionmakingand the demand for prison beds specifically
within the context of the new TIS guidelines. The program has the flexibility to model
a wide variety of alternative sentence ranges and recommendations.

The original forecasts proved to be inaccurate because of errors in two basic
assumptions: (1) declining arrests for violent crime and (2) slower than expected
growth in total arrests. However, because the simulation model itselfwas comprehensive and conceptually sound, the basic assumptions could be altered and the
forecasts re-estimated. The bottom line is that Virginia made the investment in a
valid simulation model to estimate the impact of sentencing reform on the expensive resource of prison space.

3. j k a & M c o t i z p k e with
g@ktm. Mrginia uses compliance rates as a key
measure of sentencing guideline performance and makes the results public on a regular

Tmtb-in-Sentencingin Krginia

basis. High levels of statewide Compliance indicate that sentences are being meted out
consistently and, as a consequence, reduces concern over unwarranted sentencing disparity. In addition, high mmpliance rates, especially in a volunrary setting like Virginia,
indicate judicial acceptance and approval of &e sentencing recommendations.
In 1998, compliance rates were 83% in terms ofdispositional compliance (type of
sentence), and 76% in terms of durational compliance (length of sentence). The
overall compliance rate is 75% for over 42,000 cases sentenced between January,

1995 and March 30, 1998. The aggravation rate (sentences more severe than the
guidelines) is currently 13% (generally homicide and sexual assault cases) and the
mitigation rate (sentences less severe than the guidelines) is 11% (generally rape
--

and burglary cases). Since 1994, judges have been requited to articulate their reasons for departure from the guidelines. The most common mitigation reasons
have been the availability of an alternative sanction or community punishment;
for aggravation reasons, judges most often cite historic criminal lifestyle and previous conviction for the same offense.

4. Preventable crime estimates under TIS. Virginia’s General Assembly wanted to
know how the extended incapacitation of violent offenders under TIS would affect crime rates. Specifically, they asked for information on how Virginians would
benefit from locking up violent offenders for longer periods of time. Is the cost
associated with giving certain offenders lengthier sentences justified through a reduction in the amount of crime they might otherwise commit if they had been
released earlier? Is there a beneficial “incapacitation effect” associated with TIS?
Certainly attempting to answer such questions is speculative because there is no
generally accepted method for determining the amount of crime prevented through
longer prison sentences. Moreover the analytic techniques are complex and can be
rather mysterious to the layman. However, the VCSC deserves credit for taking
on the challenge and attempting to quantify an important aspect of the impact of
sentencing reform. Other states may wish to build on the thoughtfully conceived
approach employed by Virginia. The approach benefits from making the most out
of available data and producing estimates that are conservative in nature.
5. Impact on recidivism. A critical issue confronting Virginia’smove to TIS was whether
the reform would reduce the level of offender recidivism in Virginia. Criminological
research shows that a relatively large share of crime is committed by a small pool of
known and repeat offenders. If TIS policies are successful in reducing offender recidivism, then it is likely that these policies will help reduce the crime rate generally.
Other states may wish to consider both Virginia’s efforts to inform inmates exiting
the prison system about changes in the state’s sentencing laws and the long-term
strategy for measuring offender recidivism. First, the Offender Notification Release Program (ONRP) was developed in 1996 as a joint effort of the VCSC and
the Department of Corrections (DOC) to educate inmates leaving Virginia prisons specifically about the TIS reforms. The program provides exiting inmates an
overview of the new sentencing system that abolishes parole and increases time
Conclusion 8 1

served for violent and repeat offenders. O n average, a returning violent offender
sentenced under the new guidelines should expect to serve two to six times longer
than under the state’s previous guidelines system.
Second, to determine whether TIS and ONRP policies have afFected offender
recidivism, baseline recidivism rates have been calculated for the offender population released from prison prior to the TIS reforms. The long-range plan is to
compare the recidivism rate of offenders released pre-TIS (phase one) with the
recidivism rate of offenders released post-TIS (phase two).The VCSC is now deliberating on when the second phase, measuring recidivism for those released afier
..exposure to TIS and the ONRP, should begin.
The NCSC evaluation team believes that one of the best design decisions made by
policymakers in Virginia was the retention of sentencing guidelines. The benefit of
the sentencing guideline approach is that it allows for a more accurate assessment of
the likely impact of a change in sentencing and/or parole policy. Guidelines systems
are arguably the most cost-effective means of providing rational structure, relevant
data, and the ability to accurately monitor and forecast sentencing outcomes.

82 Tmth-in-Sentencingin Virginia

Bibliography
@Risk Version 3.0. Palisade, Inc., Newfield, NY.
Austin, J., S. Cuvelier, and A. McVey. 1992. “Projecting the Future of Corrections.”
Crime and Delinquenry 38 (3):285-308.
Austin, J., and J. Irwin. 1993. Does Imprisonment Reduce Crime? A Critique of “Eodoo” Criminoloa. San Francisco: National Council on Crime and Delinquency.
--

Baird, C. 1993. “The ‘Prison Paf Studies: Research or Ideology.” N C C D Focus
(March): 1-7.
Block, M. 1997. “Supply Side Imprisonment Policy.” In F. Zimring and M. Block,
Two Views on Imprisonment Policies. Research report. Washington: National Institute
of Justice.
Blumstein, A,, J. Cohen, and D. Farrington. 1988. “Criminal Career Research: Its
Value for Criminology.” Criminology 26 (1): 1-36.
Bowerman, B., and R O’Connell. 1993. Forecasting and Time Series: An Applied
Approach. Belmont, Calif: Duxbury Press.
Bureau o f J u s t i c e P . 1996.“NationalAssessment of S

d Sentend& (February).

Chabotar, K. 1987. “Analyzing Costs in Courts.” Washington: National Institute of
Justice Issues and Practices.
Cohen, J. 1986. ‘‘Researchon Criminal Careers: Individual Frequency Rates and Offense Seriousness.” In A. Blumstein, J. Cohen, J. A. Roth, and C. A. Visher, eds.,
Criminal Careers and “Career”Criminah: El. 1.Washington: National Academy Press.
Cohen, J. 1978. “The Incapacitative Effect of Imprisonment: A Critical Review of
the Literature.” In A. Blumstein, J. Cohen, and D. Nagin, eds., Deterrenceand Incapacitation: Estimating the Effects of Criminal Sanctions on Crime Rates.
Cohen, M. 1988. “Pain, Suffering,and Jury Awards: A Study of the Cost of Crime to
Victims.” Law and Society Review 22: 537-55.
Commission on Sentencing and Parole Reform. 1995. “Report of the Commission
on Sentencing and Parole Reform.” Richmond, VA.
Creech, J. 1997. “Summary of Inmate Simulations in Virginia.” Working paper,
Virginia Sentencing Commission, Richmond, VA.
Criminal Justice Research Center. 1994a. Ertimated Costs OfRecidivism in Virgnia.
Richmond, VA.
Criminal Justice Research Center. 1994b. State Responsible inmate Population Forecats 1995-2005.Richmond, VA.
Criminal Justice Research Center. 1994~.Study of Preventable Crime Under the
Commission>Plan: Review ofkfethohjoo. Richmond, VA.
Department of Public Safety. 1997. Inmate Population Forecats: FY 1998 to FY2007.
Richmond, VA.
DiJulio, J. 1990. Crime and Punishment in Wisconsin:A Survey of Prisoners. Milwaukee, WI: Wisconsin Policy Research Institute.
Bibliography 83

Ditton, Paula, and Doris Wilson. 1999. Truth in Sentencing in State Prisons. BJS
Special Report (January).

Flango, V., and B. Ostrom. 1996. Assessing the Needfir Judges and Court Support
Stafl Williamsburg, VA National Center for State Courts.
Gabor, T. (1986). Theprediction ofm’minalbehauior Statrkticaiapproachtx Toronto:
University ofToronto Press.
Gottfried, B. 1984. Elements ofStochasticSirnuhtion. Englewood Cliffs, NJ: PrenticeHall, Inc.
Gottfredson, M., andT. Hirschi. 1986. “The True Value of Lambda Would Appear
to Be Zero: An Essay on Career Criminals, Criminal Careers, Selective Incapacitation, Cohort Studies and Related Topics, Criminology 24 (2): 213-34.
Gottfredson, S . (1987). Prediction: An overview of selected methodological issues.
In D. M. Gottfredson and M. Tonry (Eds.), Prediction and chsijGcation: Criminal
justice decision making. Chicago: The University of Chicago Press.
Gottfredson, S. and Gottfredson, D. (1987). Screening for risk among parolees:
Policy, practice, and method. In D. P. Farrington and R. Tarling (Eds.), Predictiun in
Criminology. Albany, Ny: State University of New York Press.
“Governor Allen Signs Sweeping Anticrime Package.” 1994. News Release, April 6.
Governor’s Commission on Parole Abolition and Sentencing Reform. 1994. “Final
Report” (August).
Governor’s Commission on Parole Reform. 1994. “Parole Abolition Sentencing Reform Proposals.” Unpublished working papers.
Governor’s Task Force on Sentencing. 1983. “A Report to the Governor ofVirginia”
(December).
Green, E 1998. “For Lease: 3,290 Prison Beds in Virginia.” Richmond Emes-Dispatch, July 6, Al, A5.
Greenberg, D. 1979. Mathematical Crz’minolou.New Brunswick, N.J.: Rutgers University Press.
Greenfeld, LawrenceA,Allen Beck, and Darrell Gilliard. “Prisons: Population Trends
and Key Issues for Management.” CriminalJustice Review 21 (Spring 1996).
Gupta, D. 1994. Decisions ly the Number. Englewood Cliffs, NJ: Prentice Hall.
Haynes, P., and C. Larson. 1984. “Financial Consequences of Incarceration and Alternatives: Burglary.” Crime and Delinquency 30: 529-50.
Horney, J,, and I. Marshall. 1991. “Measuring LambdaThrough Self Reports.” Cn’rninobgy29 (3): 471-95.
Joint Legislative Audit and Review Commission, “Review of Virginia’s Parole Process,” Commonwealth of Virginia, 1992.
Joint Subcommittees on Public Safety of the House Appropriations and Senate Finance Committees. 1994. “Felony Sentencing Practice and Time Served.” Unpublished analysis, Virginia Criminal Justice Research Center (May 10).
Joyce, N. 1992. “A View of the Future: The Effect of Policy on Prison Population
Growth.” Crime and Delinquency 38 (3):357-68.
Judicial SentencingGuidelines Committee. 1994. “Viinia SentencingGuidelines Manual”.

Twth-in-Sentencing in Virginia

Judicial Sentencing Guidelines Committee. 1993. “Voluntary Sentencing Guidelines: Report to the Judicial Conference of Virginia.”
Judicial Sentencing Guidelines Oversight Committee. 1989. %luntary Sentencing
Guidelines, Pilot Program Evaluation.
Judicial Sentencing Guidelines Oversight Committee. 1987. Working papers presented to committee (October).
“Justice For All.” 1983. Richmond Times-Dispatch
Kauder, Neal B., Brian J. Ostrom, Meredith Peterson, and David Rottman. 1997.
Sentencing Commission R.ofiles. Williamsburg, VA: National Center for State Courts.
--

Gamer, J., R. Lubitz, and C. Kempinen. 1 9. “Sentencing Guidelines: A Quantitative Comparison of Sentencing Policies in. innesota, Pennsylvania, and Washington.” Justice Qwrterh 6 (4): 565-87.
Levitt, S’. 1996. “The Effect of Prison Population Size on Crime Rates: Evidence
from Prison Overcrowding Litigation.” QuarterlyjournJofEconomics 11 1: 319-5 1.
Marvell, T., and C. Moody. 1996. “Determinate Sentencing and Abolishing Parole:
The Long-Term Impacts on Prisons and Crime.” Criminology34 (1): 107-28.
<

Marvell, T,and 6. Moody. 1994. “Prison Population Growth and Crime Reduction.” Journal ofQuantitatiue Criminohgy 10 (2): 109-40.
Microsoft Excel Version 4.0. Microsoft Corporation, Redmond, CA.
Miller, T. 1990. “The Plausible Range for the Value of Life: Red Herrings Among
the Mackerel.” Journal of Forensic Economics 3 (3):75-89.
Miller, T., M. Cohen, and S. Rossman. 1993. “Victim Costs ofviolent Crime and
Resulting Injuries. Health Afairs 12 (4):186-97.
Miller T., M. Cohen, and B. Wiersema. 1996. Victim Costs and Consequences:A New
Look. Research report. Washington: National Institute of Justice.
Mitchell, R., and R. Carson. 1989. Using Survqs to W u e Public Good: The Contingent Valuation Method. Washington: Resources for the Future.
National Council on Crime and Delinquency. 1995. “The 1995 NCCD Prison Population Forecast: The Cost of Truth-in-Sentencing Laws.” NCCD FOCUS (July).
National Council on Crime and Delinquency. 1992. Criminal/Elstice SenmcingPolicy
Statement. San Francisco.
National Institute of Corrections. 1995a. State Legiskztiue Actions on Truth in Sentencing: A h i e w of Luw and Legislation in the Context of the ViolPnt Crime Control
and Law En$rcement Act of 1994.Longmont, CO.
National Institute of Corrections. 1995b. TIS Law and Indeterminate Sentencing
in Eighteen States: A Review in the Context of I935 Congressional Decisions.
Longmont, CO.
Ostrom, Brian J., Neal B. Kauder, David Rottman, and Meredith Peterson. 1998.
Sentencing Digest: Examining Current Sentencing Issues and Policies. Williamsburg,
VA: National Center for State Courts.
“Report of the Ad Hoc committee on Sentencing Guidelines.” 1985. (August 15).
‘Report of the Joint Legislative Audit and Review Commission on Review ofVirginia’s
Parole Process to the Governor and General Assembly of Virginia.” 1992. Senate
Document No. 4.
Bibliography 85

“Sentencing and Parole Reform.” 1997. House Document No. 19.
Spelman, W. 1994. Criminal Incapacitation. New York: Plenum Press.
Survey Research Laboratory, Center for Public Service, Virginia Commonwealth
University. 1993. “A Survey of Virginia Citizens Concerning Violent Crime Issues,
January 8, 1993.”
Tonry, M. 1997. Intermediate Sanctions in Sentencing Gudelines. Washington: Bureau
of Justice Assistance.
Tonry, M. 1993. “Sentencing Commissions.” In M. Tonry, ed., Crime andJustice: A
Review of Research. Chicago: University of Chicago Press.
2

Tonry, M. 1991. “The Politics and Processes of Sentencing Commissions.” Crime
and Delinquency 37 (3): 307-29.
Tonry, Michael. 1996. Sentencing Matters. New York: Oxford University Press.

U.S. Department of Justice, Office of Justice Programs, Bureau of Justice Statistics.
1994. Felony Sentences in the United States. Washington.
Virginia Criminal Sentencing Commission. 1998a. 1998AnnualReport.Richmond, VA.
Virginia Criminal SentencingCommission. 1997.1397 AnnualReport Richmond, VA.
Virginia Criminal SentencingCommission. 1996.199G AnnualRport. Richmond, VA
Virginia Criminal SentencingCommission. 1995a. 1995AnnualRPport.Richmond, VA.
Virginia Criminal Sentencing Commission. 1995b. “Sentencing Guidelines Compliance Results, 7/ 1193-3/3 1/94.” Unpublished compliance analysis.
VirginiaCriminal SentencingCommission. 1998b. “ViiginiaSentencingGuidelinesManual”
Visher, C. 1986. “The Rand Inmate Survey:A Reanalysis.” In A. Blumstein, J. Cohen,
J. A. Roth, and C. A. Visher, eds., Criminal Careers and “Cmeer” Criminalr: EL 2.
Washington: National Academy Press.
Von Hirsch, Andrew. 1976. DoingJurtice: The Choice ofPllnishments, New York: Hill
and Wang.
Westing, Peter. 1982. “The Empty Idea of Equality.” Harvard Law Review 95: 537.
Wilbanks, W (1987). Predicting failure on parole. . In D. I?Farrington and R Tarling
(E&.), Prediction in Criminohgp Albany, NY: State University of New York Press.
Wilkins, William W. Jr., Phyllis J. Newton, and John R. Steer. “Competing Sentencing Policies in a ‘War on Drugs’ Era.” Wake Forest Law Review.28 (Summer 1993).
Wooldredge, J. 1996. “Research Note: A State-Level Analysis of Sentencing Policies
and Inmate Crowding in State Prisons.” Crime and Delinquency42 (3):456-66.
Wright, R. 1998. Managing Prison Growth in North Carolina Through Structured
Sentencing. Washington: National Institute of Justice.
Zedlewski, E. 1987. Making Conjhement Decisions. Washington: U.S. Department
of Justice.
Zimring, E, and G. Hawkins. 1988. “The New Mathematics of Imprisonment.”
Crime and Delinquency 34 (3):425-36.
..
’

.. . ...-.I

,. ,

Truth-in-Sentencing in Virginia